Datasets:

mimiklee

/

masterthesis-longt5-sum-8k

like 0

Introduction And Background Chronic kidney disease (CKD) impacts millions of people worldwide and is growing in prevalence. In the United States, 800,000 end-stage renal disease (ESRD) patients require regular kidney dialysis treatment [1,2]. CKD is a cause of morbidity and mortality, as well as a contributor to rising healthcare costs [3]. The global mortality rate of CKD has increased by over 40% since the 1990s [1,4]. Nationwide, kidney disease creates millions of dollars in healthcare costs, and the economic burden continues to rise [5]. Social determinants of health (SDOH) may contribute to disparities in CKD. The United States Centers for Disease Control and Prevention (CDC) defines SDOH as "the conditions in the environments where people are born, live, learn, work, play, worship, and age that affect a wide range of health, functioning, and quality-of-life outcomes and risks" [6]. By examining the SDOH that relates to barriers to care in CKD, we can help prevent future disparities. Factors identified in the background review of literature that may affect care in CKD may include socioeconomic status (SES) [2,7,8,9], membership in vulnerable populations (including prison inmates, homeless individuals, and undocumented immigrants) [4,10,11], race and sex [12][13][14][15][16], barriers created by the coronavirus disease 2019 (COVID-19) pandemic [17][18][19], insurance coverage [20], relationship with primary care provider (PCP) [21], transportation access [20,22,23], and utilization of at-home dialysis [24]. There has not yet been a scoping review addressing the SDOH that affects access to renal dialysis for ethnic minority patients. The aim of this scoping review is to identify existing literature on the different SDOHs that create barriers for ethnic minorities with CKD in receiving hemodialysis care. The findings of this study were earlier presented at the Nova Southeastern University Office of Graduate Medical Education Spring 2023 poster competition. To be included in the scoping review, research publications had to explore the social factors affecting access to renal dialysis treatment in racial and ethnic minority patients with ESRD. The inclusion criteria were determined using a Population, Concept, Context (PCC) framework (Figure 1). Peer-reviewed journal publications were used if they were published between 2012-2022 and written in English. The research studies needed to look specifically at patients living and receiving care in the United States. The decision was made to limit the analysis to studies in the United States because healthcare and SDOH differ greatly throughout the countries of the world. Papers discussing clinician biases when dealing with ESRD were included. --- FIGURE 1: PCC framework organization of the inclusion criteria for this study. --- PCC: Population, Concept, Context --- Information Sources An initial draft literature search was conducted on August 5, 2022, that yielded 349 articles. The last comprehensive literature search was executed on October 9, 2022. All information was sourced from online resources in databases. Only peer-reviewed published articles were included in the study. --- Search Strategy A systematic search was carried out following Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines on Embase (Excerpta Medica Database), PubMed, and CINAHL (Cumulative Index to Nursing and Allied Health Literature)(EBSCOhost) for all reviews published from 2013 to 2022 using a combination of keywords and medical subject headings (MeSH) related to disparities in healthcare, hemodialysis, ESRD, ethnic minority, United States of America, and SDOH. Additionally, one gray literature source (Google Scholar) identified no new articles. Boolean operators AND and OR were used to produce more focused and productive results. The search strategy was initially conducted in PubMed and later adapted to other databases to identify articles from the four databases. Studies that were included addressed physical, informational, and interpersonal barriers to care. --- Selection of Sources of Evidence The search generated 349 articles that were uploaded to Rayyan, an organizational tool used for sorting articles for review papers (Rayyan Systems, Inc., Boston, Massachusetts, United States). Following a screening and selection process, 10 final articles were selected for analysis to be included in our scoping review. The selection of articles was summarized using a PRISMA 2020 flow diagram for transparency, represented in Figure 2. --- Critical Appraisal of Individual Sources of Evidence An assessment of the quality of the 10 articles that were selected in Tier 2 was performed. In order to obtain a qualitative score for each article in terms of measures taken to prevent author bias and poor study design, each article was evaluated using the critical appraisal guidelines produced by the Joanna Briggs Institute (JBI), 2015. The checklist published by JBI was used to stratify the articles into categories to ensure that the data from all articles selected was based on statistically significant evidence with control of confounding factors. Each article was assessed by different standards in accordance with the experimental design of the study. Upon examination, six articles were found to meet 100% of the JBI criteria, one was found to meet 90%, two met 81%, and two met 75%. The threshold score determined by JBI to be classified as having a "low risk of bias" is 70%. With these findings, all of the articles that had been selected in the Tier 2 review were found to have a low risk of bias and were included in the study. --- Data Charting Process The following data were extracted from the 10 articles identified: authors, year of publication, study design, journal of population, study population, study aim, recommendations, and limitations. Key SDOHs that were referenced by the studies were recorded and listed by the number of articles that acknowledged each social determinant. The research articles were then grouped based on the recognition of each SDOH in order to synthesize the results of our study. Table 1 Relatively small sample size for a crosssectional study. In --- Results Following the screening process, a final total of 10 of the most relevant articles were selected utilizing the eligibility criteria. This collection of articles contained a variety of different types of study designs, including one special report, three cross-sectional studies, one survey, one case-control study, one cohort, two retrospective studies, and an ecological study. The populations of the studies we included ranged from 40 to over 800,000 participants. The collective pool of studies included data from over 2.1 million participants. The age of the studies ranged from one year to nine years, with the average being 4.6 years. The main SDOH that were discussed by the articles in this selection are listed below. --- Neighborhood of Residence as an SDOH Lee et al. (2020) investigated the impact of the racial composition of a patient's neighborhood on their access to hemodialysis treatment facilities [25]. The study found that the neighborhoods with the longest commute time to dialysis facilities were predominantly Hispanic, while the neighborhoods with the best access to dialysis facilities were either predominantly Black or mixed communities. They noted that the process of gentrification can have both positive and negative effects on access to care. [26]. The study points out that after controlling for age and sex, African American CDK patients have a lower life expectancy than what would otherwise be expected compared to White CKD patients, and this relationship is dependent on neighborhood SES. These findings may be attributable to hemodialysis facility access and health insurance access. Patients from other racial demographics did not experience the same decrease in life expectancy even if socioeconomic factors were controlled in both populations. --- Johns et al. (2014) examined the effects of neighborhood composition in renal dialysis patients --- Mental Healthcare Access as an SDOH While other studies focused mainly on access to hemodialysis care in CKD patients, Ashrafi et al. (2022) demonstrated that mental health may pose a barrier to treatment in CKD [27]. Although mental health access and nephrology care access are two separate issues, mental health can be viewed as an SDOH that may in itself impact a patient's access to nephrology care. The study found that while over nine out of 10 patients of all racial demographics lacked mental health care access, the racial demographic most likely to lack mental healthcare access was Asian patients. They found that Hispanic CKD patients were the most lacking in overall healthcare access. Improvement in mental health access may improve patient compliance and adherence to treatment plans. --- GFR Methodology in Medicine as an SDOH As a calculated measure used as a component of health care services, GFR methodology can be viewed as an SDOH because it influences access to care. A special report by Delgado et al. (2022) discusses the formation of a special council formed by national kidney organizations with the objective of reassessing the equations used to determine objective markers of kidney function [28]. The previously accepted equations for determining kidney function included the social construct of race as a variable. This may have delayed the diagnosis of CKD and therefore created barriers to dialysis and transplant access for African Americans. Patients may experience disparities depending on which physicians they see, and the type of equation that those physicians prefer to use. Over the course of more than 40 meetings, the council brainstormed a total of 26 alternative methods to measure kidney function by GFR while avoiding the creation of such disparities. The final consensus by the council is that the most effective method of GFR estimation that is racially unbiased is the removal of the race variable from the equation. Further recommendations from this report included widespread adoption of the estimation of renal function by use of Cystatin C as a biomarker, as well as future research to identify other potential biomarkers. Because race is not a variable in the equations that utilize Cystatin C, these equations are less likely to contribute to racial disparities, and they are also regarded to be more accurate [28]. --- SES as an SDOH While there was some overlap, two articles were identified that mainly addressed SES as a social determinant of healthcare access in chronic kidney disease. Nee et al. ( 2016) examined access to renal hemodialysis care among racial minority patients affected by poverty, as determined by zip code [29]. The study found that patients who were affected by poverty were much less likely to have access to care, especially those belonging to African American and Hispanic demographics. They point out that this disparity may compound and increase the rate of CKD progression that African American patients experience compared to White patients. Plantinga et al. (2014) investigated the relationship between patients from neighborhoods of lower socioeconomic status and access to renal hemodialysis facilities [30]. The study found that many dialysis facilities are located in low socioeconomic neighborhoods, and they did not find any barriers to renal dialysis access for patients from these neighborhoods. The study concluded that patients from low SES neighborhoods already have adequate access to renal hemodialysis. However, patients from these neighborhoods may have a smaller selection of facilities to choose from than patients from higher socioeconomic areas. --- Primary Language Spoken as an SDOH Even if a CKD patient is physically able to get to a treatment facility, they will encounter a barrier to accessing care if they are unable to communicate with the facility staff. Along with disparities in healthcare access among different racial demographics, Velez-Bermudez et al. ( 2022) examined the impact of language barriers in CKD patients undergoing hemodialysis treatment through patient interviews [31]. Spanish and other non-English speaking patients reported a lack of access to translation services in dialysis facilities, which created a barrier to their comprehension of the treatment process and options. In addition to information about their treatment, Spanish-speaking patients are also prone to missing important information about legal medical documentation and may be asked to sign paperwork that has not been translated into their language [31]. The article pointed out that hospital visitor restrictions imposed by the global COVID-19 pandemic in 2019 inhibited patients with language barriers from bringing family members with them to dialysis appointments to serve as translators, highlighting a potential contribution to barriers of care [31]. --- Immigration Status as an SDOH In addition to language barriers, CKD patients from countries outside the United States may experience barriers to dialysis healthcare access due to their immigration status. Madden and Qeadan (2017) report that Hispanic undocumented immigrant patients are often barred access to health insurance coverage, and often present to the Emergency Department for dialysis treatment rather than going to a dialysis treatment facility [32]. The study found that undocumented immigrant Hispanic patients are even more at risk of seeking emergency dialysis care if they also have a concurrent language barrier. Because these patients wait until their situation becomes an emergency before seeking treatment, they do not receive the proper nephrology care needed to manage the progression of their CKD. The Hispanic population has twice the likelihood of the White population of developing CKD in their lifetime, so these additional barriers to treatment can further worsen the already poor prognosis in this demographic [32]. The authors state that further research is required to investigate whether immigrants of other ethnicities experience similar barriers to renal dialysis access [32]. --- Military Rank as an SDOH Norton et al. (2021) investigated disparities in CKD within the context of the military healthcare system. The results of their cross-sectional study suggest that there is an increased incidence of CKD in older enlisted service members as compared to senior officers [33]. Because the military provides all its members with adequate insurance coverage through the TRICARE Health Plan, this disparity is unlikely to be caused by differences in insurance coverage. The authors conclude that the findings of their study may be attributable to social and cultural factors [33]. --- Discussion From the thesis described in the introduction, some of the questions presented at the beginning of the study have been investigated and answered, while others still represent gaps in the literature that future studies should address. Topics that were well covered included SES, language barriers, and immigration status. While Nee et al. ( 2016) supported the hypothesis that lower SES communities have less access to care, it conflicted with the findings from Plantinga et al. ( 2014), which reported that many low-income communities have high enough numbers of renal dialysis facilities to ensure adequate access to care [29,30]. It may be that SES might be more of a barrier to access in some locations than others, which may vary by geographic location. A lack of proper pre-dialysis care may promote the demand for access to dialysis in these communities. The studies reviewed supported the fact that both immigration status and language are social determinants of renal care access and suggest that they have significant overlap [30,31]. However, there were many social determinants that were not addressed in the articles that met the study inclusion criteria. The topics of levels of education, incarceration status, limitations in personal or nonemergency medical transportation, and changes in healthcare delivery brought on by the COVID-19 pandemic were not covered by these studies, so the research questions on those topics as SDOH remain unanswered by this review. This study presents implications for future research and guidelines to be used in practice. It is recommended that dialysis centers make note of the potential for language barriers to create disparities in access to care. --- Strengths and Limitations The articles selected for this study were found to be reliable in the quality assessment that was conducted using the guidelines published by the JBI. Therefore, there is a high level of confidence in the accuracy of the studies that addressed the review questions. However, there are several limitations to the study that must be discussed. While the study mostly identified articles that focused on African American and Hispanic populations, many studies did not focus on other racial minority groups, such as Asians, Native Americans, or Pacific Islanders. Furthermore, the categorization of racial groups can be problematic to define, overencompassing, and may obscure disparities found in smaller groups that are often lumped into larger categories, such as the Hmong population being a sub-demographic of Asians. In addition, the study only examined articles focusing on adult populations. While kidney disease requiring dialysis treatment is more common in adults, there may be similar disparities in the access to dialysis care found in pediatrics. The study only included peer-reviewed articles published in large databases, which may exclude information from articles whose findings were not significant enough to warrant publication. --- Conclusions Neighborhood of residence, mental healthcare access, GFR methodology, SES, language barriers, immigration status, and military rank were identified by this study as being major SDOHs discussed by existing literature as impacting access to renal dialysis care. The aim of this study was to identify existing literature on social factors contributing to disparities in access to renal dialysis care for racial minorities in the United States, and while the study identified articles discussing these social determinants, future research is needed to determine if other social determinants of health exist that contribute to disparities in this setting. In the identification of these contributing social factors, the hope is that future action can be taken to reduce disparities in kidney health across all demographics. Because CKD is a debilitating condition for millions of people in the United States, even small improvements to healthcare access have the potential to greatly improve the quality of life for countless underserved patients. Going forward, efforts to eliminate language barriers in dialysis clinics, improve mental healthcare, and eliminate the social construct of race as a variable within GFR equations are important steps to take to reduce future health disparities, especially those related to CKD. accountability for the integrity and validity of its contents. --- Additional Information Disclosures --- Conflicts of interest: In compliance with the ICMJE uniform disclosure form, all authors declare the following: Payment/services info: All authors have declared that no financial support was received from any organization for the submitted work. Financial relationships: All authors have declared that they have no financial relationships at present or within the previous three years with any organizations that might have an interest in the submitted work. Other relationships: All authors have declared that there are no other relationships or activities that could appear to have influenced the submitted work.

Although widespread, the burden of disease presented by chronic kidney disease (CKD) is not equally distributed among all demographics. Examining the social determinants of health (SDOH) that relate to barriers to renal dialysis care in CKD can help to prevent future disparities. There has not been a study addressing the social factors that create barriers to care for ethnic minority patients with CKD. The aim of this scoping review is to address the SDOH that affects access to renal dialysis for ethnic minority patients in the United States. This study was based on the protocol published by the Joanna Briggs Institute. A total of 349 studies were identified from PubMed, EBSCOhost, and Embase. Each article was screened against population, concept, and context criteria in order to be considered for inclusion. The population was determined to be adults of all genders from underrepresented minority populations. The selected concept was SDOH. The context of this study was the United States population. From the articles selected by the search criteria, neighborhood of residence, mental health care access, glomerular filtration rate (GFR) methodology, socioeconomic status (SES), language barriers, immigration status, and military rank were identified as SDOH affecting access to renal dialysis care. While this study identified four social determinants, more research is needed for the investigation of other possible SDOH contributing to disparities related to CKD and access to renal dialysis care.

INTRODUCTION The participation and empowerment of targeted persons and groups are essential for successful health promotion. According to the Ottawa Charter for Health Promotion, people can only achieve their fullest health potential if they have influence over their own health determinants (1,2). Two studies, in particular, describe the possible health promotion effects of participatory interventions on (a) the individual level (e.g., strengthening of self-efficacy, health behavior, and perceived social support) and (b) the community level (e.g., strengthening the awareness of vulnerable groups) (3,4). Furthermore, some studies have pointed out that participatory approaches foster the sustainability of public health interventions due to the integration of stakeholders' resources, the division of responsibilities, the customization of interventions for intended users and the strengthening of abilities and competences (5)(6)(7)(8). A special benefit of participatory interventions lies in their potential to reduce health inequality (4). Inequalities exist because many people of low socioeconomic status have a high need for health promotion, as their often precarious life conditions place further strains on their health. However, different barriers limit their chances of accessing health promotion and prevention (9,10). Participatory interventions have the potential to reach socially disadvantaged people because they incorporate low-threshold interventions that fit the needs of vulnerable groups (5,11). To assess the public health impact of such interventions, RE-AIM, which was conceived by Glasgow et al. (12), was established as an important framework. According to RE-AIM, public health impact is achieved by reaching many people (those who are targeted), through an effective intervention at different sites over time. The public health impact of communitybased participatory research (CBPR) projects might be low in many cases, simply because sustaining such projects and transferring them to new sites have been challenging (13)(14)(15). Key to achieving the public health impact of CBPR projects is sustainability, since the effects of interventions often only occur years later (16). As such, long-term CBPR interventions are needed to achieve structural changes so as to prevent the inefficient and pointless input of resources and the disillusionment of participants (16). Further, CBPR health interventions need to be transferrable to multiple sites, since it is only in the act of transferring interventions that the potential Abbreviations: BIG, Bewegung als Investition in Gesundheit (movement as an investment in health); CBPR, community-based participatory research; PA, physical activity; RE-AIM, reach, efficacy, adoption, implementation, maintenance. of reaching more people and settings increases and limited resources are utilized more efficiently (e.g., investments for program development) (17,18). The transfer and sustainability of CBPR interventions have hitherto proven to be challenging and have not been well studied (8,16,18). Ultimately, this has led to a lack of evidence regarding the transfer, sustainability, and public health impact of these interventions (3,6,18,19). On top of the challenges involved in transferring and sustaining CBPR projects, unfavorable public health structures for funding the transfer of these projects exist in many parts of the world (20)(21)(22). For instance, the German healthcare system has been described as having a strong curative, rather than preventive, orientation (23). Recently, a law strengthening prevention and prevention research was passed and implemented (24). The so-called prevention act enhances collaboration between different stakeholders as well as the role of the Federal Center for Health Education (BzgA). However, CBPR health promotion projects had been taking place in Germany before the prevention act was passed. The corresponding projects were funded for limited time periods on a small scale by various actors in the field of health promotion, such as national, federal, and municipal agencies or sickness funds (25)(26)(27). In the absence of a national funding agency with the capacity to fund initiatives to vertically "scale-up" (growth of the project's reach) or horizontally "scale-up" (transfer to more sites) such CBPR projects, sustaining and transferring a successful project required a patchwork of funding agencies (22,(26)(27)(28). This might have limited the successful transfer and, hence, the public health impact of interventions. This case study investigates how the BIG project "Bewegung als Investition in Gesundheit" (movement as an investment in health), a CBPR project, was transferred and sustained across 17 sites in Germany. BIG is a participatory project aimed at promoting physical activity (PA) among socially disadvantaged women (e.g., of a low socioeconomic status or ethnic minority background). Based on the above-described challenges involved in achieving public health impacts of CBPR projects, the aim of this case study was to examine the mechanisms of successful transfer and sustainability strategies of this CBPR project. --- METHODS --- BIG-A CBPR Project for Physical Activity Promotion The BIG project was initiated in 2005 at the Friedrich-Alexander-University Erlangen-Nuremberg (FAU) and has now been transferred from the first community to 16 other sites. The aim of BIG is to promote PA among socially disadvantaged women. This includes women in difficult life situations, who face challenges due to their migration background, low education, unemployment, welfare dependency, low income, and/or their status as single mothers. For these women, access to PA and sport is often impeded because of barriers such as the lack of culturally sensitive exercise offers, language difficulties, high membership fees, or the lack of childcare services. Their chance of benefitting from the positive outcomes of PA are reduced (29), and evidence suggests that the rate of physically inactive individuals is above average among socially disadvantaged women (30). To counteract this inequality, the BIG project empowers women to improve their access to PA through a participatory approach. As part of this approach, socially disadvantaged women, researchers, local project coordinators, policymakers, and other local experts (e.g., sport club representatives) jointly plan and implement activities in a number of cooperative planning sessions (31). The diverse composition of the cooperative planning group is essential for the implementation of BIG, as all stakeholders provide different resources (e.g., policymakers can provide funds, and sport associations can provide access to sports facilities). Women articulate their needs in terms of exercise offers and support the implementation of these offers. --- BIG-Reach of Exercise Programs, Attendance, and (Health) Effects Exercise programs implemented through BIG address existing economic, social, cultural, environmental, and behavioral barriers to participation. This is achieved by lowering attendance fees, offering free child care, having female instructors, choosing facilities close to where women live, and the non-requirement of membership to attend classes. Women advertise the exercise programs by word of mouth. Exercise classes offer a range of activities, from aerobic fitness, Zumba, Nordic walking, and martial arts to women-only indoor-pool hours (29). The number of attendees per class ranges from 5 to 20, depending on the offer. Classes commonly take place once or twice per week. In 2019, more than 800 women regularly participated in about 60 different exercise classes at BIG sites. Previous analyses of BIG have demonstrated that its exercise classes do indeed reach women in difficult life situations (32). There is also evidence of the health-promoting effects of BIG on women. Women attending BIG exercise classes reported increased social networks and beneficial effects on their physical and mental well-being (29,32). By taking part in the cooperative planning process, the women reported increased self-efficacy, as they were empowered to voice their interests and increase their knowledge of political and administrative processes (29). On a structural level, BIG improves municipal opportunities for exercise by strengthening local networks for health promotion, establishing routines for citizens to participate in municipal decision-making processes, and removing barriers to participation in exercise for women in general (33). --- Data Collection and Analysis This case study reports on the efforts to transfer (adopt) and sustain BIG. Regarding the transfer of BIG, we investigated (1) different methods of recruiting new BIG sites and (2) different support strategies for implementation at these sites. Pertaining to the sustainability of BIG, we examined (3) the years, up to the winter of 2019, that BIG was maintained at the sites and (4) the reasons given by sites for discontinuing BIG. Further, we investigated (5) each site's costs regarding the transfer of BIG. Information on recruitment methods (1) and support strategies (2) was gathered through a data analysis of unpublished project reports and meeting protocols from the last 15 years. This included field notes from informal consultations with current and former local project coordinators and staff at the various municipalities. Over the years, different recruitment methods and support strategies were carried out. Recruitment methods varied from conference presentations and open calls to networking. Support strategies provided by FAU researchers to communities included the drafting of an implementation manual, setting up a system for the phone-based and faceto-face counseling of communities and providing extensive on-site support to municipal staff in charge of the project. In order to inquire as to whether the BIG project was (3) implemented/maintained or (4) discontinued, the municipalities were contacted by phone. To shed light on the reasons why project activities were discontinued at different sites and to validate this information, a reflection workshop was conducted with three researchers who had supervised the BIG project. Regarding the costs of transfer ( 5), grant applications were screened to extract information on the amount of funding the FAU received for carrying out and transferring BIG. From these data, the annual funding amount per site was calculated. This was achieved by dividing the funding amount by the duration of the funding phase and the number of new sites. The annual longterm costs of the transfer were also calculated. This was done by dividing the funding amount of each funding phase by the average years of the project's lifespan at the sites on which BIG was implemented during this funding phase. Costs covered the initial project development and its transfer to other sites. This included staff and travel costs at the FAU to initially develop the project, recruit, and support municipalities willing to implement BIG and evaluate the effects of BIG at the individual and structural levels. For three subprojects (BIG, BIGff, and BIG.kompetenz), funding also included seed money for the communities to set up the project and hire a local project coordinator. --- RESULTS Since 2005, six subprojects [BIG, BIGff, BIGGER(t), BIG.Manual, BIG.kompetenz, and BIG.Disseminierung] implemented and transferred the BIG approach to 17 sites. As BIGff and BIGGER(t) utilized the same approach, they are reported on jointly. The aim of each subproject was to develop more effective ways of transferring BIG in order to increase its public health impact. --- Recruitment of New Sites for Project Transfer Table 1 reports on the different recruitment strategies for all the subprojects. Across all the subprojects, most of the sites (N = 13) were recruited through open calls for applications. This recruitment strategy was used in four subprojects (BIGff, BIG.Manual, BIG.kompetenz, and BIG.Disseminierung). The researchers directed a standardized letter to all Bavarian municipalities (mayors, offices, and coordinators of sports or integration) with more than 20,000 inhabitants. Three sites were obtained, as the researchers addressed local stakeholders directly. One site (Regensburg) was recruited after presenting the project at a conference. Besides the recruitment method, the number of new sites depended on whether the funders, who supported the researchers at the FAU, also provided seed funding for the project implementation at the sites. This seed funding facilitated the decision-making of the local administrators to implement the project as well as the recruiting process of new sites. Seed funding was provided for three subprojects (BIG, BIGff, and BIG.kompetenz). Ten of the 17 BIG sites were recruited when seed funding was provided, which was used to support the setup of the project and the financing of a local project coordinator at the site. --- Support Strategies for Project Implementation In each subproject, the FAU tested new strategies to support the sites (Table 1). Within the initial subproject (BIG, 2005(BIG, -2008)), the researchers, targeted women, and persons in charge of project implementation collaborated to develop the BIG approach and went on to probe and readjust it in practice. They coordinated the project onsite (e.g., by conducting the cooperative planning process and building a BIG network). After 3 years, the municipality took over the coordination of the project. From the beginning of the subprojects BIGff and BIGGER(t), each site had its own local BIG coordinator. The objective was to relocate the coordination and strategic planning of the project from the researchers to a responsible person at the site. The researchers provided extensive support to the local project coordinators by holding weekly exchanges, assisting with network-building and moderating the cooperative planning process. In order to reduce the level of assistance and enable local coordinators to conduct the project more independently, the researchers developed a manual (34) in collaboration with the practice partners. This manual-structured in six sections: (1) about BIG, (2) discovery phase, (3) preparation, (4) cooperative planning process, (5) implementation management, and (6) sustainability assurance/networking-was intended to describe every phase of the project implementation in easy-to-understand language and with the use of many examples, practical advice, and working aids (e.g., questionnaires and quality assessment forms). Many practice partners and other researchers reported that the BIG manual was helpful, not only for the implementation of BIG but also for other participatory projects. Nevertheless, in addition to the BIG manual, the local project coordinators needed consultation. In order to provide implementation support and facilitate the transfer of BIG, the so-called BIG competence center was established at the FAU in 2012. Funded by different sources, and staffed with a 0.5 researcher position, the BIG competence center consulted sites in implementing and planning, evaluated BIG at the sites, and initiated further transfer of the project. --- Maintenance of Program Activity at the Sites At seven of the sites, BIG courses were being offered up to the winter of 2019 (Figure 1). The projects' lifespan at these seven sites was, on average, 9.8 years. Another site restarted the implementation process of BIG following an initial unsuccessful attempt (Coburg). The duration of the project activities at each site is shown in Figure 1. The first BIG site (Erlangen) recorded the longest lifespan (14 years). There, about 30 PA courses were offered per semester. Four sites offered courses for at least 9 years (the number of PA courses is between 2 and 11). At the nine sites that had not maintained program activities, the projects' lifespan averaged 4.2 years. Two of these sites kept the project running for over 9 years before it came to an end. Four sites offered PA courses for 1-2 years. --- Reasons for Discontinuing BIG at the Sites The most common underlying reason for the termination of the BIG project was, in one way or another, connected to the local project coordinator. This was the case at six of the 10 sites where project activity was discontinued. Two of these sites had to terminate the BIG program because the local coordinator retired, and no successor could be found. At four sites, the local project coordinators could not continue to manage the project due to other work tasks, maternity leave, or extended sick leave. However, at two sites where local project coordination changed, the handover of responsibilities was successful. Furthermore, five local coordinators were incapable of building a strong network due to a lack of qualified and relevant contacts within their reach. Three sites stated additional reasons for the termination of the project, including insufficient financial resources and funding. This could be linked to the low-fee courses being offered, for which external funding support through the municipality or other entities would have been necessary. Two sites further mentioned that they did not have sufficient participants or trainers for the courses. At some sites, a combination of the aforementioned issues presaged the abandonment of the project. --- Project Transfer Costs The amount of funding depended on the timespan of the funding arrangement, the extent of scientific support, and the provision of seed funding. Table 1 presents an overview of the funding amounts and transfer costs of each subproject. Funding for the first subproject was, by far, the highest, and thus the scientific support provided to the first site resulted in high annual costs. The largest expenditure was the cost of personnel relating to the development, evaluation, and implementation of the project. Given this site's long lifespan, the annual long-term transfer costs were not exceptionally high. The lowest annual costs per site were recorded for the subprojects BIGff + BIGGER(t). Even though the total funding amount for these subprojects was rather high, it was split between many new sites (N = 7), thus reducing the per-site costs. The funding amount and long-term transfer costs of the BIG manual were the lowest. Nevertheless, the annual costs covering support per site were the second highest, as only two sites could be recruited for this subproject. For BIG.kompetenz and BIG.Disseminierung, the long-term transfer costs were striking. These high costs did not result from actual higher transfer-related expenditures. Instead, they resulted from the more recent start of the subprojects at these sites, which meant that the transfer costs could not yet be amortized. It is apparent that more support [for BIG, BIGff + BIGGER(t)] did not automatically result in higher annual costs per site. The costs depended on the extent of support, the timespan of funding, and the number of new sites during the funding phase. Although these three factors varied in each subproject, the annual cost of support from the researchers was approximately EUR20,000 per site. The annual long-term transfer costs of BIG and BIGff + BIGGER(t) show how a rather high amount of funding can amortize through sustainability of program implementation. --- DISCUSSION Over the years, the transfer of BIG to 16 other communities yielded several valuable experiences: (1) in order to recruit new communities, BIG mainly utilized open calls for applications and contacting municipal administrations. This likely resulted in the participation of communities with a genuine and intrinsic interest in the BIG project. Of those, about 35% were able to sustain BIG over time, with an average lifespan of 9.8 years. In other communities, the BIG project ceased, on average, after 4.2 years. In one municipality, the transfer did not result in the promotion of PA activities, and thus, the project failed to implement. (2) Due to the German healthcare system's (scarce) resources for conducting CBPR (28), BIG was forced to rely on a patchwork of funding agencies and funding schemes to transfer the project. In some subprojects, external seed funding for communities was budgeted to stimulate the uptake of BIG. This scarcity of resources made the researchers focus on effective ways to transfer BIG-a quest that is still ongoing. (3) These efforts resulted in different approaches in terms of how to support communities in implementing a CBPR project. The experiences portrayed in this study exemplify that, in addition to seed money, hands-on support in the local implementation of BIG was essential. The implementation of BIG using the manual-only approach, while highly cost-effective (lowest funding amount and lowest long-term transfer costs), did present some limitations. However, the manual proved to be useful when the communities also received additional support from researchers. This study's tentative results suggest that more support from researchers (e.g., by counseling local project coordinators on whom to invite to the planning process as well as attending the planning process) had a positive influence on the sustainable implementation of the projects and did not necessarily incur higher costs in the long run. (4) Aside from the methodology employed in this study, vital factors beyond the researchers' control also played a determining role in the success of project transfers. The two main reasons for the termination of BIG in some communities was the resignation or retirement of the local coordinator and the impossibility of finding a sufficient number of supportive stakeholders or key community members to advertise the project to targeted persons. According to these results, a local project coordinator and his or her competences and capacities are of fundamental importance for the project's success. To facilitate successful handovers of responsibility, a strong network supporting the implementation of BIG appears favorable. (5) Consequently, this study identified the following necessary elements within local coordinators' set of competences and capabilities to facilitate a successful transfer: (a) a local BIG coordinator who already has relevant stakeholder contacts; (b) knowing how to network with the target group, relevant stakeholders, and policymakers; (c) having the scope and opportunities to act (i.e., time, funds, etc.) and the competences to network and implement offers; and (d) being able to draw on support from experienced CBPR researchers. These elements correspond to the results of other studies on the role of program champions (those advocating for the program and its continuation) and their capacity to act (e.g., funding and community/stakeholder support/involvement) (8,19,35). --- Transferring and Sustaining CBPR Interventions A major challenge in achieving public health impacts through CBPR is the sustainability of a local CBPR project and, often simultaneously, transferring it to other communities. As such, researchers face two conundrums: How do we transfer a CBPR intervention that is, by nature, highly context-bound? How can we balance efforts to sustain a project at one site with efforts to transfer it to other sites? Based on the experience in BIG, we suggest that a sustainable transfer requires strong and individual support from municipal administration staff in order to adapt the intervention to the implementation setting and, thereby, increase the likelihood of a sustained implementation. Achieving a public health impact requires researchers to work with communities to "normalize" CBPR projects, i.e., incorporating it in routine practice (36). Under scarce financial resources, researchers need to balance the efforts needed to sustain CBPR interventions. This means that the scientific support is limited in time, and the number of new sites that can receive this support is restricted. This, in turn, limits the transfer of the intervention on a broader scale. Therefore, long-term funding mechanisms are needed (21) for the sustainable transfer of evidence-based and successful health promotion interventions. In BIG, this issue was overcome by utilizing a patchwork of funding agencies to transfer the project to other sites. The current development of the German public health funding system provides an opportunity to establish structures that strengthen the long-term funding of research transfer and, thus, foster the public health impact of research. In addition to the scientific influence, there are a number of factors in the implementation setting that are beyond the control of researchers, which appear to partially determine the success of transferring CBPR, as other models on scalability have demonstrated (14,37). To cope with the challenging process of transferring and sustaining CBPR, there is a need for more research into the interaction and interdependence between sustainability and transfer. There are studies, frameworks, and models that depict sustainability as an indicator of a successful scale-up (14). However, this disregards the influence of sustainability on subsequent horizontal and vertical scale-up. For instance, it is only through long-term implementation that interventions can enfold their potential reach (vertical scale-up) and demonstrate effectiveness in increasing the prospect of their transfer to additional sites (horizontal scale-up) (38). Moreover, such research could help identify factors that similarly foster transfer and sustainability (e.g., a competent local coordinator). --- Limitation and Strengths Some study limitations should be considered. This case study focused on the recruitment of new sites, support strategies, the maintenance of program activity, reasons for discontinuing program activity, as well as transfer costs as the sole relevant factors in the sustainable transfer of the BIG project. As they can have a huge impact on sustainability (8,14), it would be interesting and meaningful if future research could investigate contextual factors relating to the implementation setting, such as political support or external funding for a CBPR project. Furthermore, this case study focused on the cost of transfer and included expenses relating to research and seed funding. To determine the cost of the project's implementation, further research should consider funding from municipalities and other financial contributions from external parties. Moreover, we examined the BIG project as an example of sustainably transferred CBPR. Hence, only data from one project were included. Therefore, it cannot be said with certainty that the results are representative of other CBPR projects. Future research on the sustainability and transfer of CBPR projects is required in order to identify strategies that support the sustainability and transfer of evidence-based interventions and maximize their public health impact. --- CONCLUSION Although the sustainable transfer of CBPR is challenging, the BIG project was successfully transferred to 16 sites over the last 15 years. This study identified factors in the implementation setting that are of central significance for sustainable transfers and, thus, for the public health impact of CBPR, which are beyond the control of research (capacities of local coordinators, funding stability, and the support of local networks). Therefore, it might be necessary to examine contextual factors and, where needed, to appropriately focus on further developing the requirements in the setting. However, other factors within the control of researchers do facilitate sustainable transfers. In BIG, the high extent of scientific support for the sites, seed funding, and open calls for applications positively influence the long-term transfer of the project. To foster the public health impact of interventions, we recommend more research on the interrelationship of sustainability and transfer as well as increased long-term funding for research on sustainably transferring effectproven interventions on a broad scale. --- DATA AVAILABILITY STATEMENT The data generated and analyzed in this study is available upon reasonable request from the corresponding author annika.herbert-maul@fau.de. --- ETHICS STATEMENT Ethical review and approval was not required for the study on human participants in accordance with the local legislation and institutional requirements. The patients/participants provided their written informed consent to participate in this study. --- AUTHOR CONTRIBUTIONS AH-M collected, analyzed, and interpreted the data and drafted the manuscript. KA-O participated in the interpretation of the data and helped in the drafting and revision of the manuscript. AF and AS participated in the data collection. AF, AS, and AR critically reviewed and revised the manuscript. All authors read and approved the final manuscript. Environment and Health (LP00178), and a statutory health insurance company. The funders were not involved in the design of the study, the collection, analysis, and/or interpretation of the data or in writing the manuscript. --- Conflict of Interest: The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Introduction: Community-based participatory research (CBPR) is an effective approach to health promotion, especially in relation to socially disadvantaged groups. However, the long-term implementation of CPBR-based projects on a broad scale is often challenging, and research regarding the sustainable transfer of participatory research is lacking. This limits the scaling-up and public health impact of CBPR. Therefore, this study examines the mechanisms utilized to transfer and sustain the BIG project, a multifaceted CBPR project aimed at promoting physical activity among women in difficult life situations. Materials and Methods: Borrowing from the RE-AIM framework, we analyzed project documentation and conducted a reflection workshop to investigate methods of transferring BIG to new sites as well as strategies from researchers to support project implementation and the maintenance of program activities at those sites. Moreover, we analyzed the reasons for discontinuing program activities at some former BIG sites and the costs involved in transferring BIG. Results: Since its establishment in 2005, BIG was transferred to and implemented at 17 sites. As of the winter of 2019, the program activities were maintained at eight sites. The average duration of sites that continue to offer program activities was more than 9 years. Discontinued sites maintained project activities for an average of 4 years. According to the study findings, the extent of scientific support, the provision of seed funding, and the local project coordinator, the person managing the project at the site, all have a significant impact on the sustainability of the transfer. A patchwork of funding agencies was needed to finance scientific support and seed funding in BIG. The transfer of BIG projects accrued annual costs of approximately EUR20,000 per site; however, long-term project implementation resulted in a decline in the annual transfer costs of BIG.The sustainable transfer of CBPR is challenging but possible, and increased support of research and seed funding can facilitate long-term transfer. Nevertheless, other factors in the implementation setting are beyond scientific Herbert-Maul et al. The Sustainable Transfer of BIG control. With scarce financial resources, researchers need to carefully balance the efforts of the sustainability and transfer of CBPR projects. To address this issue, there is a need for further research into the interrelationship of the sustainability and transfer of CBPR projects as well as increased long-term funding.

Background The Human Immunodeficiency Virus /Acquired Immune Deficiency Syndrome (HIV/AIDS) remains one of the most stigmatizing pandemics worldwide [1]. The fight against HIV/AIDS remains one of the major points in the sustainable development agenda [2]. Though great efforts have been implemented at different levels to limit the spread of HIV/AIDS, the incidence of HIV in the developing world and more precisely Sub-Saharan Africa (SSA) is still relatively high [3]. By the end of 2015, an estimated 36.7 million people were living with HIV/AIDS worldwide, with over two-thirds residing in sub-Saharan Africa, which includes Cameroon [4]. Despite the recently reported drop in the prevalence of HIV in Cameroon from 5.5 to 4.3% over the last decade [5], it still remains amongst the highest in West and Central Africa [6]. Tackling AIDS-related stigma and discrimination is crucial in the effective prevention of HIV/AIDS, the care of PLHIVA (People Living with HIV and AIDS) and goes a long way to significantly help in containing and managing this pandemic [1,[7][8][9][10]. HIV-related discrimination is not only a human right violation according to the United Nations General Assembly Special Session on HIV/AIDS, but problems of HIV-related stigmatization and discrimination need to be properly addressed to successfully meet public health goals [11,12]. The prevalence of HIV-related stigma varies with the setting and method of evaluation. Even though it has no standardized method of evaluation, HIV-related stigma is reported to be highly prevalent among PLHIVA [11][12][13][14][15]. At every level of the HIV/AIDS management ladder (the prevention, care and treatment), stigma and discrimination has seriously impeded its success [12]. Unlike other chronic conditions such as diabetes, hypertension and chronic liver disease, HIV testing and status disclosure are seriously limited by stigma and discrimination [13][14][15][16][17][18][19]. Negative predictors of HIV-related stigma include, but are not limited to: a good knowledge on HIV/AIDS, past participation in HIV anti-stigma campaigns, low level of education, and religion [19]. HIV-related stigma and discrimination is a major barrier limiting participation in prevention of mother to child transmission programs, HIV treatment, and adherence to ART [19]. HIV-related stigma is therefore a serious barrier towards the successful attainment of the ambitious triple 90 goal (90% diagnosed, 90% on treatment and 90% with suppressed viral load) of the Joint United Nations programme on HIV/AIDS (UNAIDS) by the year 2020 [20]. In Cameroon, despite diverse efforts put in place by the Cameroon health system to combat HIV-related stigma and discrimination, HIV-related stigma and discrimination remains a significant obstacle to the fight against HIV/AIDS and the adherence to HIV/AIDS related counseling [5]. To the best of our knowledge, no study has been done to quantify HIV-related stigma and identify its associated factors in Cameroon. A quantification of HIV-related stigma will not only serve as a baseline for future comparison, but will help to evaluate the anti-stigma strategies integrated in the management guidelines. Identifying factors associated with HIV-related stigma might help redesign or reinforce these interventions. Therefore, we sought to measure HIV-related stigma, and identify factors associated with high HIV-related stigma among PLHIVA on antiretroviral therapy (ART) at the HIV-day care unit of the Bamenda Regional Hospital. --- Methods A hospital-based cross sectional analytic survey was conducted from February to April 2016. All eligible and consenting persons living with HIV/AIDS (PLHIVA) on ART at the HIV-day care unit of the Bamenda Regional hospital were targeted and recruited. The HIV-day care unit of the Bamenda Regional Hospital is a reference HIV treatment unit in the North-west Region of Cameroon, and receives patients from both the rural and urban areas of the Region. We excluded PLHIVA who were mentally incapacitated, below 15 years of age and patients who arrived critically ill and unable to respond to the questionnaire. ART-na<unk>ve patients were not included because according to the Test and Treat recommendation, any patient not yet on treatment should just have been awaiting confirmation of his or her status. All eligible and consenting participants who visited the treatment unit during the study period were consecutively included in the survey (n = 308 PLHIVA). With the aim of limiting surveyor induced stigmatization and discrimination, the principal investigator undertook a two-months internship at the HIV-day care unit of the Bamenda Regional Hospital, and recruited colleagues as interviewers. Training sessions were organized during which interviewers were trained on the consent process and data collection procedures. When the protocol, the questionnaire and informed consent form were completed, the data collection tools were pretested on a sample 15 PLHIVA in a treatment centre in Yaoundé (Centre region of Cameroon) and validated after analysis by a team of experts. Berger HIV Stigma Scale validated by Feyissa et al. in a resource-limited setting (Ethiopia) [21] and Jeyaseelan et al. in India [22], contains four subscales (domains): Enacted Stigma, Disclosure Concerns (a form of anticipated stigma), Negative Self-Image (e.g. internalized stigma), and Concern with Public Attitudes towards HIV (e.g., anticipated stigma) [21]. This culturally validated, HIV-Stigma scale has been used with good results particularly in clinical settings to identify patients in need of psycho-emotional support and assess post-intervention changes in stigma in many settings [23,24]. It should be noted that onto the elements of the Berger scale was added sociodemographic data and some information that allowed the identification of factors associated with HIV-related stigma. The questionnaire was paper-based and is attached as (Additional file 1). Interviewers were of both sexes and were allowed to interview willing participants irrespective of gender. During data collection, the interviewer identified eligible participants, and after thorough explanation of the information notice form of the study, obtained informed consent from the participant or legal guardian. The data were collected face to face by interviewers in private rooms within the facility (one participant at a time) who asked questions and took down the responses of each participant. Participants received no remunerations for taking part in the survey. --- Data analysis Data from validated questionnaires were entered into a predesigned data entry sheet. After cleaning, the data were analyzed using the statistical software Epi-Info version 3.5.4. Categorical variables are reported as proportions with their corresponding 95% confidence intervals (CI), while means were calculated for continuous variables. HIV-related stigma level was categorized into "high" and "low". Participants with "high" HIV-related stigma were those for which the Berger stigma scale classified them into at least a moderate level of stigma (that is, participants with a mean score value greater than or equal to 80). The strength of association between the selected covariates (level of education, estimated monthly income, marital status, number of years from detection of HIV status, place of residence) and the stigma level (high or low) was determined by calculating the Odds Ratio (OR) and their 95% CI. A multivariable logistic regression analysis was used to identify factors associated with high HIV-related stigma (with gender, religion, age, and occupation considered as potential confounders; solely based on literature). The threshold of statistical significance was set at a p-value less than 0.05. --- Results A total of 331 PLHIVA were contacted for the study and 23 refused to participate giving a final sample size of 308 PLHIVA and a non-response rate of 6.9%. The mean age of the 308 participants was 40.1<unk>10.2 years with a mean age of 43.9 <unk> 8.4 years and 38.1 <unk> 10.6 years among men and women respectively (p <unk> 0.001). The population was dominated by women (65.3%). About 2/5th (40.9%) of the participants were married, 35.1% single and 54.9% had acquired at least secondary education. Christians were most represented (95.8%) with the rest Muslims. More than 2/3th (69.9%) of the participants reported estimated monthly revenue of less than 50000FCFA (80.75 United States dollars). Table 1 presents the ranges and means of the different domains of stigma experienced by the participants. The overall mean HIV-related stigma score was 88.3 <unk> 18.80 (with a possible range of 40-160) which corresponds to a moderate level of stigma according to the Berger stigma scale. The mean HIV-related stigma score was 87.65 <unk> 18.62 and 88.64 <unk> 18.89 in men and women respectively (p-value = 0.6608). Table 2 shows the frequency distribution of respondents according to HIV-stigma score level. Overall, 37.3, 58.4% (180/308) and 4.2% experienced mild, moderate and severe HIV-related stigma respectively. Following a subgroup analysis, irrespective of gender, it was noted that most patients suffered from moderate forms of personalized (49.8%), disclosure (66.4%), negative self-image (50.0%) and public attitude (52.1%) stigmatization. The proportion with high level of stigmatization among the population was 62.7% (95% CI = 57.8-68.9%). Table 3 presents the factors associated with high HIV-related stigma following bivariable and multivariable logistic regression analysis. When controlled for gender, religion, age, and occupation; level of education below tertiary (Adjusted Odd Ratio [AOR] = 0.70 [95% CI 0.44-0.91]; p = 0.036) and duration from diagnosis below 5 years (AOR = 1.74[1.01-3.00]; p = 0.046) were independently associated with high HIV-related stigma. There was no statistically significant association between high HIV-related stigma and place of residence (rural/ urban), monthly revenue and marital status. --- Discussion The mean overall HIV/AIDS related stigma score was 88.3 <unk> 18.80 which corresponds to moderate level of stigma according the Berger stigma scale. In addition, most participants suffered from moderate forms of the different subtypes of stigma including: personalized (49.8%), disclosure (66.4%), negative self-image (50.0%) and public attitude (52.1%) stigmatization. Over 60% of the participants lived with high levels of HIV-related stigma. A level of education below tertiary and duration from diagnosis below 5 years were significantly associated with high HIV-related stigma. The sex ratio reported in this study is similar to that reported among PLHIVA in Africa with a female preponderance [6,25]. In addition, national demographic and health survey report in 2011 shows that in Cameroon, the ratio is about two women infected with HIV for every one man [5]. This is in correlation with data on the prevalence of HIV/AIDS in Cameroon and more precisely in the North West Region of Cameroon where surveys have shown that more women are infected with HIV/AIDS than men [5]. The higher prevalence in women simply brings to light vulnerability of women to the HIV/AIDS infection, mainly due to biological factors, which are usually favoured by socio-behavioural practices and socioeconomic differences [26]. The overall mean HIV-related stigma score was 88.3 <unk> 18.80 (with a possible range of 40-160) which corresponds to a moderate level of stigma according the Berger stigma scale. Even though women seem more likely to be infected by the virus (sex ratio of about 2:1), the mean HIV-related stigma score was 87.65 <unk> 18.62 and 88.64 <unk> 18.89 in men and women respectively (with no statistically significant difference. p-value = 0.6608). This means that on average, most clients in the HIV-treatment unit experienced moderate HIV-related stigma and that both genders were similarly affected. This is true as each of the estimated means of the stigma sub-types was found to fall at least into the moderate stigma category. HIV-related stigma has been reported high and highlighted as a barrier to HIV/AIDS management by a series of African surveys [5,25,27,28]. This study indicates that HIV-related stigma is still very common among PLHIVA taking treatment from the Bamenda Regional Hospital. A study carried out in Ethiopia reported similarly high levels of perceived HIV-related stigma mean score values [29]. Following a subgroup analysis, irrespective of gender, it was noted that most patients suffered from moderate forms of personalized (49.8%), disclosure (66.4%), negative self-image (50.0%) and public attitude (52.1%) stigmatization. The proportion with "high" level of stigmatization among the population was 62.7% (95% CI = 57.8-68.9%). This is indicative of the high rate of HIV-related stigma perceived by these participants. Multiple studies in the African context have reported similarly high rates of HIV-related stigma among HIV positive patients [29][30][31]. These high rates of stigma could be associated with multiple pitfalls in the effective management of HIV/AIDS in Africa. According to investigators of the 2011 national DHS in Cameroon, 46 and 58% of women and men respectively had never been screened for HIV [5]. High rates of HIV-related stigma have been associated with decreasing level of HIV voluntary testing [14,[32][33][34], status disclosure [25,32], adherence to ART [32,35], educative programs on HIV transmission and increased probabilities of maintaining the transmission chain [32]. If patients received at the treatment centres are this stigmatized, it gives us an idea of the level of stigmatisation and the effect the effect that this can have among patients lost to follow-up and others in the community. Indeed, up to 66.4% of the participants had moderate problems with disclosure. This form of HIV-related stigma could be associated with very low rates of HIV sero-status disclosure and consequently maintenance of the transmission chain [7]. This high level of HIV-related stigma can seriously impede on the attainment of the ambitious triple 90 HIV management goals by the UNAIDS, set for 2020 [20]. In Cameroon, anti-HIV-stigma interventions have been adopted at the national, local (Health units and the community) and individual level. Interventions at the national level include, but are not limited to: legislative protection of the rights of PLHIVA, facilitation of access to family planning services, emergency obstetric care and ART, and organisation of trainings to reduce stigma and discrimination in schools and among health care providers. Interventions at the community level include among others: improving the knowledge of community dwellers on the transmission of HIV/AIDS, correcting perceptions through educative talks, and encouraging leaders to create a climate of tolerance with no prejudice. The health care provider can reduce HIV-related stigma by avoiding stigmatizing statements, adoption of attitudes void of judgement, preserving and respecting patient-provider confidentiality, and adopting universal standard precautions for all patients regardless of their HIV-status. The fight against this pandemic in Africa and more precisely in Cameroon requires much more focus in redesigning educative programs for both HIV positive and negative individuals in order to break this underground barrier. HIV-related stigma maintains the epidemic underground and remains one of the major explanations why people do not wish to know their HIV status [7,36]. After controlling for gender, religion, age and occupation, a level of education below tertiary and a duration from diagnosis below 5 years were significantly associated with high HIV-related stigma. The findings in this study suggest that PLHIVA with at least a tertiary education are less likely to be victims of high HIV-related stigma compared to their counterparts with lower level of education. A high level of education is associated with a better understanding of HIV/AIDS [5] and consequently, reduced disease-related stigma [14,16,36,37]. A facility-based cross-sectional Ethiopian study in 2015 reported a high level of education to be significantly associated with reduced HIV-related stigma among PLHIVA [29]. It is clear that a high level of education and awareness among the population on HIV/AIDS could limit the amount of stigma and discrimination associated with the AIDS pandemic [28]. Research findings suggest that improvingliteracy rates and a focus on correcting incorrect conceptions on HIV/AIDS might go a long way in fighting HIV-related stigma [14,31]. Clients who had been diagnosed within the last 5 years were more likely to be victims of high levels of stigma compared to those who had lived with their diagnosis for more than 5 years. Some level of "normalization" of the HIV status and stigma reduction has been reported in surveys by provision of ART to PLHIVA [32,38]. This finding is coherent with that of Fido et al., where a decreasing trend of HIV-related stigma was associated with increasing number of years on ART [29]. Similarly, Tzemis et al. reported a decreasing trend of HIV-related stigma with the years of living with HIV/AIDS and years on antiretroviral therapy [37]. Intensified psychological counselling and emotional support in the early days following HIV diagnosis can help PLHIVA to gradually accept and live with their status. Our study found no significant associations between place of residence (urban/rural), estimated monthly revenue, and marital status and high level of HIV-related stigma. Contrasting findings were reported by Tzemis et al. concerning these variables [39]. This could be explained by a relatively lower number of patients per variable modality in the current study (compared to the study of Tzemis et al.), with consequential lower statistical power. In addition, the sociocultural and financial The findings herein should however be interpreted with care. The cross-sectional design of this study only allows development of hypotheses and cannot establish any cause-effect relationships. In addition, no randomization or power calculation was conducted. Also, recruitment of participants was essentially convenient as data collection was dependent on the study duration. Even though our interviewers were well trained, the responses of some participants could be made socially desirable rather than expressing their actual feelings, thereby underestimating the true burden of stigmatization in PLHIVA. Although the Berger HIV stigma scale has a few limitations, it has been reported as an effective measure of stigma levels especially for future comparisons. In addition, the findings from a single treatment unit in Cameroon cannot fully paint the picture of the reality in Cameroon and sub-Saharan Africa. Also, our study enrolled only patients on ART and systematically excluded those who are not taking treatment. This can possibly lead to an underestimation of the level of HIV/ AIDS-related stigma in the whole population since we think that patients who are loss to follow-up would be possibly more stigmatized than the study participants. However our findings were consistent with results of studies carried out in Africa and SSA and can be used as a baseline for future estimates and comparison. --- Conclusion About 3 out of every 5 PLHIV receiving ART in the HIV-day care unit of the Bamenda Regional Hospital still experience high levels of HIV-related stigma. This occurs more frequently in participants with low educational status, and who may have known their HIV status for less than 5 years. The overall mean HIV-related stigma score among these patients remains high. Anti-HIV-stigma programs in the North West Region need strengthening with intensified psychosocial follow-up of newly diagnosed cases. Improving literacy rates may also help in reducing HIV-related stigma. Large scale studies that quantitatively measure HIV-related stigma in more treatment centers are required; so as to establish a clearer picture of HIV-related stigma in Cameroon. --- Additional file Additional file 1: Questionnaire for the evaluation of HIV-related stigma among PLHIVA and associated factors. (DOCX 32 kb) --- Availability of data and materials The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request. --- Abbreviations AIDS: Acquired immune deficiency syndrome; AOR: Adjusted odds ratio; ART: Antiretroviral therapy; CI: Confidence interval; DHS: Demographic and health survey; HIV: The human immunodeficiency virus; PLHIVA: People living with HIV and AIDS; SSA: Sub-Saharan Africa; UNAIDS: Joint United Nations Programme on HIV/AIDS Authors' contributions ABA and PNN and NEN conceived the study and guided field data collection; ABA and MNY analyzed data; ABA drafted the manuscript; MJE, VNA, and KB contributed in supervising data collection and revising the manuscript; all authors read and approved the final manuscript. --- Ethics approval and consent to participate Ethical clearance for this study was obtained from the ethical review board of the Faculty of Medicine and Biomedical Sciences of the University of Yaoundé I, and administrative authorization obtained from the Director of the Bamenda Regional Hospital. Only consenting participants were included in the study and for participants who were aged less than 18 years, consent was obtained from their legal representatives. The interviewers were well trained on the consenting and data collection procedure and participants who were mentally incapacitated and critically ill were not included in the survey. --- Competing interests The authors declare that they have no competing interest. --- Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Background: The Human Immunodeficiency Virus /Acquired Immune Deficiency Syndrome (HIV/AIDS) is not just a medical problem but its social impact is increasingly affecting its effective management. The fear of HIV-stigma constitutes a major barrier to HIV testing, prevention, uptake and adherence to antiretroviral therapy (ART). We aimed to quantify HIV-related stigma, and identify the factors associated with high HIV-related stigma among persons living with HIV and AIDS (PLHIVA) and on ART. Methods: A hospital-based cross sectional analytic survey targeting PLHIVA on ART at the HIV-day care unit of the Bamenda Regional Hospital of Cameroon was conducted from February to April 2016. A total of 308 eligible and willing participants were consecutively included in the survey. Data were collected using a pretested questionnaire designed from the Berger HIV stigma scale and analyzed using Epi info 3.5.4. Results: The mean age of the 308 participants was 40.1±10.2 years. The mean overall HIV/AIDS related stigma score was 88.3 ± 18.80 which corresponds to a moderate level of stigma according to the Berger stigma scale. Further analysis revealed that most participants suffered from moderate forms of the different subtypes of stigma including: personalized (49.8%), disclosure (66.4%), negative self-image (50.0%) and public attitude (52.1%) stigmatization. It was estimated that 62.7% (95% confidence interval [CI] = 57.8-68.9%) of the participants lived with high levels of HIV-related stigma. After controlling for gender, religion, age and occupation, level of education below tertiary (Adjusted Odds Ratio [AOR] = 0.70 [95% CI = 0.44-0.91]; p = 0.036) and a duration from diagnosis below 5 years (AOR = 1.74 [95% CI = 1.01-3.00]; p = 0.046) were significantly associated with high HIV-related stigma. Conclusion: About three out of every five PLHIVA receiving ART in Bamenda Regional Hospital still experience high levels of HIV-related stigma. This occurs more frequently in participants with low educational status, and who may have known their HIV status for less than 5 years. Anti-HIV-stigma programs in the North West Region need strengthening with intensified psychosocial follow-up of newly diagnosed cases.

Enhanced emotional reactivity during adolescence. An important determinant for individual differences in developmental outcomes is emotional reactivity 5. Emotional reactivity can be seen as one's emotional response to an event or changes in the environment 6. An individual with higher levels of emotional reactivity typically shows stronger responsiveness to environmental influences. It has been well documented that adolescence is a period characterized by liable emotional states and increases in frequency and intensity of emotional reactivity 7,8, with a peak during mid-adolescence causing heightened sensitivity to social and emotional changes [9][10][11]. This heightened adolescent emotional reactivity includes increased fluctuations, compared to childhood and adulthood, in both positive and negative emotions. In addition, across adolescence there is an increase in the experience of negative emotions and a decrease in the experience of positive emotions (i.e. the mean levels of emotions) 12, but how these developmental patterns are affected by the COVID-19 pandemic is currently unknown. Although during young adulthood there is no longer a peak in enhanced emotional reactivity, there are still experiences of more frequent daily stressors, compared to older adults, which can have a negative influence on stress-related affect 13. Numerous studies have linked heightened emotional reactivity in adolescents to negative developmental outcomes such as risk-taking behavior, alcohol and substance abuse, and affective disorders [14][15][16]. However, this rise in emotional reactivity, along with an increased salience of peer relationships, also may serve an adaptive function for adolescents 8. For example, recent neuroimaging studies have shown that ventral striatum reactivity, a neural indicator of reward, plays a role in the sensitivity to social connections, which in turn may stimulate otherbenefiting behavior and healthy relationships 17. In other words, this framework poses that increased emotional reactivity during adolescence can possibly also promote greater prosociality and help to successfully navigate challenging times, depending on whether adolescents grow up in a supportive or unsupportive environment. Therefore, an additional goal of this study is to shed light on the impact of the COVID-19 pandemic on young people's mood and to use the abovementioned theoretical framework to predict under which circumstances heightened emotional reactivity aids or harms behavior during the pandemic. --- Social and socioeconomic stressors during the COVID-19 pandemic. Here we test emotional reactivity as a differential susceptibility marker for responding to challenges in the COVID-19 pandemic 18. Depending on the specific stressors within the environment, higher emotional reactivity could lead to maladaptive behaviors in some individuals but in positive outcomes for others 19,20. Adolescents may experience environmental stressors during the COVID-19 pandemic that interact with their emotional reactivity to impact their mental health. Environmental factors that are likely to impact adolescents' mental health can be categorized in two types of stressors: social stressors and socioeconomic stressors. Two important social stressors that are likely to impact adolescents' mental health are family stress and emotional maltreatment. The COVID-19 outbreak has impacted some families more than others, which, in combination with adolescents being forced to spend more time at home with their parents and siblings, may threaten family wellbeing 21. As a consequence, some adolescents have experienced more negative changes in their family-and social environment than others. It is important to acknowledge that there is heterogeneity in whether the increased time spent with family members during the first wave of the COVID-19 pandemic has positive or negative outcomes on both parents and adolescents. For example, some adolescents have experienced stress and tension within the family during the COVID-19 pandemic 22, and heightened levels of stress within a family place some adolescents at risk for maltreatment 23. A recent study on potential child abuse during the COVID-19 pandemic, indicated that parents with higher symptoms of anxiety and/or depression were at increased risk for maltreating their children 24. Prior research has shown that the impact of chaotic circumstances in a household on a child's wellbeing is influenced by the emotional reactivity of the child 25. Children who grew up in a chaotic household and had a lower emotional reactivity showed fewer behavioral problems during adolescence, while higher emotional reactivity was related to more behavioral problems during adolescence 25. Besides social stressors, there are also socioeconomic stressors that may threaten adolescents' mental health, such as limited educational opportunities and economic instability within households. A recent COVID-19 related study among Canadian adolescents revealed that 72% of the adolescents were very worried that the pandemic would impact their school year 26. Issues related to proper internet connection or the lack of physical space to concentrate on schoolwork compose inequal opportunities for online home schooling and may contribute to experiences of stress. Moreover, some adolescents are confronted with sudden financial concerns within the family as the COVID-19 pandemic also has enormous economic consequences 4. The same Canadian COVID-19 study also reported that 58% of the adolescents were a little or somewhat worried about their family's finances, and 36% was very worried 26. Although financial concerns could indeed negatively affect adolescent's mood, it might also lead to other-benefiting behavior, especially to family members with whom they share these concerns 27. Together, home schooling inequality and financial concerns in the family are socioeconomic stressors that may threaten adolescents' mood, specifically during the COVID-19 pandemic. --- The present study. In the present study we tested the impact of the COVID-19 pandemic on young people's mood, emotional reactivity, self-oriented and other-benefitting behavior. We conducted a two-week daily diary study during COVID-19 lockdown (two waves; May 2020 and November 2020) in which participants aged 10-25-years (i.e. adolescents and young adults) received online questionnaires every weekday, resulting in ten assessments per wave. We operationalized mood as mean levels of three emotions: vigor, depression, and tension, and operationalized emotional reactivity in terms of fluctuations in daily vigor, depression and tension. Self-oriented behavior was operationalized as disobedient behavior with potentially harmful consequences and other-benefitting behavior was defined as behaviors related to helping and comforting others. For this purpose, participants filled in questionnaires regarding their mood, stressors within the family context (i.e. family stress, emotional maltreatment, inequality of opportunity in online home schooling and financial concerns), the COVID-19 governmental rules (i.e. perceived unimportance of the rules, disobeying the rules) and societal commitment (i.e. emotional support to family, friends, and the willingness to help others during the pandemic), respectively. The aims of the present study were two-fold: Aim 1 was to explore the mean levels and fluctuations of vigor, tension, and depression, to unravel state specific effects for mood and emotional reactivity among two age samples: adolescents and young adults, at two time points: May 2020 (T1) and November 2020 (T2). We also explored whether mood and emotional reactivity showed age-related differences across adolescence and young adulthood. We conducted a follow-up two-week daily diary study in November 2020, to assess the long-term effects of the COVID-19 pandemic on the mean levels and fluctuations of the three mood states (preregistration: https:// osf. io/ d62y7/) and to explore whether changes in mood and emotional reactivity were age-dependent. We expected a decrease in vigor mean levels and an increase in mean levels of tension and depression 28,29. It was hypothesized that fluctuations in all three mood states would increase between May 2020 (T1) and November 2020 (T2) 30. Aim 2 was to test the hypotheses that the associations between heightened emotional reactivity and adolescents' self-oriented and other-benefitting behaviors are moderated by social and socioeconomic stressors within the family context (preregistration: https:// osf. io/ uf9dn/). Consequently, aim 2 is restricted to the adolescent sample, as the young adult sample consisted of college students who were less likely to live with their family. Aim 2 was solely tested using data from May 2020 (and not November 2020) for several reasons. First, we aimed to specifically test the effect of social and socioeconomic stressors during the period in which young people could not attend schools and were forced to spend more time within the family context, which was the case in May 2020 but not November 2020. Second, moderation analyses to examine the effects of individual differences require relatively large sample sizes for adequate statistical power, which is another reason why the data acquired in May were more suitable to answer the second research aim (N May = 462, N Nov = 238). We examined pre-registered associations between emotional reactivity and self-oriented and other-benefitting behavior. We expected no main effects of mood fluctuations (i.e. vigor, tensions, and depression) on self-(i.e., perceived unimportance of the rules, disobeying the rules) and other-benefitting behavior (i.e., emotional support to family and friends, and the willingness to help others during COVID-19). Instead, we expected interaction effects with social and economic stressors. We expected that emotional reactivity would serve as a differential susceptibility factor. That is, we tested whether individuals with higher levels of emotional reactivity were more likely to show negative outcomes, such as harmful self-oriented behavior, compared to individuals with lower levels of emotional reactivity when in a family environment in which they are dealing with social and socioeconomic stressors. At the same time, we expected that when in an environment with less exposure to social and socioeconomic stressors, individuals with higher levels of emotional reactivity will show positive outcomes, such as prosocial other-oriented behavior, compared to individuals who are less emotionally reactive. --- Methods Participants. Two samples participated: an adolescent and a young adult sample. With regard to the adolescent sample, 511 adolescents initially applied for the present study, of whom 485 actually participated. Of these 485 adolescents, 23 adolescents were excluded because they reported mood states on less than three days. Hence, the final sample included 462 adolescents (M age = 15.27 years, SD = 1.79, age range 10-20 years, 64% females) at the first timepoint. 431 of the 485 adolescents gave permission to be invited for the second timepoint of the study in November 2020. In total 258 adolescents reapplied for the second wave of the present study, of whom 255 also participated. 17 adolescents were excluded as they reported mood states on less than three days. The final adolescent sample consisted of 238 participants (M ageT1 = 15.54 years, SD T1 = 1.77, age range T1 10-19 years, 75% females). In the young adult sample 473 college students applied, 463 of whom also participated. We recruited college students specifically as they were in a comparable situation as the adolescents being restricted to attend education. We excluded 22 participants as they were not college students. Additionally, 37 participants were excluded for being older than 25 years (i.e., they did not fit our pre-defined young adult age range) and 33 participants were excluded for not having three or more days of mood state measurements. This resulted in a final sample of 371 young adults (M age = 21.49 years, SD = 1.91, age range 17-25, 81% females). Those young adults who attended a university programme and who gave permission for follow-up requests were invited for the second wave (N = 404). Initially, 266 young adults participated in the second wave of the present study. We excluded 10 participants as they reported mood states on less than three days, and 25 for not fitting our predefined age range at T1 (i.e. 17-25 years). In the final sample 231 young adults were included (M ageT1 = 21.26 years, SD T1 = 1.89, age range T1 17-25 years, 84% females). Table 1 shows detailed information on the sample compositions. The adolescent sample was representative of the Dutch population in terms of ethnicity (i.e. 25% of the current sample did not exclusively identify as Dutch, nationwide this is 28%) 31. However, in terms of educational level, both the adolescent and young adults sample were attending above average education levels 31. We explored potential differences between the attrition and non-attrition group. We performed two separate one-way ANOVAs, per age sample, to assess whether the attrition group differed from the non-attrition group on vigor, tension, and depression mean levels and fluctuations during T1. The results revealed no significant Procedure. For the recruitment of adolescents, we reached out to various secondary and high schools in the Rotterdam area in the Netherlands through existing contacts with the university. All adolescents from participating schools received a notification about our study through their schools communication platform. Those who were interested in participation could subscribe via an online form. Inclusion criteria were: living in the Rotterdam area and being aged 10 -22 years. The young adults were university students who attended a college program at Erasmus University Rotterdam (EUR). They were approached through the EUR website, email and social media platforms. Inclusion criteria were: attending EUR (whether as a national or international student) and being aged 17-25 years. Participants received a monetary reward of €15 for their participation, regardless of the amount of completed questionnaires. Participants gave permission by means of informed consent. In line with the guidelines from the ethics committee, parental informed consent was ensured for participants aged 15 years or younger. The present study was approved by the ethics committee of the Erasmus School of Social Table 1. Overview on the sample characteristics for the adolescents (N = 462) and the students (N = 371) in May 2020 (T1). Percentages do not necessarily add up to 100, due to missing data. Responses with "I do not know" were reported as missing data. --- Adolescents Students --- Variable N (percentage of sample) N (percentage of sample) Age (years) and Behavioural Sciences at the EUR (application 20-036) and conducted in accordance with the guidelines and regulations from the ethics committee. Participants received daily online questionnaires for two consecutive weeks from Mondays to Fridays, resulting in ten questionnaire days in total, in both May 2020 and November 2020. Data were collected using survey software platform Qualtrics. Based on their application date participants were assigned to two different batches in May: batch one starting on May 4, 2020 and batch two on May 11, 2020. Participants who applied on May 4th or later were assigned to batch two. In November 2020 participants were invited in one single batch. The COVID-19 related restrictions during May 2020 and November 2020 in the Netherlands were: social distancing (1.5 m physical distance), no large gatherings in groups, limited visitors allowed at home, staying home in case of symptoms, and closing of high schools and universities (latter applied to May 2020 only, however in November 2020 there were signals of re-closing the schools). Daily invitations to fill out the questionnaire were sent by email at 12:00 AM. In addition, a text message was sent at 19:00 PM as a reminder to participants who had not yet filled in the questionnaire of that particular day. On day 1, 5 and 10 the duration of the questionnaires was 15-20 min due to extra measures. All other days the duration was 5 to 10 min. Participants were instructed to fill out the questionnaire on the day of the invitation. They were also encouraged to respond to the new invitations even after missing one or more days. See S1 for an overview of the data preparation steps. --- Measures. Mood and emotional reactivity. The shortened Dutch translation of the Profile of Mood States Scale (POMS) was used to assess mood levels and mood fluctuations 32. For the present study, we focused on three subscales: vigor (five items, Cronbach's <unk> Day1 = 0.77 and 0.80 for adolescents on respectively T1 and T2; Cronbach's <unk> Day1 = 0.84 and 0.84 for young adults), tension (six items, Cronbach's <unk> Day1 = 0.84 and 0.86 for adolescents on respectively T1 and T2; Cronbach's <unk> Day1 = 0.86 and 0.87 for young adults), and depression (eight items, Cronbach's <unk> Day1 = 0.91 and 0.91 for adolescents on respectively T1 and T2; Cronbach's <unk> Day1 = 0.90 and = 0.92 for young adults). Participants were instructed to indicate to what extent they felt that the descriptions represented their current mood state. The questionnaire used a five-point Likert scale ranging from 1 ("not at all") to 5 ("extremely"). Mean scores were computed for each subscale as a measure of mood level. Mood fluctuations for each subscale were measured using the within person standard deviation approach 33. Social stressors. We examined two types of social stressors: family stress and emotional maltreatment. Family stress was assessed with two items of the recently developed Pandemic Questionnaire (van de Groep et al., https:// osf. io/ kgcdm/) (Cronbach's <unk> Week1 = 0.70): "I think my family and I live too much in each other's pockets since the schools are closed and most people work from home" (1) and "I would rather be as little as possible at home, since there is too much tension going on" (2). This selection of the Pandemic Questionnaire used a sevenpoint Likert scale, ranging from 1 ("not at all") to 7 ("totally true"). The items were administered weekly. A mean score was computed of the two items of both weeks. Emotional maltreatment was assessed as a lifetime experience using the Childhood Trauma Questionnaire-Short Form, a questionnaire proven to be reliable and valid among adolescents and adults 34. In the present study, we used the emotional neglect (5 items, Cronbach's <unk> Week1 = 0.84) subscale and a selection of the emotional abuse items (3 items, Cronbach's <unk> Week1 = 0.61), including, "I feel loved", "People in my family look out for each other" (both reversed scored), "I feel that someone in my family hates me", and "People in my family say hurtful and insulting things to me". The questionnaire used a five-point Likert Scale, ranging from 1 ("(almost) never") to 5 ("(almost) always"). We combined the items of the two subscales and computed standardized mean scores. The questionnaire was administered weekly. Socioeconomic stressors. The Inequality of Opportunity in Home schooling Questionnaire developed for the Covid-19 pandemic (Green et al., https:// osf. io/ 9g8cq/) was used to assess inequality of opportunity with regard to home schooling during the COVID-19 pandemic crisis. The questionnaire consisted of five items (Cronbach's <unk> Week1 = 0.68) and used a five-point Likert scale ranging from 1 ("not at all") to 5 ("totally true"). Participants were asked if they were dealing with difficulties in online home schooling. They were requested to indicate to what extent situations such as: "stable internet connection in order to take online classes", "a room in the house where you can do your schoolwork", and "a proper functioning laptop/computer/tablet in order to take online classes" applied to them. The questionnaire was administered weekly. Mean scores were computed. A single-item from the Pandemic Questionnaire (van de Groep et al., https:// osf. io/ kgcdm/) was used to assess financial concerns within the family due to the COVID-19 pandemic crisis, namely: "Are you worried that there will be financial problems in your family due to the COVID-19 pandemic?". The Pandemic Questionnaire consisted of open and multiple-choice questions that were specifically targeted to measure contributions to selfish behavior and risk behavior during COVID-19 pandemic. The selected item had a response scale from 1 ("never") to 5 ("almost always") and was administered weekly. We computed the mean score of the week 1 and 2. Self-oriented behavior. Self-oriented behavior in adolescents was operationalized as the unwillingness to follow the advised protective rules. We used a selection from the Pandemic Questionnaire (van de Groep et al., https:// osf. io/ kgcdm/). The items were grouped into two subscales: perceived unimportance of the rules (3 items, Cronbach's <unk> Week1 = 0.63), including items like "I would at this moment go to a party if my friends were organizing one at home". This subscale had a seven-point Likert scale, ranging from 1 ("entirely disagree") to 7 ("entirely agree"). Responses were reversed, hence a higher score represents more perceived unimportance of the rules. The disobeying the rules (2 dichotomous items) subscale contained items like "Did you meet with a friend last week while www.nature.com/scientificreports/ you were coughing, experiencing a sore throat, a runny nose, or some other health complaint?". The items were administered weekly. For both subscales mean scores were computed. Other-benefitting behavior. Other-benefitting prosocial behavior was conceptualized as: emotional support to family, to friends, and the willingness to help others during COVID-19. We used two measures from a prior study to assess these three forms of prosocial behavior 35. An adapted subscale from the Opportunities for Prosocial Actions questionnaire was used to assess emotional support to family and friends during the COVID-19 pandemic. Per target (i.e. friend and family member, respectively Cronbach's <unk> Day1 = 0.74 and Cronbach's <unk> Day1 = 0.74) the following three items were administered: "I comforted a family member today", "Today I did my best to spend my time with family", and "Today, I called and/or send a message to a family member". Responses were given on a scale from 0 ("not at all") to 5 ("a lot"). Mean scores were computed per target. The items were administered daily. The Contributions to Society questionnaire was used to assess the willingness to help others during the COVID-19 pandemic. Three items from this questionnaire were adapted to be suitable to the current COVID-19 pandemic crisis (Cronbach's <unk> Day1 = 0.74): "I have been committed to society for the past 24 h", "I have helped others in the past 24 h", and "I've been working for the people around me for the past 24 h". Responses on the statements were given on a seven-point Likert scale ranging from 1 ("not at all") to 7 ("totally true"). The items were administered daily. Mean scores were computed. --- Results Aim 1: unravel the short term and long-term effects of COVID-19 on mood and emotional reactivity. Both the adolescent and student sample from May 2020 (N T1 = 833) and November 2020 (N T2 = 469) were included in Aim 1, in which we explored the initial and long-term impact of the COVID-19 pandemic on young people's mood and emotional reactivity. In addition, we explored whether the exposure to social and socioeconomic stressors changed over time between T1 and T2. Differences in mood states and age-related effects at T1. Assumption checks and correlations among variables can be found in S2. Separate repeated measures ANOVAs were performed to explore (1) differences in mean levels of the three mood states and (2) differences in mood fluctuations. The first results showed a main effect of mood mean level, F (1.16, 966.30) = 236.99, p <unk> 0.001, <unk> 2 = 0.22. To test how these mood states differed across age, linear and quadratic age were entered to the model in a stepwise manner as continuous predictor. The results showed a significant linear age <unk> mood state interaction effect, F (1.16, 966.30) = 94.52, p <unk> 0.001, <unk> 2 = 0.10, as well as a significant quadratic age <unk> mood state interaction effect, F (1.17, 967.34) = 14.09, p <unk> 0.001, <unk> 2 = 0.02. Follow up analyses revealed that the quadratic models explained significantly more variance in the levels of the three mood states than the linear models (vigor <unk>R 2 = 0.02, p <unk> 0.001; tension <unk>R 2 = 0.01, p = 0.026; and depression <unk>R 2 = 0.01, p = 0.003). As can be seen in Fig. 1a, there was a quadratic age effect for vigor, F (2, 830) = 45.55, p <unk> 0.001, showing that vigor decreased with increasing age after which it reached a plateau in early adulthood, and a quadratic age effects for tension (F (2, 830) = 42.67, p <unk> 0.001) and depression, F (2, 830) = 23.44, p <unk> 0.001, showing that depression and tension increased with increasing age after which it reached a plateau in early adulthood. Next, we tested differences in mood fluctuations. There was a main effect of mood fluctuation, F (1.71, 1418.89) = 123.09, p <unk> 0.001, <unk> 2 = 0.13. Pairwise comparisons with Bonferroni correction revealed more fluctuations in vigor (M = 0.86, SD = 0.25), compared to tension (M = 0.69, SD = 0.35) and depression (M = 0.58, SD = 0.35), and more fluctuations in tension than in depression (Fig. 1c). Significant interaction effects were found first for linear age <unk> mood fluctuation, F (1.71, 1418.89) = 66.01, p <unk> 0.001, <unk> 2 = 0.07) and then for quadratic age <unk> mood fluctuation, F (1.72, 1429.27) = 14.88, p <unk> 0.001, <unk> 2 = 0.02). Post hoc analyses revealed a negative linear effect of age on vigor fluctuations, F (1, 831) = 11.91, p = 0.001, vigor fluctuations decreased with age. The quadratic model did not explain additional significant variance in vigor fluctuations. In contrast to vigor, adding quadratic age to the model did result in greater explanation of the variance in tension and depression fluctuations (tension <unk>R 2 = 0.02, p <unk> 0.001; depression <unk>R 2 = 0.02, p <unk> 0.001). We found quadratic effects of age on tension, F (2, 830) = 34.48, p <unk> 0.001, and depression, F (2, 830) = 35.76, p <unk> 0.001, fluctuations, with a peak in early adulthood around age 20 (Fig. 1c). Differences in mood states and age-related changes at T2. To examine whether the T1 (i.e., May 2020) mood patterns would be replicated at T2 (i.e., November 2020), we performed repeated measures ANOVAs. For T2, like T1, we found a main effect of mood state, F (1.16, 539.24) = 60.36, p <unk> 0.001, <unk> 2 = 0.11. Adding age, both as a linear and quadratic factor, to the model resulted in significant interaction effects, F (1.16, 539.24) = 18.5, p <unk> 0.001, <unk> 2 = 0.04, and F (1.16, 539.06) = 6.38, p = 0.009, <unk> 2 = 0.01 (respectively linear and quadratic). Post-hoc analyses revealed that compared to the linear model, the quadratic model explained significantly more of the variance in vigor (<unk>R 2 = 0.03, p <unk> 0.001). As can be seen in Fig. 1b, we found that vigor levels increased during adulthood after an initial dip during late adolescence/early adulthood, F (2, 466) = 18.93, p <unk> 0.001. Furthermore, we found a positive linear age effect for mean levels of tension during the second wave, F (1, 467) = 5.70, p = 0.017. Adding the quadratic age factor did not lead to an increase in the explained variance in tension (p > 0.05). Finally, the results showed a positive linear age effect for depression, showing an increase in mean levels across adolescence, F (1, 467) = 5.84, p = 0.016 (see Fig. 1b). Again, adding the quadratic age factor did not lead to an increase in the explained variance in tension (p > 0.05). In sum, the mood mean level results at T2 overall replicate the findings at T1. Next, we tested the difference in mood fluctuations at T2. We found a main effect of mood state, F (1.76, 821.85) = 22.50, p <unk> 0.001, <unk> 2 = 0.05. We found an interaction effect with age as linear factor, F (1.76, 821.85) = 8.98, p <unk> 0.001, <unk> 2 = 0.02. Adding age as quadratic factor to the model did not result in a significant interaction effect (p > 0.05). Post-hoc tests, revealed no linear age effect in vigor fluctuations (p > 0.05). We found positive linear effects of age on tension, F (1, 467) = 3.94, p = 0.048, and depression fluctuations, F (1, 467) = 7.82, p = 0.005 (see Fig. 1d). Overall, mood fluctuation levels in November 2020 partially overlap with the pattern found in May 2020. That is adolescents showed more fluctuations in vigor, at both time points, but in November 2020 the peak in negative emotions was no longer present. --- Long-term effects of COVID-19 on mood and emotional reactivity. To test the hypothesis that the continuation of the COVID-19 pandemic has a negative impact on adolescents' mood and emotional reactivity, we performed repeated measures ANOVAs to assess effects of time on mood mean levels and fluctuations. In both models, we added linear and quadratic age as continuous predictor in a stepwise manner to test for covariate effects. Post-hoc analyses included linear and curve estimation regression analyses per outcome (i.e. vigor, tension, and depression). Of the 494 adolescents and young adults at T2, 35 participated at T1 but were later excluded due to an insufficient number of mood state reports (<unk> 3 days). Hence, we included measurements at both time points for 459 participants in total. The analyses described below were performed within this sample. We found a main effect of time, F (1.00, 457.00) = 7.45, p = 0.007, <unk> 2 = 0.02, and an interaction effect between time (i.e. T1 vs. T2) and mood state, F (1.27, 578.59) = 25.82, p <unk> 0.001, <unk> 2 = 0.05. Post-hoc tests revealed that there was a decrease in vigor levels from T1 (M = 2.96, SD = 0.78) to T2 (M = 2.90, SD = 0.79), t = 2.32, p = 0.021. In contrast, we found increases in tension (M T1 = 1.74, SD T1 = 0.72; M T2 = 1.85, SD T2 = 0.77) and depression Figure 1. Graph A: adolescents and young adults reported higher levels of vigor compared to tension and depression in May 2020 (T1). All three mood levels showed quadratic age effects, with a drop in vigor and a peak in tension and depression in young adulthood. Graph B: adolescents and young adults reported higher levels of vigor compared to tension and depression in November 2020 (T2). Vigor showed a quadratic age effect with a drop in young adulthood, whereas tension and depression mean levels showed positive linear effects. Graph C: Adolescents and young adults reported more fluctuations in vigor, compared to tension and depression, which was specifically pronounced among younger adolescents at T1. Fluctuations in vigor showed a negative linear effect as a function of age, while tension and depression fluctuations peaked during late adolescence/young adulthood. Graph D: Adolescents show more vigour fluctuations compared to tension and depression fluctuations at T2. This disparity is less profound among young adults. There was no age effect of vigour fluctuations. However, tension and depression fluctuations showed a positive linear age effect across adolescence. levels (M T1 = 1.56, SD T1 = 0.65; M T2 = 1.63, SD T2 = 0.71) between T1 and T2, respectively t = -4.37, p <unk> 0.001 and t = -2.75, p = 0.006. We also found an interaction effect between time, mood state, and linear age, F (1.27, 578.59) = 20.05, p <unk> 0.001, <unk> 2 = 0.04. The decrease in vigor levels and the increases in tension and depression levels between T1 an T2 were greater among the younger compared to older adolescents/young adults, see Fig. 1a,b. There was no interaction with age as quadratic factor (p > 0.05). The same analysis was performed for changes in mood fluctuations over time. We found a main effect of time, F (1.00, 865.78) = 7.45, p = 0.002, <unk> 2 = 0.02, which was qualified by an interaction effect between time and mood state, F (1.89, 578.59) = 6.87, p = 0.001, <unk> 2 = 0.02. Post-hoc tests revealed that there was a trend towards a decrease in vigor fluctuations between T1 (M = 0.85, SD = 0.24) and T2 (M = 0.82, SD = 0.26), t = 1.94, p = 0.053. We found increases in tension fluctuations between T1 (M = 0.70, SD = 0.34) and T2 (M = 0.73, SD = 0.33), t = -2.23, p = 0.027 There were no significant differences in depression fluctuations between T1 and T2 (p > 0.05). Finally, we found an interaction effect between time, mood state, and linear age, F (1.89, 865.78) = 4.90, p = 0.009, <unk> 2 = 0.01, such that the increase in tension fluctuations between T1 an T2 was greater among the younger adolescents, see Fig. 1c,d. There was no significant interaction effect with age as quadratic factor (p > 0.05). Aim 2: examine emotional reactivity as susceptibility marker within the family context. For Aim 2 we used the May 2020 measurements of the adolescent sample (N = 462) to examine the effects of family household experiences and emotional reactivity on self-and other-benefitting behaviors. Associations between emotional reactivity and stressors. Assumption checks on all measures as well as a complete overview on the correlations between all independent and dependent variables can be found in S3. To investigate the relations between emotional reactivity and the social and socioeconomic stressors, we performed 15 partial correlational analyses. Multiple tests may lead to an increase in Type I error, therefore we used

Adolescence is a formative period for socio-emotional development which is threatened by the COVID-19 pandemic. The current longitudinal study examined two aims: (1) the short-and longterm effects of the pandemic on young people's mood (i.e. vigor, tension, and depression levels) and emotional reactivity (i.e. fluctuations in daily mood), and (2) the impact of stressors on mood, emotional reactivity, self-oriented (i.e. maladaptive behavior towards COVID-19 rules) and otherbenefitting behaviors (i.e. behavior aimed at helping and comforting others). We conducted an online two-week daily diary study among 462 Dutch adolescents (M age = 15.27 years, 64% females) and 371 young adults (M age = 21.49 years, 81% females) in May 2020, with a follow-up in November 2020 (N = 238 and 231, respectively adolescents and young adults). In May 2020, young adults and older relative to younger adolescents showed higher levels and more fluctuations in tension and depression and lower levels of vigor. Vigor levels decreased and tension and depression levels increased between May 2020 and November 2020, especially for younger adolescents. There were positive associations between instability of negative emotions (i.e. tension and depression fluctuations) and the exposure to stressors (i.e. family stress and inequality of online homeschooling) in the adolescent sample. Together, this study demonstrates vulnerability regarding young people's mood and emotional reactivity during the COVID-19 pandemic, especially for adolescents who experience more stressors. Adolescence and young adulthood are formative periods for emotional and social development 1,2 . The COVID-19 pandemic threatens this development, as it required adolescents and young adults to deal with marked life changes like fully transitioning to online education, having limited to no physical contact with peers or friends, and being constrained to stay at home 3 . Both adolescents and young adults are likely to have experienced remarkable alterations in most of their social relationships, due to social distancing restrictions 4 . The aim of this study is to test the emotional and social consequences of the COVID-19 pandemic and how these influence the self-oriented and other-benefitting behavior of individuals in these two formative phases for social development. While for some the COVID-19 pandemic might be a window of opportunity for positive other-benefitting behavior, such as helping and comforting, others may be prone to more self-oriented behaviors with possible negative consequences, such as not following societal rules 2 . The main aims of the current study are to: (1) test for age-related changes in daily levels and fluctuations of vigor, tension and depression early and later in the pandemic, and (2) test the pre-registered hypothesis that the interaction between emotional reactivity

between T1 and T2 (p > 0.05). Finally, we found an interaction effect between time, mood state, and linear age, F (1.89, 865.78) = 4.90, p = 0.009, <unk> 2 = 0.01, such that the increase in tension fluctuations between T1 an T2 was greater among the younger adolescents, see Fig. 1c,d. There was no significant interaction effect with age as quadratic factor (p > 0.05). Aim 2: examine emotional reactivity as susceptibility marker within the family context. For Aim 2 we used the May 2020 measurements of the adolescent sample (N = 462) to examine the effects of family household experiences and emotional reactivity on self-and other-benefitting behaviors. Associations between emotional reactivity and stressors. Assumption checks on all measures as well as a complete overview on the correlations between all independent and dependent variables can be found in S3. To investigate the relations between emotional reactivity and the social and socioeconomic stressors, we performed 15 partial correlational analyses. Multiple tests may lead to an increase in Type I error, therefore we used Bonferroni correction for multiple comparisons, p = 0.003 36,37. All analyses were controlled for mean mood levels. We found no significant associations between vigour fluctuations and the social and socioeconomic stressors. For tension fluctuations, the findings showed a positively association with experiences of family stress (r = 0.16, p = 0.001), see Fig. 2a. With regard to depression fluctuations, we found significant positive association with family stress (r = 0.19, p <unk> 0.001) and the inequality of opportunity in online home schooling (r = 0.19, p <unk> 0.001). As can be seen in Fig. 2b, adolescents with more exposure to these stressors reported more fluctuations in their feelings of depression. --- Associations between emotional reactivity and self-oriented and other-benefitting behaviors. To test the effect of emotional reactivity on self-oriented and other-benefitting behaviors, we performed hierarchical regression analyses in which fluctuation in vigor, tension, and depression were entered as independent variables, and the opportunity to provide emotional support to family and friends, the willingness to help others during the pandemic (i.e. benefitting-oriented behavior), the perceived unimportance of the rules, and disobeying the rules (i.e. self-oriented behavior) as dependent variables. To control for mood, mean levels were entered in a second step in each relevant model. We performed hierarchical regression analyses per mood fluctuation (vigor, tension, and depression) and dependent variable (five measures of self-oriented and other-benefitting behaviors), resulting in 15 regression analyses in total. We only report effects that survived Bonferroni corrections, resulting in p <unk> 0.01, controlling for dependency between dependent measures 36,37. The independent variables were mean centered prior the analyses in order to diminish multicollinearity when testing main effects 38. The results revealed no main effects of fluctuations in vigor, tension, and depression on the perceived unimportance of the rules, disobeying the rules, the opportunity to provide emotional support to family and friends, and the willingness to help others during the COVID-19 pandemic (all ps > 0.05). Detailed information on the associations between the social and socioeconomic stressors and self-oriented and other-benefitting behaviors can be found in S4. Interaction effects between emotional reactivity and social and socioeconomic stressors on self-oriented and other-benefitting behaviors. The macro PROCESS by Hayes was used to perform moderation analyses in order to test for interaction effects between the social and socioeconomic stressors and emotional reactivity (i.e. mood fluctuations) on self-and other-oriented behaviors 39. Correlation analyses revealed moderate to high correlations between the measures of each stressor type, respectively r = 0.51 and r = 0.40 for the social and socioeconomic stressors (see Table S3b). We, therefore, combined emotional maltreatment and family stress to create a social stressor composite score, and inequality of online home schooling and financial concerns to create a socioeconomic composite score. Composite scores were computed by calculating the mean of the standardized z-scores. This section includes multiple tests, we therefore only report effects that survived Bonferroni corrections, for multiple comparisons with an alpha of <unk> 0.002 36,37. No effects were found with regard to self-oriented behaviors (i.e., perceived unimportance of the COVID-19 rules and disobeying the COVID-19 rules) and emotional support to family (see S5ab). Next, we assessed the hypothesis that social and socioeconomic stressors moderated the associations between mood fluctuations and the opportunity to provide emotional support to friends. Results did not survive the Bonferroni correction, and hence none of the effects were significant (see Table 2 for all p-values that did not survive the correction, see Table S5ab for all other non-significant findings). Lastly, we tested the hypothesis that social and socioeconomic stressors moderate the associations between mood fluctuations and the willingness to help others during COVID-19. As can be seen in Table 2, no significant effects were found as the findings did not survive the Bonferroni correction. Long-term effects of COVID-19 on social and socioeconomic stressors. For descriptive exploratory purposes we examined whether the social and socioeconomic stressors changed over time. We performed paired sample t-tests, per variable, to test for changes over time. We found an increase in exposure to emotional maltreatment by family members between T1 (M = 1.72, SD = 0.65) and T2 (M = 1.84, SD = 0.76), t = -3.51, p = 0.001. At the same time, adolescents reported overall a decrease in family stress between T1(M = 3.32, SD = 1.64) and T2 (M = 3.06, SD = 1.65), t = 2.54, p = 0.012. No significant changes over time were found for inequality of opportunity online home schooling and financial concerns (ps > 0.05). --- Discussion In the present study we investigated (1) the impact of the COVID-19 pandemic on adolescents' and young adults' daily mood during the first months of the crisis, and (2) whether mood states differed across adolescence and young adulthood (i.e., as a function of age), (3) long-term effects of the pandemic on daily mood and mood fluctuations, and (4) the moderating role of social and socioeconomic stressors on the associations between emotional reactivity and self-oriented and other-benefitting behaviors among adolescents. During both the first and second wave (May and November 2020) of the pandemic in the Netherlands, adolescents and young adults showed higher levels of vigor compared to tension and depression, along with more fluctuations in their daily vigor levels. In May 2020, these differences between mood states were smaller in older adolescents, but from May 2020 to November 2020 these differences particularly decreased for younger adolescents (i.e., they were most sensitive to mood changes over time). A closer examination of how heightened emotional reactivity during adolescence interacts with COVID-19-induced social and socioeconomic stressors, revealed no evidence Table 2. Interaction effects between emotional reactivity and stressors on self-oriented and other-benefitting behaviors among adolescents in May 2020 (T1) that did not survive the Bonferroni correction. The social stressor is a composite score of the measures emotional maltreatment and family stress, the socioeconomic stressor is a composite score of the measures inequality of opportunity in online home schooling and financial concerns. www.nature.com/scientificreports/ that emotional reactivity acts as a differential susceptibility marker. We did find associations between tensions and depression fluctuations on the one hand, and the social and socioeconomic stressors on the other hand. On average adolescents and young adults showed higher levels of vigor (a positive emotion) compared to tension and depression (i.e., negative emotions), and this difference was particularly pronounced in younger adolescents. With increasing age, differences between mean levels of vigor, tension, and depression diminished. This suggests that younger adolescents showed possibly higher resilience in the first months of the pandemic. Recent COVID-19 studies, in which adolescents who were also followed prior to the pandemic, reported either no changes in negative affect, or a decrease in tension and an increase in vigor levels in comparison to pre-pandemic conditions 22,35. Besides the well-documented vulnerabilities, adolescence is also a period characterized by goal flexibility, the capacity to quickly shift priorities, and flexibility in social behavior 2,40,41. This flexibility enables young people to adapt adjustments in challenging situations 2. In the current study this seemed to be primarily applicable to relatively young adolescents, and restricted to the early months of the pandemic. Indeed, other COVID-19 studies showed an increase in anxiety and depressive symptoms, and reduced life satisfaction among adolescents 42,43. These mixed findings may be a result of the dynamic and changing effects of the pandemic, which is likely to differ from nation to nation. As a result, policy-responses to COVID-19 are continuously being altered worldwide. In addition, individual differences in pre-pandemic mood levels, coping mechanisms, and emotion-regulation might contribute to the varied findings. In the present study we also examined the long-term effects of the pandemic. Literature on previous economic crises and long-lasting stressful events indicate that such situations tend to worsen mood and psychological wellbeing in general [28][29][30]. In line with our expectations, the present findings indeed indicated that the positive affect of adolescents and young adults decreased, while negative affect increased during the course of the pandemic. Interestingly, we found that particularly the younger adolescents, who at T1 still displayed greater resilience in terms of mood, experienced stronger changes in mood over time in the direction of less positive and more negative mood. Past research on adolescents' resilience in situations of adversity have shown that their wellbeing depends on the complex interplay of systems consisting of peers, family, school, and communities, rather than individual factors alone 44. Support from family and friends has been shown to positively influence resilient behavior 45,46. These systems are therefore likely to contribute to the stability of young people's resilience during the COVID-19 pandemic. As the COVID-19 pandemic continues, with unique challenges that may differ between nations, the current findings underline the need to invest in promoting resilient systems surrounding adolescents and young adults to improve their mental health 45. We found that adolescents with more fluctuations in all three mood states were more likely to experience family stress and inequality of opportunity in online home schooling, and that exposure to financial concerns and emotional maltreatment were positively related to depression fluctuations. Hence, we replicate findings showing associations between emotional reactivity and stressors 25,28,30. A recent study has shown that difficulties during online home schooling and familial conflict are associated with an increase in depressive symptoms during the pandemic as compared to pre-pandemic situations 42. In another study family income stability has been shown to be a protective factor against anxiety among Chinese college students 47. The present findings are valuable for implementing effective interventions, as they highlight the importance of a healthy social environment and stable systems. Interventions should not only be aimed at increasing resiliency at the individual level, but also on making the systems surrounding individuals, such as households and schools, more resilient. For example, this can be achieved by providing family-based interventions aimed at improving parental skills and restoring parent-child relationships 48. Although more complex, it is also crucial to focus on how systematic exposure to stressors, such as socioeconomic inequity, can be reduced at societal level, as it has been linked to poorer mental health among children and adolescents 49. Re-assessing our current public health and education strategies could be one step forward. Future research should further investigate which psychological mechanisms and environmental factors may mitigate the adverse effects of COVID-19 related social and socioeconomic stressors on adolescents' mental health, especially for vulnerable groups. In contrast to literature on adolescents' emotional reactivity, we did not find a peak in fluctuations in negative emotions during mid-adolescence [9][10][11]. Instead the present cross-sectional findings of May 2020 suggested a peak in early adulthood, possibly indicating a shift and pointing to prolonged emotional instability as a result of the current pandemic 50. Together with our findings on the mean mood levels (i.e. less resilience with increasing age), the present study suggests that older adolescents and young adults were struggling more compared to the early-and mid-adolescents during the first months of the pandemic. A possible explanation for this could be the interference of COVID-19 with one's transition to adulthood, which opens a new door of concerns and challenges, such as becoming an independent member of society 51. There is one other study in which the developmental trajectories of certain mood fluctuations were not found to peak during mid-adolescence 52. In this longitudinal study among Dutch adolescents, linear declines in sadness and anger fluctuations were found across adolescence. Hence in contrast to our results, these findings do not indicate greater emotional instability in older adolescents, making it more likely to assume that in the present study the COVID-19 pandemic may have influenced emotional instability. Maciejewski and colleagues (2015) observed an initial increase for anxiety related fluctuations during early adolescence, followed by a decrease in mid-adolescence and slight increase again at the end of adolescence, which they prescribed to the possible involvement of stress when entering adulthood 52. The same authors reported a peak in mood fluctuations during mid-adolescence in a more recent study, however only in a minority of their sample 53. This finding highlights the importance of examining individual differences. The current study is novel in covering two periods and a wide age range during the pandemic. We show that as the pandemic continues to interfere with our daily lives, tension and depression fluctuations no longer peak (compared to T1) but increase linearly as a function of age. However, longitudinal data covering a longer period are needed to assess whether the pandemic indeed leads to an extension of adolescent emotional instability or whether interindividual variability in the intraindividual mood fluctuations could also play a role. We also examined whether the exposure to social and socioeconomic stressors changed over time. The experience of socioeconomic difficulties was stable, however we observed that family stress and emotional maltreatment changed. Interestingly, the latter two constructs showed an opposite pattern, as family stress decreased but emotional maltreatment increased over time. An explanation could be that due to the reopening of schools, families experienced less stress, since they spent more time independent of each other. At the same time, the pandemic was not over and other restrictions still applied. Hence in those families with more problems, the longevity of the pandemic may have outweighed the reopening of schools, resulting in more severe household hardship such as emotional maltreatment towards adolescents. We further aimed to test the hypothesis that emotional reactivity acted as a differential susceptibility marker, resulting in more self-oriented behavior and less other-oriented behavior in an environment with more social and socioeconomic stressors. In line with our expectations emotional reactivity alone was not associated with self-or other-oriented behavior. However, contrary to our expectations, we did not find interactions with the social and socioeconomic stressors. Hence, the present study provides no evidence for the assumption of emotional reactivity as differential susceptibility marker. A possible explanation could be that in the present study we operationalized emotional reactivity, stressors, and self-oriented and other-benefitting behaviors in too many ways. The number of different variables resulted in multiple testing, making it impossible to interpret the results without proper corrections, and therefore reducing the likelihood of robust effects. Future research should target specific variables when investigating the complex interplay between emotional reactivity, self-oriented and other-benefitting behaviors, and the social environment. The present study has several limitations which have to be taken into account when interpreting the results, and which should be addressed in future research. First, the present study has no pre-pandemic measures, which are needed to assess to what extent daily mood has changed within individuals as a function of the pandemic. Although we did not observe changes in mood levels during the pandemic, it remains unknown in the present study whether adolescents' and young adults' mood were different before the outbreak of COVID-19. A second limitation concerns the research sample. The samples consisted of educated participants, of whom the majority also has a medium to high socioeconomic status, which is not representative of the Dutch population. The lack of heterogeneity, in terms of educational and socioeconomic background, limits the generalizability of the present findings to the whole adolescent and young adult population. Access to participate in the study and initial eligibility could have contributed to selection bias and thus partially explain our lack of heterogeneity. Our study was done fully online and recruitment of participants was conducted through online platforms. Hence, individuals with limited access to online resources may have been missed out from our study. Our young adult sample was not representative as we focused on university college students only. In addition, adolescents from schools who did not wish to collaborate with our project did not have access to our study as they did not receive an invitation to participate. Ideally including participants who experience more variation in exposure to social and socioeconomic stressors during COVID-19 would be beneficial for our understanding of its impact on adolescents' mood, especially as they might more severely impact adolescents with a lower socioeconomic status. As we found no differences in mood mean levels and emotional reactivity between the attrition and non-attrition group, it is unlikely that missingness for the second wave due to drop-out influenced the current findings. Lastly, the present study did not contain measures on personality or coping mechanisms, which might also have a mediating or moderating effect on the relationship between emotional reactivity and self-oriented and other-benefitting behaviors in general, as well as during the COVID-19 pandemic. Future research should include these measure as they might give us a better understanding on how to protect adolescents' mental health during challenging times. In conclusion, in the present study younger adolescents relative to older adolescents showed higher levels of vigor and lower levels of tension and depression in both May 2020 and November 2020. Paradoxically, the longevity of the pandemic primarily affected younger adolescents: feelings of vigor decreased, while feelings of tension and depression increased between May and November 2020, particularly among younger adolescents. We furthermore found evidence for a link between vulnerability factors (i.e. family stress and inequality of opportunity in online homeschooling) and instability in negative affect (i.e. tension and depression fluctuations) during the first months of the pandemic. Finally, together, these results highlight the complex interplay between emotional reactivity and influences from the social environment during emotional development in adolescence. --- Data availability The data generated and analysed during the present study, along with the computer code, are upon reasonable request from the authors, available in the EUR Data Repository. --- Author contributions --- Competing interests The authors declare no competing interests.

Adolescence is a formative period for socio-emotional development which is threatened by the COVID-19 pandemic. The current longitudinal study examined two aims: (1) the short-and longterm effects of the pandemic on young people's mood (i.e. vigor, tension, and depression levels) and emotional reactivity (i.e. fluctuations in daily mood), and (2) the impact of stressors on mood, emotional reactivity, self-oriented (i.e. maladaptive behavior towards COVID-19 rules) and otherbenefitting behaviors (i.e. behavior aimed at helping and comforting others). We conducted an online two-week daily diary study among 462 Dutch adolescents (M age = 15.27 years, 64% females) and 371 young adults (M age = 21.49 years, 81% females) in May 2020, with a follow-up in November 2020 (N = 238 and 231, respectively adolescents and young adults). In May 2020, young adults and older relative to younger adolescents showed higher levels and more fluctuations in tension and depression and lower levels of vigor. Vigor levels decreased and tension and depression levels increased between May 2020 and November 2020, especially for younger adolescents. There were positive associations between instability of negative emotions (i.e. tension and depression fluctuations) and the exposure to stressors (i.e. family stress and inequality of online homeschooling) in the adolescent sample. Together, this study demonstrates vulnerability regarding young people's mood and emotional reactivity during the COVID-19 pandemic, especially for adolescents who experience more stressors. Adolescence and young adulthood are formative periods for emotional and social development 1,2 . The COVID-19 pandemic threatens this development, as it required adolescents and young adults to deal with marked life changes like fully transitioning to online education, having limited to no physical contact with peers or friends, and being constrained to stay at home 3 . Both adolescents and young adults are likely to have experienced remarkable alterations in most of their social relationships, due to social distancing restrictions 4 . The aim of this study is to test the emotional and social consequences of the COVID-19 pandemic and how these influence the self-oriented and other-benefitting behavior of individuals in these two formative phases for social development. While for some the COVID-19 pandemic might be a window of opportunity for positive other-benefitting behavior, such as helping and comforting, others may be prone to more self-oriented behaviors with possible negative consequences, such as not following societal rules 2 . The main aims of the current study are to: (1) test for age-related changes in daily levels and fluctuations of vigor, tension and depression early and later in the pandemic, and (2) test the pre-registered hypothesis that the interaction between emotional reactivity

Introduction Compared to more equitable societies, those with wider income gaps between rich and poor have worse population health outcomes [1,2], including obesity [3], cancer [4], cardiovascular disease [5], and mortality [4,6]. A 2018 longitudinal study found a strong association between state-level income inequality and life expectancy (LE) [7], a commonly used summary measure of population health representing the average number of years a person in an area can expect to live, based on current age-specific mortality rates [8]. The consensus in public health is the more equitable society is, the better the population health [2]. However, empirical evidence suggests that the association between income inequality and population health is more nuanced. For example, research has determined that the strength of these associations between LE and income inequality in the USA was significantly stronger in areas of higher overall income [4] and in rural areas compared to urban areas [6]. Results of a Canadian study further suggest that rural populations may be more vulnerable to the influence of income inequality on health due to certain population and geographic characteristics, such as reduced access to basic health care services and greater socioeconomic and demographic homogeneity compared to urban areas [9]. Nonetheless, the reasons for these observed nuances are not fully understood. Although rural-urban status is often viewed and conceptualized for analysis as a dichotomy, it is more accurately understood as a continuum [10], with important implications for population health [10,11]. Dichotomous measures of rural-urban status may be easier to interpret; however, important nuances between less urban and rural areas may be missed, especially in those areas of intermediate rural-urban status. There are a number of rural-urban status measures available on the national scale that attempt to capture a more detailed gradation of what it means for a geographic area (e.g., neighborhood, county, state, etc.) to be rural or urban [12]. Furthermore, the magnitude and direction of associations between income inequality and health depend on the geographic unit of analysis [13]. The association between higher income inequality and poor health outcomes is well established [1-4, 7, 9]. However, the majority of evidence demonstrating an association between income inequality and worse population health was conducted on large geographic scales, such as the national, state, or regional levels [14]. On a finer geographic scale, such as the neighborhood or municipal level, the processes related to inequality may operate differently [15]. It has been posited that income inequality at smaller scales may be less likely to reflect the degree of social stratification and endogenous inequality in the wider society, and as a result, be related to population health outcomes [14]. A 2015 study conducted in Switzerland found that mortality was actually lower in neighborhoods with highincome inequality than those with lower income inequality. This finding has been deemed the "Swiss paradox" [16]. Studies examining income inequalities and health on a fine geographic scale are lacking in the USA, however. Despite this, there is growing evidence that neighborhood and community-level factors play a unique and critical role in population health outcomes. According to an analysis by Woolf and colleagues, a "web of conditions" on the neighborhood or community level that may be difficult to disentangle contributes to individual behaviors and, therefore, population health and health inequalities [17]. These include, but are not limited to, race/ethnicity, education, socioeconomic status, the built environment, access to critical resources of everyday living (e.g., healthy foods, recreation, etc.), and social support and cohesion. Recent evidence supports the importance of examining population health inequalities on a fine geographic scale. A 2020 analysis found several key associations between population health outcomes and measures of healthcare access and social determinants of health that would have been masked had the analysis been conducted on a higher level of geospatial aggregation, such as the county or state level [18]. Another study identified small-scale associations between neighborhood-level socioeconomic status and premature mortality [19] and HIV status [20], associations which may have been masked had a higher level of spatial aggregation been used. Therefore, there is a support for and a need to investigate more nuanced relationships between socioeconomic status, inequality, community type, and health outcomes on a fine geographic scale. To facilitate such investigations, improvements to the quality of methods and availability of national data on LE on a fine geographic scale, such as the census tract, allow researchers and policymakers to better understand the subtle but important differences of the impact of potential small-scale geographic and place-based characteristics on population health. However, to date, few studies have directly examined the potential for other socioeconomic and demographic factors to moderate or attenuate the established association between income inequality and LE. Therefore, the objective of this empirical study was to determine how income and rural-urban status potentially moderate the associations between LE and income inequality on a fine geographic scale (i.e., census tract). --- Methods --- Measures LE at birth (2010-2015) for each US census tract was the main outcome variable and was obtained from the CDC Wonder database [21]. Census tracts are small subdivisions of a county or statistically equivalent entity, such as a city [22] and can be considered as a geographic cluster of neighborhoods or small communities. Each census tract generally contains between 1200 and 8000 people, with an optimum size of 4000 [22]. LE was then linked by census tract to socioeconomic and demographic data from the 2010 US Census [23] and the 2010 American Community Survey (ACS) [24]. The main explanatory variable was Gini index, which measures the extent to which the income distribution among units within an area deviates from a perfectly equal distribution and ranges from 0 for perfectly equitable distributions to 1 where all income is concentrated in one individual [25]. Key moderator variables from the 2010 ACS included median household income and population density in each census tract. Population density is a continuous measure and one of the most commonly used proxy measures for rural-urban status in the population health literature. All three variables-Gini index, median household income, and population density-were converted into quintiles (Q1-Q5) to capture the continuous nature of each element and to aid in interpretation [16,[26][27][28]. Covariates used in the analysis were percent of the population that identified as Black or African American (% Black), percent identifying as Hispanic or Latino/a/x (% Latino/a/x), percent with a bachelor's degree or higher, and median age of the population in each Census tract. Descriptive statistics (means, standard deviations, medians, and interquartile ranges [IQR]) were obtained for all study variables. Checks of normality on all study variables were conducted using Kolmogorov-Smirnov statistics and visually using Q-Q plots. Bivariate associations were estimated for all pairs of study variables (e.g., LE and Gini index) using Spearman's rank correlation tests for non-normally distributed measures. Partial correlations were used to estimate the adjusted bivariate associations between LE and Gini index, adjusting for covariates. These partial correlations were conducted for all census tracts combined and stratified by quintile of income and rural-urban status to assess differences and linear trends in the magnitude of the associations between LE and Gini index by these factors. Average LE was also calculated for groups of census tracts cross-classified by quintile of income, rural-urban status, and Gini index. Lastly, multivariable models were used within each quintile of income inequality to examine the adjusted associations between median household income, population density, and their interaction with census-tract LE. All data were aggregated, and no individual-level data were used for this analysis. SAS version 9.4 (Cary, NC) and IBM SPSS version 27 (Armonk, NY) were used for analyses. --- Results There were 66,857 census tracts in the analytic sample. The mean tract-level LE was 78.3 years with a standard deviation (SD) of 4.0 years, a minimum LE of 56.3 (tract # 40001376900 in Adair County, Oklahoma) and a maximum of 97.5 (tract # 37037020104 in Chatham County, North Carolina) (Table 1). The mean tract-level values for median household income were $66,792 (SD $32,688) and 0.426 (0.064) for Gini index. The mean tract-level population density was 5231 people/square mile, with a standard deviation of 11,729. Spearman correlation coefficients between LE and Gini index, median household income, percent Black population, and percent with a bachelor's degree or higher were -0.132, 0.672, -0.364, and 0.608, respectively (Table 2). All were statistically significant (p <unk> 0.001). Furthermore, LE was significantly correlated with population density (Spearman's <unk> = 0.035, p <unk> 0.001). Gini index was significantly correlated with median household income (Spearman's <unk> = -0.329, p <unk> 0.001), % Black population (Spearman's <unk> = 0.108, p <unk> 0.001), and population density (Spearman's <unk> = 0.054, p = 0.001). Substantial differences in the associations between LE and Gini index by quintile of median household income were evident (Fig. 1). Among the tracts in the lowest income quintile (Q1), LE was negatively associated with Gini index and LE decreased monotonically from 74.9 years in the tracts with the lowest Gini index to 73.3 years in the tracts with the highest Gini index (p = 0.003). However, among the tracts in the highest income quintile (Q5), the association between LE and Gini index was reversed and positive: higher LE was observed in areas with higher Gini indices (LE 81.4 in Q1 of Gini and 82.2 in Q2 of Gini, p = 0.011). Additional variability was observed when stratifying the tracts further by rural-urban status. The lowest average LE (72.3 years) was observed in the tracts with the lowest income, highest Gini index, and with intermediate (Q3) rural-urban status. The highest average LE (82.3 years) was observed in the tracts with the highest income (Q5), second-highest Gini index (Q4) and second most urban (Q4). The monotonic trends with increasing LE with decreasing Gini index in the tracts with the lowest income generally held for all levels of rural-urban status, but the reverse trends (increasing LE with increasing Gini index) in the highest income tracts varied by rural-urban status. The complex associations between Gini index and LE are further illustrated in Fig. 2. This figure shows the partial correlation coefficients between LE and Gini index in each group of census tracts, cross-classified by rural-urban status quintile and median household income quintile, as well as average LE in each class of census tracts. For the four lowest quintiles of income (Q1-Q4) in the four most rural quintiles of census tracts, the associations between LE and Gini index were significant and negative (Spearman's <unk> between -0.198 and -0.041, p between <unk> 0.001 and 0.021). In contrast, the associations between LE and Gini index were significant and positive for the census tracts in the highest income quintiles, regardless of rural-urban status. For the most urban census tracts, those in the lowest quintiles of income (Q1-Q4) there were no significant associations between LE and Gini index. Lastly, multivariable models were used to assess the associations between LE and rural-urban status and median household income, overall and by quintile of Gini index (Table 3). Overall, each one-quintile increase in population density was associated with an average decrease in life expectancy of 0.13 years (95% CI 0.11, 0.15). The association between population density and LE remained significant across all quintiles of Gini index, yet the magnitude of the association was significantly stronger in areas of low inequality than in high inequality (p value for trend 0.022). For median household income quintile, each one-quintile increase in income was associated with an average 1. 11-years increase in life expectancy and did not vary significantly by Gini index. The interaction terms for quintiles of rural-urban status and median household income were significant overall and for the areas with the lowest Gini index (Q1-Q3), but not significant in the census tracts with higher levels of the Gini index (Q4 and Q5). With interaction term -0.17 (-0.21, -0.13) 1.06 (1.02, 1.10) 0.015 (0.004, 0.027) --- Q1 (lowest Gini) Main effects -0.13 (-0.17, -0.09) --- Q4 Main effects -0.12 (-0.16, -0.07) 1.12 (1.07, 1. With interaction term -0.13 (-0.21, -0.05) 1.11 (1.02, 1.20) 0.005 (-0.020, 0.030) --- Q5 (highest Gini) Main effects -0.07 (-0.12, -0.03) 1.00 (0.94, 1.06) With interaction term -0.01 (-0.09, 0.08) 1.07 (0.97, 1.17) -0.023 (-0.049, 0.002) --- Discussion In this study, the association between LE and Gini index in most US census tracts was negative, supporting the vast literature concluding that lower income inequality is associated with better population health [1][2][3][4][5][6][7]. However, those associations were not present in more urban areas. Furthermore, in the census tracts with the highest income, the association was reversed: higher LE was associated with higher Gini indices. These findings are partially antithetical to the "income inequality thesis", which states that increasing wealth is associated with improved population health, but only to a certain level of wealth [14]. Once this threshold of wealth is reached, reducing income inequality is the most important driver of improving population health [15]. Study findings support the tenet of the income inequality thesis proposing that increasing wealth is associated with better population health, as measured through LE. Increasing median household income was associated with increased tract-level LE. However, the findings contradict the part of the income inequality thesis that for sufficiently high levels of wealth, reducing income inequality is associated with increased LE. For the wealthiest census tracts, increasing LE was associated with increased income inequality, consistent with the Swiss study described earlier [16]. However, more research is needed to determine if these unexpected associations are observed for other population health outcomes [29,30]. The findings for rural-urban status and LE were more complex. The association between income inequality and LE was strongest in areas outside the most rural (Q1) and most urban (Q5) census tracts. The reasons for this are not well understood. One explanation for this finding is that areas with greater poverty could be more vulnerable to the deleterious effects of income inequality on health, especially when those areas lie in the intermediate areas of rural-urban status. Broadly speaking, while not specific to urban areas, over the past century, urban areas are more likely to have adequate housing, access to primary health care, sanitation, and resources [31], which may temper the associations between LE and inequality, even in lower income areas. Likewise, there are potential health benefits to living in highly rural and remote areas, such as the lower cost of living, access to green space, pace of life, improved environmental factors (e.g., pollution) [32][33][34]. However, like the benefits of urban living, these attributes are not unique to the most rural and remote areas. Therefore, the reasons for these highly nuanced differences in the strength and direction of association between LE and income inequality jointly by wealth and rural-urban status remain unclear and merit further research. The empirical findings of the current study should be interpreted in the context of several limitations. First, estimating LE on a small geographic scale such as census tract is subject to systematic errors [35]. Second, the partial correlations only adjusted for two factors, percent Black population and percent with a bachelor's degree or higher, and therefore the observed associations are subject to residual confounding. This analysis only considered moderation of the association between income inequality and LE by two factors, income and rural-urban status. Factors, such as race/ethnicity, education, and built environment, likely moderate these associations. The analysis also did not consider potential regional differences in the association between income inequality and LE [36]. Only one measure of each main predictor variable was used, largely due to limited variables available at the census tract level. Patterns of association and moderation may vary based on which measure of health was used [37]. This study used Gini index as the main measure of income inequality [38]. As with all measures of income inequality, the Gini index is only somewhat sensitive to income inequality occurring in the middle of the income distribution [39]. Future studies could determine if the observed associations are sensitive to the type of income inequality measure used, such as the Atkinson Index, which is less sensitive to differences in the middle of the distribution. Also, the analysis considered only one measure of rural-urban status-population density. As there is no universally accepted and utilized measure of rural-urban status, it is possible that the observed associations would also change if a different measure of rural-urban status, such as distance to the nearest metropolitan area or percent urban population, were used [40]. Lastly, for analysis and interpretation, continuous measures-area-level income, income inequality, and population density-were categorized into quintiles. Choosing other cut-points (i.e., quartiles) may result in different patterns of associations. Despite these limitations, this study provides empirical evidence that the widely established principle that areas with lower income inequality generally experience better population health may not extend to all areas and, in fact, may be reversed among high-income populations. The rationale for these findings is unclear and requires further research. While this analysis found that for most census tracts, the established associations held, the variation by income suggests that any efforts to improve population health through reducing income inequality must be tailored to the needs of distinct populations to maximize effectiveness. Future research should focus on identifying and addressing these nuanced associations that lead to critical health inequalities that may be masked when examining such associations on a higher geographic level of aggregation [41]. --- Availability of data and materials The datasets used and/or analyzed during the current study are available from the corresponding author upon reasonable request. --- Author contributions SAC conceived the idea for the study, obtained the data, conducted the analysis, and wrote the majority of the manuscript. CCN wrote the Introduction and provided critical edits for the remainder of the manuscript. ENB developed an annotated bibliography that was used in writing the manuscript. MLG oversaw the writing process and was a major contributor in writing and critically editing the manuscript. All authors read and approved the final manuscript. --- Funding The authors did not receive funding to conduct the research presented in this article. --- Declarations Ethics approval and consent to participate Not applicable. --- Consent for publication Not applicable. --- Competing interests The authors declare that they have no competing interests. Cohen et • thorough peer review by experienced researchers in your field • rapid publication on acceptance • support for research data, including large and complex data types • gold Open Access which fosters wider collaboration and increased citations maximum visibility for your research: over 100M website views per year --- • At BMC, research is always in progress. --- Learn more biomedcentral.com/submissions Ready to submit your research Ready to submit your research? Choose BMC and benefit from:? Choose BMC and benefit from: --- Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Background A preponderance of evidence suggests that higher income inequality is associated with poorer population health, yet recent research suggests that this association may vary based on other social determinants, such as socioeconomic status (SES) and other geographic factors, such as rural-urban status. The objective of this empirical study was to assess the potential for SES and rural-urban status to moderate the association between income inequality and life expectancy (LE) at the census-tract level. Methods Census-tract LE values for 2010-2015 were abstracted from the US Small-area Life Expectancy Estimates Project and linked by census tract to Gini index, a summary measure of income inequality, median household income, and population density for all US census tracts with non-zero populations (n = 66,857). Partial correlation and multivariable linear regression modeling was used to examine the association between Gini index and LE using stratification by median household income and interaction terms to assess statistical significance.In the four lowest quintiles of income in the four most rural quintiles of census tracts, the associations between LE and Gini index were significant and negative (p between < 0.001 and 0.021). In contrast, the associations between LE and Gini index were significant and positive for the census tracts in the highest income quintiles, regardless of rural-urban status.The magnitude and direction of the association between income inequality and population health depend upon area-level income and, to a lesser extent, on rural-urban status. The rationale behind these unexpected findings remains unclear. Further research is needed to understand the mechanisms driving these patterns.

Introduction Deprivation and other area-based socioeconomic indices are used extensively in public health in a number of countries [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18] including Canada. [19][20][21][22][23] Despite their widespread use, they have seldom been explicitly validated, except in a few mainly British studies. 7,[24][25][26][27] Validating a deprivation index means verifying whether it adequately reflects the reality being measured. Validation is a complex exercise because the index must respond to a number of criteria and have certain properties that are useful in its field of application (in this case, public health). The purpose of this study is to subject Quebec's material and social deprivation index 23 to these validation criteria and properties. The Quebec index was developed at the end of the 1990s and has since been used in Quebec and Canada in various contexts. In this paper, we first describe the index and then present the validation criteria and properties, first with reference to the international literature, then to the Quebec index. Finally, we discuss the nature of the Quebec index and make proposals for additional validations. --- Quebec material and social deprivation index The Quebec deprivation index was designed to illustrate social inequalities in health and in the use of health services. Its objectives are primarily exploratory and descriptive in nature. It applies to the entire Quebec population, by place of residence. The design and creation of the index is based on Peter Townsend's ideas on deprivation and the international literature on social determinants of health. The index has two dimensions, material deprivation and social deprivation. The index is also geographical: it is based on the smallest standardized Canadian census unit, composed of one or more blocks of neighbouring houses with a population of 400 to 700 persons. This unit is the enumeration area (EA) for the 1991 and 1996 censuses and the dissemination area (DA) for the 2001 and 2006 censuses. 28 The Quebec deprivation index is made up of six socioeconomic indicators by EA or DA: the proportion of people 15 years and older with no high school diploma or certificate; the employment:population ratio of people aged 15 years and older; the average income of people aged 15 years and older; the proportion of people aged 15 years and older living alone; the proportion of people aged 15 years and older who are either separated, divorced or widowed; and the proportion of singleparent families. All but the last are adjusted according to the age and sex of the Quebec population. We extracted two components from these indicators using principal component analysis (PCA): the material component, which is associated with employment, education and income, and the social component, which is associated with marital status, living alone and single-parent families. For each component, the PCA produces a factor score by EA or DA, indicating its relative level of deprivation. Depending on this score, Quebec EAs or DAs are grouped into quintiles (population groups of 20%) from the most privileged (quintile 1, Q1) to the least (quintile 5, Q5). Thus, it is possible to follow variations in deprivation for each dimension separately (Q1 to Q5) and for both dimensions simultaneously (Q1Q1 to Q5Q5). --- The validation of deprivation indices Validation of deprivation indices, including the Quebec material and social deprivation index, is based on proposals in the literature 7,[24][25][26][27] and, more specifically, on work focused on the surveillance and measurement of deprivation and social inequalities in health. 24 After reviewing the deprivation indices used in the United Kingdom, Carr-Hill and Chalmers-Dixon 24 suggested using three criteria to evaluate this type of index (validity, reliability and responsiveness) and also suggested considering other properties useful for health policies. While recognizing that the scientific community identify other criteria and properties, 29 we used the definition proposed by Carr-Hill and Chalmers-Dixon. 24 We used three approaches to measure the validity of the deprivation indices. These three approaches are usually referred to as content validity, criterion validity and construct validity. --- Content validity Content validity refers to the agreement between the general concept of deprivation, its main dimensions and the indicators selected to illustrate them: 24 Are the dimensions and indicators appropriate? Do they represent all the facets of deprivation that the index is attempting to reflect? The conceptual foundations of the Quebec material and social deprivation index are mainly based on the proposals set forth by Peter Townsend, 30 for whom deprivation is a ''state of observable and demonstrable disadvantage, relative to the local community or the wider society or nation to which an individual, family or group belongs.'' The author distinguished between two forms of deprivation: material and social. The first, material deprivation, refers to the lack of the normal goods and amenities of modern living in various areas, such as food, housing, the environ-ment and work. The second, social deprivation, which according to Townsend, is more difficult to define, refers to the fragility of social ties. This fragility may occur within the family unit or it may extend to close relationships, friends, confidants, neighbours and others who provide emotional and material support (social support). It can also reflect the difficulties associated with integration and participation in social relationships and other common activities within the local community, such as recreational or educational activities. This brief definition of deprivation forms the basis for a number of deprivation indices. 7,9,20,25,26,[31][32][33] The authors of these indices highlighted the relative character of deprivation, its subjective and objective aspects, and its material and social dimensions. The analysis of deprivation can, however, involve more than two dimensions or different fields 13 and overlap with other concepts, such as poverty, disadvantage, socio-economic status or position, 1,6,10,15,16,26 marginalization, 22 or social isolation or fragmentation. 34,35 In all cases, the concepts beneath these area-based deprivation indices and other socio-economic indicators remain underdeveloped. [25][26][27] The area-based scale is, however, a fundamental element of deprivation indicators that distinguishes them from indicators related to individuals, even though they often serve as a substitute or proxy for each other and are sometimes compared. 1,5,11,16,26,27 An area-based indicator reflects a specific reality 6,13,36 that varies according to the scale considered. 36,37 --- Criterion validity Criterion validity is used to verify whether the variations in a deprivation index correlate highly with those of an external measurement of deprivation. 24 Criterion validity is not used extensively because it is commonly accepted that there is no gold or reference standard for deprivation. Nevertheless, certain practices are similar. For example, some authors have compared the area-based variations of different deprivation indices with one another 25,27,37 or with those of measurements involving individuals, even though they are different realities. 1,16,26 Moreover, certain authors have compared the area-based variations of a new index to indices already in use, such as Townsend's. 6,7,15,16 Because there is no standard or reference measure for deprivation, we preferred to discuss the Quebec index in terms of convergence validity, as will be discussed later. --- Construct validity Construct validity of a deprivation index in the health sector can take on a number of forms. 24,29 Above all, it aims to determine whether the construction is consistent with the concept of deprivation. Construct validity is also expressed through consistent relationships between the index and other measurements related to the concept of deprivation, on the one hand, and various health measures and the use of health services, on the other. These forms of validity will be more specifically addressed through convergence validity and predictive validity, respectively. To operationalize his vision of deprivation, Townsend reviewed various indicators used in Great Britain, some from administrative bases and others from health surveys, 30 and proposed a material deprivation index combining four indicators. 24 Other authors added a social dimension by creating a separate social deprivation index, 26 or social isolation index, 34 combining a number of indicators, all from censuses. To construct the Quebec index, we took into consideration these indicators and also conducted a literature review on the social environment and social inequalities in health. 34,[38][39][40][41] We then selected our indicators on the basis of theoretical and practical criteria: affinity with one of the two forms of deprivation, known link with health, availability at a fine geographical scale in the census 28 and a limited number of indicators in the composition of the index (parsimony) to simplify comprehension. We selected six indicators through this process. $ 13 Vol 34, No 1, February 2014 -Chronic Diseases and Injuries in Canada The integration of these indicators in the form of an index was not the subject of any explicit hypothesis. The intention was to let the ''natural'' area-based variations of the indicators express themselves without a priori grouping. For this, we used principal component analysis (PCA), an exploratory synthesis method widely used in the creation of geographically based indices, 3,6,7,13,16,18,20,22,32,33 while recognizing the relevance of using groups of experts 8,19 or equally weighted sums 5,25,27 for the integration of indicators related to certain indices. The PCA revealed the presence of two components. In the 2006 census, the first component reflected the variations in education, employment and personal income 42 (see Table 1). The second component reflected the variations in the proportion of individuals who were living alone, separated, divorced, widowed or living in single-parent families. These results are similar to Townsend's proposals concerning the two dimensions (material and social) of deprivation. However, they differ in terms of education, which according to Townsend, is associated with social deprivation. Moreover, these two components do not appear to be very explicit with respect to the forms of deprivation. Work connecting the two dimensions of the Quebec index with other indicators from censuses by EA or DA makes it possible to clarify these dimensions. 43,44 For example, social deprivation is closely associated with residential mobility (frequent moves) and the proportion of renters, two indicators used in the construction of social fragmentation and isolation indices. 34,35 The fact remains that the census is a limited source of data for reporting on the fragility of social networks. --- Convergence validity It is therefore necessary to compare the index to external measures (not from censuses) that reflect deprivation and its various dimensions. We conducted three exercises of this kind. We first compared the spatial variations in the deprivation index to those in the proportion of children living with families receiving last-resort financial assistance from the Government of Quebec (see Table 2). Such assistance is given to families whose liquid assets (cash, etc.) are less than a particular amount that corresponds to the size and needs of the family. It is the only source of income the family has to meet its basic needs (e.g. housing and food). Two-thirds of the families receiving this assistance are single-parent families. 45 Therefore, we expected material and social deprivation to increase with the proportion of children living with families receiving this assis-tance, which is the case according to the statistics provided by Quebec's Department of Employment and Social Solidarity. 45 The other two exercises made it possible to better define the social dimension of the deprivation index. One linked the variations in the Quebec index with those observed in an in-depth study of three areas in the Quebec City region. [46][47][48] Two of the areas had different health reports. The material deprivation index was similar in these areas, whereas the social deprivation index differed significantly. A telephone survey of 600 respondents in each area collected data on health and perceptions of the local environment. The use of a social cohesion index, 49 addressing the appeal of the local environment and sense of neighbourhood and community, produced coherent results with those obtained from the social deprivation indices. Where social deprivation was high, social cohesion was low, and vice versa. Qualitative interviews with residents revealed that being born in the area and having family members in the area were cohesive factors. The last exercise was based on an analysis of a number of cycles of the Canadian Community Health Survey 50 and explored the links between certain social support measures at the individual level 51 and the social deprivation index in urban Quebec. 52 The exercise revealed that an increase in social deprivation went hand in hand with a decrease in three social support measures, that is, affection, positive social interactions, and emotional or informational support. These associations are independent from the age, gender, lifestyle, education and household income of the survey respondents. In summary, not only do the indicators used in the construction of the social dimension of the index reflect family structure and marital status, the dimension also captures a broader reality. At the individual level, this reflects the fragility of social support for single-parent families and those who are living alone or who are separated, widowed or divorced. At the local scale, it reflects residential instability (very frequent moves 34,35 ), which does --- Predictive validity As we have seen, the primary objective of a deprivation index is to identify social inequalities in health and, therefore, the associations between deprivation and health. 24 These associations must be plausible, corroborate observations made in the literature, or be supported by credible explanations or hypotheses. Predictive validity is by far the most widely used approach to demonstrate the quality of a deprivation index. 24 It is seen as ''proof'' of its performance. For example, links have been made with overall mortality, 10,12,14,27 premature mortality (0-64 years), 4,18 cause of death, 3,18 the incidence of cancer 10 (including lung cancer 14 ), long-term disability, [25][26][27] perceived health, 1,37 smoking and nutrition, 5 low birth-weight, immunization status and lead poisoning among children, 11,14 sexually transmitted infections, tuberculosis and violence, 54 myocardial infarction, 7 hospitalization, 14,27 and use of medical 8 and psychiatric services. 16 Moreover, the strength of the relationship between deprivation and health varies according to the size of the basic spatial unit of the index. The smaller the spatial unit, the stronger the relationship. 1,10,11,26,54 The Quebec deprivation index accounts for various health and social situations. It is linked to global health indicators, namely, life expectancy and health expectancy at birth and different ages 23,44,55,56 and mortality, including overall mortality, mortality by medical cause (e.g. cancer, circulatory disease, trauma and stroke), mortality related to lifestyle (e.g. smoking), premature death (less than 75 years), death among young people (18 years or less) and survival. 23,[55][56][57][58][59][60][61][62][63][64][65][66][67][68][69] For example, an increase in the rate of premature deaths was observed both in the early 1990s and the mid-2000s as a function of material and social deprivation (Figure 1). The same is true for other indicators, such as disability, 56,64,[70][71][72] the incidence or prevalence of diabetes and high blood pressure, [72][73][74] self-reported health, 70 and protective and risk factors for health: flu vaccination, premature birth or low birth weight, smoking and exposure to secondhand smoke, obesity, food insecurity and physical inactivity. 23,61,70,[75][76][77][78] Social issues, such as teenage pregnancy and cases of abuse, neglect and behavioural problems among young people, are also associated with deprivation. 23,44,61 Such relationships were also observed in use of health services. An increase in visits to general practitioners was noted with increased deprivation, but an opposing trend was sometimes found for certain medical specialties. 44,61 This opposing trend was also true for certain free services available for young people aged under 18 years (eye exams) and under 10 years (dental appointments) (Figure 2). However, the use of local community service centres (CLSCs), as well as hospitalization, day surgery and stays in long-term care facilities increased with material and social deprivation. 44,61,70,79 A recent example is the rate of hospitalization following influenza A(H1N1) infection (Figure 3). In summary, the Quebec deprivation index accounts for significant inequalities in health, even though their magnitude may vary depending on the theme under consideration. The two forms of deprivation (material and social) usually act independently. 23,44,[56][57][58][59][60][61][63][64][65][66][67][68][69][71][72][73][74][75][76]78,79 --- Reliability The reliability of a measurement tool is defined as its ability to produce the same result under the same circumstances. 24 For deprivation indices, this ability can be expressed through strong correlations between the indicators that form the index. These correlations are often tested using Cronbach's alpha. Some authors refer to an index's internal consistency. 6,7,26 This internal consistency, however, is not relevant when the index has more than one dimension. 24 The reliability of a deprivation index can also be expressed through correlation structure stability in time and space. The goal is to verify whether the correlation structure remains, regardless of the period and environment being considered. The reliability of the Quebec deprivation index can be seen from the perspective of internal coherence for each dimension of deprivation. As seen in Table 1, close correlations exist between the indicators that make up each of the two dimensions (material and social) of the index. This fundamental structure of the index can be seen throughout Quebec and Canada 42,68 at various levels: regional, census metropolitan areas, cities of varying sizes and rural environments. It is also present for each census year between 1991 and 2006. Although the correlations between the indicators may vary slightly according to the location and period considered, the two-dimensional structure of the Quebec index is maintained. 42 This fundamental structure seems to be permanent, an essential quality for monitoring the social inequalities in health in time and space. --- Responsiveness Responsiveness reflects the ability of a measurement tool to detect differences or changes according to the location, time and individual characteristics. 24 Variations in the deprivation index are observable at the national, regional and local levels, through the use of maps, for example. 2,7,8,26,37 They are also observable in relation to various health characteristics. The relationships vary according to the age and gender of the population, 3,4,18,27 with adults (aged 25-64 years) usually showing the highest inequalities in health. The inequalities change over the years (reducing or increasing) or with the area 3,4,11,16 and fluctuate according to the health issue under study (e.g. cause of death). 10,16,27 The Quebec deprivation index was used to create an interactive atlas 44,80 that shows wide variations in deprivation at the provincial level and at a smaller level, in both urban and rural environments. These variations in the Quebec index are also associated with inequalities in health that relate to gender and age, with adults having the highest mortality ratios between groups at the extreme ends of material and social deprivation (Figure 4). Moreover, as is the case elsewhere, 18,81-84 the Quebec index has identified an increase in relative health differences in Quebec. According to the data presented (Figure 1), the premature mortality ratio between groups at the extreme ends of deprivation increased from 1.8 in 1989-1993 to 2.4 in 2004-2008. The Quebec index identified health inequalities of varying magnitude according to geographical area and fluctuating over time. 62,64,66 Thus, inequalities are growing throughout Quebec, except in the Montreal area, where they are actually bigger than in the rest of the province. Such health differences have also been demonstrated elsewhere in Canada. 63,67,68 Other properties In the context of the development of public health policies or programs, deprivation indices must respond to requirements beyond those that are purely technical or statistical. 24 This is the case for the comprehensibility of the index for an audience made up of decision makers and stakeholders in the field. The index must be easy to understand, appeal to common sense and be conducive to reasonable, unambiguous explanations. Thus, the contribution of the indicators to the index must be precise, clear and, if possible, quantified. The index must also be objective (cannot be manipulated) and be applicable to every part of the area being considered, at the national, regional and local levels. Finally, the index must respond to practical requirements. It must be possible to update it regularly, using the same method, and be manageable in terms of time and cost; it should also be possible to introduce it into health databases. As we have seen, the Quebec deprivation index remains a simple measure, made up of two components and six indicators that are well known as being connected to health. Its structure is clear, and the weighting of the indicators in the index reflects their correlation with the components (Table 1). Its use demonstrates its comprehensibility for an audience made up of stakeholders and decision makers in the health and social service sectors in Quebec. Local variations in the index corroborated the perception of CLSC stakeholders, 79,85 and, at a provincial level, --- Material and social deprivation Source: A(H1N1) surveillance record, MED-E <unk>CHO hospitalization records, Ministe <unk>re de la sante <unk>et des services sociaux du Que <unk>bec. --- Note: The relative risk is adjusted by age, gender, geographical area and other form of deprivation. a From Q1, the most privileged quintile, to Q5, the least privileged quintile. $ 17 Vol 34, No 1, February 2014 -Chronic Diseases and Injuries in Canada these variations were used to develop departmental policies 61 and to allocate health resources among regions. 86 A recent compilation indicates that most of Quebec's regional health and social services agencies use the deprivation index to identify variations in their areas and the connections with various health and social issues. 87 Although groups of experts were not involved in the design or initial construction of the deprivation index, many health experts (stakeholders and managers) at all geographical levels have since commented on, used and adapted the index to their needs and work contexts, contributing to its validation and evolution. For example, a local version of the index and an interpretation grid of the inequalities in the use of services were developed jointly with local CLSC stakeholders. 79,85 The grid compares the variations in the index and the knowledge of stakeholders regarding their organization directions and practices (e.g. target clientele, service access criteria), resources available locally (e.g. medical clinics, self-help groups and associations) and hard-to-reach populations (e.g. the homeless or individuals with mental health issues). Finally, the relevance of the Quebec index depends on its availability over time and space. We have seen that the index exists for 1991, 1996, 2001 and 2006, and that it covers all of Quebec (and Canada) in different versions: national, regional and local. There are supporting products (e.g. interactive maps, population tables, index assignment programs), which are all free and available online. 80,88 Tables and figures illustrating the health inequalities in Quebec using the deprivation index are regularly produced and posted online. 89 --- Conclusion Despite the widespread use of deprivation indices, there have been few formal validation exercises. On the basis of the validation criteria proposed by Carr-Hill and Chalmers-Dixon, 24 it can be concluded that the Quebec material and social deprivation index responds favourably to various requirements for validity, reliability, responsiveness and use in public health. However, there are limitations related to the geographical nature of the index. The index characterizes the socio-economic attributes of all residents of small areas. Although it is often used as a substitute for measurements related to individuals, the index is a measurement linked to an area. Studies, some of which are from Quebec and Canada, 56,64,67,90 show that the magnitude of health inequalities is underestimated through geographical measurement, espe-cially in small cities and rural environments. They also reveal that health inequalities are associated with both types of measurements (those related to area and those related to individuals), independently, which signifies that they result from both geographical and individual realities. 56,64,67,[91][92][93][94][95][96][97] A better understanding of these geographical realities is therefore necessary to identify all the content and construct elements associated with a deprivation index. To achieve this, a research strategy at the local level combining theories, concepts, methods and indicators is necessary. [98][99][100][101] Reference frameworks on ''contextual'' factors associated with health must be used. 53,98,102,103 The social dimension of the index would particularly benefit from being associated with concepts and measurements of social cohesion and capital as well as their components (e.g. values, social support, informal social control and community participation). The material dimension would benefit from being associated with various fields, such as the physical environment (e.g. water and air), the built environment (e.g. housing and access to services), and public (e.g. schools, green space and public transportation) and private (e.g. food stores) infrastructure. This roadmap should be followed for future validation exercises of the Quebec index. Finally, it should be noted that this index was designed to illustrate the existence of social health inequalities and that its purposes are exploratory and descriptive. The index is not an explanatory framework for these inequalities. For example, it does not consider dimensions related to health, such as immigration or Aboriginal status, even though these dimensions can be accounted for. 63,66 Rather, the Quebec index constitutes more of a marker of social and health inequalities and, as a result, is a relevant starting point toward more in-depth studies and increased understanding of these inequalities.

Introduction: Despite the widespread use of deprivation indices in public health, they are rarely explicitly or extensively validated, owing to the complex nature of the exercise. Methods: Based on the proposals of British researchers, we sought to validate Quebec's material and social deprivation index using criteria of validity (content, criterion and construct validity), reliability and responsiveness, as well as other properties relevant to public health (comprehensibility, objectivity and practicality).We reviewed the international literature on deprivation indices, as well as publications and uses of the Quebec index, to which we added factual data.Based on the review, it appears that the Quebec index responds favourably to the proposed validation criteria and properties. However, additional validations are required to better identify the contextual factors associated with the index.

Introduction Analysing the relevance of family background characteristics to individuals' levels of income is important to understand how individuals' income is determined by factors beyond their responsibility and, consequently, the extent to which a society is able to guarantee equal opportunities for all. This article aims to analyse the role family background characteristics play in shaping individuals' income and its subsequent effect on inequalities in Spanish regions. For this purpose, we get estimates of inequality for the Spanish regions and compute the importance of family background characteristics to the observed levels of income Additionally, we analyse the differences in inequality levels between the Spanish regions trying to disentangle the main drivers of the levels of inequality. Ensuring equal opportunities in a society would mean that all individuals are able to achieve a certain level of income regardless of the level of income, wealth or education attained by their progenitors or any other circumstances beyond their own control. The analysis of equal opportunities in economics has been popularised by Roemer (1993Roemer (, 1998)), following the work of several authors (Arneson, 1989;Cohen, 1989;Dworkin, 1981aDworkin,, 1981b;;Nozick, 1974;Rawls, 1971;Sen, 1979), whose contributions incorporate the idea of equal opportunities from different perspectives. Likewise, educational inequalities and income inequalities are closely related (Checchi and van de Werfhorst, 2018;Palmisano et al., 2022;Solga, 2014) For instance, Solga (2014) found that education could be a great equalizer for income inequality when accompanied with redistributive policies. Furthermore, Checchi and van de Werfhorst (2018) find that inequality in the quality of education affects income inequality, indicating that their results are to some extent consistent with human capital and functionalist theories, which argue that education is rewarded because of the skills needed in the labor market, so inequality in the quality of education is expected to affect income inequality more. Palmisano et al. (2022) show that inequality of opportunity in education is strongly correlated with inequality of opportunity in income. The research available in this regard for Spain shows a strong relationship between the income and social class of the parent and the school performance of individuals (Bernardi and Cebolla, 2014;Peraita and Pastor, 2000). Although there are some works that analyse income inequalities and mobility in Spain (Ayala and Jurado, 2011;Budr<unk>a and Moro-Egido, 2008;Cabrera et al., 2021;Cantó, 2000;Pijoan-Mas and Sánchez-Marcos, 2010;Suárez <unk>lvarez and López Menéndez, 2020) most of them study inequality before the year 2000. In this sense, it is important to perform an analysis with a more recent period in order to have a more adequate picture of the current reality, which will allow policy makers to carry out public policies that are more in line with current needs. Likewise, the role of parental background in education has previously been analysed in several investigations in equality of opportunity. It has been proved that parental education and parental occupation are two of the most relevant circumstances in shaping individual outcome in income (Belhaj Hassine, 2012;Bourguignon et al., 2007;Brunori et al., 2018;Pestieau, 1989;Singh, 2012;Suárez <unk>lvarez andLópez Menéndez, 2018a, 2018b). Moreover, focusing on the Spanish case (Cabrera et al., 2021;Suárez <unk>lvarez and López Menéndez, 2018a), found that parental education significantly affects individuals' income. Our objective is to analyse how different educational and family background variables at regional level are related to the observed levels of inequality. Inequalities and educational achievements can experience great variations between regions. In the Spanish case this fact could be especially relevant and interesting due to several reasons: (1) there are significant income inequalities between regions (2) the distribution of income growth among the population differs between regions and (3) the education systems are different since the competencies in education are shared by the Spanish regions and the central government, resulting in differences in the educational systems across regions. To this end, the analyses carried out in this article try to answer two main questions: (1) To what extent are family background characteristics important in shaping inequalities? And (2) Which are the main drivers of inequalities across regions? We will consider inequalities in income, and for this purpose, we will use the disposable income of individuals as an outcome variable to measure both income inequality and inequality of opportunity. Taking into account that Spanish regions are very different from one another, one could expect to find different answers to these questions depending on the region. In fact, Suárez <unk>lvarez and López Menéndez (2020) showed the importance of the regional dimension for the analysis of inequalities and opportunities. Therefore, we perform our analyses at regional level, considering the seventeen Spanish NUTS-II units (Autonomous Communities, CCAA). Likewise, since the analysis is at the regional level, it is important to note that there was no devolution of government responsibilities during the period under analysis. The rest of the article is structured as follows. Section two presents the database and the variables used, also providing some descriptive statistics. Section three shows the estimates of inequality and inequality of opportunity for the Spanish regions and the contribution of different circumstances to the level of inequality of opportunity. Section four provides evidence on the main determinants of the different levels of inequality observed across regions, with a focus on educational variables. Finally, Section five summarises the conclusions of the article. --- Database description The analyses performed in this article rely on the European Survey of Income and Living Conditions (EU-SILC), a survey that contains datasets at individual and household level. Furthermore, data are classified by regionsin particular, EU NUTS-II regions, which correspond to the Spanish Autonomous Communities (CCAA). They are first level political and administrative divisions and each one has its own organic laws known as Statutes of Autonomy, which determine the powers they have. They are also divided into provinces (EU NUTS-III regions). Specifically, in this article we use the EU-SILC surveys conducted in 2011 and 2019 which provide income variables referred to years 2010 and 2018. The use of these two surveys is particularly relevant since they include an ad-hoc module which incorporates variables regarding individuals' family background. Our analysis is focused on the seventeen Spanish Autonomous Communities and the sample is restricted to individuals aged twenty five to fifty nine years because parental background variables are only provided for individuals within this range of age. The variables we use refer to individuals' circumstances, i.e. factors that are beyond individuals' responsibility. For this purpose, we use variables regarding individuals' family background when they were children. The first is Parental Education, which is constructed using the maximum level of education of both progenitors classified into three categories: low if both progenitors have at maximum a degree of compulsory education, medium, if at least one progenitor has completed secondary education and high, when at least one progenitor has completed tertiary education. Similarly, we use the variable Parental Occupation, which refers to the occupation of individuals' progenitor when they were children. This variable is also divided into three categories, according to the ISCO-08 jobs classification. More specifically, these three categories are low skilled when both parents work in elementary occupations (groups eight and nine of ISCO-88 classification), medium skilled when at least one parent works in an occupation comprised in groups five to seven of the classification and high skilled if at least one progenitor works in a high skilled occupation (groups one to four in ISCO-08) 1. We also include the variable Childhood Financial Difficulties, which is a perception-based measure that considers respondent's feeling about the financial situation of the family when the respondent is around fourteen years old. It is divided into four categories: Bad/very bad, Moderately bad, Moderately good and Good/very good. The variables used as circumstances correspond to a period prior to the one for which the income of individuals is analysed. These variables correspond to the childhood period of the individuals, since the idea is that they provide an approximation of their social and family environment and that they are circumstancesthat is, that they do not depend on the responsibility of the individuals and, clearly, individuals cannot occur to them due to situations that occurred during their childhood. Finally, we include a variable referred to income, which will be used to measure inequalitywe use the equivalised disposable income of households, which is equivalised by considering the structure of each household. Specifically, this variable is constructed according to the equivalence scale used by EU-SILC, which is the Organisation for Economic Cooperation and Development (OECD)-modified scale: e 1 0:5 N 14 1 0:3N 13 where N 14 is the number of household members who are over fourteen years and N 13 the number of members who are thirteen or younger. The equivalised disposable income is considered a good proxy of the available income individuals have. In addition, this variable has been used in other studies with EU-SILC data referred to education and inequalitiessuch as Brzezinski, 2015;Marrero and Rodr<unk>guez, 2012;Palomino et al., 2016;Suárez <unk>lvarez andLópez Menéndez, 2018b, 2018a. Table 1 provides a summary of the descriptive statistics at the regional level and for the two years analysed, 2011 and 2019. More specifically, it shows the share of the population within each category of the above-mentioned variables. Additionally, Table 1 shows the sample size for each region and year. It can be noticed that in some regions the sample size is not very large, the survey used (EU-SILC) provides weights to make the results representative. Nevertheless, caution is advised when interpreting the results. --- Estimating inequality of opportunity This Section is devoted to estimate income inequality and inequality of opportunity (IOp) and to compute the contribution of the variables used to circumstances to IOp with the aim of showing at what extent family background characteristics and other circumstances are affecting individuals' performance in terms of income. For this purpose, in addition to estimating income inequality, we compute inequality of opportunity (IOp henceforth) for the Spanish regions and then we analyse the contribution of the variables used as circumstances to this indicator. The basis of the study of IOp can be found in Roemer (1993) and Yitzhaki and Schechtman (2013); later on Roemer (1998) formalise this concept and distinguish between circumstances, understood as factors beyond individuals' control, and efforts, that can be attributed to individuals' performance and commitment. In this context, the analysis of IOp tries to compute the part of inequality due to the first kind of factors, in which parental background characteristics are included. To know the effect that parental background characteristics have in inequality and more specifically in inequality of opportunity, we estimate the indices of both sorts of inequalities for each of the seventeen Spanish regions analysed before computing the contribution of these characteristics. Firstly, to estimate IOp, in addition to previous variables referring to individuals' family background, we include two more variables regarding individuals' personal circumstances in order to obtain reliable estimates and reduce the omitted variables bias. These two variables are: Gender, with two categories, female and male and Country of birth with two categories Foreignborn persons and Spanish-born. To estimate IOp we use the ex ante parametric method (Ferreira and Gignoux, 2011), although there are many other alternative methodologies (see Ferreira and Peragine, 2015;Ramos and Van de Gaer, 2016;Roemer and Trannoy, 2016). We have decided to use the ex-ante parametric method since this is the most widely used procedure in the IOp literature and thus allows comparability with other studies. Furthermore, this method has the additional advantage that it does not require including information referred to the level of effort. In order to compute IOp we divide individuals into types T, being each type formed by individuals that share the same categories of each circumstantial variable. In our case we have five variables Parental Education, Parental Occupation, Childhood Financial Difficulties, Gender and Country of birth with three, three, four, two and two categories each, thus leading to a total of 144 different types of individuals. These are the variables considered to be beyond individuals' responsibility and therefore treated as circumstances. Then, we estimate individual incomes following the expression: lny i C i <unk> u i, where C i denotes the different variables used as circumstances and <unk> represents the effect these circumstances have on income. Once the equation is estimated we get the fitted values: <unk>i exp C i <unk> which are a counterfactual distribution of income that depends only on the circumstances. (Tables A1 andA2 in the Appendix section show the results of the regressions). The inequality of opportunity indices can then be obtained in absolute (IO pA ) and relative (IO pR ) terms quite straightforward: IOp A I <unk>i and IO pR I <unk>i I y. Table 2 shows the results of both income inequality and IOp in the equivalised disposable income through two indices, the Gini and the GE (0). We use the Gini index due to its easy interpretation and because it is widely used to measure inequality. Generalised Entropy GE (0) is also computed since this is the only measure with a path-independent decomposition (Foster and Shneyerov, 2000) using the arithmetic mean as reference and can be easily used to compute the contribution of the circumstances to IOp. According to the obtained results, North-western regions (Galicia, Asturias, Cantabria and Basque Country) and also two central regions (Madrid and La Rioja) experienced an increase in both income inequality and IOp whilst for the remaining regions we observe a decrease in both inequalities, with the exception of Castile and Leon and Castilla-La Mancha for which we observe an increase. Likewise, in most regions we can see an increase in relative IOp. That is to say, the share of inequality of opportunity in overall inequality increases; and consequently, factors that are beyond individuals' responsibility become more important in shaping individuals <unk>well-being and performance. To answer our question 'to what extent are family background characteristics important in shaping inequalities?', once we get the estimates of IOp we estimate the contribution of the circumstances and pay attention to the amount of IOp due to Parental Education and Parental Occupation. To this end, we decompose IOp using the Shapley value procedure (Shorrocks, 2013), and we compute the marginal effects each variable has in IOp in terms of the GE (0) index. Table 3 shows the average contribution of family background characteristics to IOp, these circumstances are of great relevance in shaping inequalities, the three of them account for about half or more of the total IOp. The exceptions are four regions of the northeast (Navarre, La Rioja, Aragon and Catalonia) where the relative importance of family background characteristics is lower. As it can be seen in this Section, both inequalities, income inequality and inequality of opportunity, both in levels and evolution, differ between regions. Additionally, the importance of family background characteristics to determine the levels of inequality of opportunity also varies greatly. For this reason, it is relevant to analyse the determinants of these regional differences. --- Drivers of inequalities The aim of this section is to know which are the main drivers of the observed inequalities across regions. We analyse the reasons behind the differences in inequality levels between regions and how these differences can be reduced. As a first step, we look at the association between inequality levels and several variables. We analyse the correlations between inequality variables and potential determinants of inequality, we include economic, demographic and educational variables. Table 4 shows the pairwise correlations, it can be seen that, as expected both sorts of inequality are positively and strongly correlated, as Figure 1 also illustrates. Regarding the economic variables included, income inequality is positively and significantly correlated with the unemployment rate, therefore, those regions with higher rates of unemployment also suffer from higher levels of income inequality. Nevertheless, its effect on IOp is not significant. The other economic variables included are the per capita Gross Domestic Product (GDP), the average wage, the wage of the tenth percentile and the wage of the ninetieth percentile. The GDPpc, the average wage and the wage of the 10 th percentile are negatively correlated with income inequality and positively correlated with IOp, whilst the wage of the The Role of Family Background and Education in shaping inequalities ninetieth percentiles is positively correlated with both sorts of inequalities. Nonetheless, these correlations are weak and non-significant. With regard to demographic variables, we include the share of population over sixty five years and the dependency rate, both variables are negatively correlated with the levels of inequality. Moreover, the correlation between the population shares over sixty five years and the inequality variables is quite strong and significant, as can be also seen in Figure 2. It implies that regions with aged population have less inequalities, possibly because retirement benefits an equalising effect on income. We also include the net inter-regional migration rate but its correlation with inequality is weak and not significant. Educational variables seem to have the highest levels of correlation with inequality variables. As educational variables, we include five variables at regional level. The first of them is the share of individuals with tertiary education, which is negatively correlated with both income inequality and IOp, although this correlation is not significant. Then, the educational drop-out rate 2, which is positively and significantly correlated with income inequality and the success rate of highsecondary school which is negatively and significantly correlated with income inequality. The two educational variables that show the highest level of correlation with the inequality variables are the average student/teacher ratio and the education expenditure per student 3. The students/teacher ratio correlation implies that the higher the average number of students per teacher, the higher the levels of inequality and IOp. Additionally, the correlation between the education expenditure per student and the inequality variables shows that the greater the education expenditure the lower the inequality levels, this relationship is also illustrated by Figure 3. These results show that there is a significant association between the levels of inequality and educational resource endowments, both economic (educational expenditure) and human capital (students/teacher ratio). It is observed that those regions with a greater endowment of educational resource exhibit lower levels of inequality. In addition to the pairwise correlations, we perform several regressions to see if the previously observed associations also entail causal relationships. Table 5 shows the results of the regressionsas expected not all positive and significant pairwise correlations are significant in the regression analysis. Moreover, regression results are not robust when comparing fixed effects and pooled Ordinary Least Squares (OLS). Still, the regression analysis shows that an increase in unemployment rate would cause an increase in the levels of income inequality. Additionally, the expenditure on education is significant to explain IOp, showing that an increase in education expenditure would reduce inequality of opportunity. Summarising, the analysis carried out in this Section highlights the great relevance of education in reducing inequalities, especially the expenditure on education and the number of teachers. --- Conclusions Throughout this article we show the relevance of family background characteristics in shaping individuals' income and the determinants of regional differences in levels of both income inequality and inequality of opportunity. We wonder if familiar background characteristics have an impact on individuals' disposable income. To test that, we compute inequality and IOp indices for the different regions. Our analysis reveals that family background has a great impact on individuals' income and it is a crucial source of unequal opportunities. Moreover, this analysis reveals that, for the Spanish regions, familiar background characteristicsin particular, parental education, parental occupation, and the financial situation during childhoodare of great importance in shaping individuals' performance and opportunities of achieving a certain level of income. Additionally, we analyse the main drivers of the regional differences in income inequality and IOp, finding that educational variables are highly associated to inequality levels. Moreover, the regression analysis shows that there is a causal relation between educational resource endowments and both income inequality and IOpthe analysis shows that regions that invest more in education experienced lower levels of inequality. The obtained results suggest that different regional education policies, reflected in differences in expenditure per student, are an important determinant of regional inequalities, especially of inequalities of opportunity, which are the sort of inequalities beyond individuals' responsibilities and therefore, considered unfair. Consequently, some policy implications can be drawn from this empirical evidence. The results seem to indicate that an increase in social expenditures could have an equalising effect on income. In this sense, it is important to mention the work from Qui<unk>onez (2022), who found that in the case of Latin America social spending increases are associated with reduced levels of income inequality. He distinguishes four areas of social expending (social protection, education, health, and housing and community services) and shows that each of them has different effects on income inequality. Additionally, Ellison and Fenger (2013) show that the implementation of socially equitable and effective state policies or interventions depends on a thorough prior analysis of the relationship between labour market structures, inequality, social investment, and social protection in specific contexts. This underlines the importance of specific policies that can directly and effectively address existing inequalities and specifically target the groups that need them. It is important when designing public policies to carry out analyses that clearly identify those individuals or groups susceptible to being targeted by the policies. The analysis carried out in this article has allowed us to determine which are the most vulnerable groups and which are more likely to receive lower income and be, therefore, in a position of material disadvantage. These are individuals from disadvantaged family backgrounds, with parents who have a low level of education and who work in low-skilled occupations. Therefore, it would be advisable to implement public policies targeting this vulnerable group. Additionally, given that this research shows that the greater the educational resource endowments (both human and material) the lower the levels of income inequality, it would be advisable to combine specific policies for vulnerable groups with policies aimed to increase educational resources. More specifically, redistributive policies could be combined with policies aimed at improving the quality of education and guaranteeing access to education regardless of socioeconomic factors and family background. To sum up, policies aimed at 'levelling the playing field' seem to be necessary in order to improve the situation of those individuals who suffer a lack of opportunities in both educational and economic dimensions and who are more vulnerable to economic shocks. --- Data availability statement. The data that support the findings of this study are available from the corresponding authors upon reasonable request. --- competing interests to report. --- Notes 1 Groups 1 to 4 are: 1 Managers, 2 Professionals, 3 Technicians and Associate Professionals and 4 Clerical Support Workers. Groups 5 to 7 are: 5 Services and Sales Workers, 6 Skilled Agricultural, Forestry and Fishery Workers, 7 Craft and Related Trades Workers. Groups 8 and 9 are: 8 Plan and machine operators, and assemblers and 9 Elementary Occupations. This classification is made by the International Labour Organization (ILO) 2 Share of population aged 18-24 who have not completed upper secondary education and are not in any form f education or training. 3 Both are referred to non-university education.

Family background characteristics and educational resources are crucial in shaping individuals' income and therefore a potential source of income inequality and inequality of opportunity. This article analyses inequalities in the Spanish diverse regions using data from the European Survey of Income and Living Conditions (EU-SILC). Along this article we analyse the relevance of family background characteristics in shaping individuals' opportunities as well as the role education plays in income inequality. Our findings show that family background characteristics are of great importance in shaping individuals' performance and opportunities of achieving a certain level of income, well-being and education; also, educational variables at regional level, such as the education expenditure per student, are highly correlated to the levels of inequality. In view of the results, it would be advisable to implement public policies targeting vulnerable groups to ensure equal opportunities and invest in education given its potential as equalizer of income inequality.

BACKGROUND The improvement in breast cancer survival conferred by chemotherapy is the same whether it is delivered before definitive surgery (neoadjuvant) or after (adjuvant) [1,2]. There are advantages to neoadjuvant treatment, however, particularly for certain groups of patients. For patients with larger cancers or smaller breast size, neoadjuvant chemotherapy increases the likelihood of successful breast conservation and can make inoperable disease resectable [3]. Furthermore, unlike adjuvant chemotherapy, neoadjuvant chemotherapy both predicts prognosis and allows assessment of treatment response, and therefore has the potential to alter the course of treatment. This information about treatment response and prognosis is especially important for patients with higher recurrence risk, and guidelines by the National Comprehensive Cancer Network (NCCN) recommend considering neoadjuvant chemotherapy for women with Stage 2 cancer or above [3]. Given these advantages, neoadjuvant chemotherapy is now used in some sites for up to a quarter of breast cancer patients receiving chemotherapy [1,2,4]. However, its use varies substantially even between large centers [4][5][6][7][8]. Research into adoption of other new medical technologies has shown that adoption often varies strongly by nonclinical factors, including patient socioeconomic status. Variation by socioeconomic status would be a particular concern in breast cancer care. On average, patients with lower socioeconomic status have higher-stage breast cancers [9] and therefore have a greater chance of benefitting from neoadjuvant therapy. Variation in use of neoadjuvant chemotherapy could also stem from differences in patient preferences. There appear to be few disadvantages of neoadjuvant chemotherapy, but care may appear more complex to the patient, and patients may find the delay to surgical excision concerning [10]. Little is known about patients' experiences with neoadjuvant compared with adjuvant chemotherapy, the value patients place on the advantages and disadvantages of neoadjuvant timing of chemotherapy, or whether these might vary by socioeconomic status. To address these gaps, we examined a sample of women with breast cancer who participated in a cross-sectional multicenter study at eight centers and received chemotherapy, examining patterns of chemotherapy timing (neoadjuvant vs adjuvant) by socioeconomic status. We also describe patients' experiences with neoadjuvant compared with adjuvant chemotherapy, including their perceptions of the risks/benefits, the decision-making process, and quality of care. --- METHODS --- Study Source and Population The study was performed within the Greater Plains Collaborative (GPC), one of 13 Patient Centered Outcomes Research network (PCORnet) clinical data research networks (CDRNs). The study sample included eight GPC sites (of a total of nine that treated adult breast cancer patients) from seven Midwestern states. In brief, each center extracted data from their North American Association of Central Registries-formatted tumor registry into its Informatics for Integrating Biology and the Bedside (i2b2) system, and women aged 18 or older with a first diagnosis of histologically-confirmed Stage 0-3 breast cancer made between 1/1/2013 -5/1/2014 were identified. A random sample of 250 of these women from each site was then selected for a mailed survey sent in May 2015, with up to 10 replacements for any surveys that were returned with evidence that a subject was ineligible (subject died, mailing was returned by the postal service unopened, or subject reported that the 2013-2014 breast cancer was not her first breast cancer). In addition to the survey, all subjects were asked for additional consent for investigators to examine their medical records, including information from the tumor registry. The study was approved and monitored by the Institutional Review Board (IRB) at the coordinating center (University of Iowa) with the other centers ceding review. Patients were asked in the survey to report timing of their chemotherapy, specifically whether they had chemotherapy before surgery only (neoadjuvant), after surgery only (adjuvant), or both before and after surgery. Prior studies suggest that patient self-report of chemotherapy is highly consistent with other measures, including administrative (billing) data [11]. Subjects with invasive (stage 1-3) breast cancer who reported having received either neoadjuvant or adjuvant chemotherapy were eligible for this study. Because extent of disease is important in cancer treatment decision-making, only those women who also agreed to use of tumor registry information were included in the final sample. --- Variable definitions Subject demographics (age, race/ethnicity, education, employment, income, and marital status) were obtained from survey responses. Rural residence was identified using the ruralurban continuum categories from 2013 US Census data for the subject's zip code. Extent of disease (stage) and tumor markers (hormone receptor (HR) and human epidermal growth factor receptor 2 (HER2)) were obtained from cancer registry data. Three Likert-scaled items were also developed for the survey regarding the importance patients placed on the advantages and disadvantages of neoadjuvant vs adjuvant chemotherapy. These items were based on summary statements from physician treatment guidelines [3] and used wording that was similar to previously developed items regarding importance of factors in breast cancer treatment decision making [12]. Participants responded to "When decisions were being made about chemotherapy, how important was it that the timing of the chemotherapy (before or after surgery or both)....(a).....would allow you to know that your cancer had shrunk; (b).....would improve your chances of having a lumpectomy; and (c)......would make sure the cancer was removed by surgery as soon as possible" with 5 response levels, with labels below the first as "not at all important" the third "somewhat important" and the last "very important". Patient experiences with decisions related to chemotherapy were assessed using several measures. To assess subjects' reports of shared decision-making, we utilized two instruments with Likert-scaled items, [13] the first of which evaluated patients' preferred participation in breast cancer treatment decisions by asking "Which of the following best describes the way you would prefer to make a decision about your cancer treatment?" with five responses ranging from "I prefer that I make the decisions about treatment with little or no input from my doctors" to "I prefer that my doctors make the decision about treatment with little or no input from me." The second instrument then asked about subjects' actual experiences specific to the chemotherapy with responses substituting, for example, 'I made' for 'prefer that I make' [13,14]. Using an adaptation of the actual experience item, subjects were also asked about family participation in the decision. Patient-reported experiences with quality of physician communication (2 items) and coordination of care (6 items) were measured through Likert-scaled items based on the Consumer Assessment of Health Care Providers and Systems. These items were previously used in a longer measure of patient-perceived quality of cancer care [15,16] and are strongly correlated with overall perceptions of quality of care [16]. --- Analysis Subjects who reported that they had chemotherapy both before and after surgery (n=4) were not included in the study. Because of relatively high missingness of information for variables for both HER2 status and income, missing categories were created for all variables. Initial examination of the overall sample showed that there was no association of education or income with overall rates of chemotherapy use. Furthermore, few African-American or Hispanic women were treated for breast cancer at our study sites. We thus focused our examination on the potential association of neoadjuvant chemotherapy use (vs adjuvant chemotherapy use) with socioeconomic status as measured by education and income. In a bivariate examination of the association of income with neoadjuvant receipt, there were no differences between the lowest 5 groups we examined, so in further analyses they were classified as highest income (>$100,000) vs all others. In bivariate examination of the association of education with neoadjuvant receipt, there were no difference in neoadjuvant receipt in the highest three groups we examined, so in further analyses they were classified as lowest group (high school/GED or less) vs all others. Our primary analyses included either income or education examined in a regression model which included tumor markers, stage, and demographic variables. To examine the robustness of these results, we also examined several alternate models of SES: a model with both education and income included together, and one using a variable which combined income and education into four groups: Less than $100,000 and high school or less, less than $100,000 and more than high school, greater than $100,000 and high school or less, and greater than $100,000 and more than high school. Finally, we examined the inclusion of a measure of health literacy [17,18] which in models with education was not statistically significant, so was removed from further analyses. We also examined subjects' beliefs/preferences and experiences with care (importance of neoadjuvant chemotherapy advantages/disadvantages, decision-making roles, and perceived quality of care) by timing of chemotherapy. Results for the new variables for importance of neoadjuvant advantages/disadvantages were examined either in three categories (1-2, 3 and 4-5) or dichotomized with 1-3 vs 4-5) ; results were similar so the two-category variable was used. Following Sulayman, decision-making variables were recategorized as passive, shared, and active [14]. Finally, we performed exploratory analyses examining whether differences in patients' beliefs and/or experiences were mechanisms for the association of income and education with chemotherapy timing. Based on unadjusted findings for the association of perceptions of neoadjuvant chemotherapy advantages/disadvantages with chemotherapy timing, we examined this relationship in adjusted models, as well as whether any differences by income in perceptions of advantages might explain differences in receipt of neoadjuvant chemotherapy. All analyses were conducted with SAS statistical software, version 9.4 (SAS Institute, Cary, NC) and two-tailed P values. The datasets during and/or analyzed during the current study are available from the corresponding author on reasonable request. --- RESULTS The overall sample of 1,235 subjects and the details of the calculation of survey response rates (overall response rate 62.2%) have been previously described [19]. Among the 877 survey respondents who provided consent to view their cancer registry information, 331 received either neoadjuvant or adjuvant chemotherapy and were included in further analyses. About 40% of this sample was age 50 or under, 19.9% had a high school education or less, and 29.2% had a household income that was less than $50,000 yearly. As shown in table 1, 29.6% of the women (n=98) received neoadjuvant chemotherapy. --- Factors Associated with Neoadjuvant Chemotherapy Receipt of neoadjuvant chemotherapy among the chemotherapy sample (i.e., only patients who received chemotherapy) varied substantially by age, tumor markers and extent of disease, so that 50% of HR-and HER2+ received neoadjuvant chemotherapy compared with 22.8% of those with HR+ and HER2-(table 1). Neoadjuvant chemotherapy receipt ranged from a low of 13.8% for those with Stage1 cancer to 40.0% for those with Stage 3 disease. Although differences in use of neoadjuvant chemotherapy by site of care were not statistically significant, there was substantial site-to-site variation, from a low of 20.0% of chemotherapy given as neoadjuvant at one site, to a high of 43.6%. Use of neoadjuvant chemotherapy among those receiving chemotherapy also varied substantially by income and education (table 1), so that 41.9% of patients with household income (>$100,000 (approximately the highest quartile) received neoadjuvant chemotherapy. Only 18.2% of those with a high school education or less received neoadjuvant chemotherapy, vs 32.7% of those with at least some college. While women residing in rural zip codes were less likely than those in urban zip codes (i.e. zip codes in metropolitan statistical areas) to receive neoadjuvant chemotherapy, this difference was not statistically significant. In models adjusted for demographics, stage, and biomarker status, patients with lower household income were only slightly more than half as likely (OR 0.56, 95% CI 0.2-0.9) as those with income >$100,000 to receive neoadjuvant chemotherapy (table 2). The differences in neoadjuvant use by education that were seen in unadjusted analyses (table 1) were no longer significant in adjusted analyses of education alone (p=0. 15) or analyses that included both income and education. When income and education were combined into 4 categories (see analysis) neoadjuvant use by those with low income/low education differed from other categories, but no other differences were found (data not shown). --- Patient experiences with and preferences regarding chemotherapy Patient reports of beliefs/preferences and experiences regarding chemotherapy are shown by timing of chemotherapy in Table 3. Subjects who received neoadjuvant chemotherapy differed from those with adjuvant chemotherapy in their response to two of three questions about the importance of specific advantages and disadvantages of neoadjuvant chemotherapy. Women who received neoadjuvant chemotherapy were more likely than those who received adjuvant to report that "know(ing) the cancer had shrunk" was important (4-5 on a Likert scale of 1-5) and less likely to report that "making sure the tumor was removed by surgery as soon as possible" was important. Women who received neoadjuvant chemotherapy were slightly more likely to report that "increasing the chances for lumpectomy" were important, although this difference was not statistically significant (p=. 21). Also shown in table 3 are patient reports of chemotherapy decision-making for the sample overall and by timing of chemotherapy. Subjects who received neoadjuvant chemotherapy were slightly more likely to report that they desired a more passive role in decision-making about their treatments (either the doctor makes the decision and considers my opinion or the doctor makes the decision with little input from me), but this difference was not statistically significant (p=0.22). The difference was larger, though still not statistically significant, when subjects were asked about the role in the decision they actually played, with 34% of patients who received neoadjuvant chemotherapy having a more passive decision-making role as compared with 24% of those who received adjuvant chemotherapy (p=.06). There were no differences by timing of chemotherapy in response to a similar item asking about the actual role of family participation. Subjects who received neoadjuvant chemotherapy reported similar quality of communication regarding cancer treatment compared with those receiving adjuvant chemotherapy. They also reported similar quality of coordination of care overall, and in response to each of the individual items that made up the coordination score. --- Potential mechanisms for variation in chemotherapy timing by income Given the differences in subjects' ratings of the importance of neoadjuvant advantages/ disadvantages by timing of chemotherapy, we explored their potential role in the variation of chemotherapy timing by income. In all three cases, high-income subjects were less likely to consider the advantage/disadvantage as important. Specifically, 54.8% of high-income subjects reported that treatment that "allowed you to know the cancer had shrunk" was important or very important compared to 67.8% of lower-income (p=0.04), 19.7% of highincome reported that "improving your chances of a lumpectomy" was important or very important compared with 46.9% of lower-income (p<unk>.001), and 60.3% of high-income reported that "making sure the tumor was removed by surgery as soon as possible" was important vs 79% for lower-income (p=.005). Only the result for "removed by surgery" would be consistent with a preference for neoadjuvant chemotherapy among higher-income patients (that is, higher-income patients were less likely to report that a benefit of adjuvant chemotherapy is important/very important). As would be expected given these differences by income, in models that adjusted for the responses "improving your chances of a lumpectomy" and "making sure the tumor was removed by surgery as soon as possible" the difference in neoadjuvant receipt between higher-income and lower income subjects was actually widened (appendix table). In the model that included the subjects' rating of importance of having the cancer "removed by surgery as soon as possible," the differences in neoadjuvant use by income shown in table 2 were narrowed but not fully explained (appendix table ). --- DISCUSSION In this large, multicenter sample of women with invasive breast cancer, over 29% of patients who received chemotherapy received it before surgery. Receipt of neoadjuvant chemotherapy varied as expected by stage and cancer biomarker status, but women were also more likely to get neoadjuvant chemotherapy if they were older or in the highest SES groups (household income >$100,000. Patients who received neoadjuvant chemotherapy rated the importance of its advantages higher, and the importance of its disadvantages lower, than those who received adjuvant chemotherapy. Although there were some differences regarding perceptions of the importance of benefits of neoadjuvant chemotherapy by income, these did not appear to be major factors in the association of income with chemotherapy timing. Patients' reports of decision-making and quality of care were similar by timing of chemotherapy. Our study findings regarding differences in treatment by income appears to conflict with one earlier report regarding neoadjuvant chemotherapy [1]. This study in the National Cancer Database found a lower likelihood of neoadjuvant chemotherapy with nonmetropolitan residence and older age, but residents of higher-income zip codes in that study were actually slightly -but not statistically significantly -more likely to receive neoadjuvant chemotherapy [1]. Our study used individual-level measures of income that were not available in the National Cancer Database, but it is not known whether that might explain the discrepancies in our findings. It is also notable also that our study is consistent with other studies showing that lower-SES patients appear to be at higher risk of slow adoption of other advances in chemotherapy treatment, in some studies either not receiving chemotherapy despite indications for it, [20] or being underdosed [21,22]. We also were able to measure novel and detailed information about patient beliefs and preferences regarding neoadjuvant chemotherapy in our sample. Many patients valued the advantages of both neoadjuvant and adjuvant chemotherapy (figure), but the relative importance of these advantages (and disadvantages) differed with the chemotherapy received. That these preferences also differed by income is concerning, though perhaps not that surprising given that information about neoadjuvant chemotherapy is probably new to many patients [23,24]. Nonetheless, differences in preference by income did not appear to explain differences in chemotherapy timing by income; in fact, when perceived importance of the two neoadjuvant advantages, were accounted for, differences by income increased. Future research that measures variation in preferences regarding neoadjuvant care prospectively, and with careful attention to health literacy and the information provided to patients, is needed. There were no statistically significant differences by timing of chemotherapy in reports of desired or actual decision-making in our sample, although an association of more passive decision-making role with receipt of neoadjuvant chemotherapy was suggested (p=.06). A more passive decision-making experience may be expected for a procedure that many patients are not aware of until it is actually offered to them. Many patients want an active role in breast cancer decision-making [14] so these findings should be examined in other settings and larger samples. Our results also showed similar patient-reported quality of communication and coordination by timing of chemotherapy. This is reassuring, given that the treatment experiences might be expected to be quite different, particularly since patients who had neoadjuvant chemotherapy had higher-stage disease that generally requires more complex treatment. Our study has some limitations. Our reports of timing of chemotherapy were based on selfreport, and it is possible some patients were unsure or mistaken in their reports. However, there is high validity to patient reports of chemotherapy receipt generally [11], and recall of its timing relative to surgery might be expected to be high one to two years after the experience as well. Our study was cross-sectional. It is possible that a prospective observation of counseling sessions (e.g., by use of audiotapes) would have had different findings about perceived advantages and disadvantages of neoadjuvant therapy. Our examination of specific measures of SES were somewhat limited by power, so that our ability to tease out differences between low-income and low-education was limited. There is evidence that a growing number of patients do not need cytotoxic chemotherapy [25][26][27], and it is possible that factors associated with timing of chemotherapy will change as the cohort of patients receiving chemotherapy shrinks Finally, our results may be affected by survival bias or selection bias if patients with shorter survival or who did not participate in the study would have reported a different relation between chemotherapy timing and experiences than women who did participate. However, early mortality is low in breast cancer [28] even among patients with stage 2 and 3 disease. Despite these limitations, our study provides important information regarding patients' preferences and experiences with neoadjuvant chemotherapy compared with adjuvant chemotherapy. Our study's findings regarding large differences in use of neoadjuvant chemotherapy by income have important implications. While these results are consistent with the patterns seen for adoption of other new technologies and therapies, it is particularly important in breast cancer given the restriction of many new breast cancer trials to neoadjuvant users. Variability in neoadjuvant use by income could perpetuate or even worsen disparities in clinical trial enrollment, and deserves further research and if confirmed, urgent attention. * Because those with missing data for these variables were excluded from the models, results for income vary slightly from --- Appendix

BACKGROUND-There are advantages to neoadjuvant chemotherapy in operable breast cancer, particularly for those with higher-risk cancers, but little is known about factors that are associated with the use of neoadjuvant chemotherapy outside of clinical trials. We examined whether use of neoadjuvant chemotherapy instead of adjuvant chemotherapy varies by nonclinical factors such as patient socioeconomic status or rural residence. METHODS-Women diagnosed with breast cancer in 2013-14 at eight medical institutions were surveyed by mail regarding their experiences with breast cancer treatment, and this information was linked to hospital-based cancer registries. We examined the use of neoadjuvant chemotherapy among women with histologically-confirmed invasive stage I-III breast cancer and used regression models to examine the association of socioeconomic status with chemotherapy timing. We also explored potential mechanisms for those differences. RESULTS-Over 29% of the sample overall received neoadjuvant chemotherapy. Neoadjuvant receipt was lower for those with income <$100,000 (AOR 0.56, 95% CI 0.2-0.9) even with adjustment for other demographics, stage, and biomarker status, and findings for education and a variable for both lowest education and income<$100,000 were similar. Rural/urban residence was not associated with neoadjuvant receipt. Differences by income in perceptions of the importance

INTRODUCTION Since the beginning of the pandemic, human rights, feminist, patient, and maternity advocacy groups have voiced their concern over fundamental rights violations in maternity care 1. Early in the pandemic, COVID-19 was identified as a risk factor for obstetric violence 2, often accompanied by rollbacks in quality of care 3-5. In their commentary, our colleagues from COST Action 18211point 5 points to mounting evidence of adverse consequences that COVID-19 containment measures enforced in facility maternity care worldwide. They call for the priority implementation of evidence-based, human-rights-informed care, including during times of crisis such as the current pandemic. The commentary authors go on to highlight how the reaction to health services and the way they were reorganized reveals'something about the underlying ethos of maternity care provision around the world, raising serious questions about how it should be reframed when services are rebuilt once the pandemic is finally over' 5. Patient advocacy organizations, in this case, national parents' groups monitoring and advocating for improvements in maternity services, are relevant stakeholders in healthcare 6 because they work at the intersection of parents' lived experiences, policies and rights, using their unique insight to catalogue and report patients' lived experiences in general [7][8][9], Policy case studies but also in maternity services specifically 1,10,11. As in any crisis, caution is advised when reactions include sudden practice changes to maternity services that are not grounded in the best available evidence. These can bring unforeseen consequences for women, pregnant and birthing people, and families. Respectful maternity care cannot be seen as secondary to pandemic containment measures in maternity facilities; it is imperative that all containment measures are proportionate to the threat (in this case, the spread of COVID-19) and that undue burdens are not put on pregnant, birthing, and postnatal families. Practice guidelines for maternity care during COVID-19 in Europe during the first three waves of the pandemic, in countries where they existed, revealed erratic, contradictory and inconsistent approaches to scientific evidence 12. Moreover, there is emerging evidence that containment measures implemented in maternity services as a response to the COVID-19 pandemic negatively impacted maternal and perinatal outcomes 13,14, including maternal mental health [15][16][17] but also healthcare professionals' well-being 18. Although some authors have reported on parents' experiences in neonatal care 11,19, to our knowledge, no other published research has used a human-rightsbased approach to survey national pandemic-related changes to care for pregnant and birthing women from the perspective of parents' organizations. Organizations representing patients overall, or specific patient groups (e.g. cancer patients) have been conducting research and publishing reports using similar methodology throughout the pandemic 7-10. The perspective of patient organizations, who collect and monitor information from patients receiving care directly but also through monitoring activities (e.g. social media, policy monitoring) is unique and, in a crisis, is a valuable source of real-time information. This article aims to provide an overview of different changes made to maternity services implemented in some or all maternity facilities over the first 17 months of the pandemic, roughly corresponding to the first three waves of the pandemic in Europe, as reported by maternity service (parent) organizations who are monitoring the situation in their countries. The article reports on 16 core services and containment measures implemented in maternity services which directly impacted women, birthing people and families. Based on the findings, we propose a strategy for reimagining post-pandemic maternity services across Europe, including these 16 issues and consider the implementation challenges. --- STRATEGY FOR POST-PANDEMIC MATERNITY SERVICES We used a descriptive qualitative and participative approach to gather input from parents' organizations across Europe by preparing the protocol, research questionnaire and semistructured interviews in a participative manner. According to Sandelowski 20, a 'Qualitative description is especially amenable to obtaining straight and largely un-adorned [...] answers to questions of special relevance to practitioners and policymakers'. In November 2020, a poll of COST Action 18211 network members was conducted online with six questions that collected information on the most significant changes in maternity services during the pandemic. Seven COST member organizations completed the survey, identifying some of the major COVID-19 containment measures implemented in their countries, which women and families had reported to them as problematic. The authors compared and analyzed the survey responses by clustering the data according to the phases of provision of obstetric care (antenatal, intrapartum and postnatal care), which identified 16 core issues from the survey for national maternity (parent/patient) organizations, as detailed in Table 1. Based on these results, we prepared a questionnaire with the 16 core issues and conducted semi-structured interviews with representatives of national maternity service (parent) organizations, recruited through COST Action 18211 and the European Network of Childbirth Associations (ENCA). Five of the organizations from the COST Action had also participated in identifying the 16 core issues, while the others did not. The interviewees were invited to comment on the identified issues and to detail the situation in their countries, based on reports they had received from parents on the ground and information they had gathered nationally. The focus was on generalized tendencies observed in the individual countries, leaving space for single (major) facilities or regions differing from this overall trend. The information gathered through the interviews is given in Table 1. Upon completion of an interview, the information was shared with the interviewees, who were invited to review and revise the recorded responses if needed. Data collection, processing and storage, conformed to the General Data Protection Regulation (GDPR) and the Declaration of Helsinki. Consent was given at the beginning of each interview, with the possibility of exiting the interview at any time. No financial or other incentives were offered to the interviewees. Table 1 summarizes the data collected during the semistructured interviews for each country. A discussion on general trends across the countries continues below. --- Antenatal care Interviewees reported that in-person antenatal care appointments were reduced overall across the countries studied, with the change being quite drastic in some countries. In Ireland, antenatal and postnatal visits were capped at 15 minutes regardless of the reason for the appointment, and in Slovakia, in-person appointments were banned during the first pandemic wave. Routine tests and scans remained available but were affected by shorter appointment times (Ireland) and were difficult to access during the peaks of the first three COVID-19 waves (Portugal, Spain, Slovakia). Companions at antenatal appointments were banned in most countries, except the Netherlands and the Czech Republic, where interviewees stated that some facilities allowed companions. One country (Croatia) did not have the practice of companions at antenatal appointments prior to Eur J Midwifery 2022;6(December):71 https://doi.org/10.18332/ejm/156902 --- Policy case studies --- Table 1. Reports on maternity care during the height of the first, second and third waves of COVID-19 in 13 European countries <unk> <unk> <unk> <unk> <unk> <unk> x* / <unk> ( <unk>) <unk> <unk> Companion at antenatal appointment x x x x (<unk>) x x x (<unk>) x x - x Access to routine tests and scans Policy case studies the pandemic. Overall, interviewees commented that the quality of care was compromised as a result, especially in countries where the quality of care was already low or varied before the pandemic. Antenatal mental health services, where they existed prior to the pandemic, were reduced. This is particularly problematic considering that the pandemic triggered the compromise of the mental health of all people, especially pregnant, birthing, and postnatal families 21,22. Parents' organizations reported observing higher stress levels among pregnant women in their countries because of these difficult circumstances. One organization quoted a pregnant woman's experience in Germany: (<unk>)* (<unk>) <unk> (+) 15 min maximum <unk> <unk> <unk> <unk> <unk> <unk> (<unk>)* / <unk> <unk> <unk> <unk> Antenatal mental health services available - -/ x <unk> Reduced (<unk>) - - - - - - - - Intrapartum care Companion at vaginal birth x* / <unk>** (<unk>) / <unk>** <unk>** <unk>** <unk> <unk>* / <unk>** x* / (<unk>) / (<unk>)** x <unk>* x* / (<unk>) <unk>** / x x <unk>** / <unk>* Companion at CS birth x / - x <unk> (+) (<unk>) / x <unk> <unk> x / - x / - (<unk>) - - - <unk> ( 'My gyn offered me this test for malformations and my partner was not there but I just did not know what to do and how could I decide this without asking his opinion? It is his baby, too'. --- Intrapartum care Companionship Parents' organizations reported differences in the possibility of having a birth companion. This measure was subject to the broadest variety over time, type of facility and country. All interviewees reported a reduced possibility of having a birth companion or the institution of outright bans. Another experience from Germany showed the rapidly changing situation: 'The ward that allowed my partner with me was closed the day before I gave birth because they focused staff on COVID patients. I had to go to a hospital with a very different policy where my partner stayed in the parking lot the whole time'. In those situations where a birth companion was allowed, interviewees reported severe restrictions, including limiting when the companion was allowed to be present or the requirement that the birth companion have a negative COVID-19 test, which before the advent of rapid tests usually meant a minimum 24-hour wait. Only one country (Netherlands) stood out for being consistent in maintaining the same level of a right to a birth companion at a vaginal birth throughout the pandemic. Women who give birth by cesarean section are not always afforded the same rights to companionship as those who give birth vaginally. Prior to the pandemic, reports by interviewees showed that the situation regarding companionship at a cesarean varied widely across the countries. In three countries, interviewees reported having a companion at a cesarean section was not the norm before the pandemic (Slovakia, Hungary, Croatia). In four countries where companionship at a cesarean was mostly the norm prior to the pandemic, bans were introduced as part of COVID-19 containment measures (Portugal, Spain, Italy and Poland), with differences between public and private facilities in Portugal, even though companionship at cesarean section is guaranteed by law 23. Doulas have appeared as second-birth companions over the past twenty years across Western Europe. Despite the overwhelming evidence that the presence of a doula can improve birth outcomes for mothers and newborns, especially for racialized and poor women, migrants and refugees [24][25][26], interviewees reported that second companions, and therefore doulas were banned as part of containment measures in the majority of the countries included in the survey, with only two countries offering the possibility of doula companionship in some facilities (Spain, Netherlands). Visitor bans were reported by most parents' organizations in the study, whereby women were not able to see their partners or other family members after birth and for the duration of their stay in-hospital, at least during the peaks of COVID-19 waves. The exception to this were three countries (Poland, Czechia, Slovakia). Later research linked these types of bans to increased postnatal traumatic stress response 21,27. It is important to note that this study did not collect data on the length of the average hospital stay after vaginal or cesarean birth, which may vary across the countries and alter the long-term effects of visitor bans. --- Mode of birth While most parents' organizations surveyed did not report that women who were suspected of having or were COVID-19 positive (COVID-19+) had their labors induced without obstetric indication, in two countries (Portugal and Croatia) there were reports of this happening in some facilities. In two countries, interviewees reported that most facilities required that women with suspected or confirmed COVID-19 give birth by cesarean section without obstetric indication (Poland, Cyprus). In Poland, women who were COVID-19+ were required to birth in special facilities, where cesarean section was mandatory for COVID-19+ mothers. In five other countries, interviewees reported mandatory cesarean for COVID-19+ women for some facilities (Portugal, Germany, Italy, Hungary, Croatia), during certain pandemic waves. This echoes results published for the United States, which showed increased medicalization of childbirth for women suspected or confirmed of having COVID-19 28. --- Availability of home birth services During the pandemic, more women sought home birth services 29. In this study, organizations in six surveyed countries stated that only private-sector home birth services continued to be available as normal during the pandemic. In Portugal, where home births are only offered in the private sector, their number nearly doubled in 2020 compared to 2019 29, which can partially be explained by people wanting to avoid the restrictions implemented in facilities, and perceived safety in avoiding COVID-19 infection. As the demand increased substantially, some women were not able to find midwives who could attend a home birth. Two organizations stated that home birth services remained available as normal in both public and private sectors (Netherlands, Germany). Interestingly, the Irish parent-organization reported that the health service increased the availability of home birth services as a result of increased demand. In four of the countries, organizations reported that home birth services were not available before Eur J Midwifery 2022;6(December):71 https://doi.org/10.18332/ejm/156902 Policy case studies the pandemic, and were therefore not available during the pandemic either (Czechia, Slovakia, Croatia and Cyprus). Despite not being officially available, the number of home births in Croatia increased by 30% between 2019 and 2020, although the number remained low 30. --- Separation from newborn and visiting bans Early in the pandemic, the WHO issued recommendations clearly stating that if the mother is COVID-19+, the mother and newborn should be kept together, provided the mother feeling well enough to care for the newborn 31, a recommendation mirrored by the guidelines by the Royal Colleges of Midwifery and Obstetrician-Gynecologists 32. Despite this, one of the major problems with COVID-19 response policies was the separation of mothers who were suspected or confirmed COVID-19+ and their newborns, especially in the first three to nine months of the pandemic; this was reported by most interviewees. Notable exceptions to these were reports from organizations from Ireland and the Netherlands, which stated that mother-newborn dyads were kept together as normal, regardless of the mother's COVID status. In some countries, the health authorities were slow to issue evidence-based recommendations. In Portugal, the first recommendation by the College of Obstetrics and Gynecology in March 2020 stated that women who are COVID-19+ must be separated from their infants and prevented from breastfeeding 33. As more evidence and international guidelines were published resulting in increased pressure from advocacy groups, the General Directorate of Health updated the guidelines, albeit one full year later (March 2021). Despite the new guidance, the previous policies continued across facilities 34. In Croatia, women who gave birth before the results of their routine PCR test were available, were separated from their newborns as a precautionary measure until a negative result was returned. This practice was prevalent in some facilities throughout the pandemic waves but also between waves, and was especially a problem in smaller facilities where PCR tests took longer to analyze. Visiting restrictions for premature or sick newborns were also reported by most national organizations, with some interviewees reporting the implemented total visitation bans in most facilities (Czechia, Hungary, Croatia, and Cyprus). These measures were implemented despite a lack of evidence that they contribute to containing the spread of COVID-19 and despite being in contradiction with readily available professional and WHO guidelines that were being updated in real-time 32,35. Visitor bans were also reported for hospitalized mothers and newborns by all interviewees except those from three countries (Poland, Czechia, Slovakia), where parents' organizations reported that visitation of mothers was the same as in pre-pandemic times. In the other countries, organizations reported that women hospitalized during pregnancy did not have visitation from the time of their hospitalization to post-partum, regardless of the duration of that stay (from two days to several weeks, in the case of complications). A mother from Germany stated: 'I stayed at the hospital because of late pregnancy complications but after my cesarean birth I left as soon as they let me -I missed my older daughter so much! I hadn't seen her in 2 weeks'. Evidence shows that these visiting restrictions can have a negative effect on postnatal mental health 14,21,36. --- Postnatal maternal mental and physical health Despite increasing awareness about the importance of perinatal mental health during the pandemic, interviewees reported that mental health support services were only rarely available, either because they did not exist pre-pandemic, because they were restricted due to containment measures or because demand was so high that services were not accessible. Interviewees reported that in-person postnatal care, usually organized around six-weeks after birth continued to be mostly available as normal, much as antenatal care was. However, they also reported that pandemic measures resulted in increased use of telehealth consultations, which depend on women and families having access to reliable, affordable internet and devices capable of video calls, which is not always possible. Later evidence about the efficacy of telehealth maternity care during the pandemic showed that this type of care was not always optimal 5,37. The results we found were similar to those described by others 19,38,39, who defined similar themes: less family involvement, reduced emotional and physical support for women, compromised standards of care, increased exposure to medically unjustified cesarean section, and staff overwhelmed by rapidly changing guidelines and enhanced infection prevention measures 40, as well as a decrease in respectful care due to fears surrounding COVID-19 transmission 2,38,41. At the same time, interviewees reported that facilities that were known for providing more motherfriendly and baby-friendly care made great efforts to maintain that level of quality, sometimes at considerable staff effort, whilst otherwise less friendly facilities more rapidly downgraded their services. --- Maternity care guidelines from international and national organizations Throughout the pandemic, international and national organizations have made efforts to create and regularly update guidelines for COVID and maternity care starting. Even the European Parliament was concerned about rollbacks in maternity care in member states 42. Despite these efforts, our data have shown that facility policies did not change much between the first and third waves of the pandemic, even after more evidence was available. A specific example is the benefits of labor and birth companions, which have been widely studied and recognized 22,24. Guidance available very early in the pandemic and updated in real-time as new evidence became available 31,32 clearly stated that companionship at vaginal birth was important and could safely continue with the implementation of basic pandemic practices. Despite Eur J Midwifery 2022;6(December):71 https://doi.org/10.18332/ejm/156902 Policy case studies this, the political will to maintain proportionally appropriate limitations to companionship among policymakers and professional organizations varied widely throughout the various countries and pandemic waves. An example of good midwifery practice was the UK during the second wave, where professional organizations reiterated that attendance of partners during labor and birth must be a priority, with reliable testing and appropriate PPE available to both, ensuring the safety of all those using maternity services, and must be proportionate to policies being used in other departments of the facility 32,43. A second good practice came from France, which did not have a parents' organization representative included in this study but is a major European country. There, the Ministry of Health issued guidance for hospitals in the first month of the pandemic outlining how to include companions in labor and childbirth, while lowering the risk of infection 44. The German Society for Gynecology and Obstetrics issued a similar statement 45 on the inclusion of fathers, also in the first month of the pandemic. The real-time guidelines by the Royal Colleges of Obstetrics and Gynecology and the Royal College Midwives and World Health Organization, referred to earlier in this article, stated that elective induction or cesarean should be avoided for women who have symptoms of or are positive for COVID-19, without obstetric indication 31,32. Other European professional societies issued similar statements, including the German Obstetric Society 46 and the Italian Obstetric Society 47. Conversely, the Portuguese College of Obstetrics and Gynecology recommended a shorter threshold for the decision in the use of epidural analgesia and instrumental birth, meaning that women were more likely to have either of these interventions if they were suspected of or confirmed to be COVID-19+ 33. --- Maternity evidence-based human-rights care The differing responses to the COVID-19 crisis have shown that national and local decision-makers' opinions often impact maternity facility policies more than scientific evidence and international guidelines. As a result, there must be a significant change to the prevailing paradigm and overmedicalization of maternity care in Europe, especially during crises, to one where a proportional, evidence-based response prevails. A healthier perinatal period sets babies and families on a long-term beneficial health trajectory 48. So, while 'classic' impediments to system changes such as financial and human resources must not be underestimated, it seems that the biggest challenge to good quality services lies in an ethical decision: 'Is the well-being, physical and mental health of mothers, babies, and families important to our societies?'; and 'Is an evidence-based, human-rights informed, salutogenic approach to maternity care politically desired?'. If so, policymakers need to take appropriate steps to enforce such an approach, as the current political and medical establishments alone seem unable or unwilling to birth the necessary change. A woman-centered, positive birth experience (with no unnecessary interventions, in a space that feels physically and emotionally safe to the mother) is a valuable, long-term investment for any society. These birth experiences protect the physical and mental health of mothers 49, and result in higher breastfeeding rates 50, which should be an integral part of public health planning. --- Limitations All the countries included in the survey have a mix of public and private facilities and insurance, which may have had different policies during the peaks of the pandemic waves. These have been noted where appropriate, but the mix of public and private facilities varies widely across the countries. Additionally, the results reflect general national trends, as they were reported to and collected by parent (patient) maternity rights organizations with a national presence in their countries. Although most of the organizations are large, experienced organizations with national networks, there is a potential for reporting bias. The multi-national nature of the data collection and importance of collecting these experiences, which may never be reported in official data, make them a valuable source of information. Future quantitative research will shed more light on the nuance in changes in maternity services during the pandemic. Furthermore, the research questions did not take into consideration access to medical abortion or medically assisted fertility services, and therefore did not map these aspects of reproductive healthcare. Finally, the research only includes information for the first 17 months of the pandemic, roughly corresponding to the first three COVID-19 waves in Europe, with data collection happening from May to July 2021, and for this reason does not include information on vaccines, rollout to pregnant individuals or data on the fourth wave, which began in Europe in September-October 2021. --- CONCLUSION Crises such as pandemics are a litmus test for health services and societies, requiring a balance between containment measures and quality care. The responses to the COVID-19 pandemic mapped in this article have shown that in some countries, pregnant, birthing and postnatal women and their families were expected to bear a disproportionate burden of the pandemic response. From a parents' perspective, the solution lies in a paradigm change, towards a respectful, woman-centered and familycentered approach 51. The prevailing biomedical model of maternity services focused on pathology, must give way to a neuro-psycho-social model of care where maternity care services are shaped according to woman's needs, based on scientific evidence, and focused on the promotion of human health 52. The long-term mental and physical health of mothers, newborns and families must be considered when measuring outcomes and reshaping services, and mothers and their families must be recognized as stakeholders and be involved in all levels of decision-making. This is critical as countries are still dealing with, and will likely continue to deal with COVID-19 waves and new variants of the disease, which may disproportionately affect pregnant women. Other --- CONFLICTS OF INTEREST The authors have each completed and submitted an ICMJE form for disclosure of potential conflicts of interest. The authors declare that they have no competing interests, financial or otherwise, related to the current work. DD reports receiving a virtual mobility grant to complete the research on this article (COST Action DEVOTION CA18211) and being a member of board of Human Rights in Childbirth (unpaid) and being a paid employee and an unpaid member of board of Roda -Parents in Action. KH reports receiving grants from Mother Hood e.V. (National Guideline project (AWMF No. 015-083), CEDAW dissemination project) and payment for lectures and presentations from University of Cologne and First Emotional Aid conference (Brixen, Italy). KH also reports being an unpaid science officer and responsible for international network for Mother Hood e.V. CB reports receiving a virtual mobility grant to complete the research on this article (COST Action DEVOTION CA18211) and being an unpaid member of board of Associaç<unk>o Portuguesa pelos Direitos da Mulher n Gravidez e Parto (APDMGP). RT reports receiving consulting fees for Project: 'Transitioning motherhood, mother2mother support' under the registration number 2021-2-HU01-KA210-ADU-000050329 to the call entitled Strategic Partnerships announced in the framework of the Eramus+ as part of the project team from 'El Parto es Nuestro' in delivering the research services to run the fieldwork activities in Spain. --- FUNDING This research was partly financed through a grant from EU COST Action 18211. --- ETHICAL APPROVAL AND INFORMED CONSENT Ethical approval was not required for this study. Participants gave informed consent. --- DATA AVAILABILITY The data supporting this research are available from the authors on reasonable request. --- AUTHORS' CONTRIBUTIONS DD, CB and KH worked together to conceive the article's premise and to collect initial information, based on a survey by Olga Gouni, who participated in the initial mapping. DD and KH mapped indicators based on survey responses and conducted interviews with parents' organizations. DD, CB, KH and RT collaborated on writing the manuscript. --- PROVENANCE AND PEER REVIEW Not commissioned; externally peer reviewed.

We surveyed changes to maternity care services in the first 17 months of the COVID-19 pandemic in 13 different European countries, from the perspective of national maternity service (parent) organizations advocating for a human rights approach to maternity services. A qualitative study was conducted in November 2020. An open-question survey was sent to national maternity service (parent) organizations and members of COST Action 18211 in Europe, asking about COVID-19 measures in maternity services (antenatally, intrapartum, postnatally, and overall satisfaction). From the open answers, 16 core issues were extracted. Between February and August 2021, semi-structured interviews with the national representatives of 14 parent member organizations in Europe were conducted, collecting details on overall national situations and changes due to COVID-19 measures. The reported experiences of parent organizations from 13 European countries show wide variations in epidemiological containment measures during the first 17 months of the COVID-19 pandemic. Practices differed between facilities, resulting in emotional disquiet and confusion for parent-patients. Most countries maintained antenatal and postnatal care but restricted psychosocial support (antenatal and birth companions, visitors). Organizations from nine countries reported that women had to wear masks during labor, and all but two countries saw separations of mothers and babies. Most parent organizations described a need for more reliable information for new parents. During the pandemic, non-evidence-based practices were (re-) established in many settings, depriving women and families of many factors which evidence has shown to be essential for a positive birthing experience. Based on the findings, we consider the challenges in maternity services and propose a strategy for future crises.

Introduction In recent years, a movement, led by humanities undergraduates, to 'decolonise' the traditional university curriculum, has gathered momentum. Its aims are to provide a more inclusive education, one that looks beyond the traditional Eurocentric white male syllabus to include more women and black and minority ethnic authors. Decolonising the curriculum is an emerging issue for all types of higher educational curricula as seen in the University College London (UCL) initiative entitled 'Why is my curriculum white' [1]; Priyamvada Gopal's (University of Cambridge) provocative article in the Guardian newspaper, 'Yes, we must decolonise: our teaching has to go beyond elite white men' [2], and SOAS, University of London blog 'Decolonising the curriculum: what's all the fuss about?'. [3] Efforts by the humanities departments for inclusivity finds its parallels in medicine's discussions on how a more inclusive curriculum can improve disease diagnosis and treatment, patient management, professionalism and quality of care. Stone et al speaking from a planetary health education curriculum perspective also say: "Understanding the differences between equality and equity in theory and practice, and concepts of marginalisation, vulnerability, resilience, and who benefits and is harmed in a given scenario, is a core objective of planetary health teaching. Since the effects of environmental change on human health are heterogeneous and mediated by factors such as geographical scale, temporal scale, socioeconomic factors, and political and cultural context, students should think critically about whose health is at stake and how it is measured." [4] As a result, more recently, medical educators are beginning to think of how this process of decolonisation can apply to medical education. Medical schools must address the equality and diversity agenda, which requires that the education system produce doctors who can meet the complex needs of a diverse population. Marginalised groups have poorer healthcare outcomes than patients at the top of the post-colonial hierarchy, such as patients who are white, male, high social class and heterosexual. This paper examines this professional and legal impetus to decolonise the medical curriculum and how elements of decolonisation can be applied to the modern medical curriculum. --- The colonial backdrop of dehumanised care Decolonising the history of medicine promotes awareness and questions the traditional narratives and power imbalances in order to disrupt the legacy of white male patriarchal colonisation of medicine. By reassessing past medical colonial legacies, it could be possible to re-frame, reorient and reform the profession. Porter, a cultural historian, says that the 18 th Century European Enlightenment and subsequent professionalism of medical practice encouraged the school of scientific enquiry which privileged experimentation and discovery at the expense of empathy. [5] It encouraged the practitioners to perceive the patient as a subject first and a human second. Over the course of centuries this emboldened the European male practitioners and moulded a patriarchal medical system, allowing a state of medical cultural arrogance to grow. The benefits of the 18 th Century Enlightenment were to favour reason over superstition and further scientific advancement. However, in the process of establishing medical criteria, it entrenched power hierarchies, excluded other ideas of healing, and other types of practitioners such a midwives, folk healers or indigenous healing systems in the colonies. Medical pluralism was squeezed out of the growing medical power hierarchy. Proponents of medical decolonisation argue that the story of Dr. Marion Sims (who has been named as the 'father of gynaecology' due to his development of the Sims speculum and surgical techniques for the repair of vesico-vaginal fistulae) which is omitted currently, should be included in medical curricula. Sims's motivation was to repair fistulae in black women slaves, so they could return to the workforce rather than to heal them -representing the 18 th century enlightenment's aspect of dehumanisation. The fistulas in this population were a result of acquiring the skills of forceps delivery by obstetricians whilst disregarding and ignoring indigenous practices of midwifery. The black women slaves that were experimented on to improve surgical technique did not give their consent to the procedure and nor were they given any sort of analgesia. The names of three of these women, Anarcha, Betsy and Lucy, have now found equal fame from a human rights perspective. After protests regarding their treatment, Sims' statue in Central Park, New York, USA, has been relocated. [6] Likewise, in the history of the Tuskegee syphilis experiments in the USA [7], where 600 black men were recruited to eugenics experiments to follow the natural history of untreated syphilis. They were intentionally denied treatment though a lack of honesty. The US government has since apologised, and compensated the families. Proponents of decolonisation argue that this story should have a place within ethics and law syllabi of medical schools. --- Figure 1. Useful definitions 'UNCONSCIOUS BIAS refers to a bias that we are unaware of, and which happens outside of our control. It is a bias that happens automatically and is triggered by our brain making quick judgments and assessments of people and situations, influenced by our background, cultural environment and personal experiences.' [8] 'CULTURAL COMPETENCE education for health professionals aims to ensure all people receive equitable, effective healthcare, particularly those from culturally and linguistically diverse backgrounds.' [9] 'CULTURAL HUMILITY was used in a variety of contexts from individuals having ethnic and racial differences, to differences in sexual preference, social status, interprofessional roles, to healthcare provider/patient relationships. The attributes were openness, self-awareness, egoless, supportive interactions, and self-reflection and critique. The antecedents were diversity and power imbalance. The consequences were mutual empowerment, partnerships, respect, optimal care, and lifelong learning. Cultural humility was described as a lifelong process. With a firm understanding of the term, individuals and communities will be better equipped to understand and accomplish an inclusive environment with mutual benefit and optimal care.' [10] The process of dissipating cultural arrogance in favour of cultural humility (Figure 1) as part of decolonisation would entail disrupting elements of the hidden medical curriculum where the legacy of colonial norms operates as unconscious bias in professional behaviours. By directly tackling the history of medicine through the lens of decolonisation within the ethics and law, anthropology or global health frameworks of the syllabus, medical educators can critically evaluate the evolution of patient centred care, and reflect on who, why and how the profession chooses to commemorate knowledge. Equally educators can then contextualise the legal advancements of human rights in healthcare. --- Professional and legal impetus to decolonise the medical curriculum Professional impetus The case for decolonising the medical curriculum stems from a social justice perspective on the persistent legacy of colonialism which results in power imbalances and healthcare inequity. [11] Decolonising the curriculum could equip future doctors with the professional literacy and confidence to deal with the complexity and uncertainty of growing 'grass-roots' decoloniality activism by the patients, academics and public. This debate has started difficult conversations which we will explain further in this article. 'White Fragility' is the refutation of the existence of structural racism as described in Robin DiAngelo's bestselling book 'White Fragility: Why It's So Hard for White People to Talk About Racism' [12]. Defensive responses regarding the concept of decolonisation, from doctors who have a socio-cultural background which we could call 'white coat privilege could be regarded as demonstrations of 'privilege fragility' a term that we have coined for this article. Privilege fragility is at odds with the embodiment of the diversity and equality agenda within professional behaviours, such as displaying 'cultural humility' (Figure 1) for the purposes of neutralising power imbalances in education and healthcare. Susan Goldsmith, Chief Operating Officer and Deputy Chief Executive of the General Medical Council (GMC) said: "We understand and value the diversity within the medical profession and recognise the importance of supporting doctors to serve a diverse population across the UK... we have to be compliant with equality and human rights legislation, but being fair goes way beyond just meeting legal requirements. It is an integral part of everything that we do." [13] These ideas are further re-enforced in the GMC's Outcomes for Graduates 2018 report. [14] Legal impetus The legal incentives for promoting decolonisation of the medical curriculum are inextricably tied up with the risks of litigation and the relevant laws in statute, case law, as well as international legal instruments. The European Convention on Human Rights [15] has a number of provisions which protect an individual's rights in a healthcare setting. In particular Articles 2, 3, 8, 12 and 14 enable patients to "make choices in line with their own opinions and values -even if those choices seem irrational, illadvised or rash to others." [16] However, in addition to the above Convention rights, the Article 9 right to freedom of thought, conscience and religion, is also pertinent because some modalities of treatment, from the perspective of the patient, may also have a significant spiritual/religious component. Examples of these include the many international cases relating to Jehovah's Witnesses refusing blood transfusions [17]. It follows that restricting indigenous modalities of treatments, which have religious/spiritual elements, may also breach the patient's Article 9 rights. In Montgomery v Lanarkshire Health Board [2015] the supreme court said that a patient's decision on whether or not to incur a risk regarding their care "does not depend exclusively on medical considerations" and the "patient is entitled to take into account her own values, her own assessment of the comparative merits." Further the "relative importance attached by patients to quality as against length of life, or to physical appearance or bodily integrity as against the relief of pain, will vary from one patient to another." This is because many reasons "may affect their attitude towards a proposed form of treatment and the reasonable alternatives. The doctor cannot form an objective,'medical' view of these matters, and is therefore not in a position to take the 'right' decision as a matter of clinical judgment." [18] A possible explanation for the doctors not being able to attach a similar level of significance as patients do to issues of relevance to them, may be because such a decision does not fit with doctors' paradigm of medicine or health. [19] If doctors are unable to act in a person-centred manner, which, in diverse populations would entail being culturally competent by exercising cultural humility (Figure 1), it may leave their actions falling short of required standard of duty of care and therefore open to litigation. Therefore as Lord Kerr and Lord Reed in the Supreme Court put it, it is "necessary to impose legal obligations, so that even those doctors who have less skill or inclination for communication, or who are more hurried, are obliged to pause and engage in the discussion which the law requires." [18] Furthermore, Outcomes for Graduates 2018 specifies that graduates "must demonstrate knowledge of the principles of the legal framework" as well as "the importance of the links between pathophysiological, psychological, spiritual, religious, social and cultural factors for each individual." [14] Decolonisation and medical education --- Recent decolonial public engagement events University College London (UCL) Medical School In effort to improve curricula delivery, the UCL Medical School Clinical and Professional Practice team held two public engagement events. The first event, 'Practically Creating an Inclusive Curriculum', 2017 [20], received feedback that there should be an exploration of how to decolonise the medical curriculum. This in turn inspired a second public engagement event in 2018 named 'Decolonising the Medical Curriculum' [21] which focused on: decolonising the body; decolonising the curriculum; decolonising learner experience; decolonising educational space; decolonising professional behaviours; and decolonising ideas of healing. The event broadly covered the areas of: identifying symptoms and signs in varying skin tones in the context of over diagnosis/underdiagnosis; feminist pedagogy describing sessions using patient narratives e.g. session on violence against women; mapping the diversity curriculum; the attainment gap and students experience of racism; the UCL eugenics enquiry; anthropology as a vehicle of improving professional behaviours; biomedicine's colonisation of indigenous healing systems. [21] Wellcome Collection The 'Ayurvedic Man: Encounters with Indian Medicine' accompanied an associated 'Decolonising Health symposium' at the Wellcome Collection, in 2018. [22] The exhibition explored Ayurvedic medicine in South Asia through its museum antiquities collection. The exhibition asked how medicine has been transformed by biomedical and cultural exchanges and what the impact of colonialism has been on medical heritage. As the Wellcome says: "we are at a time of growing popularity of pluralistic approaches to health, societies around the world are at increasing risk of losing natural resources, medicinal plants and traditional knowledge. Ayurvedic Man highlighted the delicate balance between sharing and protecting heritage, cultural resources and environmental biodiversity" [23]. --- UCL Anthropology Department -Flourishing Diversity The 'Flourishing Diversity Series' [24] were a series of high-profile events in September 2019 which promoted actively listening to the voices, stories and wisdom of the representatives from remote indigenous communities around the world gathered in London. These listening encounters promoted key themes about sustainability, climate change, indigenous land rights, indigenous ideas of healing, and collective sharing of indigenous knowledge. The messages from these public engagement events attracted the attention of major news outlets such as the BBC World Service [25] and TIME magazine [26]. --- SOAS University of London The --- Learning from active listening to decolonise medical education Following the active listening of the above events, we have distilled concepts that may have been omitted from conventional top-down organisational equality and diversity policy in the context of medical education. These concepts may feel uncomfortable, however, the emerging 'grass-roots' interest in decolonisation is a liminal space for educational transformation. In this article, we have further explored areas specific to the concept of healing in the individual and in the population. --- Figure 2. One man in traditional religious clothing. Credit: Wellcome Collection [28]. --- Medical Pluralism The tension between western and traditional healing systems The image (Figure 2) used to publicise the Wellcome [23] exhibition and associated symposium, of two men meeting -one in western dress and the other in traditional Indian attire with a halo around his head -stimulated some thoughts on definitions of the term 'enlightenment' and its different meanings in Western and Eastern contexts. In the East (India, Sri Lanka, China etc), traditional healing ideology about overcoming illness through indigenous medicine (e.g. Ayurveda, Traditional Chinese Medicine etc) can be viewed, on a superficial level, as symptom relief/cure, but their philosophies have a deeper endpoint of paving the road to 'enlightenment' of mind, body, and spirit, through attempting to create ecological balance by working with nature. In the West, the '18 th -century Enlightenment' refers to the overturning of superstition with science and logic; in European colonies this contributed to the subjugation of indigenous medicine in favour of Western health practices. This meant that Eastern practices, such as Ayurveda, Traditional Chinese Medicine and other indigenous healing systems, were regarded through the sceptical lens of a Western medical perspective, and their efficacy rejected. The exertion of colonialism led to Europeans establishing their own biological, structural and cultural superiority and their own epistemology on the narratives of healing as the only legitimate knowledge base of medicine in the countries they colonised. [29] However, the unique context of decolonisation of healing practices means that the process depends not only on dismantling colonial epistemologies and aspects of patriarchal institutions, but also on simultaneously re-centring the displaced indigeneity. Eastern ideas of healing, as with many other indigenous healing systems, encompass a range of multiple outcomes of mind, body and spirit on receiving nuanced individualised treatments. Such outcomes are awkward to capture with standard epidemiological mathematics and 'one-size-fitsall' study design (which prefer to look at one primary endpoint). Hence it is challenging to develop an evidence base according to Western criteria. However, as conversations relating to decolonisation in healthcare circulate and contribute to ongoing health activism, it will be interesting to see what will unfold and what patients want from healthcare systems. The decolonising agenda to're-centre' displaced indigenous healing systems will be challenging for those in biomedicine due to the paucity of epidemiological evidence. However, equality and diversity training in professionalism within education, could have an anthropological component that includes the concept of 'cultural humility' (Figure 1), where, in this case, there could be awareness that Western medical practice is just one of many global healing systems. This could throw light on biomedical unconscious bias. However, it should be noted that the Outcomes for Graduates 2018 says that: graduates should "respect patient choices about the use of complementary therapies, and have a working knowledge of the existence and range of these therapies, why patients use them, and how this might affect the safety of other types of treatment that patients receive". [14] Often the unconscious bias is based on the narrative of the assumed superiority of Western medicine based on evidence-based research. However, reviews of the evidence base of medicine guidelines, such as Wright et al, say that "Practice guidelines by specialist societies are surprisingly deficient" and indicate that only a minority of medical guidelines are based on grade A evidence and the majority of medicine practice is not based on evidence (shaped as best practice) or lesser grades of evidence. [30], [31] Analyses of evidence levels in obstetrics and gynaecology and have found patterns of predominantly low grades of evidence for the Royal College of Obstetricians and Gynaecologists guidelines [32] and the Society of Obstetricians and Gynaecologists of Canada guidelines [33]. Subsequently, it could be postulated that biomedicine may be more belief based than generally presumed as explored in Charles et al's paper 'The evidence-based medicine model of clinical practice: scientific teaching or belief-based preaching?'. [34] Given that both western and indigenous healing systems have deficits of high-quality scientific evidence, there could be a case for looking beyond the western vs. traditional antagonism, in order to explore opportunities for synergy and co-delivery in a medical pluralistic, person-centred approaches to healthcare. --- Transgenerational trauma and indigenous ideas of healing Demographic studies show ethnic differences in diagnosis of psychiatric disease. [35] However, attention needs to be paid to research into inter-generational trauma originating from the legacy of slavery and subjugation of an indigenous way of life. [36] Decoloniality academics have proposed that those affected should be approached by being both respectful of the indigenous system of beliefs and healing systems of their ancestors as well as making available standard psychological therapies. Publications in humanities such as Fay's in 'Decolonising mental health one prejudice at a time: psychological, sociological, ecological and cultural considerations' [37], Davis's PhD thesis [38], Quijano's 'Coloniality and Modernity/Rationality' [29] and Linklaters' 'Decolonizing Trauma Work: Indigenous Stories and Strategies' [39] have highlighted the inadequacies of current psychological and psychiatric practices in dealing with this reported phenomena. The Office of Hawaiian Affairs is presently backing a Bill promoting native Hawaiian concepts of well-being, culturally-grounded health methodologies and traditional healing and health practices. [40] This was a governmental response to findings that colonised style of western healthcare did not meet the mental health needs of indigenous people in Hawaii. The whitepaper to the Bill says that the "health programs and services that are aligned with Native Hawaiian cultural identity, values and beliefs can significantly increase the number of Native Hawaiians who utilize mental health services." [41] An underlying factor relevant may be the stigma attached to psychiatric disorders which are seen differently from an indigenous treatment perspective. For instance, psychiatric conditions are commonly seen in indigenous communities as an invasion of an external entity which can be remedied in rituals or looking at psychiatric conditions as a symptom of disharmony in the community as opposed to the individual. Another perspective might be that psychological conditions are a result of westernisation and labelling of conditions. It is probably due to the above reasons that the Diagnostic and Statistical Manual of Mental Disorders (DSM-5) incorporates a greater cultural sensitivity in its latest update. See also Youngs and Nadeau's article on Decolonising the Body: Restoring Sacred Vitality which describes their experience in setting up a program in Canada to help native women who have suffered sexual, racist and colonial violence, makes an interesting contribution in this area. [42] Molecular medicine has also started to accrue epigenetic evidence [43] that suggest the transgenerational transmission of DNA methylation changes from parents to children derived from the experience of trauma. It follows that there could be cross linkage between biomedicine and humanities and to act synergistically to explore new conversations on inter-generational trauma and its treatment. Furthermore, we introduce these ideas into this paper within the context of decolonising medical education and propose inclusion of this aspect of cultural psychiatry within the intersect between the Mental Health and Social Determinants of Health syllabus. An ideal educational vehicle for creating professional awareness and cultural sensitivity around indigenous systems of healing would be for all medical students to have a basic knowledge of medical anthropology. For instance, the UCL Medical School curriculum gives students the opportunity to take a special study component which is an anthropological project looking at mental health from an indigenous perspective in a 'Culture, ethics, and religion in the clinical encounter' mini course. The course explores some of the complexities of clinical practice with culturally diverse patients for a selection of students. Medical schools in general could develop medical anthropology as part of the core curriculum, as a way of decolonising the medical curriculum and improving cultural competence and a cultural humility in professional behaviours (Figure 1). This viewpoint was expressed by a student participant in the Decolonising the Medical Curriculum event in terms of culture in the clinical encounter and her essay and talk can be viewed at the project's website. [44] Elements of a modern medical curriculum It is invariably difficult to define the areas that should be included in regard to decolonisation within the medical curriculum. However, in addition to the areas discussed above, the following areas have been highlighted as important by the public who attended the prior mentioned decolonising public engagement events. Due the nature of public engagement events, the public's contribution identified through active listing to the public's voice, does not at times emerge in a logical and structured way. --- Decolonising global health and research At the Wellcome Decolonising Health symposium [22], Branwyn Poleykett, anthropologist and postdoctoral research associate in the Centre for Research in the Arts, Social Sciences and Humanities, University of Cambridge pointed out that decoloniality is rising as a challenge to the average biomedical view of global health and biomedical research funding. She referred to Richard Smith (who was the editor of the BMJ until 2004) who said in his BMJ blog that there was a popular way of talking about developments in global health circles at the moment, which said that the global health '3.0' model is being currently upgraded to the more modern global health '4.0' model [45]. But Poleykett, reasoned that there are decolonising arguments that say that if we are to use a 'computer upgrade' type terminology for global health development, then it should be acknowledged that global health '1.0' was colonialism (an epoch of discrimination) and that global health '2.0' was post-colonial international health. The global health '2.0' model had contained many unresolved aspects of discrimination, inequality and power imbalances between former colonising nations and those colonised. From her anthropological viewpoint, the legacy of models '1.0' and '2.0' are still eroding and undermining the ideologies of the global health '3.0' and the aspirations of version '4.0'. Thus, Global Health '4.0' is still riddled with the legacy of colonisation and not a completely virtuous crusade for health improvement. Hence this complexity needs to be incorporated from a decoloniality perspective into medical syllabus of global health education. Poleykett's arguments at this event were relevant as the Wellcome Trust is a large funder of global health research and the Wellcome Collection is a large repository of colonial antiquities related to health and medicine. Also circling back to the arguments of global health and the power imbalances of the able and disabled, Meekosha's article on 'Decolonising disability: thinking and acting globally' [46] highlights the lack of global health focus on the challenges that impaired peoples face from a multiplicity of phenomena including: war and civil strife, nuclear testing, the growth of the arms trade, the export of pollution to 'pollution havens' and the emergence of sweatshops. Poleykett also pointed out that research methodology is often accused of being dominated by western perspectives of looking at the individual in isolation from other elements such as mind, spirit or nature, which in turn distorts the outcome of the research as well as research being dictated by funders objectives which may not have a direct public health benefit. [47] Scholars attending the SOAS decolonising event [27] drew attention to global geographic disparities in knowledge production as seen in Pan et al's global map within their publication 'World citation and collaboration networks: uncovering the role of geography in science' where by authors from the global south are grossly underrepresented. [48] Consequently, funders of global health research should have the cultural competence and decolonising knowledge to deal with this complex area, which is fraught with power imbalances inherent between research centres of the global north and global south. Imperial College London is leading the way in examining these biases in their educational development unit as seen by their work described on the website 'Examining geographic bias in our curricula'. [49] Ecology and public health The terminology around ecology and human health is developing and is discussed in Buse et al's paper on 'Public health guide to field developments linking ecosystems, environments and health in the Anthropocene'. The Anthropocene is defined as the new geological epoch of humanity's own making based on evidence of damaging human influence on the biosphere. This paper describes seven emerging fields, which are: occupational and environmental health; political ecology of health; environmental justice; ecohealth; One Health; ecological public health; and planetary health. [50] Planetary heath [51] is the most recent development in this area, and it is significant that the Lancet has a standalone journal dedicated to the topic titled the 'Lancet Planetary Health'. This shines the light of importance on the public events such as the UCL 'Flourishing Diversity Series' (mentioned earlier) in which displaced indigenous ideas were re-centred and foregrounded to highlight the importance of human health within planetary ecology. Rebekah Jaung in her article on how decolonising health could save the planet says: "Indigenous people have always had ecological perspectives on health, which have only recently entered'mainstream' health discourse. The scope now is planetary healthapproaches which benefit all people and the natural environment. Ideas we have learned from indigenous people include seeing climate breakdown as a symptom of non-reciprocal and exploitative relationships with land and acknowledging that such a relationship exists. Ways of honouring the land will not only restore it, it will lead to good health for the people who live on it. This is not just a nice sentiment but the approach on which cutting edge thinking on global climate action is structured." [52] Presently there is a social tide of interest in climate change and health as popularised by the extinction protests seen in major cities and activists such as Greta Thunberg. These issues are likely to challenge the present and future generations of doctors. It is therefore important to incorporate non-biased critical thinking about these issues on global health and ecological public health and raise awareness of the emerging field of planetary health within the medical education curriculum. [53] Cultural appropriation Cultural appropriation occurs when members of one culture (often a historically dominant culture) adopts or misappropriates elements of another culture (often a historically disadvantaged culture). In the context of health, some biomedical therapeutic discoveries have an indigenous knowledge bases. [54], [55] Maridass & De Britto in 'Origins of Plant Derived Medicines' [54], say that the discovery of some new medicines are credited to traditional approaches which in some cases were based on "trial and error over many years in different cultures and systems of medicine". They go on to say that natural products have been the source of many drugs such as quinine, morphine, paclitaxel camptothecin, etoposide, mevastatin, and artemisinin. [54] Similar accusations of cultural appropriation are also seen in respect of meditation, acupuncture, yoga and other indigenous modalities of treatment which are used and re-branded in the west. [56], [57] The fact that this information is not conveyed in medical education exemplifies an aspect of medical cultural appropriation, which involves indigenous cultural knowledge not being acknowledged and respected by a historically dominant group. --- Intersectionality Intersectionality [58], [59] is used to describe the multiple 'interlocking' or'simultaneous' oppressions of race, class, gender, sexuality and disability. It was originally coined by Kimberlé Crenshaw, a feminist, in describing the experiences of women of colour. Social activism has highlighted the health inequalities and higher mortality rates for black women who are disadvantaged by their race and their gender. These health disparities are noted in both the UK and the USA in routine national data sets [60]- [63]. Indeed, a new report from MBRRACE-UK states that "it is striking... we seem to be making little impact. Research is urgently needed to understand why black women are five times more likely and Asian women twice as likely to die compared to white women" [63]. We recommend that the curriculum should utilise the term 'intersectionality' as part of the social determinants of health, in order to teach cultural competency, in a way that integrates self, social and global awareness as demonstrated at Columbia University, USA, and described by Cushman et al. [64] They recognised the association between the power and privilege with intersectional identities, which often go unrecognised and unacknowledged by the individuals and groups that hold them and ultimately can affect professional behaviours towards patients and the public. Also using the voices of marginalised individuals or groups who explain their intersectional experiences within teaching sessions would fit in with the concept of social accountability as defined by the World Health Organisation's report that there is an "obligation for medical schools to direct their education, research and service activities towards addressing the priority health concerns of the community, region and/or nation they have a mandate to serve" [65]. In this way marginalised groups experiencing simultaneous oppressions can act as patient educators and become future changemakers finding decolonising solutions for this aspect of inequality. [66], [67] Decolonising symptoms, signs and investigations Decolonising diagnosis of disease requires an awareness of studies that have shown ethnic differences in routine chemistry, fertility, endocrine, cancer, and haematological markers, as well as in vitamins and carotenoids, in children, adolescents and adults. This variation in markers is extensively discussed by Tahmasebi et al in 'Influence of ethnicity on population reference values for biochemical markers' which has a useful Table on the available literature evidence for ethnic differences in biomarker levels. [68] It follows that if white European biochemical normal values are used by laboratories and clinicians with diverse populations, then BAME populations are likely to be placed at an increased risk of misdiagnosis. This in turn results in BAME patients being either treated for conditions they don't have or not receiving the treatment in a timely manner and therefore affect their ability to heal. For instance, if medical students and doctors are not trained in skin diagnosis in darker skin tones, then delays in diagnosis or misdiagnosis in detecting conditions such as skin cancer [69] or cyanosis become more likely. Indeed, the lack of teaching on the appearance of cyanosis in darker skin tones in global north medical schools could feasibly contribute to a BAME patient intersectional excess of mortality. --- Decolonising reflective practice In the UK, the Academy of Medical Royal Colleges and COPMeD define reflective practice as "the process whereby an individual thinks analytically about anything relating to their professional practice with the intention of gaining insight and using the lessons learned to maintain good practice or make improvements where possible". [70] However, we suggest that it would be prudent to incorporate the concept of 'cultural safety' which arose originally as a decolonising nursing concept from New Zealand to address power imbalances between healthcare practitioners and indigenous New Zealanders. The process of cultural safety would essentially entail that a nurse, doctor or student, as part their reflective practice prior to a patient encounter, quickly checking (reflect) on their privilege status and any potential power imbalances between themselves and the patient. This quick reflection should give insight and situational awareness, aspiring to create a fairer clinical encounter. [71] Cultural safety could also be incorporated into human factors training to improve patient safety. Human Factors training is defined as "Enhancing clinical performance through an understanding of the effects of teamwork, tasks, equipment, workspace, culture and organisation on human behaviour and abilities and application of that knowledge in clinical settings". [72]- [74] This type of reflective practice could be developed within the medical reflective practice syllabus as a healthcare human rights checklist, thus improving professionalism [75]. --- Conclusion Our article demonstrates that there is guidance from professional regulatory bodies to improve equality, diversity and inclusion as well as legal imperatives to improve person centred care which links to our recommendation that it is important for healthcare educators, health professionals, and researchers/funders to adopt a 'decolonising attitude' and consider the areas that are highlighted in this paper. A system of education that encourages doctors to recognise and challenge their own implicit biases and cultivate an attitude of open-minded professionalism, is vital to address the complex power imbalances of the legacy of colonisation within health care. Decolonisation includes the tolerance of medical pluralism. We can learn from patients as educators and through public engagement with active listening. We can acknowledge the emerging field of planetary health, which stretches healthcare beyond the individual, public health and global health, and which indigenous peoples have proposed regarding living harmoniously with the planet, pre-colonisation. This is pertinent in light of the growing phenomenon of human migration as consequence of a number of factors including economic opportunities, economic crisis, conflict, and natural disasters stemming from climate change.

The legacy of colonial rule has permeated into all aspects of life and contributed to healthcare inequity. In response to the increased interest in social justice, medical educators are thinking of ways to decolonise education and produce doctors who can meet the complex needs of diverse populations. This paper aims to explore decolonising ideas of healing within medical education following recent events including the University College London Medical School's Decolonising the Medical Curriculum public engagement event, the Wellcome Collection's Ayurvedic Man: Encounters with Indian Medicine exhibition, and its symposium on Decolonising Health, SOAS

practice syllabus as a healthcare human rights checklist, thus improving professionalism [75]. --- Conclusion Our article demonstrates that there is guidance from professional regulatory bodies to improve equality, diversity and inclusion as well as legal imperatives to improve person centred care which links to our recommendation that it is important for healthcare educators, health professionals, and researchers/funders to adopt a 'decolonising attitude' and consider the areas that are highlighted in this paper. A system of education that encourages doctors to recognise and challenge their own implicit biases and cultivate an attitude of open-minded professionalism, is vital to address the complex power imbalances of the legacy of colonisation within health care. Decolonisation includes the tolerance of medical pluralism. We can learn from patients as educators and through public engagement with active listening. We can acknowledge the emerging field of planetary health, which stretches healthcare beyond the individual, public health and global health, and which indigenous peoples have proposed regarding living harmoniously with the planet, pre-colonisation. This is pertinent in light of the growing phenomenon of human migration as consequence of a number of factors including economic opportunities, economic crisis, conflict, and natural disasters stemming from climate change.

The legacy of colonial rule has permeated into all aspects of life and contributed to healthcare inequity. In response to the increased interest in social justice, medical educators are thinking of ways to decolonise education and produce doctors who can meet the complex needs of diverse populations. This paper aims to explore decolonising ideas of healing within medical education following recent events including the University College London Medical School's Decolonising the Medical Curriculum public engagement event, the Wellcome Collection's Ayurvedic Man: Encounters with Indian Medicine exhibition, and its symposium on Decolonising Health, SOAS

Introduction Transgender is a general term that refers to people with a variety of identities. In the broadest sense it refers to people who do not adhere to the cultural definitions of gender [1]. This term encompasses transsexual men and women, or those that have altered their genitals to match the gender they choose to express, but it also includes transgender men and women who express a gender identity other than the one assigned at birth but do not choose to have genital alteration surgery [2]. Both transgender and transsexual people may use hormones, clothing, makeup, wigs and/or hairstyles, and other types of surgery (such as facial feminization surgery or mastectomies) in their presentation of gender. Transgender also may refer to drag kings and queens, women and men who dress as the opposite gender for the purposes of performance, or people who identify as genderqueer-people who choose to express both or neither gender in their presentation. Transgender women experience depression, suicidal ideation, and suicide attempts at rates much higher than in the general population: estimates of the lifetime prevalence of depression in transgender women have been reported as high as 62% [3], while the lifetime depression rate for the general United States population is 16.6% [4]. The high rate of depression is not surprising given the discrimination transgender women face. However, it is important to understand the factors influencing depression in this population as well as how depression influences other factors affecting transgender women. Such an understanding will aid in the development of effective modes of treatment for depression in transgender women. This paper reviews the burgeoning literature on depression in transgender women. The author searched the databases PSYCINFO, MEDLINE, and CINAHL using the keywords transgender and depression to find articles and also culled citations within the articles found for other relevant papers. The majority of research has focused on the factors influencing depression and suicidal ideation (Table 1). Second, theoretical constructs for the high rates of depression in this population are discussed. Finally, three known studies have included depression as a variable that influences health behaviors in this population; findings of these studies are presented. The paper concludes with implications for clinicians working with transgender women. All articles that included transgender women in the sample pool and discussed relationships between depression and other variables were included. A total of fourteen articles were found and are discussed within. Most of these studies either did not mention how they chose their sample of transgender women [13,15] or allowed women to selfidentify as transgender [2,3,5,9,10,12,16,17]. Budge and colleagues [11] included only transgender people (men and women) who identified on what they called a "binary gender identity, " in other words, participants who used to identify as one gender and now identified as the other gender. This has the potential of excluding many individuals who consider themselves transgender but are along a spectrum of transition. In a series of studies, Nuttbrock and colleagues [7,8,14] used a definition for inclusion that a participant must have been assigned the sex of male at birth but did not consider themselves "completely male" in all areas of life. This allows much more inclusion for people who may be early in transition or for people who are living as a female in only certain social spheres or times. Study recruitment methods varied, but only two studies [12,15] were recruited among high-risk groups of sex workers. The other studies recruited via venues that are not likely to yield only high-risk participants such as a conference [9] and online [2,11]. The majority of the other studies recruited as widely as possible, targeting community groups (some of which do cater to HIV-positive individuals) but also using respondent-driven sampling to reach transgender females who would not be reached otherwise. --- Factors Predicting Depression Variables that are proposed to predict depression in transgender women include both interpersonal and intrapersonal constructs. The author has discussed variables that do not show a relationship with depression in the sample where the lack of relationship was indicated in the studies. Social support is a measure that examines the availability of people in one's life who provide emotional and mental resources for coping. Often this is broken into categories such as family support and peer (or friend) support. Family social support is often less available for transgender women as families react to the transition and may reject the transitioning family member [15,18]. Peer support varies, as transition might lead to rejection by a women's pretransition peer group but transition often leads to acceptance by other transgender women, particularly in urban areas [19]. Social support does seem to reduce the risk of depression in transgender women. Several studies have found a direct relationship of family and peer support associated with better mental health [2,11,15]. Nemoto et al. [12] suggest that it is not just the presence of social support in the past month (from family and both transgender and nontransgender friends) but rather the satisfaction with this support that is protective against depression. Though it is not conclusive, peer support may not be as important as family support in predicting depression [2]. However, this study measured family and peer support differently; family support was assessed via items regarding perceived supportiveness of parents, siblings, partners, and children while peer support was assessed via measures of the proportion of time spent with other transgender people and perceived feelings of isolation in being transgender. The inequity of these measures is problematic in terms of comparing these measures. Another study suggests that the influence of social support, at least superficially in terms of accepting versus rejecting the transgender woman's gender identity, may vary based on life stage. More specifically, acceptance/rejection by parents and siblings matters until middle age, when acceptance/rejection by a sexual partner matters from young adulthood [14]. Violence. There are three categories of violence often discussed in the literature of transgender women: physical violence, sexual violence, and verbal harassment (sometimes labeled as stigma or discrimination). Transgender women are at risk of violence, including physical violence and discrimination, at rates between 2 and 3 times those of people who are not transgender women [20]. Violence against transgender women can damage the woman's sense of self and being, further intensifying the damage [8]. Studies that examined effects of physical and verbal violence in aggregate indicated that violence against transgender women is associated with increased depression [8] and may be a predictor of depression [7]. Physical and verbal abuse may have different impacts on victims than sexual abuse. Testa and colleagues [10] found that physical violence was associated with suicidal ideation and attempts, while sexual violence was associated with increased substance use. A study examining sexual violence found an association between sexual partner violence and depression [5], indicating that the relationship between the transgender woman and the perpetrator of the violence may be an important factor. Nemoto et al. [12] found that despite high rates of physical and sexual violence (half of participants reported victimization), only verbal violence predicted depression in a multiple regression model. Bockting and colleagues [2] found similar results. These results contrast with an earlier study that found that physical and sexual violence, but not verbal violence, predicted attempted suicide [6]. Perhaps the dynamics behind depression are different than the ones behind suicidal attempts. Sex Work. Many transgender women engage in sex work. Often it is the only viable source of income due to workplace discrimination and it is certainly the most lucrative form of income for transgender women who need to finance expensive surgeries and/or hormone treatments [21,22]. Sex work is very common in this population: as many as 67% of transgender women between 15 and 24 have engaged in sex work at some point in their lives [23]. The relationship, if any, between sex work and depression is unclear. Several studies show no relationship between sex work and depression [5,7,12]. Yet several factors leading to depression, such as a lack of social support and a lower education status, are associated with engagement in sex work as well [23]. In addition, transgender sex workers have stated during interviews and focus group-based studies that they engage in drug use to escape the stress of sex work [21,24], indicating that mental distress is indeed occurring. Though interviews with transgender sex workers indicate that high rates of drug use among transgender women are due to an attempt to deal with mental stress and depression [21], no quantitative studies to date have examined whether drug use is associated with depression in transgender women. Gender Identity. The very term transgender implies movement, a crossing (trans) of gender identities. This crossing is commonly called transition and takes place over a span of time that may last for several years. The process is affected by the social, cultural, developmental, and individual factors and economic and social resources of the person transitioning. Part of this transition entails resolving issues of conflict regarding gender identity, often by altering one's behavior and presentation to better fit a gender identity that did not mesh with one's gender at birth [15]. Gender identity is a complex, multidimensional concept and researchers have conceptualized it in many ways. Bockting and colleagues [2] examined several aspects of gender identity, including gender dysphoria, investment in being perceived as their gender of expression, and openness and pride about being transgender. Of these, only transgender pride and openness (feeling that being transgender made her special, and that she was comfortable revealing that she was transgender to others) was negatively associated with mental distress. In a study of the association between feelings about being transgender and psychological distress (anxiety, depression, and somatization), Sánchez and Vilain [9] found that fear about how being transgender would affect life was associated with psychological distress, but other thoughts and feelings about being transgender were not predictive of distress. Budge et al. [11] examined the feelings of loss a transgender person has regarding aspects of life that have deteriorated since transitioning, including job, housing, family relations, friendship, and finances. Though this variable was not predictive of depression in their final structural equation model, they did find that transgender women experienced more loss than did transgender men, indicating that transgender women may experience more discrimination than do their male counterparts. Sociodemographic Variables. Most studies included demographic variables that moderated the main variables of study. Many of the results assessing these variables are mixed. Age is a common demographic control; several studies indicate that being younger is associated with more depression [2,6,7], while still other studies found no association between depression and age [8,12]. A younger transgender woman may not have developed the capacities to cope with the stressors of life as transgender or may be more entrenched in the stressful transition process, leading to greater depression among younger women in some studies. The majority of studies assessing it found no association between race/ethnicity and depressive symptoms [6,12] but Nuttbrock, and colleagues [8] found that Black transgender women had lower rates of clinical depression than did other racial/ethnic groups, indicating that depressive symptoms and clinical depression may have different correlates. Less education predicted more depression in two studies [2,5] but was not correlated in a third study [6]. At first glance education is a proxy for income, but the relationship between education and income does not only hold for transgender populations due to widespread discrimination. Indeed, income appears to be unrelated to depression [2,5,11]. However, unemployment was a risk factor for depression in a sample of transgender women in Ontario, Canada [13]; perhaps a lack of steady income is a stressor regardless of the amount one's income is in general. --- Theoretical Constructs Explaining Transgender Women's Depression Researchers studying depression in transgender women have utilized several sociopsychological theories to account for the choices of variables to study. These constructs are presented next, roughly in order from the most to the least researched as of this point. Minority Stress. This concept is derived from research literature into mental health of lesbians and gay men [25,26], which in turn was based on similar theories for other minority groups. In our context, the definition of minority stress is that the discrimination and stigmatization transgender people face lead to stress, which in turn leads to depression and other negative mental health outcomes. It seems clear that discrimination, both actual in the form of verbal, physical, and/or sexual violence and perceived stigma, does lead to depression and suicide attempts [2, 5- Conversely, positive feelings about being transgender are associated with less depression [9]. Transgender discrimination differs from other forms of discrimination in several ways. Transgender women may find themselves facing rejection from the family of origin due to their transition [15,18], while a person of a given race/ethnicity will continue to have acceptance from their family on the basis of homogeneity alone. Second, though exact numbers are unknown for several reasons (including reluctance of transgender people to report being victims of violence and a lack of identification of transgender status in police data), transgender women face high rates of violent discrimination, leading to assault, sexual violence, and murder [27]. Coping is the process by which people use various psychological and behavioral mechanisms to manage reactions to stress. Coping mechanisms are generally divided into two categories: avoidant and facilitative coping. Avoidant coping is considered an unhealthy way of dealing with stress, as the person minimizes or avoids thinking about the problem via cognition, distraction, overeating, or substance use, which can lead to more stress or additional health problems. Facilitative coping, or addressing the problem via support of others, learning new skills, or changing behavior is considered healthier, as the problem is resolved in a positive manner via social support, changing behavior, or learning new skills [9,11,28]. Greater use of facilitative coping would therefore be predicted to be protective against depression while avoidant coping would place one at risk of depression. Studies of transgender women have indicated that approaches indicative of avoidant coping, such as substance use, were predictive of depression [6], while facilitative coping mechanisms like utilizing social support led to less depression [11,13,15]. Avoidant and facilitative coping may interact in terms of their role in predicting depression, as Budge and others [11] found that avoidant coping mediates the relationship between social support and depression. In addition, depressed individuals may reject social support, further deepening their negative mood state [12]. While these studies do lend support to a model of avoidant and facilitative coping affecting mental effect, most studies are examining only a small sample of the variations in each type of coping. Avoidant coping is more than substance use and facilitative coping is more than calling on social support. To date no studies have examined the effects of avoidant and facilitative coping in totality. Identity control theory originated in social psychology and centers on the balance between a person's identity and the labels and expectations of others in the person's life. According to this theory, a person compares her identity against the perceptions of her from others in her life. If there is a discrepancy between the identity and the perceptions then the person changes her behavior to bring these into alignment [29]. A transgender woman would be balancing her identity as female with the perceived attitudes of friends and family (among others), which would be expressed as rejection or support of her as a transgender woman. Rejection of her identity would be associated with greater risk of depression. Only one research team to date has examined identity control theory in the context of depression among transgender women. Following a preliminary study indicating that both friend and family supports were protective against depression [15], Nuttbrock and colleagues [14] examined conflict and affirmation reactions to disclosures of gender identity among parents, siblings, classmates, friends, coworkers, and sex partners of transgender women and whether these reactions influenced depression. They found that conflict was a risk factor for depression and affirmation was protective against depression, supporting identity control theory and indicating the importance of a transgender woman's social network in her mental health. --- Depression as a Factor Predicting Health-Related Variables Due to extremely high rates of HIV among transgender women (approximately 27% of transgender women are HIV positive, according to Herbst et al. [30]), researchers have examined factors contributing to higher risk of HIV or of behaviors that put one at risk for HIV, such as unprotected receptive anal intercourse (URAI). Three known studies have included depression as a predicting variable (Table 2). It is unclear whether depression contributes to HIV, as Garofalo et al. [16] found no association between depression and HIV while Nuttbrock et al. [7] found that depressive symptoms increased the risk of contracting HIV or other sexually transmitted infections. Similarly, results indicating a relationship between URAI and depression are also mixed, with Nuttbrock et al. [7] reporting a positive association between URAI and depression and Operario and Nemoto [17] reporting no association (though they did report an association between suicide attempts and URAI). --- Conclusions and Implications Though the literature regarding depression in transgender women is still sparse, a few key findings do emerge. It is clear that interactions with other people, whether positive in terms of social support or negative in terms of rejection or violence, are important factors predicting depression in transgender women. Second, the very process of transitioning is stressful and this stress may contribute to depression Depression Research and Treatment in transgender women. Results regarding gender identity, social support across the lifespan, and identity control all support this assumption. Finally, the relationship between sex work and depression is not clearly understood. Given the relationship of sex work to other risk behaviors which lead to higher risk of HIV [19], and given the pervasive workplace discrimination suffered by transgender women which contributes to engagement in sex work, it is important to continue research into how sex work affects the mental (as well as physical) health of transgender women. The variables surrounding a transgender woman's physical and emotional health are complex. To date nobody has tested a model that attempts to describe the dynamic relationship between sexual risk behaviors, transgender identity and support, and mental health, though the author has proposed a preliminary model [31]. Still other studies are needed to fully elucidate the many relationships between depression and predictor variables in the population. Though these studies are ostensibly about depression in transgender women, only two of these studies [8,14] measured clinical depression among study populations. The other studies assessed prevalence of depressive symptoms, which may not be measuring clinical depression. As suggested by the different associations for Black transgender women for clinical depression, studies indicating associations of variables with depressive symptoms may not mean that clinical depression will be associated with these variables. Of the fourteen studies described here, eleven were cross sectional, which limits the conclusions that can be drawn regarding factors that precede depression because of both the lack of time as a factor and the possibility that third variables may be driving the variable relationships seen. The studies examining participants longitudinally [7,8,14] are important additions to the body of knowledge, as transitioning is a dynamic process that almost necessitates the inclusion of time when studying health within transgender populations. Future studies should consider a longitudinal study design while addressing the interaction of inter-and intrapersonal variables influencing depression in transgender women. Such studies will elucidate the variables affecting health in transgender women, providing indications to be used in creating effective interventions to reduce depression, drug use, and HIV among this population. --- Conflict of Interests The author declares that there is no conflict of interests regarding the publication of this paper. Submit your manuscripts at http://www.hindawi.com --- Stem Cells International Hindawi

Rates of depression are higher in transgender women than in the general population, warranting an understanding of the variables related to depression in this group. Results of the literature review of depression in transgender women reveal several variables influencing depression, including social support, violence, sex work, and gender identity. The theoretical constructs of minority stress, coping, and identity control theory are explored in terms of how they may predict depression in transgender women. Depression and depressive symptoms have been used to predict high-risk sexual behaviors with mixed results. The implications of the findings on treating depression in transgender women include taking into account the stress of transition and the importance of supportive peers and family. Future studies should explore a model of depression and high-risk behaviors in transgender women.

Interdisciplinary research conducted in the environmental field is often confronted with the complexity of functional processes, the absence of appropriate methods and the dissipation of skills and data in an area of study which is both vast and tackled by many different institutions. In addition, research that aims to contribute to sustainable development not only involves a systemic and interdisciplinary approach, but also requires the empowerment of civil society stakeholders in acquiring knowledge and learning to speak the same language. Indeed, the collective action advocated by this concept is not limited to the scientific realm alone. It also involves finding synergies with local management policies and users, and implies helping them in their procedure by designing methods and tools and making them available. The objective is to promote the enhancement of goods and services of a territory to maintain local populations and ensure their well-being. The methodology must therefore be part of an interactive process during which stakeholders with diverging interests progressively construct a shared representation of reality, give it meaning and set goals for themselves. For effective deliberation, these stakeholders must be able to access information and produce knowledge. Public awareness-raising and education are also sought, for individuals to become aware of their own role in the ambitious political project that sustainable development represents (Brodhag 2004). --- 1.2 In the first phase of our study, we examined the interactions between socio-economic and natural dynamics in Ushant Island (West Brittany, France) [1] by using companion modelling (ComMod 2005;D'Aquino et al 2001). Ushant Island is part of a regional natural park and was designated part of the Mer d'Iroise Biosphere Reserve in 1988. The international biosphere reserves network (UNESCO MAB program) has three main objectives: biodiversity conservation, sustainable development, and environmental education, training and involvment of the local population (Batisse 1990). --- 1.3 This small island (1541 ha), located off the western coast of France, has been experiencing landuse and landcover changes (Gourmelon et al 2001) for several years. Major land use changes resulted in fallow land encroachment and its consequences on landscape, traditional activities and biodiversity. --- 1.4 Within the companion modelling approach, a conceptual model of the "Ushant Island system" was constructed, thereby providing the basis for the development of a prototype using the CORMAS multiagent platform (CIRAD) (Bousquet and Le Page 2004; Bousquet et al 1998; Gourmelon et al 2008; Levrel et al 2009; Rouan et al 2010). It was used as the basis of several scenarios to explore trends in the environment and biodiversity, according to various socio-economic options (e.g. decline or increase of grazing, maintenance of coastal sward, etc.) (Kerbiriou et al 2009; Gourmelon et al 2008). This concrete application together with companion modelling enhanced interdisciplinarity (Wiesmann et al 2008). Concurrently, a role-playing game (RPG) run by the platform was initiated to improve the model and raise the stakeholders' awareness of the dynamics underway [2] (Charles et al 2008; Chlous-Ducharme et al 2008). In the second phase, the discussion turned to transferring knowledge to the civil society [3] by adapting research outputs to two operational contexts, in order to promote their mastering and adoption by two types of publics: multi-agent simulation and RPG as management support to a territory submitted to constraints (designed for management purposes), RPG to help raise awareness and help young people learn about sustainable development issues (designed for educational purposes). --- 1.5 This paper refers to the second aspect, devoted to transferring a RPG created in a co-modelling experiment in a multi-agent environment to an audience of schoolchildren. Objectives 2.1 Sustainable development education involves making links between disciplines and aims at providing students with an understanding of their surroundings and developing their abilities to get involved (Vergnolle Mainar 2008; 2009). Yet despite definite awareness [4] which occurred some thirty years ago, the French high-school system still favours disciplinary mechanisms with no actual contribution to a cross-disciplinary object (Tutiaux-Guillon 2009). In this context, the goal set by other organisations such as local authorities, scientific and technological culture centres, programmes including that of the UNESCO "Man and the Biosphere" as well as environmental protection groups is environmental education for youth, which involves knowledge transmission methods differing from those traditionally implemented by the Ministry of Education. 2.2 Among these transmission methods, a play-based approach would raise young people's awareness of the environmental processes taking place on several temporal and spatial scales and develop their ability to implement the sustainable development concept, in the same way Minassian and Rufat (2008) showed that certain video games had educational relevance in teaching geography. Since this is related to deliberations between diverse stakeholders, the game's objective is not just to promote the learning required for a dialogue that would not exclude those who most lack either knowledge or skills, but that would also allow new standards and values to be acquired. In this context, the game, for which Ushant Island was taken as a model, should be adapted to a broader geographic framework to raise public awareness of environmental issues and deliberation processes. Therefore, the proposed method must be flexible enough to be transposed to any coastal area; since the final objective is to ensure a transfer of tools which can be re-utilised in the educational realm without scientists and regardless of the initial application site to raise youth's awareness of sustainable development. The Educational Workshop "Between Land And Sea: Coasts" Familiarisation to the RGP for the educational team --- 3.1 The team in charge of this part of the study is the educational department of Océanopolis [5], its two relay advisers from the Ministry of Education (educational team) and three researchers (scientific team) involved since the beginning of the project. The first step for new participants (the educational team) is to become familiarised with the RPG and understand how it works. To this end, the whole team met several times to analyse the game proposed by the scientists, define how it could be incorporated into a broader educational project and provide any modifications required. After analysing a --- 3.3 The educational workshop is based on three work sessions which take place at Océanopolis under the supervision of an instructor: the RPG, the guided tour of Océanopolis' temperate pavilion and a "Summary outline" session. The study of a local case (Iroise Sea) provides an extension of the school curricula, while broadening their outlook. The issues could partly be tackled by finding information on the various exhibition displays in the temperate pavilion at Océanopolis. The summarising exercise can be provided as a sketch or outline, in compliance with the official guidelines of the Ministry of Education and with high-school final examinations (A-levels) in prospect. The content of this workshop ties in with: fifth-form geography courses ("coastalisation" of activities, study of a seafront, development of tourism and recreational activities, urban development in coastal areas, changing landscapes, protection schemes); curricula for civic, legal and social subjects in high school (citizenship classes [using role-playing games], decision-making, speaking to a group, collective action and debating); fifth-form life and earth sciences curriculum (teaching sustainable development, relationships between humans and animal species). Adaptation of the RPG for the Educational Workshop "Between Land And Sea: Coasts" --- 4.1 The RPG aims to improve schoolchildren's awareness and learning of sustainable development stakes. Originally developed for adults playing their own role and provided with knowledge about activities and the environment (Charles et al 2008; Chlous-Ducharme et al 2008), and towards a mediation goal, the RPG's adaptation to this new context required various modifications, mainly as regards the conceptual model, the game board and rules, and computer interface. --- 4.2 The initial game uses the knowledge acquired in the research project and the possibilities presented by the CORMAS platform (D 'Aquino et al 2002). The transfer of the prototype described in Rouan et al ( 2010) to a RPG involved simplification mainly regarding land, leading to abstract representation, and rules. Indeed, vegetation dynamics only were taken into consideration, as the game aimed at dealing with relationships between different human activities and environmental degradation, and correlatively, with the latter's impact on such activities. --- 4.3 Each session, about two hours long, included three phases: game presentation (goals and rules), game play with about ten turns simulating twenty years, and game analysis by all protagonists. The conceptual model --- 4.4 In order to elaborate this new RPG an adapted conceptual model has been built (Fig. 1). Figure 1. Conceptual model of the natural and social interaction in the new RPG 4.5 This model highlights the modifications that were required to adapt the RPG to meet its new target. A new class of vegetation has been added: the impassable scrub (blackthorn bushes). It was included in order to bring in a strong constraint as regards accessibility and visibility, and to generate potential collective decision-making. In the game, it is indeed impossible to clear this type of vegetation using the default means available to players. --- 4.6 Eight roles (out of nine initially) were retained: local elected representative, natural park representative, cattle farmer, restaurant owner, president of the hunting society, turf collector, second-home owner and permanent resident (sheep farmer). In the previous configuration, two second-home owners were represented but due to the relative passivity of this role, only one was maintained. --- 4.7 It is important to note that most of the roles involve several actions. For instance, the hunter, restaurant owner and the turf collector also keep sheep. The role of the visitor is played by the second-home owner who organises walks on the island. --- 4.8 A protected bird species that nests in coastal cliffs was also included in the game in order to introduce the concepts of biodiversity and conservation. Two pairs present at the beginning of the game may breed, persist or disappear depending on the evolution of the environment and of the players' activities. The position of nests is set randomly at the game's initialisation. The park representative is the only one to be aware of this species' status and possibly location. Its protection may be the subject of a collective decision or may not arouse any specific interest. The game board and rules 4.9 The game board is similar to a draughtboard with 100 cells representing 5,000 m 2 (Fig. 2). Each cell represents land including coastal sod, heathland, grassland, scrub and impassable scrub. A game board was specially created by an artist, taking account of the educational department's comments to make it more attractive for a young public: colours of board and squares, squares with holes for easier handling, and shape and appearance of the playing pieces. The content of the game cards has also been adapted (Fig. 2). Figure 2. Presentation of the board, roles and actions 4.10 The game rules are structured as follows (Fig. 2). 4.11 Gathered around a game board representing an area of land, eight players are given action cards (turf-stripping, mowing, clump ordering, etc.). The players' actions cause environmental evolution that may lead, for instance, to scrub encroachment on ungrazed plots, management of mowed plots, deterioration of the coastal area, etc. The computerised prototype is used at the end of each turn to determine the land changes induced by the players' actions. The scrub encroachment process on open land is voluntarily accelerated in order to create reactions, from the first turns, which may involve various individual or collective strategies to maintain (or not) an open area. Six to ten rounds can be played per session, simulating the evolution of the environment over a timeframe of 12 to 20 years. Computer application interface 4.12 A specific Graphical User Interface (GUI) was developed in collaboration with an IT consultant from Oxiane [6] in order to adapt it to the workshop and to make it easier to use for the facilitator. In the previous configuration, the prototype required three people to operate it: one acted as a game master, another transmitted the players' moves to a third person who was in charge of entering the data in a text file and running the model. Finally, the game board had to be updated with the results of the simulation. --- 4.13 The challenge was to design the application in order to make it usable by the game master alone. Therefore, the Oxiane consultant had to take into account some constraints, i.e. simplifying the operations to be performed, checking the rules and making the interface more user-friendly. The system had to be robust, intuitive and simple to use. A peripheral device (graphic tablet) was opted for to allow the facilitator to remain mobile and always present for the players. 4.14 Obviously, in order to make the application thoroughly "tablet-friendly", it was necessary to design it with a unique window, eliminating any sort of drop-down or contextual menu. Regarding the stylus' behaviour, it was decided to use a drag & drop metaphor, trying to make it as intuitive as possible. --- 4.15 The application is based on Cormas, a powerful framework developed with the VisualWorks implementation of Smalltalk language (Howard 1995). This framework offers many interesting tools for programming multi-agent systems, including some components for visualising the space model, but it had rarely been designed for a role-playing game purpose (e.g. http://www.negowat.org/). In concrete terms, the standard Cormas tools used in the application consist of three windows. One is the main menu, one is opened to configure the simulation and the last is a visualisation of the space model. --- 4.16 The first difficulty was to gather these three windows into one and to eliminate unnecessary functionalities in order to satisfy the ergonomic constraints of the game. http://jasss.soc.surrey.ac.uk/14/4/21.html 3 12/10/2015 4.17 Smalltalk language is particularly suitable for this kind of activity: components are programmed with the concepts of object and class and it is easy to directly reuse, modify or refine them. The main window was simply derived, its menus were removed, the initialisation step was automatically activated and finally the SpaceInterface and its SpaceView [7] were also derived and integrated in the right part of the main window (Fig. 3). Figure 3. The interface 4.18 The second challenge was to actually make it user-friendly. It should be usable after a couple of hours of coaching. For this purpose, it was decided to use a metaphor of the real play on the board. Players take their playing pieces and drop them on the squares of their choice. They may take their pieces back and change their choices. The game master has to report their moves in the system (Fig. 4, 00:31, 00:49, 01:49). provides playing pieces that can be carried, provides an icon and its offset (used to display a representation of the current mode that follows the stylus), provides an action symbol used to easily test the current mode, provides an 'interact' method, called in response to a user click. --- 4.20 The second trick is the use of a very powerful method called#follow:while:on:. 4.21 This method is used to continually display an icon at the stylus' position until a stylus' click occurs, giving visual support for a drag & drop feeling. It is used in the#displayMode: method frequently called in the main window (further technical details are available in supplementary material). --- 4.22 In addition, a number of operations were created to optimise the handling and supervise the sessions by automatically verifying the game rules. For example "cow playing-pieces must always be paired on 2 successive boxes and cannot be dropped on the same boxes as the previous round" or "only undergrowth can be cleared", etc.. The same visual features as the game board and colour codes of the players are used in the graphic interface of the application. After several uses, this interface allows the scientific facilitator to work alone and take on the game master's missions. 4.23 To supplement the game, appropriate teaching tools and documents were also developed. A film called "The coast as a medium for a collective, multidisciplinary approach" was designed by the scientists working in collaboration with the Océanopolis relay advisers (Fig. 4, 00:01). 4.24 It shows students the interdisciplinary work carried out by researchers, opens and closes the session. An educational document drawn up for students allows them to report on the different activities in the workshop and a file assists teachers in preparing their day at Océanopolis. --- Evaluation --- 5.1 In order to evaluate the running of the game by educators, the involvement of students and teachers and whether the workshop has met its aim of raising awareness of shoreline issues, 33 test sessions were set up. These sessions, involving 590 high school students, from 16 classes and diverse backgrounds, were monitored from January to June 2010. --- Method --- 5.2 In order to analyse the progress of sessions and their ability to meet the initial goals, the whole team (educational and scientific) developed a monitoring process. Each session takes place in the presence of an observer provided with an evaluation grid taking into account various criteria: --- 5.3 In addition, a satisfaction survey was conducted among teachers on the workshop's organisation and content. --- Results --- 5.4 After 33 sessions, the team was gathered together through a workshop to analyse the observations made. Three observations made respectively by the teachers, Océanopolis educators and scientists were listed. --- 5.5 Half of the teachers completed the survey by Océanopolis. Out of the 7 surveys completed, 5 reported the informant to be very satisfied with the workshop, its organisation and its content. Two of the teachers were satisfied, criticising two points: the length of the game (considered too long) and its aims, which were considered to be too vague. --- 5.6 The unanimous view of the 4 Océanopolis educators involved in the 33 sessions was as follows. While familiarisation with the game and the handling of the board are not considered problematic, the educators experienced difficulties in introducing and concluding (debriefing) the session. The game's success with students varies according to classes. Certain sessions are very lively, with the students getting very involved in the game, while others are more passive. Teachers' participation is just as variable. Some are keen to join in, others are absent. --- 5.7 During these test sessions, the scientists observed both the students and educators. The goal was to identify how players/high school students integrate knowledge specific to environmental management, natural and socio-economic dynamics, as well as conservation issues. This also entails the analysis of the skills involved as regards the wording of a problem, codes for taking part in the discussion, the capacity to listen to others and comprehend differences, the development of "sensible" arguments. The students quickly understand the rules and, by the end of the session, have a critical approach to the game. Some complain of the inactivity of certain roles (the second-home owner for instance), the lack of money and the limited means to control scrub encroachment. In response to the question "what is the purpose of this game", asked during the debriefing, two types of answers emerge: it raises awareness of land management and it demonstrates the difficulties involved in cooperation. The complexity, the challenges and the dynamics of the shoreline are rarely mentioned. --- 5.8 In terms of the workshop organisation, the students consider the film to be slightly too long and complicated, in particular the section describing the conceptual model. On the other hand, the introduction to the game is considered useful, as is its conclusion, which also links in with the debriefing. The scientists observed that the debriefing is often shortened, or even omitted, because the rounds of the game take too long. Furthermore, when it takes place, the educator has difficulty in running it. Improvement of the RGP 5.9 The analysis of this monitoring effort enabled the team to improve the running of the game, mainly so as to facilitate the leadership of the session by the educator. The improvements include the drawing up of specifications, the simplification of the introductory film and the preparation of a PowerPoint presentation for the debriefing. No alterations were made to the game itself or its rules. 5.10 Specifications were drawn up for the educator by the team. They guide the educator through the different stages of the session, specify the aims and expected outcomes of the game, enabling them to face unforeseen situations and answer various technical questions. 5.11 The film, considered to be too long and complicated, was shortened, to focus specifically on the geographical context (the shoreline and its current challenges) and the introduction to the game. 5.12 Finally, a PowerPoint presentation provides a framework for the debriefing (Fig. 5) and helps to conclude the session. A multi-agent model devoted to interactions between social and natural dynamics on Ushant Island was used as the medium for a RPG adapted to the school curriculum of fifth-form classes on the theme of sustainable development and coastal areas (History-Geography, Life and Earth Sciences and Civics). Adapting it required close collaboration between the scientists behind the project, the educational department at Océanopolis and teachers. --- 6.2 Transferring the game required transforming both its substance and form, running it with an easy-to-handle ergonomic platform, as well as support from scientists in presenting the approach to teachers and a film to introduce the game. --- 6.3 In terms of methodological aims, the RPG allows students to play an active part in their learning process through hands-on experience, and to concretely set out the issues involved. Yet what is the real aptitude of this type of tool to raise awareness and facilitate the learning experience of young people with respect to the environment, forecasting issues, and integrated and participatory management? This question is difficult to answer as it begs the question of how to assess the efficiency of these participatory tools. Nevertheless, the likelihood is that without the involvement of the teachers, resulting in subsequent work in class on the notions conveyed by the game, its aptitude to raise awareness will only last as long as the session... It is simply yet another experimental element.

The study refers to the interactions between socio-economic and natural dynamics in an island biosphere reserve by using companion modelling. This approach provides scientific results and involves interdisciplinarity. In the second phase of the study, we transferred knowledge by adapting the main research output, a role-playing game, to young people. Our goal was to introduce interactions between social and ecological systems, coastal dynamics and integrated management. Adapting the game required close collaboration between the scientists and educators in order to transform both its substance and form and to run it with an easy-to-handle ergonomic platform.

Introduction Cervical cancer is the second most common cancer among women worldwide at a yearly death rate of 274,000; 88% of these cases were reported in developing countries (WHO, 2008). According to the World Health Organization (WHO, 2013) cervical cancer is highly curable, therefore, the main culprits for high mortality rates were lack of effective prevention, early detection and treatment programs, and access to available preventive programs. Epidemiological studies found that cervical cancer was highly correlated to Human papillomavirus (HPV) after controlling for other risk factors (Ferlay et al., 2015;Al Zaabie et al., 2015). Several risk factors were studied worldwide to find that smoking, sexual relationships with multiple partners, multiple marriages, and dietary habits were leading in cervical cancer diagnoses (Hosono et al., 2010;Basu et al., 2014;Patel et al., 2018;Sharma and Pattanshely, 2018). Early marriage and childbirth and multiple pregnancies were also found to be high risk for cervical cancer (Basu et al., 2014). Muradi et al., (2017) stated that vaginal inflammation was a high risk factor for developing cervical cancer, as chronic infection could develop into cervical --- RESEARCH ARTICLE --- Knowledge, Attitudes and Practices of Women in the Southern Region of Saudi Arabia Regarding Cervical Cancer and the Pap Smear Test Enas A Dhaher* dysplasia and cervical cancer, in addition to poor hygiene and low socioeconomic status (Velenciuc, 2009). Since the time of introducing the conventional Pap smear test and early detection programs, mortality rates have shown dramatic decreases in developing countries by 70% (Salslow et al., 2002). Reports from developing countries showed that the main barriers to receiving Pap smear tests were lack of overall knowledge about the disease, its signs and symptoms, the importance of prevention and the preventive measures, women's attitudes towards the preventive measures for cervical cancer, and general myths and beliefs (Nwankwo et al., 2011;Esin et al., 2011;Reisetal, 2012). The American College of Obstetricians and Gynecologists (ACOG) urged women who had been or were still sexually active, and those at least 21-years of age to have an annual Pap smear test and pelvic examination. For those 30 years and above who have had three successive normal pelvic examinations and Pap smear tests, the ACOG advised for monitoring on longer intervals (Ozan, 2005) According to Saudi Arabia's cancer statistics, of the 152 cervical cancer cases detected yearly, 55 die with a 1.3 incident rate (Addar, 2017). According to Alsbeih (2014) in Saudi Arabia cervical cancer placed at the eighth common cancer among women in the age group 14 to 44. There were some studies conducted for HPV detection; 100 women were tested in Jeddah at King Fahad University Hospital resulted in a 94% negative rate for HPV (Gazzaz, 2007). Another study of volunteer, cervical cytology screening was conducted in the western region of the Kingdom with 485 women at King Abdulaziz University Hospital. Only 5.6% were found to be high-risk HPV (60 years and above), and 188 women from the 40-49 age group were found more likely to accept HPV testing (Bondagji et al, 2013). While a study in Riyadh revealed that 31.6% of 120 women had HPV 16/18, and 10 had cervical abnormality out of which six were positive HPV 16/18 (Al-Muammar et al., 2007). Al Sheikh's (2014) study found that the HPV prevalence in the Saudi native population was still unknown. One recent study conducted in Riydh by Jaradi and Bwazir (2019) assessing women's knowledge, attitudes and practices towards cervical cancer through focus group discussion with 77 women revealed that women were lacking knowledge regarding cervical cancer and pap smear test neither immunization, and screening was not necessary since they do not complain of any diseases. The main objective of this study is to measure women's knowledge, attitudes and practices of cervical cancer and its preventive measures in the southern region of Saudi Arabia. --- Materials and Methods --- Study design A cross-sectional survey of 255 Saudi women of at least 15 years of age was conducted at the Obstetrics and Gynecology Outpatient Clinic of the Armed Forces Hospitals Southern Region (AFHSR) in Saudi Arabia regardless of the visit reason. The study was conducted between March and April 2017. The clinic was selected, as it provided reproductive health care services to the entire southern region of Saudi Arabia, which included antenatal care, high-risk pregnancy, gynecology, infertility treatment, postnatal care, and family planning. Furthermore, the clinic was connected to the Obstetrics and Gynecology of the hospital where, on average, about 370 patients receive care daily. --- Study instrument Based on literature review self-administered questionnaire was developed, it consists of two main parts demography (15 questions) and knowledge, attitudes and practices of cervical cancer and the HPV vaccine. (15 questions). All questions were formulated in a simple direct language and the answers were given in a list of options where women supposed to choose one option only, an option was provided for the knowledge question which is" I don't know ". The questionnaire was pilot tested (test-retest) with 25 women whom were not included in the study. The questionnaire was reliable with person's correlation of 0.80. The alpha value for the overall questionnaire was 0.90. --- Sample and data collection Sample size: The sample was designed assuming the total population visiting the clinic in one month was 8000 women based on clinic statistics and appointments. A sample of 250 women with a confidence level of 95% will provide 50% of women answering the questions of the entire relevant population with a 50 <unk> 6.1 confidence interval and 80% power to check for differences. Given the excellent response rate, the questionnaire was distributed to 275 women, and 255 questionnaires were returned with a response rate of 93%. Data was collected daily for one month. Questionnaires with additional health education materials (healthy life style and nutritional habits) or educational materials without questionnaires were distributed in sealed envelopes to ensure randomization to all the women (simple randomization was done in that every other woman in the waiting area received a questionnaire that accounts around 50% of the attending women receive the questionnaire at the time of data collection). Women with at least 15 years of age, and all present in the clinic regardless of their reason for attendance (a client or a companion) were the target subject of the study. The researcher was available to assist illiterate women and those with questions regarding the purposes of the study. --- Ethical consideration This study was approved by the Research Ethical Committee at the Armed Forces Hospital in the Southern Region, prior to the research being conducted and furthermore, consent forms were signed by all participants in the research where it included the purpose of the study and ensure anonymity and privacy. --- Statistical analysis The Predictive Analytics SoftWare 18 (PASW) was used to enter and analyze the data. Frequency distributions were used to present women's demographic characteristics, knowledge, attitudes and practices related to cervical cancer and the Pap smear test. Cross tabulations were used to assess the demographic factor relations regarding the same. --- Results --- Sample demographic and obstetric characteristics The mean age of women was 32 years, with the majority above 25 years. Approximately 11% of the women were illiterate, and 32% were university graduates. The educational levels of the husbands were half-attained secondary schooling, and 19% were diploma and university degree holders. Among the women's professional careers, 2.7% were in health professions and 19% were teachers and educators. Overall, only 15% of the women participating in the study were currently employed (Table 1). Women's obstetric and marital characteristics revealed half were married for more than 10 years, and the majority had four pregnancies. The median and mean number of deliveries was 2.95 and 2.80 respectively. When excluding women with no children, the mean number of children factor, but almost none knew the direct cause of cervical cancer. Regarding attitudes of prevention, approximately 57% responded that cervical cancer (HPV) could be prevented, only two women had ever conducted Papsmear test based on doctors referral but close to all women had never heard of the HPV vaccine and, consequently, only one woman had received the vaccination in the United States. A new variable was created with the total number of risk factor recognized by women in four categories (0; 1-2; 3-4; > 4) to find out that the highest percentage was with those recognize no risk factors (Table 3). Number of risk factors was highly significant (p=0.001) with women's level of education as those with higher education recognize more risk factors. Whereas none of the demographic factors was correlated with number of risks recognized neither the availability of cervical cancer immunization. Regarding the Pap smear test, 43% of women had ever heard of the test through various media channels, 32% had received health and wellness information from female medical professionals, approximately 64% were increased to 3.57. Nine percent of women reported having more than six living children, and nearly half of the women had never used family planning methods. Interestingly, only 23% of women had ever conducted a comprehensive reproductive health assessment (Table 2). --- Women's knowledge, attitudes and practices of cervical cancer and Pap smear test. While close to three quarters of the women had ever heard of cervical cancer, the primary source of information was from social media, mainly Facebook and Twitter. Responding to further questions related to the risk factors of cervical cancer, around two thirds of women responded with either do not know or not a risk factor. More than half of the women recognized vaginal inflammation as a risk unaware that the test could discover asymptomatic lesions, and 56% recognized that early detection could lead to better outcomes. Regarding attitudes towards the test, only 38% of women said they would participate in a screening if they were properly informed. The reasons mentioned for those not willing to participate ranged most from no complaints to least husband disagreement, 45.5% and 2.6% respectively. Two third of those willing to participate in the screening preferred served by a female gynecologist at an OB/GYN hospital (Table 4). The relationship between women's characteristics and their KAP towards Cervical cancer and Pap smear test. Based on cross tabulation a significant relationship was found between women's level of education and knowledge about cervical cancer (P=0.000). In addition, Women of higher education levels showed significant acceptance of participation in screening tests when properly informed (P= 0.009). Significant relationship was found between the belief that cervical cancer can be prevented with educational level (P= 0.010). Those with medical professions or those with any other professional careers were more knowledgeable about the disease than other areas. Number of deliveries showed a significant relationship between women's knowledge of both cervical cancer and the Pap smear test (P=0.029 and P=0.041). None of the other demographic characteristics showed a significant relationship between women's knowledge, attitudes and practices.(Table 5). --- Discussion In the southern region of Saudi Arabia, there were no This study demonstrated that while close to two third of women recognize cervical cancer, few knew of its risk factors and almost none knew of the direct cause, which represents the studied women having less knowledge than participants in the Jeddah study and Kuwait (Sait, 2009;Al Sairafi and Mohamed, 2009 ). Previous studies found that lack of knowledge regarding risk factors of cervical cancer might be the most important factor for women not obtaining screening tests (Were et al., 2011;Aswathy et al., 2012;Nadarzynski et al., 2012). The lack of national screening programs and public education campaigns could explain why women were not aware of the disease's risk factors. Regarding the Pap smear test, less than half the women were aware of it and about two thirds had no knowledge of the purpose of the test, which, again, represents a stark contrast to the Jeddah study (Sait, 2009). Additionally, half of the women thought early detection could lead to better outcomes. Our study showed the majority of women's source of information regarding both cervical cancer and the Pap smear test was from social media, which is considered the most popular form of communication, including the communication of health information. Eight out of 10 internet users seek health information online, and two thirds seek social media to access health information in the United States (Child and Martin, 2012;Von Muhlen and Ohno-Machada, 2012). Social media could help improve and promote healthy communities when used properly (Ventola, 2014), increase access to educational health resources, and even track personal health progress (Lambert et al., 2012;Mac Millan, 2013;Grindrod et al., 2014). However, use of social media as the main source of health information could be dangerous in terms of content reliability and accuracy, currency, affiliation and authority (Moorhead et al., 2013;Pirraglia and Kravitz, 2012). Eventually in this study, we found that women recognized cervical cancer and the Pap smear, but were not familiar with risk factors or preventive measures. Despite the dangers of social media, it could be used as an effective educational tool in the southern region of Saudi Arabia. Well-organized media campaigns featuring famous regional figures or movie stars could be beneficial. Hyacinth et al., (2012) suggested a similar recommendation with the dissemination of information regarding knowledge and awareness of cervical cancer and the Pap smear test. Level of education and type profession were found to be highly correlated to women's knowledge about cervical cancer and the Pap smear test, as well as their attitudes toward preventing cervical cancer. Level of education was found to be a positive factor in other studies in Saudi Arabia and Kuwait (Amarin et al., 2008;Ravichandren et al., 2011). Regarding attitudes toward preventive Pap Smear tests, more than half of the women agreed that early detection could prevent cervical cancer, and 38% agreed to participate in a screening program similar to the Akanb et al., (2015) and Almobarak et al., (2016) studies. The main barrier for not getting the Pap smear test was having no complaints, which could be explained by lack of knowledge regarding the disease and prevention, as discussed above. Moreover, if women had no symptoms and did not feel sick, it was assumed that diagnostic procedures like the Pap smear test were unnecessary, the study in Riyadh supports the same result (Jaradi and Bawazir, 2019). Similar beliefs were found in studies in Indonesia, Qatar and Iran (Wong et al., 2008;Al-Ali et al., 2016;Ashtarian et al., 2017 respectively). Fear of pain and embarrassment, which could be considered a misconception of the test, were barriers found in studies in Malaysia and Ghana (Wong et al., 2008;Ebu, et al., 2015 respectively). Another barrier mentioned by the women, no information about where to go (26.6%), considered a very important and devastating barrier, as found in the Ghana study (Ebu et al., 2015). Other barrier as waiting for the physician to recommend the Pap smear test was found in the Kuwait and Jeddah studies (Al Sairafi and Mohamed, 2009;Sait, 2009). The aforementioned barriers could be explained as lack of awareness about cervical cancer and its prevention techniques, and lack of knowledge about the Pap smear test, and its implications and procedure. Several other studies posited the same (Asgharnia et al., 2009;Nwankwo et al., 2011;Farshbaf-Khalili et al., 2015). Our study showed that women with a higher level of education, and those with medical backgrounds or other professional careers were more likely willing to have Pap smear tests compared to those with less education, no education, or other educational backgrounds. This would support the thinking that awareness and knowledge played a significant role in women's attitudes. Medical backgrounds and high levels education were previously found to be significant with women's attitude of having the Pap smear test in a KAP study in The Sudan study (Almobarak et al., 2016). Conversely, among those who were willing to have the test in this study, 94% preferred a female health care provider, which was also found in a Kuwait KAP study (Al Sairafi and Mohamed, 2009), and among older American women (Guilfoyle et al., 2007). The study also found that a high number of deliveries were significantly associated with more awareness regarding cervical cancer and the Pap smear test. Which could be related to frequent contact with a medical and health staff during the reproductive life span and more chance for health education or health related information to be given. Alsbeih (2014) concluded that neither national screening programs nor national vaccination programs against the HPV virus would be cost-effective in reducing cervical cancer in Saudi Arabia, as the reported incidence was already low. While the same study discussed the real prevalence of HPV was not recorded among native women in Saudi Arabia, our study showed that women were lacking knowledge regarding the Pap smear test, and cervical cancer risk factors, implications, and methods of prevention. Despite the recommendation from Alsbeih, women should have the right to choose screening tests, in which case, information on preventive measures need to be available to all women. Freedom of choice could be accomplished when women are well-informed of the disease and its available preventive measures. Therefore, it's recommended in this study to organize a mass media campaign educating women about cervical cancer, the Pap smear test, and prevention measures targeted to women of the southern region of Saudi Arabia. --- Funding Statement No funding source. --- Study strength and Limitations High response rate is considered one of the study strength, variation in between the sample is also a strength but since the study is taking place at the obstetric clinic social desirability bias might be significant here. --- Statement conflict of Interest No conflict of interest declared.

The main barrier for women to receive Papanicolaou (Pap) smear tests and immunization is lack of knowledge about the disease's signs and symptoms, women's attitudes toward prevention programs and cultural myths and beliefs. Therefore, the main objective of this study is to measure women's knowledge, attitudes and practices about cervical cancer and the Pap smear test in the southern region of Saudi Arabia and to assess the findings in relation with women's demographics. Methods: A cross sectional survey was conducted at the Armed Forces Hospital Southern Region Obstetrics and Gynecology Clinic using a self-administered questionnaire with a sample size of 255 women between the ages of 15 and 65 years. Results: Forty-three percent of the women in this region are aware of cervical cancer, but do not recognize its risk factors, implications, timing or main cause, which is Human papillomavirus (HPV). In fact, the primary source of information was obtained through social media. Only two women conducted Pap smear test and that was based on doctor's referral, where women's main reason from not conducting the test was feeling good and no need. Conclusions: There is a need, therefore, to create awareness programs for cervical cancer, its causes and risk factors, as well as its preventive measures for women in the southern region of Saudi Arabia.

Introduction Child marriage, referring to a marriage of a young person less than 18 years is still widely practiced in many parts of the World and predominantly affects girls. It is estimated that, if nothing changes, a 100 million young girls aged 15 years or less will be married within the present decade 1. Regional estimates of its occurrence for girls include 48% in Southern Asia, 42% in Africa and 29% in Latin America and the Caribbean 2,3 with wide differentials across the countries. In the West African sub-region, the proportions of girls affected vary from 28% -43% (Ghana, Togo, Cote d'Ivoire, Senegal, Benin, Nigeria) to 60-80% (Burkina Faso, Guinea, Mali, Chad and Niger). In Nigeria, the practice of child marriage is deeply entrenched in tradition, culture and religion and the country has one of the highest rates of child marriage in the world, with estimated 42% of girls married before 18 years; and while this is found among many ethnic groups across the country, its predominance is clearly in the Northern part of the country 4. While nation-wide, 20% of girls are married by age 15 and 40% are married by age 18, child marriage is extremely prevalent in some regions such as among the predominantly Mus-lim Hausa-Fulani of the Northwest and North-East (of which Gombi is a part) where 48% of girls are married by age 15, and 78% are married by age 18 5,4. While average age at first marriage is 17 years nationwide, average age of marriage for girls is just over 11 years in Kebbi State and about 15 years in most other parts of Northern Nigeria. Among the Yorubas, made up of a mixture of Christians and Muslims in the South-West, child-marriages are no longer arranged and for the Igbo in the South-East that are predominantly Christians, the practice has declined considerably and restricted to few communities 6 and situations where pregnant teen-agers are forcefully married off, to prevent family shame or restore honour 6. The reasons for child marriage are based on a mix of cultural, social, economic and religious factors. Poverty is observed to be at the core of decisions and practices related to early marriage, more in low-income societies than in their high-income counterparts, as they lack resources to support healthy alternatives for girls, such as prolonged schooling and skill acquisition to secure their future. The girls in turn have higher chances of being poor and remaining poor and of facing serious social and health consequences inimical to their personal growth and development. A study 8 makes it clear that the widespread practice of child marriage makes it increasingly difficult for families to escape poverty in the developing world, thereby undermining critical international efforts to fight poverty, HIV/AIDS and other development challenges, and making huge investments in development assistance less effective. It is therefore very costly in terms of the consequences for these societies. Girls who marry early are more likely to experience abuse and violence than others, with inevitable psychological as well as physical consequences. Studies indicate that women who marry at young ages are more likely to believe the justification for wife battering as a corrective measure and therefore acceptable for a husband to do so and are therefore more likely to experience and accept domestic violence themselves 9,10,11. Child marriage for girls often means a confinement to a helpless lifetime of domestic and sexual subservience. The psycho-social requirements of marriage (family and wife-mates politics and diplomacy) are often beyond child brides who are largely unprepared for their roles in complex family settings. Young brides are more often than not subjected to forms of abuse such as psychological trauma, domestic violence, forced sexual acts, marital rape by the husband and in-laws and subjected to domestic slavery, given her position as the young wife. Evidences of the links of early marriage with divorce, abandonment, separation, widowhood and denial of property rights abound in contemporary literature 12. Child marriage is widely and globally regarded as a surviving form of social discrimination and is challenged by the Nigerian constitution and several human rights laws. In the last Century, particularly, intense global opposition to child marriage has deepened and the concerns are expressed in various conventions and charters. Under the Nigerian Law, the concept of the child is based exclusively on calendar age. The legal age at marriage is 21 years and anyone under this age is considered a minor and would require a parental consent before legally entering into a marriage in the country 13. The rights of the girl-child in Nigeria are protected by a legal framework, including national laws and international and regional conventions which the country has ratified. The persistence of the practice of child marriage has been linked to lack of sufficient political will to engage the problem. In many countries, and despite the ratification of the international treaties and agreements, existing laws against child marriage are not enforced or upheld and sanctions are either not clearly spelt out or enforced at all 14. However, it is evident that the key national and international instruments that challenge child marriage focus on age, consent, equality within marriage, personal and property rights of women. The study therefore examines the implications of child marriage on maternal health and freedom. --- Data and methods The study was conducted in Gombi LGA in Adamawa State of Nigeria which is one of the twenty one LGAs in the state. The TRF is 6.7 while the mean number of children ever born to women of aged 40-49 is 7.1. Women in the reproductive age group 15-49 years are 32,527 (National Population Commission,2010). The Gombi LGA has a population of 146,429 (Population Census, 2006). The LGA is made up of Gombi and other surrounding villages such as Fota, Fadchi, Walawal, Garkida, Guyaku, Balwhona, Wushipra, Gaanda, Lala, Bokki, Tsakuwa, Gudawi, Gombi, Ladiel, Fadchi, Kwallamba, Misamba and Ferwumarah. The unique characteristics of these villages include common language, religious beliefs and cultural practices which make the LGA homogeneous in nature. The study was carried out in rural areas. Qualitative and quantitative methods were used for data gathering. For quantitative data, the structured Questionnaire was used while Focus Group Discussion (FGD) and In-depth interview methods were used for the collection of qualitative data. Two Hundred (200) copies of questionnaire were administered. Stratified sampling technique was employed to select the sample for the study. The study area was stratified with the listing of the villages and eleven of the enlisted villages were randomly selected. The selection constitutes 50% of the total villages in the study area it will also make the study representative of the study population. From the selected villages, houses were randomly selected using the National Population Commission (NPC) house numbering. From the selected houses young women aged 15-24 years who married before age 16 years and have had at least one child were interviewed using the specially designed questionnaire (there was a recall question). The questionnaire was pre-tested for validity and reliability of the instrument. The reliability test was conducted before the administration of the questionnaire. In all, 200 questionnaires were correctly filled and analyzed for the study seven (7) were discarded. The information collected covered the socio-demographic characteristics, reproductive health issues, child bearing, sexual behaviour (questions on sexual behaviour were adapted from NDHS 2008), factors influencing child marriage in the study area and the risks of early marriage and childbearing. Six (6) Focus Group Discussions were held (50 discussants were selected across different socio-economic strata within the selected villages) to elicit information on traditional expectations regarding sexuality, impact of early marriage on well-being and consequences of early marriage. The participants were those in reproductive age 15-49 years who had at least a child as at the time of the survey. Sturctured interview guide was used for the FGDs. One focus group discussion each was carried out in six of the villages selected for the quantitative study. The FGDs were conducted by two trained research assistants from the state who are conversant with the culture and language of the people but not from the local government. Information provided was strictly maintained and the real names of the respondents were not used for the study. Issues about traditional expectations regarding sexuality, women's right regarding marriage, reactions to infected partners, and the extent to which men have control over sexuality. In-depth interviews were also conducted among 12 stakeholders (two each from health workers, community leaders, religious leaders, market women and traditional health workers. They were selected across the selected villages) in the study area. The exercise covered areas like: issues of child marriage, factors influencing child marriage, girl child education, sexual rights and choices in the community, and common maternal health risks. For each of the session of the FGDs and IDI, two trained research assistants were used (one served as note taker while the other RA served as moderator). The discussions (notes taken from each of the sessions) from the FGDs and IDIs were transcribed and translated verbatim. Analysis involved developing a system of indexing the data into sets of categories or codes that provided structure to the data based on the research objectives and the topic included in the questions guide. A qualitative software ZY-index software package for ethnographic data was used for textual data analysis. This does not allow the use of percentages and statistical analysis. Some striking expressions were for verbatim reporting. Data from qualitative survey have been extensively used inthis paper to validate quantitative analysis where and when necessary. The quantitative data collected from this survey were subjected to three levels of analysis. The first level involves an examination of the distribution of the respondents according to selected characteristics. This is because the behaviour of individuals in society is, to a large extent, determined by their personal characteristics as well as those of the environment in which they live. For this reason, it is expected that child marriage will be greatly determined by background characteristics such as education, age, marital status, type of marital union and reproductive health decision making. To this end, simple percentages were employed to describe these variables. The second level involves the examination of the pattern of relationship between the dependent variables i.e. child marriage and other independent variables such as perceived risks of HIV, maternal health risks, reproductive decision making and antenatal care. The third level of the analysis involves the use of Logistic Regression techniques to show the effect of the relationship between the dependent variable (e.g. ever experienced pregnancy complications, yes=1, and no =2) and other independent variables (such as income, occupation, education, reproductive decision making. On ethical consideration, consent forms were administered to all the participants those who could read were asked to sign the consent forms while those who could not thumb printed the consent form. These forms were strictly kept by the researchers Findings Socio-demographic characteristics Majority of the respondents ((72.5%) were in the 20-24yrs age cohort and had little exposure to formal education as 60% had attended a primary school while 30% had opportunities to attend and complete secondary school education. All the respondents were young mothers who married early at various ages before 16 years of age and had been married less than 5years (45.0%), 5-9 years (50.0%) and few (5.0%) for more than 10 years. Some of the respondents were engaged in farming (41%), trading (15.5%), salaried workers in the local administrative offices while about one in every three (35.0%) was a full-time housewife. They were more Muslims (59.0%) than Christians (41.0%). Estimated incomes of respondents reveal that more than half (55.0%) earned less than N15, 000.00 ($100) per month, about a third (35%) earned between N20, 000.00-N40, 000.00 ($250) per month while 20% of the respondents earned more than this per month (Table 1). --- Table 1: Percentage distribution of respondents' socio-demographic characteristics The respondents weretherefore young mothers characterized by little education, largely informal sector engagement, low income, early marriage and rural residences. --- Family and marriage Early marriage for girls was still generally favoured to protect the girl-child and chart a good future for her. Information from FGDs revealed that culture plays very important roles in the lives of the people. Religious prescriptions were cited by male participants in the FGDs but there was a general agreement by female participants that poverty plays a major role in decisions to marry off girls early, particularly when they are withdrawn from schools for marriage, sometimes with the hope that they may continue after marriage and sponsored by the husband. This was corroborated by the quantitative data on the factors influencing the timing of marriage in the area, 47% of the respondents agreed that poverty was the major factor influencing child marriage in the area, while 21.5% and 12.5% respectively mentioned culture and religion. Gombi and surrounding communities are largely rural and there were expressions of anxiety over the spread of certain values from the nearby urban areas which may lead to the breakdown of long-held family and community values, the need to prevent prostitution, unwanted teenage pregnancies, childlessness and other unwholesome western values that allow what is considered unnecessary freedom for young girls. Majority (87%) of the respondents expressed the wish to experience schooling beyond the level they presently were and their admiration for well-educated women, in terms of appearance, confidence and material possession. This was supported by the FGDs conducted. --- Religious leader in Lala: A girl must be in her husband's house before or at puberty. This is why parents give their young girls' hands in marriage at their younger ages. Education is good for girls but they must be properly married---------------if a good marriage arrangement that favours the girl and the family comes up, it is good for the girl to leave school and marry first, then continue later if the husband permits------" Community elder, Wushipra: "The main reason why parents allow their daughters to marry very early is to avoid shame; they try to avoid promiscuity which may lead to unwanted pregnancy. They prefer early marriage to our young girls becoming prostitutes. Our culture does not permit misbehavior by girls, even boys must marry early to be responsible. It is the duty of a girls's father to marry her of appropriately". --- Decision-making Decision -making on the choice and timing of marriage is the exclusive preserve of fathers in the study area as revealed by 79% of the respondents (Table 2) and more often than not, husbands arranged for young girls are older men within the society who can afford the bride price and other items demanded by the bride's parents. On the women's economic status, 78% of the respondents depend on their husbands for all (Table 2) their daily needs and that of their children, even when involved in some employment, trade or farming. African The status of women in the study area is revealed in the generally low educational attainment, low income and the pattern of family and fertility decision-making that is highly patriarchal. Majority (82%) responded that husbands were the sole decision makers on issues of fertility and family size and use of contraception (Table 2). This is evident in the overwhelming 82.5% of the respondents that reported that husbands determined the decision to register for and timing of ante-natal care, as well as particular health facility to use, which may be heavily de-pendent on affordability. Since husbands were the main providers in the households, wives were generally very submissive as culturally expected. This was corroborated by the FGDs in which participants revealed that fertility decisions taken by husbands were sometimes injurious to the health and wellbeing of the wives and cited decisions to have another child to ensure a male child or more male children. A 24 year old, Gombi Since we depend on our husbands for the daily needs and they are the ones that can determine the number of children. If I have the resources (e.g. land) and I make a lot of money from it, I can say I want this number of children or this is the hospital I want to use. They are the ones that own the land and the crops, we are just assisting them to support our family. --- A trader (18 years), Fadchi It is our religion that allows husbands or the heads of the family to be the one to decide on the number of children. Even if the wife desires more children and the husband says no there is nothing she can do. Likewise if the husbands desire more children the wife can't say no to him. Male children are favoured in our culture--- --- A housewife, Guyaku: It is not the fault of women not to go to the hospital, their husbands must approve of it because you can't take the decision on your own. Only 25% of the respondents ever discussed the need for family planning and use of contraceptives with their husbands. Spousal communication on the use of contraceptives as previously revealed (NDHS, 2008) is therefore still very low. The data also revealed that spousal communication on fertility issues is very low as a result of the wide age differences between the young mothers and their husbands The FGDs revealed a general lack of confidence by respondents in articulating the issues around reproductive health issues and associated with childbearing. --- Maternal health risks The data revealed that respondents commenced childbearing between 14 and 18 years of age and 71% had experienced at least one serious pregnancy or birth-related health problem which include excessive bleeding during labour (19.0%), obstructed and/or prolonged labour (49.0%), frequent miscarriages (12.0%) and pro longed sickness after childbirth (20%) as shown in Table 3. African Health Sciences Vol 16 Issue 4, December, 2016 991 In addition to these, of those that had experienced various health complications, 37 (26%) had been exposed to the risk of vesico-vagina and recto-vigina fistula and were at various stages of treatment. This and discussions in the FGDs revealed that maternal morbidity is very high among the young mothers. The uptake of opportunities for antenatal and post-natal care was reported-ly low among women due to distance from the facilities, cost, husband's attitude and ignorance. Birth-spacing is not deliberately practiced and contraceptive use is almost non-existent.While there were no locally collated figures, health personnel also confirmed a high and increasing frequency of maternal deaths involving young mothers, in and outside health facilities in the study area 12,14,17. A nurse, Gombi: "Most of the cases we witness in the health facilities are prolonged labour, VVF and excessive bleeding. Some of these women are too young for childbearing but because of our culture and religion there is nothing we can do, we are educating them still we still have new cases every day. Sexually Transmitted Infections are very common around here but many are treated at the local chemists and herbal clinics rather than in the hospitals because of shame. We know that the most important issue is the immaturity of their bodies---that is the risk". A civil servant, Gombi "It is prolonged and obstructed labour that is more common in this area, with excessive bleeding. It has led to loss of many girls in our community. Some of these women are not well informed about the pregnancy complications. Some of them will not attend anti-natal clinic. They are too young and not yet ripe for childbearing but are married and expected to have children. They are also children" A full housewife, Misamba: Miscarriage and excessive bleeding are very common pregnancy complications in our community. Also prolonged labour mainly because pregnant women do not register early for ante natal care because the young ones don't have money they don't know when its appropriate to do so. A nurse/midwife, Balwhona: "We are used to complications in this health facilities, this was common among young girls, who married before age 18 years. Some of them will not come to the health facilities for anti-natal until it has resulted to complications -sometimes it might result to death. We have witnessed several deaths of young girls that are wives and too young to safely deliver their children" A medical doctor, Gombi: Majority of the complications we have handled were as a result of early marriage and their negligence of not attending anti-natal for proper medical advice. VVF cases are also common in this area, but we are trying our best to educate them. The sample of young early-married mothers was analysed for differentials in experiences of pregnancy-related complications across categories as shown in Table 5. Religious affiliation had minimal impact on the occurrence of pregnancy complications as slightly more Muslims (72%) than Christians (69%) had experienced complications. However, the FGDs and in-depth interviews with religious and community leaders revealed that early marriage of before 18 years is the norm in the study area but child marriages are conducted mostly among Muslims because the religion permits it, with caution. The experience of serious pregnancy-related complications decrease with increasing exposure to educational opportunities with more respondents with primary level (completed, 67.8%; not completed, 78.0%) education reporting complications, compared to those with secondary (61.6%) and post secondary (50%) levels of education. Similarly, respondents' husbands' level of education was found to be associated. Of the respondents that had experienced complications, more husbands had no formal education (84%) or primary level education, (68%) compared with those that had secondary (67.0%) and post secondary (50.0%) levels. The data (Table 5) also show that pregnancy-related complications were reported more among respondents aged 15-19 years (75%), low income (less N20,000.00 per month), full housewives ((79.0%) and farmers (62.0%) than among those aged 20-24 years (69.6%), high income (N40,000.00 + per month), traders (35.0%) and civil servants in salaried employment (29.0%). Significantly, experiences of pregnancy-related complications increase with increasing number of children of the respondents from one (54%) to at least four children (78%) in the study area. The data therefore reveal that level of education of respondents and their husbands, income of respondents, --- Maternal health-seeking behaviour Place of delivery is one of the factors that influence maternal deaths and 28% of the respondents delivered their last baby at home while 23% delivered at the homes of traditional birth attendant. From the survey, 33% of the respondents indicated that the health facilities were far, more than 5kms from their homes and this may be one of the reasons for none use of the health facility. Participants from the FGDs expressed confidence in nurses, midwives and traditional birth attendants compared to doctors. This was collaborated by the responses in table 4. Delay in seeking medical treatment is one of the factors mentioned among the FGDs participants as a cause of maternal death and complications in the study area. This may be as a result of the revelation that decisionmaking regarding medical treatment is the responsibility of husbands and only 17.5% (Table 4) of the respondents reported their ability to decide on their medical treatment. Only 43.5% of the respondents used ante-natal care regularly while 11% indicated that they had never attended anyante-natal prior to their delivery. --- A medical personnel, Gombi (IDI) "We don't have enough hands in this hospital but most of these women, particularly the young ones will not report at the health facilities on time until their cases are getting out of hands. With no history of the pregnancies, delays are caused in diagnosis prior to treatment---Also many women are referred to this facility from surrounding villages and sometimes they are very young for childbearing and have not been to any facility for ante-natal care so it could be too late to save them and the babies" African --- Discussion This study has established that girls in Gombi locality are facing significant cultural and social barriers to living wholesome and healthy lives that are often strengthened by religious prescriptions on marriage. These include the facts that their marriages were arranged without their consent, the opportunities for schooling were either disrupted or not utilized at all, they were made to commence sexual relations and subsequent childbearing early, they were exposed fully to various reproductive health risks as shown in this study. Another cultural factor revealed in the study is the issue of patriarchy, husbands have the final say on almost all the household decisions including the respondents' health issues. This may be one of the reasons for the delay in seeking medical care as reported among the medical health officials who participated in the in-depth interview. Haberland, Nicole, Eric. Chong, and. Bracken. ( 2006) have revealed that girls married at a very younger age experience social, personal and psychological challenges when compared with those who marry latter. There is also greater control by their husbands with restriction on movement, ability to seek health care and family planning services, limited capacity to earn independent income and paid labour force. Financially, theydepend on their husbands and unable to ask their partners to go for test, use contraceptives or abstain from sex even when they know their partners are infected. The notion of good reproductive health encompasses all aspects of the reproductive process including the right to engage in sexual relations and to decide when to have children and these are the rights not accessible to the young women in the study area. Earlier findings 18,4 revealed that child brides are likely to commence childbearing early and there is a strong correlation between the age of mother and maternal morbidity. More than two thirds of the respondents indicated that they have experienced at least one the complications during their life births. Some of the complications me tioned in this study include; prolonged and obstructed labour which may lead to hemorrhage, excessive bleeding and prolonged sickness after birth. This reaffirms the impact of child marriage on the health of the mothers. Previous studies linked child marriage with some health consequences such as, Cervical Cancer, VVF, genital ulcer etc 19,20. These girls are expected to prove their fertility immediately they are married off which often results into pregnancies a stage they are not prepared for, eventually leading to maternal and newborn morbidity. Girls aged 14 years and younger run a risk of complication in pregnancy and child birth when compared with those of older ages 21. Poverty has also been shown to be at the core of the way young girls are regarded as economic burden and quickly married off to alleviate household expenses as reported among the FGDs participants in the study area. In these communities, educational and economic opportunities available to girls are few and girls are often quickly married off to protect them as well as improve the economic well-being of the family. Considerations for family honour also put pressure on families to marry off young daughters, to avoid embarrassment and shame that may come with loss of virginity and unwanted pregnancies as they grow up. Early marriage invariably confines young girls to a life of poverty, in a cycle that is often difficult to break and significantly deprive them of freedom and entitlements to life. A study reveals that girls are viewed as economic burden and marrying them off is a way to alleviate household expenses 22. The practice of child marriage dooms girls and women to a life of poverty. Another strong factor identified in the study is the issue of promiscuity among the girls as a strong rea-sons for their early marriage. Parents believed that instead of bring shame to the family it is better to allow suit-or to come for them. Studies have shown that girls who married early are more likely to be infected with sexually transmitted infections especially HIV and human papilloma virus (HPV). Evidences from sub-Saharan African countries revealed that half of the married girls are more likely to be infected with HIV than the unmarried 3,23 The study also affirmed the significant relationship between education and pregnancy complication. Those who have higher education are less likely to have complications. Participants from the FGDs and IDIs explained that those with higher education have knowledge about signs of pregnancy complications and they can quickly access health facilities. More than two-thirds of the respondents in the study area delivered outside government hospitals and clinics, this is also a point that can influence the high maternal mortality and morbidity in the area. In her study of utilization of delivery services in Ilesa, Nigeria observed that high stillbirth and early neonatal mortality rate have been long associated with unattended deliveries compared with hospital-based deliveries 24. Those with post-secondary education also de-lay their marriage and have the ability to make choices and this was confirmed by the FGD participants. It was stated that schooling helps girls achieve ambitions and gives them skills to discuss who and when they will marry 25,26,27,28 Behind the practice of child marriage is a failure to guarantee child and young people's rights, manifested by denial of education, low status, deep poverty and restricted mobility. While the Nigerian Child is supposed to be protected by a well articulated legal framework earlier outlined in this paper, it is evident that in practice this is not so and there is a need to engage social and cultural norms surrounding the practice of child marriage. The issues of legal sanctions are desirable but complicated as the laws of the country exist alongside customary and religious laws on marriage. Country laws are also not applicable to customary marriages. 29 Within the context of poor registration of marriage and failure to apply recommended sanctions to law breakers (parents, spouses, officials etc), child brides are generally left unprotected. It is for these that in Nigeria, inconsistencies exist between the legal minimum age at marriage, which is 21 years and actual practice, as marriages of children as young as 7 to 10 years were reported in the FGDs conducted in few communities in the study area. --- Conclusion Initial concerns about child marriage centred on its contributions to rapid population growth, as early childbearing, in the absence of contraception results in large family sizes. However, in the latter part of the 20 th Century, advocates of safe motherhood and adolescent health turned attention to child marriage, emphasizing the vulnerability of young girls to HIV/AIDS, STIs and other serious health issues. Beyond this, however, and more recently, attention has further shifted to the need for widespread education for girls and its benefits, a situation that necessitated greater focus on child marriage and the exclusion of girls from schools within a rights-based approach. In the context of widespread HIV/AIDS, very high maternal mortality and growing sexual violence against girls within permissive traditional practices, the demographic and health implications of early marriage as practised in Gombi are grave for Nigeria. This study has identified the negative impact of early marriage on young women in Gombi, expressed in form of high maternal morbidity through various pregnancy complications in the study area. With this, it is clear that the practice of child marriage in the study area and in other parts of Nigeria is directly impeding the ability of the country to achieve six (6) of the Millenium Development Goals, which are those of eradicating extreme poverty and hunger, achieving Universal Primary Education, promoting gender equality and empowering women, reducing child mortality, improving maternal mortality and combating HIV/AIDS, cervical cancer, sexually transmitted diseases, VVF, malaria and other communicable diseases by 2015. Progress in health and gender matters towards the achievement of MDGs has been scored low in Nigeria. Strategic interventions planned for gender issues, infant and maternal health, HIV/AIDs, malaria, tuberculosis and primary health and implemented since 2000 have failed to yield any appreciable results and the government at all levels are being pressured to increase allocation of resources to the sectors to ensure an overall progress in the wellbeing of the population. While increase in commitment and funding are important, there is an urgent need to refocus the development lens in Nigeria to a critical and crosscutting issue as child marriage, the reduction of which can speed up the achievement of the MDGs, if not by 2015, in as short a time as possible. There is an urgent need to focus on the cultural traps to which the practice of child marriage has confined girls in Gombi and other parts of the country, through renewed commitment to compulsory education beyond primary school level for girls, enforcement of legislation and commitments to uphold the fundamental rights of the child. In particular, men, in their capacity as fathers, community and religious leaders must be targeted for change, given their roles as custodians of tradition and decision-makers on marriage and family matters. The several elements of marriage (choice, consent, timing, etc) and maternal health must be understood afresh as entitlements and the freedom associated with them must be granted to every citizen. --- Coflict of interest

Background: Efforts toward liberation of the girl-child from the shackles of early marriage have continued to be resisted through tradition, culture and religion in some parts of Nigeria. Objective: This study therefore examines the maternal health implications of early marriage on young mothers in the study area. Methods: Multistage sampling technique was employed to obtained data from 200 young mothers aged 15-24 years who married before aged 16 years. Findings: The study reveals that more than 60% had only primary education while more than 70% had experienced complications before or after childbirth. Age at first marriage, current age, level of education and household decision-making significantly influence (P<0.005) maternal health risks in the study area. The study establishes that respondents in age group 15-19 years are 1.234 times more likely to experience complications when compared with the reference category 20-24 years. Entitlements and freedom that are highly relevant to reduction of maternal mortality, provided by international treaties are inaccessible to young women in the study area. Conclusion: Strategies to end child marriage in the study area should include mass and compulsory education of girls, provision of other options to early marriage and childbearing and involvement of fathers in preventing and ending the practice.

Introduction Migration incidences and 'crises' at the European Union's (EU) external land and seaborders have been covered by traditional media (Berry et al., 2015;De Cleen et al., 2017), and debated from various perspectives within social media (De Rosa et al., 2021;Siapera et al., 2018). France's Calais, Italy's Lampedusa, Greece's Aegean islands, and more recently, the Greek-Turkey Evros land-border and the Polish-Belarusian borderamong others -are, albeit to different extents, sites for the deployment of migration crises' spectacles, which lend themselves to the demarcation of the boundaries between us and the (migrant) other. Around them, a mediated debate is evolving, rooted in a perceived fear that the events of the summer and autumn of 2015, marking the peak of the so-called European Refugee crisis, during which thousands of people requested asylum in Europe, could be revived. This is co-constituted with the idea -which seems to be transcending the mainstream political spectrum -that new arrivals should not be 'too much for Europe to handle' (Leurs et al., 2020: 679). Progressively, the idea that migrants are pushed over by Europe's enemies, or ungodly 'friends', as we will argue here, to harm it, is consolidated, justifying more border securitization regardless of the cost to human lives. Borders and identities are underpinned by practices and discourses which circulate in various media, online and offline (Engelbert et al., 2019). In social media, the far-right is mostly outspokenly antimigrant, while the populist right-wing often uses coded antimigrant discourses which have contributed to mainstreaming European exclusionary politics through the 'politics of fear' (Wodak, 2015). EU's migration agenda falls little outside the restrictive frames proposed by the far-right, and the populist, authoritarian right-wing, as the characterizations fortress Europe and apartheid Europe suggest (Ponzanesi and Blaagaard, 2011). Naturally, the EU is not openly suggesting that deaths at the borders are acceptable side-effects of its border securitization, but the far-right on social media may in fact be uttering the unsayable when they are asking for explicit violence towards migrants, as indicated by the use of hashtags such as #rapefugees (Siapera et al., 2018) or #refugeesnotwelcome (Kreis, 2017). The knot that ties together ideologies that advocate fierce antimigrant politics and mainstream European migration politics is perhaps the elephant in the room that critical scholarship should not hesitate to pinpoint. Importantly, the connections between extreme antimigrant and mainstream EU migration discourses and politics need to be drawn out at times when fears related to migration give rise to the dehumanization of people on the move, effected by the othering of other key migration actors, the legitimization of border atrocities, and the turning of the death of migrant others into a banal occurrence. In view of the above, we offer an in-depth analysis of othering in Twitter around the hashtag #IStandWithGreece, associated with a border crisis at Europe's periphery in early 2020, concerning neighbouring countries, Greece and Turkey, with a turbulent history of antagonisms. In so doing, we reflect on the extent to which the othering processes we track in these tweets, by far-right elite and non-elite voices, offer a more overtly racist expression of official EU discourses about migration to the extent that they may be mutually constituted. Hashtag IstandwithGreece, created in the midst of the events studied, began trending on Twitter and around it a 'network of intolerance' was organized to strategically promote ideologies ranging from white supremacism to Greek nationalism, glued together by antimigrant positions (Avraamidou et al., 2021). Focus on these events, and on Twitter, allowed for an in-depth analysis of transnational boundary making, its relationship to Greek nationalism, and European border politics and discourses. --- Boundary construction: othering and the mediated European migration debate Othering is a 'process of differentiation and demarcation between "us" and "them"between the more and the less powerful -and through which social distance is established and maintained' (Lister, 2004: 101). As Lister indicates, the differentiation of us from them is done in a way that favours the ingroup and undermines the outgroup such that the outgroup's subjugation and exclusion from the ingroup follows naturally. Postcolonial approaches to othering understood it as a process of portraying the other, the Indigenous communities, as different, stressing their inferiority to the Self, the colonizers (Krumer-Nevo and Sidi, 2012). Capitalizing on the constructed inferiority of different others, discrimination and exclusion was justified by the dominant group, enabling it to maintain unequal power relationships. Thus, othering aids the preservation of hierarchies and the propagation of ingroup-domination/protection policies. In the case of migration, the othering of migrants is linked to specific policies such as border management and securitization. Representations of the other are instrumental in that they justify political decision making (Glaveanu and Laurent, 2015). This connection is achieved through a process of political imagination (Glaveanu and Laurent, 2015) that is creating imageries, future scenarios, of how life would be if the other penetrated it. This process consists of building representations of others which are then used to limit or enhance those others' abilities to achieve their (imagined) political aims (Glaveanu and Laurent, 2015). Labelling 'people on the move' as migrants versus refugees, for example, is deeply political, leading to the endorsement of certain representations of these people which give rise to opposing political demands: the support of open or closed borders (Crawley and Skleparis, 2018). Similarly, crisis imaginaries legitimize the actions of those in power, in view of utopian futures, that, as in the case of migration, capitalize on the threatening nature of the other (Krzyzanowski, 2020). Mediated othering of migrants largely reflects and reproduces the above processes and has been the focus of ample research: studies have shed light on how traditional media in Europe have represented the migration debate within and across member states during the so-called 2015 Refugee crisis (Drüeke et al., 2021;Georgiou and Zaborowski, 2017) and explored also the role of more transnational sites such as social media (Siapera et al., 2018). Pre-existing antithetical migration frames (Van Gorp, 2005) re-emerged during the European refugee 'crisis' (Georgiou and Zaborowski, 2017): the humanitarian one, representing refugees as victims and as passive actors in need of help, and the fear and securitization one, construing refugees as intruders who are agentic, opting to perform illicit actions and practices such as crossing borders without permission, and therefore threatening. All in all, the strategic use of the word 'crisis' by the media to describe migration in Europe set the ground for boundary construction between 'us' and 'them', regardless of the valence of the discourse (Georgiou and Zaborowski, 2017). Another aspect of othering as noted by Dervin (2012) is defining or redefining the Self in relation to the other. Within the migration debate, the so-called influx of non-Europeans into Europe has triggered discussions around European identity or Europeanness, in an attempt to redefine the self in juxtaposition to the other. Himmel and Baptista (2020) explored migration discourses in German and Portuguese media between 2011 and 2017 and found, for example, that when migrants or refugees were assigned as 'the Other', being European was equaled to not being Muslim. Interestingly, research illustrated the multiple and often conflicting media representations of Europe, and the EU in particular, in national media, thus shedding light on its inherent fragmentation and intra-antagonisms for and against close migration policies, and on what being European means (Avraamidou et al., 2018;Avraamidou, 2020). Othering can take particularly interesting forms in social media for a host of reasons. Social media host transnational discussions (Froio and Ganesh, 2019) while also allowing for national affiliations which means that the self is not constant and can alternate between the national self or a more overarching self, such as being European. There is, for example, research showing how transnational far-right audiences bond in social media to spread antimigrant ideas (Froio and Ganesh, 2019), as well as how nationalism emerges in discussing refugee issues on Twitter among users affiliated with specific national contexts (Bozda<unk>, 2020). Research is, however, rather undeveloped when it comes to how transnational and national debates online intertwine in spreading antimigrant stances. In addition, the presence of antimigrant and nationalistic discourse capitalizing on othering processes appears stronger online. Migrant othering in Twitter was found to be more blatant compared to traditional media in 2015, as indicated by the frequent use of openly negative hashtags such as #rapefugees, with the far-right being almost exclusively responsible for endorsing the representation of refugees as terrorists or rapists (Siapera et al., 2018). Yet, in-depth examination of how the Othering of both refugees as well as other actors is motivated in social media is rare, as the field is dominated by big data approaches (Bozda<unk> and Smets, 2017). Simultaneously, and partially perhaps due to this dominance, research is rather undeveloped when it comes to the intersections of how transnational othering in social media intertwines with ethnonationalism, and vice versa. This is a gap our study seeks to address through a qualitative analysis of antimigrant tweets during a crisis at Europe's borders. --- EU border politics and the Evros events: Quo Vadis Europe? Our study focuses on the Evros events of late February and mid-March 2020, broadly attributed to Turkey's decision to suspend a controversial agreement with the EU regarding the handling of refugees fleeing Turkey to enter Europe. Briefly, the EU-Turkey migration deal of March 2016 meant that migrants crossing into Greek islands would be relocated to Turkey and in return, Turkey would receive monetary compensation from the EU, and Turkish citizens received a visa waiver for EU travel. The deal's suspension resulted in refugees, previously locked in Turkey, moving towards the Turkish, Greek, and Bulgarian land-border demarcated by the river Evros, to enter the two EU member states. Thousands also attempted to cross to the Greek islands. Greece closed its borders and deployed security forces. Reportedly, members of the far-right from Greece and other EU member states travelled to the border in early March 2020 (Human Rights Watch, 2020). Furthermore, Greece suspended all asylum procedures until mid-May 2020, citing the COVID-19 pandemic, and progressively adopted more restrictive containment measures for refugee populations in its camps (Tazzioli and Stierl, 2021). The Evros events concerned primarily neighbouring countries Greece and Turkey, with a turbulent history of geo-political antagonisms in the area. Evros has been a place of'symbolic struggles' (Gkintidis, 2013) where Greek and Turkish nationalisms tend to compete as a Greek-Muslim minority inhabits villages around the border (Gkintidis, 2013). The events were also of European interest as Evros composes EU's external border, which rendered them relevant to the EU/Turkey migration deal. The European Commission's (2020) President visited the area and, alongside some ambiguous remarks about Turkey's role, she spoke of Greece as 'our European <unk> (English: shield)', and shared the conviction that 'we will hold the line' (Statement, 2020), thus giving the EU's decisive support to border closure and securitization. The year after, Greece proceeded to an extension and surveillance upgrade of a fence in the Evros area which coincided with the Taliban's resurgence and dominance in Afghanistan of August 2021. Greek officials argued that this border upgrade would enable it to deter Afghanis from entering, declaring that 'we will not allow a reliving of the scenes of 2015' (Reuters, 2021). At the time, the Czech Prime Minister also said that there was no room for Afghani refugees in Europe (Euractiv, 2021). Importantly, the EU corroborated these concerns: an EU Ministerial statement appearing to focus on the prevention of a new humanitarian crisis, underlined the EU's goal 'to prevent the recurrence of uncontrolled large-scale illegal migration movements faced in the past' (Council of Europe, 2021). Fears about massive numbers of Afghanis arriving in Europe are only part of a series of relevant analogies made by EU and national leaders in the post-2015 era. Earlier, the controversial Belarusian President Aleksandr G. Lukashenko was accused of using migrants at Belarus' borders with Lithuania, an EU member state, to conduct a 'hybrid' war against the EU (Euronews, 2021). 1 The situation escalated and by the end of 2021, the Polish-Belarusian border had turned into a battle-ground, resembling the Evros 2020 events, around which a blame-game between key actors evolved while hundreds of people remained trapped in deteriorating conditions and some died helpless (Al Jazeera, 2021). The European Commissioner for home affairs tweeted, among others, at the peak of the events that Europe's 'urgent priority is to turn off the supply coming into Minsk airport.' (Johansson, 2021), 2 with supply meaning people on the move. Notwithstanding the cruel instrumentalization of migrants by several international actors and authoritarian leaders, the EU's preoccupation with the threat of just another influx, and its routinely dehumanizing metaphor of migrants as weapons, has implications for attitudes and policies about migration from the global south specifically. Namely, in the backdrop of this debate, border securitization across the EU increases with severe consequences for the lives of people en route to Europe. Balkan states continue the harassment and intimidation of people on the move, Poland and Lithuania are constructing their own fences, and EU member states of the Mediterranean Sea (e.g. Greece, Croatia, Cyprus) increased 'pushbacks' in the last 5 years, violating international regulations, such as non-refoulement, in the name of national and European security. 3 Ultimately, EU border politics do cause deaths, and EU's controversial Border and Coast Guard Agency (FRONTEX) is routinely accused of playing an active role in pushbacks. 4 Migrant lives matter and it is an imperative to outline how these othering processes, that border discourses engender, are linked with, and thus feed into, the existing EU border regime. --- Research rationale, methodology and method Social media, and particularly Twitter, have become a 'powerful communication tool during many kinds of crises, political or otherwise' (Howard and Kollanyi, 2016: 1). Twitter affordances allow communities outside a specific country to debate on topics of wider appeal (Kreis, 2017). An example is migration, which has drawn the attention of a transnational audience from the far-right (Froio and Ganesh, 2019) which often bond around negative evaluative hashtags in English such as #refugeesnotwelcome, used during the so-called Refugee crisis (Kreis, 2017) or #Islamexit, used after the Brexit referendum (Evolvi, 2019). Occasionally, elite and non-elite far-right voices may coordinate to spread antimigrant positions in social media using more covert hashtags too (Avraamidou et al., 2021). Simultaneously, on Twitter, nationalism is reproduced from below when twitterers of a particular country debate refugee issues (Bozdag, 2020). Our focus is on an evaluative, yet ambiguous hashtag, #IstandwithGreece, used in the midst of the Evros events and which became the most popular hashtag on Twitter around which a social network was organized, promoting antimigrant views (Avraamidou et al., 2021). The tweets shared by #IStandWithGreece's top influencers form our research focus. Namely, we sought to interrogate the othering discourse in social media post-2015, by looking for wider reasons for othering, going beyond the well-documented representation of the threatening migrant. In so doing, we performed a qualitative thematic analysis to identify meanings promoted, approaching each tweet in our data-set as an integrated discourse comprised of text, multimedia, and other Twitter affordances (Evolvi, 2019). Below, we describe the data gathering method that resulted in the collection of 1,850,954 tweets, the sampling strategy and the qualitative analysis of a smaller sample of 818 tweets. To clarify, our qualitative sampling was purposeful, driven by the research question and not by a concern for representativeness (Mason, 2002). --- Data collection and sampling rationale When Turkey declared the suspension of the 2016 deal with the EU, at the end of February 2020, we started collecting tweets using relevant keywords (e.g. Greece refugee(s), refugeesgr, Turkey asylum) through DMI-TCAT (Borra and Rieder, 2014) which provides a real-time stream of tweets, including retweets and replies. The 31st March 2020 is the last day of our sampling period because we observed a local minimum in the 5-day rolling window variance in the frequency distribution (see Figure 1). In our data, #IStandWithGreece was the most frequently used hashtag, used in about 32% of the tweets collected. Using social network analysis (Markham and Lindgren, 2014), we identified the network's top 10 influencers and the sub-network communities in which they belonged. As explained (Avraamidou et al., 2021), ethno-nationalists, white supremacists, and other antimigrant voices almost exclusively dominated the debate under #IStandWithGreece. For example, far-right British social media persona, Katie Hopkins, and members of the French Identarian movement, were among the top 10 influencers all eventually banned from Twitter (Avraamidou et al., 2021;BBC News, 2020). Our understanding of the meanings that #IStandWithGreece influencers promoted was informed solely from co-hashtag analysis, which provided us with descriptive insights. To identify a more manageable number of tweets for an in-depth analysis in line with our current research question, we isolated the re/tweets of the top 10 influencers tweeting in English and in the Greek language, in which we are fluent. Two of the top 10, the British tabloid Mail Online and French President Emmanuel Macron, had not tweeted using #IStandWithGreece, so we added the next two top influencers (both accounts currently suspended). Number 11 did not re/tweet in either English or Greek but Top Number 12 re/tweeted 29 tweets in English. This provided us with a total of 818 tweets of which 607 had a photo or a video (URL) and which we deemed sufficient for the in-depth analysis we aimed for. Overall, they concerned re/tweets of five of the most representative communities of the network: Anglophones 1 and 2 which discussed the events, adopting a global, conspiratorial perspective; the Greek community comprising mostly Greek nationalists; the Germanic community headed by the German and Austrian farright; and the French community (its top influencers tweeted only once in English). This led to the Greek community being over-represented in our qualitative sample (see Table 1). Although this can be considered a limitation, it enabled us to have rich data to study a specific nationalist discourse, that developing among users tweeting in Greek, with more transnational antimigrant discourses in English. Furthermore, as clarified, we were not preoccupied with representativeness in this analysis. --- Thematic analysis Our thematic analysis, 'a method for identifying, analysing, and reporting patterns (themes) within data' (Braun and Clarke, 2006: 6), was driven by our primary and secondary question, who is othered and why. In identifying and reporting themes as active researchers (Taylor and Usser, cited in Braun and Clark, 2006: 7), we primarily focused on nomination and predication that is the representations of actors and their characteristics and the representation of positive self and negative other (Wodak, 2015: 52). This guided our coding too: 'Codes are tags or labels for assigning units of meaning to the descriptive or inferential information compiled during a study' (Miles and Huberman, 1994: 56). In this case, the entire tweet was coded which is a rather pragmatic choice given the rather small length of a tweet, and a choice that allowed us to avoid de-contextualization of codes. Each code represented an actor of the events, and additionally, we had other codes drawn from our data (e.g. reference to the events as 'clashes'). Coding was exhaustive as all the tweets were assigned at least one code. To identify the themes, we engaged in a close reading of coded tweets reflecting what Braun and Clark (2006: 16) referred to as a'recursive process' of moving 'back and forth as needed, throughout the phases', looking for unanticipated insights, one of the great advantages of qualitative research (Mason, 2002). Although focusing on meanings, we do shed light on discursive instances that promote Us/Other binaries when relevant (e.g. ad hominem attacks) or to highlight common antimigrant discourses or rhetoric strategies (e.g. metaphors) adhering to the tradition of critical discourse analysis. This is in line with Reisigl and Wodak's assertion (2001: 41) that 'through discourse, discriminatory exclusionary practices are prepared, promulgated and legitimized'. As we studied Twitter discourse, we also follow Jackson and Welles (cited in Evolvi, 2019: 391) paying attention to meanings provided by the combination of language with images and other affordances (e.g. videos). Finally, in reflecting on the affinity of this antimigrant discourse with the EU's, we do not aim for a strictly comparative analysis, but to rather offer a critique of how we see this empirically grounded othering and the political aims connected with it relating to broader mainstream EU's border politics, with the expectation that this will motivate further work, to foreground relevant arguments more systematically. --- Findings Turkey and migrants as imminent and a real threat: 'Erdogan's hordes of illegal immigrants' #IStandWithGreece's influencers represented Turkey and migrants as a joint, mutually reinforcing threat to Greece, Greek people, and Europe or the West. Turkey was represented as the mastermind behind the border events and migrants were construed as threatening due to their sheer size, gender composition and propensity to radicalization. An indicative tweet exemplifying the representation of Turkey and migrants as a coordinated threat against Greece, while assuming that all migrants were en route to Germany, read, The scale of Erdogan's operation against #Greece is huge. Tens of thousands of migrants have been bussed to the border & constantly attack Greek border guards in their attempts to storm the border, march through the Balkans & reach Germany #GreeceUnderAttack #IstandWithGreece. (Example 1: 3 March 2020) This tweet interpreted the events as an orchestrated attack implemented by migrants and coordinated by Turkey who, it was claimed, was literary 'bussing' them to Greece. The word 'operation' is telling, as it conveys the message that the events were preplanned. The tweet is accompanied by a video showing people (migrants) in some form of congregation with smoke visible at the background (Figure 2). Evidently, this choice of war-like images facilitates the representation of the border as a battlefield and serves to underscore the severity and the urgency of the situation. In similar tweets, the migrant other was intentionally gendered (male), whereas women and children were almost absent, perhaps because reference to them would challenge the ubiquitous image of a threatening Other. Where reference to women and children was made, what was discussed was their absence. An indicative tweet noted, Count the women and children among the migrants at the Greek border. They can't make up more than 5%. (Example 2: 3 March 2020) Thus, influencers corroborated the representation of an unquestionably threatening migrant other, merely on the basis of gender and age (young male), and assumed religion. The multiple images that the network shared of young, violent men collide with the notion of the deserving other, connoting that these migrants are not worthy of being let in, as the following suggests: Sorry but when people who seek asylum use tear gas grenades to achieve their goal [they] are not refugees they are invadors and they are not accepted. BORDERS CLOSED FOREVER. #Greece_under_attack #IStandWithGreece' (Example 3: 5 March 2020) The above Twitterer appears to be apologetic claiming that had people at the border been 'genuine' refugees, then they would be accepted. But given these people carry tear grenades instead, this strips them from any right to claim refugee status. Importantly, not being real refugees is eventually used to justify permanent border closure, probably the ultimate restrictive migration policy that a nation state can implement, and a measure that ensures that these non-refugees stay away permanently, as the capitalized 'FOREVER' denotes. Another dominating male presence in these tweets was Turkey's leader, Tayyip Erdogan, represented as the vicious mastermind tricking migrants to believe that they would be able to cross from Greece towards other European countries. Reference in one tweet to 'Erdogan's hordes of illegal immigrants' alludes to how these events are in part a one-man show controlling the migrant crowds. Moreover, a Twitterer compared Erdogan to Hitler but, funnily, also prompted Greek soldiers to kill border crossers: Using Twitter affordances, as the Hitler-Erdogan analogy is a hashtag, the tweet becomes an amalgam of paradoxes. The Twitterer considered calling someone Hitler as an insult, but then demanded the killing of anyone crossing borders and used Hitler, an ultimate figure of intolerance, to support it. This ad hominem attack against Erdogan by association with Hitler was coupled with insults that repeatedly described Erdogan as a madman, a criminal and a tyrant. Ad hominem attacks and insults are used to pathologize Erdogan, representing him as a culpable supervillain; capable of calculated acts but also unable to moderate the uncanniness of his acts due to psychological instability. Erdogan's supervillain nature can only be topped by that of the jihadists -another common representation of people at the border -and together they form very potent anti-heroes. --- Historical 'other': '#Greece guarding the Thermopylae of the Western world' The history of intergroup relations in the region as merely one of ethno-religious conflicts leading to territorial disputes between two main homogenized groups, Christians and Muslims, was deployed by #IStandWithGreece influencers to justify othering. These representations were expectably infused with Greek nationalism promoted by Greeceaffiliated influencers and diffused widely by non-Greek influencers. The tweet below is a good example of a Helleno-centric historical representation, according to which Turkey is a serial perpetrator attacking Greeks, and the Evros events form recent evidence of the omnipresent Turkish aggression: Turkey committed genocide on the Greeks -Turkey invaded Cyprus in 1974, occupied for 46 years and violate human rights and international laws -Turkey also violates Greek airspace daily -Turkey is weaponising immigrants. But apparently #Greece is the aggressor. #IStandWithGreece. (Example 5: 4 March 2020) This tweet refers to certain events in an ostensibly historical sequence to illustrate the continuity of Turkish aggression against Greece and Cyprus, an EU member state, which, according to Greek nationalists, is Greek, despite the long-standing presence of a Turkish-Cypriot community on the island. The reference to a Greek genocide probably alludes to the 1922 conclusion of the Greco-Turkish war in Asia Minor that led to -among other things -the formation of the Turkish Republic and a large compulsory population exchange between the two countries. This is linked to the 1974 Turkish invasion of Cyprus, and Turkey's alleged violations of Greek airspace occurring simultaneously with the Evros events. The connecting thread justifying lumping together Arab migrants and Turkey is Islam, which is deterministically associated to fundamentalism and terrorism. Therefore, migrants by virtue of being Muslims, fit the historical enemy narrative. The tweet below illustrates vividly the representation of these events as an omnipresent conflict between the Crescent, represented by Turkey and Muslim migrants, and the Cross, defended by Greek soldiers: @KTHopkins Erdogan says he will flood Europe with Muslims until the Crescent triumphs over the Cross. This is the border with Greece. Mosque in the distance. Soldiers of the Cross holding the line. This is biblical. #IStandWithGreece. (Example 6: 8 March 2020) In another example (Figure 3), a myth is invoked to sketch Turkey and migrants as a joined enemy not only of Greece, but of Europe more broadly. It is accompanied by a photo illustrating a number of weaponized men resembling Jihadists, getting out of a wooden horse. The tweet reads, Turkey trying to pull a trojan horse on Europe but having to get through Hellas that originally invented it proves impossible. #IStandWithGreece. (Example 7: 1 March 2020) The analogy attempted here is interesting, not least because the Trojan horse myth, which the image portrays, is of a time when neither the Greek nor the Turkish nations existed. It refers to a trick that causes the target to invite a foe into its own home. According to mythology, Odysseus, a rather canny figure, conceived the idea of using it to deceive the Trojans and enter their city during the Trojan war. The made-in-Turkey Trojan horse, as the red flag with the white crescent and the star indicates, corroborates the narrative that Turkey is the mastermind behind the border events, using stratagems of this nature to implement a secret agenda to damage Europe by sending over dangerous, radical Islamists. Despite the triumphant image of jihadists, Greece is the ultimate winner because it is historically superior, for having invented the Trojan horse. A primordial understanding of nations is evident as reference to 'Hellas' instead of Greece is the characteristic way the Greek nationalist dictatorship of the 1960s referred to Greece. Another tweet (Figure 4) clearly represented the events as a conflict of civilizations, the oriental versus the occidental, drawing from antiquity. It noted, #Greece guarding the Thermopylae of the Western world once again. This time you can help. Pressure your gvmts to provide us with military assistance & to #Sanction_Turkey now. Our A photo of a barbed wire in front of policemen standing underneath a sign with the inscription 'Greece Welcome', and the Greek flag, completes the tweet. Thermopylae refers to a place in Greece, literally meaning Hot Gates, where a battle between Persians and Greeks took place. The Persians outnumbered the Greeks but the latter demonstrated enormous bravery, according to the same narrative. In this context, the Hot Gates under siege are those of the entire Western world. The portrayal of the barbed wire, and security forces in front of the welcoming sign, is telling of the paradoxes of borders as non/entry points. Articulations of Europeanness: #Europe they are at the gates. This theme focuses on the ways the network's tweets represented Europe and Europeanness, identifying a tension between ideas about Europe at the symbolic level (e.g. Christian values), and the EU and its leadership, as a failed promoter of European values. #IStandWithGreece' influencers were sending a 'wake-up' call to the leadership, warning them of the imminent threat against the whole of Europe. One characteristic tweet noted that '#Europe they are at the gates' (Example 9: 1 March 2020), and another read, Europe, as you sleep well tonight, remember it is the Greek people who are keeping Erdogan's hordes of illegal immigrants out of your countries. We will continue to do this, but do you stand with Greece? #IStandWithGreece. (Example 10: 3 March 2020) 'Europe' here is most likely used to refer to western European countries -typically the desired countries of destination for refugees -and the EU's powerful decision makers. 'Europe' is represented as not wanting these 'illegal migrants' but conveniently expecting Greece to do the job for them. The tweet is essentially inquiring whether Europe will ever show solidarity with its ailing members, therefore casting doubts about its role and intentions. These accusations are not really meant to be othering, in the sense that the attempt to present Europe as morally inferior or lacking in bravery is not done in order to distance it from the Self. They are rather a form of provocation and a cry for help and union. Nevertheless, the opening sentence implies a level of hypocrisy from Europe because, while Greeks are in a brutal fight with an enemy keeping all Europeans safe, Europe is passive ('sleeping well'). Influencers recurrently reinforced the long-standing divide between affluent Western Europe and the more exposed and victimized South and East for allegedly suffering more from migration due to its geographical proximity to areas of conflict. In the following tweet, top influencers are cheering central Eastern Europe (probably referring to Hungary and Poland), for pulling its weight and defending Europe at large, while western Europe would not help, thus picturing Europe as a fragmented union: Last night, the migrants bussed to the Greek border by Turkish authorities continued to attack Greek Border Guards with the use of rocks, fireworks & tear gas cannisters Western Europe won't help #Greece, #CentralEasternEurope must do it! #GreeceUnderAt

This article explores how Europe's border crises in the post refugee 'crisis' years were discussed in the micro-blog Twitter, through an in-depth analysis of boundary making. Our focus is on the tweets of the top influencers of the hashtag #IStandWithGreece who strategically promoted ideologies ranging from white supremacism to Greek nationalism, glued together by an antimigrant stance during a border 'crisis' at Europe's periphery. This network of intolerance promoted a representation of migrants as 'pawns'; seen like a chess piece, with no value in their own right, literally pushed towards Europe by Turkey, who elevated them into a sizable threat. Within this, Europe was represented as a paradoxical other, the fallen Self, for not rising up to the opportunity to protect its sovereignty and identity through more securitization. Despite being diffused by extreme antimigrant Twitterers, we argue that these tweets offer a more overtly racist expression of otherwise mainstream European (Union) discourses and politics on migration. Effectively, #IStandWithGreece's influencers functioned as Europe's alter-ego mouthpiece, saying the unsayable using social media, and their affordances contributing to the normalization of an oppressive and restrictive European border management.

morally inferior or lacking in bravery is not done in order to distance it from the Self. They are rather a form of provocation and a cry for help and union. Nevertheless, the opening sentence implies a level of hypocrisy from Europe because, while Greeks are in a brutal fight with an enemy keeping all Europeans safe, Europe is passive ('sleeping well'). Influencers recurrently reinforced the long-standing divide between affluent Western Europe and the more exposed and victimized South and East for allegedly suffering more from migration due to its geographical proximity to areas of conflict. In the following tweet, top influencers are cheering central Eastern Europe (probably referring to Hungary and Poland), for pulling its weight and defending Europe at large, while western Europe would not help, thus picturing Europe as a fragmented union: Last night, the migrants bussed to the Greek border by Turkish authorities continued to attack Greek Border Guards with the use of rocks, fireworks & tear gas cannisters Western Europe won't help #Greece, #CentralEasternEurope must do it! #GreeceUnderAttack #IstandWithGreece. (Example 11: 7 March 2020) Those severely targeted by the network were certain high-level EU politicians. In the following, the EC President is characterized as a 'disgrace' and Erdogan as a 'tyrant': The @eucopresident is a disgrace expressing condolences to #Erdogen -Why is the #EU appeasing this tyrant & his rogue, blackmailing, genocidal, terror state? They should support #Greece against this Erdogan organised invasion #GreeceUnderAttack #IStandWithGreece #Cyprus #Idlib (Example 12: 1 March 2020) Simultaneously, Greek-affiliated influencers used Twitter to convince Europeans to take action to protect Greece and by extent, the entire ingroup/Self. This is illustrative of a desire to articulate the Self as genuinely European: Fellow Europeans, YOU must take a stand. Do YOU want us to be the bridge to your home? WE must find a solution to this, TOGETHER. Please apply pressure to your politicians. This is serious. #IStandWithGreece #GreeceDefendsEurope #Greece_under_attack. (Example 13: 4 March 2020) The tweet makes a clear distinction between ordinary 'fellow', Europeans and political leaders. Ordinary Europeans are represented as the real carriers of Europe's ideals for union and internal solidarity in juxtaposition to the political elites who undermine authentic Europeanness. A video showing mostly young male migrants, one of them shouting in English 'We want to go to Europe' and adding that 'Your country is a bridge to us', completes the tweet by supporting the network's main argument that Greece was protecting Europe from unwanted others who aimed at reaching destinations outside Greece. Interestingly, the speaker does not really consider Greece as Europe since he notes that they want to go to Europe, which is a striking contrast to the network's symbolic representation of Greece as the ultimate heroic European, or as one Twitterer noted the 'foundation of Europe', adding that 'If Greece fails, the rest of Europe will fail too' (29 February 2020). Christian Europe was another assertion of the network grounded around the idea that Evros was not only the border of Greece and Europe, but also of Christianity. Indicative hashtags foregrounded this assertion, such as #LetsReturnToChristianity and #ChristianityUnderAttack, while one tweet noted, 'Greece is Europe! Islam is not! #IStandWithGreece' (Example 14: 1 March 2020). Finally, a tweet read, Interestingly, Christian identity is treated by influencers as a unifying element between Greece and Europe, silencing Greece's Orthodoxy which for Greek nationalists was, in the past, viewed as a dividing element distancing Greece from other EU member states (cf. Molokotos-Liederman, 2003). --- Discussion and conclusion: the 'pawns' discourse and why it matters This study sought to answer who was othered and why in a Twitter debate around #IStandWithGreece, triggered by border incidents between Greece and Turkey in early 2020. The tweets studied belonged to the top influencers of what we have already identified as a 'network of intolerance' promoting ideologies from within the far-right, cutting across national contexts of the Global north (Avraamidou et al., 2021). Focusing on the network's English and Greek tweets, we inevitably located othering processes at the intersections of transnational far-right perspectives, which share rhetoric and ideologies (Levi and Rothberg, 2018), and of Greek nationalism, therefore bringing new insights on migrant othering in social media. Subsequently, we reflect on the extent to which the othering promoted by these antimigrant influencers may serve mainstream EU migration politics. Our findings yielded the othering of three actors of the events, Turkey, migrants and Europe. Mirroring earlier media and social media research, othering of migrants was centred around the cultural and security threat (KhosraviNik et al., 2012;Lynn and Lea, 2003) and the national threat (Bozda<unk>, 2020). While the predominant representation of migrants during the 2015 'crisis' was mostly that of a 'faceless, voiceless flood of Numbers' (Nikunen, 2019: 415), in this network's tweets, we located their representation as 'pawns'. According to this, migrants are likened to a chess piece, with no value in their own right, as they are literally pushed towards Europe by another actor, Erdogan's Turkey, who elevated them into a sizable threat by turning them into 'invaders'. This representation is evident in recurrent references across the three themes that migrants 'have been bussed to the border', that they are 'Erdogan's hordes', that Turkey was weaponizing them, or Hopkins' tweet that Erdogan was threatening to 'flood Europe with Muslims'. Moreover, the Trojan horse analogy is illustrative of this migrant pawn representation because, had it not been for the made-in Turkey wooden horse, jihadists would not have made it to Europe. This dehumanizing, non-agentic representation of migrants exacerbates an understanding of them expendable, but it does not contradict their representation as 'invaders'. In contrast, the two are strategically consolidated because migrants' agency to pose a threat depended exclusively on Turkey; they can invade only because Erdogan moves them across the chessboard towards Europe. Therefore, their fate is Turkey's responsibility, which has willingly, if not purposefully, sacrificed them as pawns (see also Hoops et al., 2016). Europe is also wiped of any quilt or responsibility for these people because they are merely seen as serving the other actor's aims. The consolidation of these representations (pawn/invaders) has implications for the future; it provides another argument to justify Europe's border securitization, foregrounding its representation as a paradoxical other, the fallen Self, for not rising to the opportunity to decisively protect its sovereignty and identity through more securitization. #IStandWithGreece influencers homogenized people at the border as Muslim fundamentalists, which is common in anti-Muslim online debates that ignore the complex identities of asylum seekers (Evolvi, 2019). The representation of 'Middle-Eastern masculinity as threatening' (Rettberg and Gajjala, 2016: 179), a common strategy of Twitterers using #refu-geesNOTwelcome in 2015, was used in this network too. This allowed Twitterers to exacerbate the bond between the people at the borders and Turkey, and the corresponding representation of the events as a religious and cultural conflict, as the Crescent versus the Cross metaphor illustrates. Similarly, the network endorsed Helleno-and Eurocentric representations of modern and ancient history in which Turkey was portrayed as an eternally malevolent actor in the geo-political chessboard. Various motivated historical analogies were mobilized (e.g. genocides and invasions) to convince readers that Turkey's enabling of would-be terrorists into European soil stems naturally from its historical desire to conquer or destroy Europe. Notably, Turkey has long been represented as one of Europe's key historical external others alongside Russia (Morozov and Rumelili, 2012). Simultaneously, Christian-Greece is seen as protecting the West from barbarian others from the East due to its position at the cross-roads between West and East (Tziampiris, 2015), as the Thermopylae battle analogy claimed. Greece is a separate nation but its Europeanness is taken-for-granted, both by Greek-affiliated and non-Greek-affiliated influencers, despite this narrative being disrupted by migrant people explaining that they want to 'go to Europe', implying that Greece, to them, is not Europe. The EU-Greece relationship is even more complex: Greek Twitterers called for solidarity, but they were rather disappointed or expected little from European leadership, as evident in the rhetorical question: 'but do you (Europe) stand with Greece?'. The othering mainly of Western Europe was motivated by representing the EU and its leadership as passive hypocrites, accused of not protecting their own people and washing their hands instead. That Europe is passive and hypocritical when it comes to migration crises matches earlier media criticism but from a pro-migrant perspective (Kadianaki et al., 2018). Effectively, the EU was failing to protect the network's racialized notion of Europeanness as the Christian, occidental self, and keep out those migrants who not only posed a tangible threat by violating its sovereignty but who on top of that were unworthy of coming to Europe, and by extent of becoming Europeans, due to their Muslim culture (Wodak, 2015: 42). This is evidence of a strategic deployment of both Euroscepticism and Eurocentrism to exclude others in far-right transnational discourse. Again, the self is personified in the image of Greece and its soldiers, who are flagging the cross and guarding Europe's borders to safeguard its space, but also its identity and values from vicious anti-heroes. Europe and European leaders are being requested to follow the lead of Greek defenders. The #IStandWithGreece network demanded border securitization in its most extreme form for the global north, and not just Greece's border with Turkey. For example, metaphors and historical analogies such as Erdogan being likened to Hitler were used to not only other Turkey, but to also authorize killings of migrants. Therefore, the network's pseudo-negation of Nazism, that we located, is rather evidence of its partial identification with it, and of the ingroup it is constructing, an intolerant Self. Migrant othering was, in this instance too, begetting border securitization and militarization (Rheindorf and Wodak, 2018), which ironically is not too far from Europe's concurrent border politics. Yet, these Twitterers assumed that Europe was too tolerant, promoting rather flexible migration policies. Thus, they construe the event as a border management problem to which the only solution is the implementation of stricter border and migration policies. Simultaneously, it is also grounding a demand to protect Europeanness, illustrating how exclusionary border politics and identity politics are mutually constituted. The clear demand of a Twitterer to Europeans to protect 'our way of life' resembles the initial naming of a controversial newly introduced portfolio of the EU Commission, 'Protecting our European Way of Life', which was altered to 'Promoting our European Way of Life', after backlash (Golesorkhi, 2021). The reference of another Twitterer to Greek soldiers as 'holding the line', is almost identical to Von der Leyen's call for Europe to 'hold the line', mentioned earlier. Effectively, the network and the EU agree that keeping unwanted others out is (also) about maintaining our racialized Europeanness (De Genova, 2017). We argue therefore that these seemingly marginal voices from within the far-right that bonded around #IStandWithGreece were serving not a marginal political agenda, but rather the concurrent EU migration agenda at a supranational and national level, pushing it to reach new extremes, such as killing Muslim border crossers in cold blood. We conclude that this network of intolerance, #IStandWithGreece, functioned rather as Europe's alter-ego mouthpiece, saying the unsayable using social media and their affordances. Social media thus provided the symbolic means for the normalization of a ruthless reality at Europe's land and sea borders. To conclude, Europe is shown in yet another mediated debate about migration as being threatened by another alleged 'influx' of migrants. This time, migrants were represented as pawns of a more vicious other, Turkey. The othering we located may be common because fierce opposition to migration, and particularly to Muslim migrants from the East, as circulated in various media and particularly by the far-right, is long-standing. In their simplicity and very obvious essentialization of identities as fixed in time, these representations can be adjusted to almost every border event; they are the archetype of a Eurocentric exclusionary antimigrant discourse infused with nationalism. Influencers regardless of their national affiliations put similar emphasis on the need for border securitization. Regarding the othering of Turkey, we observed emphasis on both its historical turbulent past with Greece, relevant to Greek nationalism, and on its peculiar relationship with Europe and the west, relevant to Eurocentric approaches across the global north and simplistic dichotomizations of the civilized, Christian self and the inferior, Muslim other. The two were reinforcing each other for the common goal discussed, border securitization. Future research could shed more light into how social media offers space for this transnational dialogue of antimigrant voices to evolve and reinforce each other, during other border events. The subsequent mediated discourse about the mid-2021 Afghan events and those at the Poland-Belarus border later the same year perhaps exemplifies this as it appears to promote seemingly non-transcended distinctions between us and the other, with the undeserving other being 'weaponized' by long-standing enemies (Turkey or Russia and its puppet states), and during which Europe's member-states, in proximity, flag their national interests. Therefore, the 'pawns' representation returns to justify atrocities, and ultimately, to legitimize the continuation of an exclusionary, repressive and oppressive European migration regime which is racialized. Future research can better warrant the affinity of these extremist discourses with the mainstream and the politics they underpin, as identifying othering processes in social media is the first step towards uncovering and therefore resisting the restrictive political agenda that they feed into. --- Supplemental material Supplemental material for this article is available online. --- Notes 1. https://www.euronews.com/2021/08/11/is-belarus-using-migrants-as-part-of-a-hybrid-waragainst-the-eu 2. https://twitter.com/YlvaJohansson/status/1457791267480342528 3. https://www.borderviolence.eu/balkan-region-report-august-2021/ 4. https://undocs.org/en/A/HRC/47/30

This article explores how Europe's border crises in the post refugee 'crisis' years were discussed in the micro-blog Twitter, through an in-depth analysis of boundary making. Our focus is on the tweets of the top influencers of the hashtag #IStandWithGreece who strategically promoted ideologies ranging from white supremacism to Greek nationalism, glued together by an antimigrant stance during a border 'crisis' at Europe's periphery. This network of intolerance promoted a representation of migrants as 'pawns'; seen like a chess piece, with no value in their own right, literally pushed towards Europe by Turkey, who elevated them into a sizable threat. Within this, Europe was represented as a paradoxical other, the fallen Self, for not rising up to the opportunity to protect its sovereignty and identity through more securitization. Despite being diffused by extreme antimigrant Twitterers, we argue that these tweets offer a more overtly racist expression of otherwise mainstream European (Union) discourses and politics on migration. Effectively, #IStandWithGreece's influencers functioned as Europe's alter-ego mouthpiece, saying the unsayable using social media, and their affordances contributing to the normalization of an oppressive and restrictive European border management.

3 In a comprehensive meta-analytic review of research on the relationship between intergroup threat and prejudice, Riek, Mania and Gaertner (2006) suggest that in order to better understand the aetiology of prejudice it is important to identify variables that increase the likelihood of interpreting intergroup situations as threatening. We propose that collective narcissism -in-group identification tied to an emotional investment in an unrealistic belief in exaggerated greatness of an in-group (Golec de Zavala, Cichocka, Eidelson & Jayawickreme, 2009) -is related to high susceptibility to signs of intergroup threat and enduring prejudice against out-groups stereotypically perceived as threatening. Previous studies indicate that collective narcissism predicts intergroup hostility in response to perceived threat to in-group's image. Collective narcissism predicts retaliatory intergroup hostility over and above such robust predictors of intergroup violence as social dominance orientation, authoritarianism, high in-group identification or 'destructive' forms of idealization of a national in-group such as blind patriotism, nationalism, or in-group glorification (Golec de Zavala et al, 2009;Golec de Zavala, 2007;Golec de Zavala & Cichocka, 2010;Imhoff, Erb & Wohl, 2010). Importantly, collective narcissism has been shown to reliably predict enduring negative attitudes towards certain, but not all, out-groups. For example, Polish collective narcissism predicts anti-Semitism (Golec de Zavala, et al, 2009;Golec de Zavala, Cichocka & Bilweicz, 2010) but it is not related to negative attitudes towards the French or British people (Golec de Zavala & Cichocka, 2010). American national narcissism predicts negative attitudes towards Arabs but not towards Asians, Europeans or Latinos. Arabs are perceived as more threatening than the other out-groups (Lyons, Kenowrthy & Popan, 2010). The above results suggest that collective narcissism may be related to prejudice because it increases sensitivity to intergroup threat. In addition, it is likely to predict negative attitudes only towards social groups stereotypically construed as threatening. Running head: COLLECTIVE NARCISSISM AND In this paper, we present results of two studies that indicate that the relationship between collective narcissism and prejudice is driven by chronic beliefs that the in-group is exposed and vulnerable in intergroup relations (the siege beliefs) and that the particular outgroup targeted by prejudice is a source of threat (the conspiracy stereotype). Our studies focus on the relationship between Polish collective narcissism and anti-Semitism: one of the most prevalent forms of prejudice in Poland and across Europe, particularity curious because in most countries it is a case of "anti-Semitism without Jews", i.e. prejudice against an almost non-existent minority, however stereotypically perceived as threatening to the national selfimage (Bergmann, 2008; see also Kofta & S<unk>dek, 2005;Krzemi<unk>ski, 2004). --- Collective narcissism and exaggerated regard for self and in-group The concept of collective narcissism extends into the intergroup domain the concept of individual narcissism, a grandiose view of self that requires continual external validation 1 (e.g. Crocker & Park, 2004;Emmons, 1987;Horney, 1937;Morf, & Rhodewalt, 2001;Raskin & Terry, 1988), and is related to unstable and defensive personal self-esteem (e.g. Bosson, Lakey, Campbell, Zeigler-Hill, Jordan & Kernis, 2008;Jordan et al., 2003;Kernis, Abend, Shrira, Goldman, Paradise & Hampton, 2005;Zeigler-Hill, 2006). Individual narcissists are emotionally attached to the belief in their own greatness and they are preoccupied with protecting it. Collective narcissists believe in unique prominence of the social group with which they identify (Golec de Zavala et al, 2009). Collective and individual narcissism, although positively correlated, are functionally separate variables. Their relationship across studies ranges from weak to moderate (from r =.15 to r =.27; Cai & Gries, 2010; Golec de Zavala et al., 2009;Lyons et al, 2010). Importantly, collective narcissism predicts inter-group attitudes and behaviors that individual narcissism does not account for and individual narcissism predicts interpersonal anger and aggressiveness that is not related to collective narcissism (Golec de Zavala et al, 2009; Studies 2 and 3). The concept of collective narcissism corresponds directly to the research that differentiates between 'belligerent' vs constructive forms of positive regard for one's nation. This research has been successful in describing several forms of "in-group love" that is systematically accompanied by "out-group hate" (see Brewer, 1999): such as nationalism (de Figueiredo & Elkins, 2003;Mummendey, Klink & Brown, 2001;Kosterman & Feshbach, 1989), blind patriotism (Schatz, Staub & Lavine, 1999), in-group glorification (Roccas, Klar & Liviatan, 2006) and essentialist national identification (Pehrson, Brown & Zagefka, 2009). However, the concept of collective narcissism is in several important ways distinct from the above conceptualizations of extensive in-group favouritism and, unlike them, it offers an insight into the psychological mechanism underlying the relationship between the preferential positivity for an in-group and out-group hositlity 2. Firstly, previous studies indicate that people can be narcissistic about various social groups, not only about their nation (e.g. Golec de Zavala, et al., 2009). To our knowledge, this is the only 'belligerent' form of in-group attachment that has been empirically demonstrated to extend beyond international context. Nevertheless, national collective narcissism, blind patriotism and national in-group glorification do overlap in the uncritical approach towards the national in-group and concern with protection of the in-group' positive image. However, unlike blind patriotism and in-group glorification that avoid criticism; collective narcissism is preoccupied with it 3. Collective narcissism, just like the narcissistic idealization of self, is contingent on external validation. Therefore, collective narcissists are constantly vigilant to threat to the in-group's image. In addition, only the narcissistic exaggerated image of the ingroup is shadowed by internal doubts regarding in-group's assumed greatness (Golec de Zavala et al, 2009). National collective narcissism and nationalism share the belief in the nation's inherent superiority. Unlike nationalist, however, narcissistic intergroup aggressiveness is defensive Running head: COLLECTIVE NARCISSISM AND ANTI-SEMITISM 6 and retaliatory. It does not serve the purpose of achieving a dominant in-group position born out of competitiveness, crucial for nationalism (Schatz et al., 1999; see also Bar-Tal, 1996;de Figueiredo & Elkins, 2003). Studies confirm that collective narcissism predicts intergroup hostility over and above other 'destructive' national attachments and it partially mediates the effects of blind patriotism, in-group glorification and nationalism on intergroup negativity (Cichocka & Golec de Zavala, 2010;Golec de Zavala, 2007;Golec de Zavala et al, 2009;Imhoff, et al., 2010). Moreover, when the common variance between collective narcissism and constructive patriotism is controlled, only collective narcissism (but not blind patriotism or nationalism) emerges as a suppressor of the negative relationship between genuine patriotism and prejudice (Golec de Zavala, Cichocka & Bilewicz, 2010). We argue that it is the very nature of the beliefs about the in-group which make up collective narcissism that inspires the sensitivity to intergroup threat and links it to out-group hostility. People with contingent self-worth exaggerate failures and underestimate successes in the domains of contingency (Baumesiter & Vohs, 2001;Crocker & Park, 2004;Kernis, 2003;Morf & Rhodewalt, 2001). Therefore, collective narcissists rarely see the acknowledgement of the in-group by others as satisfactory. They quickly develop "tolerance" to known sources of external validation and are constantly on the lookout for new signs of anything that may undermine the in-group. They retaliate against what they perceive as a threat to the in-group's positive image. --- Collective narcissism and Polish siege beliefs We propose that collective narcissism, with its extraordinary sensitivity to anything that can undermine the in-group's image, is likely to inspire the siege beliefs system about the in-group. The siege mentality prompts generally distrustful and negative attitudes towards other groups as it is "[a] belief held by group members stating that the rest of the world has Running head: COLLECTIVE NARCISSISM AND ANTI-SEMITISM 7 highly negative behavioural intentions toward them" (Bar-tal & Antebi, 1992a, pp.49; see also Bar-Tal, 2000). Siege beliefs explain and justify hardships suffered in the name of the in-group in intractable conflicts and legitimize the hostility and violence perpetrated by the in-group (Bar-Tal & Antebi, 1992a, b). Importantly, people differ with respect to how much conflict they perceive (e.g. Bar-Tal, Kruglanski & Klar, 1993;Golec & Federico, 2004) and how much they find the siege beliefs convincing. The siege beliefs offer an explanation and justification for the constant monitoring of the signs of the mistreatment associated with collective narcissism. They are also likely to satisfy narcissistic need for the sense of in-group's uniqueness, special status and moral superiority because they maintain that the misunderstood and righteous in-group stands alone against the hostile and dissolute world. Most importantly, the siege beliefs may appeal to collective narcissists because they confirm what the narcissists seem continuously to suspect: that other groups do not properly acknowledge their greatness. We propose that because of its association with the siege beliefs about the nation, Polish national narcissism will be related to anti-Semitism. Polish studies indicate that Jews (or, more specifically, Poles of Jewish origin) are perceived as an out-group (e.g. Bilewicz, 2007). More than any other minority, this out-group is salient to Poles and inspires ambivalent emotions: guilt, anger and fear. Importantly, the Jewish people are seen as a threat to the positive national image (because of the criticism regarding Polish anti-Semitism), to Poland's national interest (because of the claims of Polish Jews or their families regarding financial retributions for the properties confiscated by the communist government in Poland) or to national security (because of the conspiracy beliefs attributing hostile intentions against the Polish state to Jews) (e.g. Bilewicz, 2007;Kofta & S<unk>dek, 2005;Krzemi<unk>ski, 2004;Wójcik, 2008). Thus, collective narcissists are likely to be prejudiced against Jews because this is a particularly salient out-group in Polish national context. In addition, this out-group is seen as Running head: COLLECTIVE NARCISSISM AND ANTI-SEMITISM 8 particularly threatening. As such it is well equipped to be seen as a model example of the hostile intentions of the external world towards the in-group. --- Collective Narcissism and the conspiracy stereotype of Jews The relationship between collective narcissism and anti-Semitism is likely to be independently driven by stereotypical beliefs about the Jewish out-group. Intergroup threat can be embedded in a negative stereotype of an out-group conveying the prediction of its hostile intentions towards the in-group (Stephan & Stephan 2000;Stephan et al., 2002). Although anti-Semitic beliefs and attitudes in Europe are under-studied, several studies indicate that the Jewish minority is quite universally perceived as threat to national identity, especially the regions where the Jewish minority was large before WWII (Bergmann, 2008). No matter how well assimilated, Jews tend to be perceived as a group that stands "outside the national order of the world" remaining "essentially alien to the surrounding societies" (Bergmann, 2008;p. 346). In addition, the in-group's ambivalent position during the Holocaust is experienced as threat to the national self-image (e.g. Bergmann, 2008). Poland is a good example of the ambivalent position towards the Holocaust. There are reported cases of pogroms of Polish Jews perpetrated by Poles during the WWII and shortly afterwards (e.g. Gross, 2008). At the same time many Polish soldiers (of the Home Army, Armia Krajowa) fought in the Warsaw Ghetto Uprising and Poles represent the biggest number of people who rescued Jews during the Holocaust and were awarded the 'Righteous among the Nations' medal by Israel (YadVashem, 2009). In addition to being perceived as alien to the national in-group, in Poland and other European countries Jews are stereotypically perceived as a dangerous out-group. They are seen as motivated by a common intention to dominate the world (Bergmann, 2008;Cohen, & Golub, 1991;Kofta & S<unk>dek, 2005). The alleged dominant and controlling intentions are executed by means of indirect and deceptive methods, hidden and non-obvious ways whose negative consequences can only be observed. This makes the Jewish out-group particularly dangerous and threatening. According to the conspiracy stereotype, past, present and even future harm and hardship experienced by the in-group can be explained by the veiled actions of the Jewish out-group. The conspiracy stereotype of Jews predicts anti-Semitism in Poland and support for this stereotype increases in times of elections and intense political campaign (Kofta & S<unk>dek, 2005). We expect that collective narcissists, sensitive to signs of potential threat to the ingroup, will be likely to find the conspiracy stereotype of Jews convincing and threatening. A tendency to uphold the conspiracy stereotype of Jews will mediate the relationship between collective narcissism and anti-Semitic prejudice. The mediation through the beliefs about the out-group will be parallel to and independent of the mediation through the Polish siege beliefs that emphasize the vulnerable position of the in-group. In other words, we expect that collective narcissism will be associated with anti-Semitism either because it is related to perceived vulnerability of an in-group or because it is related to the perception of the outgroup as threatening. This assumption is different than the expectation that collective narcissism is associated with perceived vulnerability of an in-group which leads to a tendency to construe the Jewish out-group as threatening. Thus, we assume that the multiple mediations -via the siege beliefs and via the conspiracy stereotype of Jews -are parallel, i.e. simultaneous and independent, rather than chain i.e. consecutive and depended. Even though the siege beliefs about the in-group and the conspiracy beliefs about the out-group may share common components, we assume that each of these variables should have a unique ability to mediate between Polish national narcissism and anti-Semitism above and beyond the other variable. Siege beliefs and the conspiracy beliefs about Jews are stereotypical beliefs that may exist in socio-cultural repertoire separately and they are two discrete reasons why collective narcissism is likely to be linked to Running head: COLLECTIVE NARCISSISM AND ANTI-SEMITISM 10 prejudice. One reason pertains to the beliefs about the characteristics of the in-group, whereas the other pertains to attributes of the out-group. --- Overview of the studies In Study 1 we test the prediction that Polish collective narcissism is associated with the Polish siege beliefs, the set of convictions indicating that the national in-group is constantly threatened by the hostile intentions of other groups (Bar-Tal & Antebi, 1992a, b). We expect that Polish siege beliefs will mediate the relationship between collective narcissism and anti-Semitism. The Jewish out-group, with whom they share a long and complicated history of competitive relations, is particularly salient to Polish participants. In Study 2 we seek to replicate the results of Study 1. We also test the hypothesis that the relationship between collective narcissism and anti-Semitism is independently mediated by the conspiracy stereotype of Jews. The conspiracy stereotype conveys the image of Jews as a powerful and skilful out-group whose competitive intentions threaten the interests, image, if not the mere existence of, the in-group. The stereotype of the Jewish out-group poses the intergroup threat. We test the parallel against the chain multiple mediation hypothesis. In addition, in Study 2 we seek to demonstrate that collective narcissism, rather than the mere strength of national group identification (e.g. Bizman & Yinon, 2001;Branscombe & Wann, 1994), predicts the perception of intergroup threat and out-group negativity. --- Study 1 --- Method --- Participants and Procedure Participants of Study 1 were 148 undergraduate students of a large Polish university. The study was conducted on-line. Participants obtained a research participation credit and the possibility to take part in a prize draw in return for participation. The age of the participants ranged from 18 to 45 (M = 23.12, SD = 4.89). There were 135 women and 13 men among the participants. --- Measures Collective narcissism (<unk> =.77, M = 3.26, SD =.67). The 9-item Collective Narcissism Scale (Golec de Zavala, et al, 2009) was used in order to measure this construct. The items for this scale were generated based on the definition of the construct and existing inventories of individual narcissism, mostly the Narcissistic Personality Inventory (NPI) (Emmons, 1987;Raskin & Terry, 1988). Only the items that corresponded to the core aspects of the concept of individual narcissism but at the same time could be meaningfully translated onto the group level were used. For selected items, beliefs about the self were replaced with beliefs about one's in-group. The 9 items that make up the Collective Narcissism Scale were selected on the basis of their face validity as evaluated by experts, the results of Confirmatory and Exploratory Factor Analyses, and the strength of their contribution to the overall reliability of the scale assessed in three studies. The scale contains items reflecting the belief in the ingroup's greatness and lack of its proper recognition ("(1) If my group had a major say in the world, the world would be a much better place"; "(2) I wish other groups would more quickly recognize the authority of my group."; "(3) My group deserves special treatment"; "(4) Not many people seem to fully understand the importance of my group"; "(5) I will never be satisfied until my group gets all it deserves"; "(6) I do not get upset when people do not notice the achievements of my group" (reversely coded); "(7) I insist upon my group getting the respect that is due to it"; "(8) It really makes me angry when others criticize my group" or "(9) The true worth of my group is often misunderstood"). Participants were instructed to think about their national group while responding to the items of the scale. Participants were Running head: COLLECTIVE NARCISSISM AND ANTI-SEMITISM asked to indicate how much they agree with statements using a 6-point scale (1 = "I strongly disagree" and 6 = "I strongly agree"). Siege beliefs (<unk> =.77, M = 2.81, SD =.76). The 12-item General Siege Mentality Scale proposed by Bar-Tal & Antebi (1992a) was used to measure this construct. The scale was translated from English to Polish by a bilingual translator. It was then back translated by a bilingual expert in social psychology in order to ensure the equivalence of meaning of items on both scales. The scale contains items reflecting the belief that the in-group is constantly threatened (e.g. "Most nations will conspire against us, if only they have the possibility to do so." or "There have always been countries which looked for closeness and friendship with us." (reversely coded) and has to protect itself in this time of need (e.g. "Only unity will save us from external enemies."). Participants were asked to indicate how much they agree with each statement using a 7-point scale from 1 = "definitely disagree" to 7 = "definitely agree". Anti-Semitism (<unk> =.71, M = 2.37, SD = 1.02). In Study 1 we defined anti-Semitic prejudice in terms of social distance, i.e. unwillingness to engage in contacts with the Jewish out-group (see e.g. Goff, Steele & Davies, 2008;Struch & Schwartz, 1989;Bogardus, 1925). We used a 4-item scale measuring preferred social distance from Jews: "Would you like a Jew to be your neighbor?" (reversely coded), "Would you like a Jew to be your friend?" (reversely coded), "Would you mind your child playing with a Jewish child?", "Would you mind your child marrying a person of Jewish origin?". Participants were asked to respond to these items using a scale from 1 -"definitely no" to 7 -"definitely yes". --- Results In the first step of data analysis we compute zero-order correlations among variables. Collective narcissism was significantly positively related to the Polish siege beliefs (r (147) =.48, p =.001) and anti-Semitism (r (146) =.20, p =.02). Siege beliefs were significantly correlated with anti-Semitism (r (147) =.37, p =.001). Running head: COLLECTIVE NARCISSISM AND ANTI-SEMITISM 13 In order to test the main hypothesis that siege beliefs mediate the relationship between collective narcissism and anti-Semitism, we used the bootstrapping method recommended by Preacher and Hayes (2004) to obtain bias corrected 95% bootstrap confidence intervals for the indirect effect of the mediator. Bootstrapping does not require assumptions about the shape of the sampling distribution of the indirect effect and is considered as a suitable method for assessing indirect effects in smaller samples (Preacher and Hayes, 2004; for information on required sample size for detecting effects with the use of different tests of mediation see also Fritz and MacKinnon, 2007). The mediation model is presented in Figure 1. The total effect of collective narcissism on the tendency to keep one's distance from Jews was positive and significant, B =.29, SE =.12, t = 2.37, p =.02. The relationship between collective narcissism and anti-Semitism was reduced and became non-significant after the mediator was added to the model. The direct effect amounted to B =.04, SE =.13, t =.31, p =.76. Collective narcissism was positively associated with siege mentality, B =.54, SE =.08, t = 6.52, p <unk>.001. Siege mentality was positively associated with anti-Semitism, B =.47, SE =.12, t = 3.91, p <unk>.001. The difference between the total and direct effects of collective narcissism on anti-Semitism is the total indirect effect via the siege beliefs. We computed its confidence interval with 10.000 bootstrap samples. The indirect effect had a 95% bootstrap bias corrected confidence interval of.12 to.40, which indicates that the indirect effect of the mediator was significant. The whole mediation model was also significant, R 2 =.14, F (4,142) = 5.66, p <unk>.001. Although considered the most appropriate method of assessment of indirect effects (Hayes, 2009;Preacher & Hayes, 2004;Preacher & Hayes, 2008), bootstrapping is also a relatively new approach. Thus, we performed the mediation analysis also using a more familiar (although less accurate) Sobel (1982) test to provide an alternative test of the significance of the indirect effect of collective narcissism on anti-Semitism via siege beliefs. The Sobel test indicated that collective narcissism had a significant indirect effect on anti-Semitism via the siege beliefs (z = 3.30; p <unk>.001). --- Discussion of Study 1 The results of Study 1 confirm the positive relationship between collective narcissism and anti-Semitism (see Golec de Zavala, et al., 2009;Golec de Zavala, et al., 2010;Golec de Zavala, 2010). Most importantly, as hypothesized, the present results reveal that this relationship is mediated by the siege beliefs portraying the national in-group as constantly threatened by hostile intentions of other groups. Thus, collective narcissism is associated with the perception of the in-group as exposed and vulnerable in the context of intergroup relationships. Such a perception is likely to increase susceptibility to intergroup threat. This perception drives the relationship between collective narcissism and prejudice against Jews: the out-group stereotypically perceived as threatening. In Study 2 we test the assumption that the stereotypical belief that the Jewish outgroup is particularly threatening and hostile mediates the relationship between collective narcissism and anti-Semitism, over and above the mediation through the siege beliefs. We expect that a belief that the in-group is particularly threatened and a belief that the out-group is especially threatening will mediate the relationship between collective narcissism and prejudice independently. In Study 2 we use more direct indicators of anti-Semitism in order to conceptually replicate the results of Study 1, namely negative emotions and hostile behavioural intentions towards Jews. In Study 2 we also compare the role of collective narcissism and the strength of in-group identification as predictors of perceived intergroup threat and anti-Semitism. --- Study 2 --- Method --- Participants and procedure Study 2 was conducted among 89 undergraduate students of a large Polish university. Participants were asked to take part in an on-line survey in return for research participation credit. The age of the participants ranged from 17 to 24 (M = 21.17, SD = 1.51). There were 63 women and 26 men among the participants. --- Measures Collective narcissism (<unk> =.84, M = 3.51, SD =.78). The Collective Narcissism Scale (Golec de Zavala et al., 2009) was used in order to measure this variable as in Study 1. Group identification (M = 5.38, SD = 1.70). The overlap in-group identification measure proposed by Tropp & Wright (2001) was used. Participants were asked to indicate the degree of their identification with the national in-group as represented by two overlapping circles, one representing the self and the other representing national in-group. The circles formed an 8-point scale from a set of two separate circles (1 -"no identification at all") through degrees of overlap to full overlap (8 -"total identification"). Siege Beliefs (<unk> =.77, M = 3.30, SD =.87). The same Polish translation of the General Siege Mentality Scale (Bar-Tal, Antebi, 1992a) as in Study 1 was used. Conspiracy Jewish stereotype (<unk> =.90, M = 5.41, SD =1.99). This variable was measured with a 6-item Jewish Conspiracy Stereotype Scale proposed by Kofta and S<unk>dek (2005). The scale measures the belief that secret and deceptive actions of the Jewish outgroup are aimed at taking control over the world. The Jewish out-group is seen as an entity driven by one common motivation to dominate others. The conspiracy stereotype is considered a central characteristic of anti-Semitism in Poland (Kofta & S<unk>dek, 2005). The items of the scale reflected the belief that Jews strive for power (e.g. "Members of this group strive to rule the world", "Members of this group attempt to take over the world's economy", "Members of this group would like to hold decisive votes in international financial institutions") and that their actions are secretive and well co-ordinated (e.g. "Members of this group meet secretly to discuss important issues", "Actions of members of this group are often clandestine", "Members of this group reach their goals through secret agreements"). Participants were asked to indicate how much they agree with each statement using a 9-point scale from 1 = "definitely disagree" to 9 = "definitely agree". Negative evaluation of Jews (<unk> =.95, M = 3.57, SD = 1.41). This variable was measured following the procedure proposed by Wright, Aron, McLaughlin-Volpe, & Ropp (1997; see also Butz, Plant, & Doerr, 2007). Participants were asked to indicate their feelings towards Jews using six semantic differentials: cold -warm, unfriendly -friendly, trustfuldistrustful, positive -negative, respect -contempt, admiration -disgust. Scores could range from 1 to 8. Higher scores indicated greater out-group negativity. Hostility towards Jews (<unk> =.92, M = 1.45, SD =.99). Aggressive behavioral intentions against Jews were measured by four items adopted from Struch and Schwartz (1989). Hypothetical aggressive acts towards Jews were listed, e.g. refusing to hire Jews because of their origins, convincing friends not to rent flats to Jews, listening to noisy music in order to irritate a Jewish neighbor. Participants were asked to indicate how much they agree with each action and would perform it themselves, using a scale from 1 = "definitely reject it" to 7 = "agree with the action and would perform it myself in certain conditions". The negative evaluation of Jews and support for aggressive actions against them were positively correlated (r =.38, p <unk>.001). We created a composite score of anti-Semitism that encompassed the two direct indicators of anti-Semitism. Because the two components of the composite score were measured on different scales, responses were first transformed into z scores before the composite indicator of anti-Semitism was computed. --- Results --- Running head: COLLECTIVE NARCISSISM AND ANTI-SEMITISM 17 Correlational analyses presented in Table 1 indicate that collective narcissism is positively related to siege mentality, the conspiracy stereotype of Jews and anti-Semitism. National in-group identification is related to siege beliefs but not to the conspiracy stereotype or to prejudice against Jews. In order to control for the common variance between collective narcissism and in-group identification we regressed collective narcissism and group identification on siege mentality (controlling for age and gender). Collective narcissism was positively and significantly associated with siege mentality, B =.73, SE =.11, t = 6.83, p <unk>.001, whereas the relationship between in-group identification and perceived threat from hostile intentions of others was reduced and became non-significant, B = -.04, SE =.05, t = -.078, p =.44; R 2 =.40, F(4,83) = 14.04, p <unk>.001. Once the common variance of the strength of in-group identification and collective narcissism is controlled, collective narcissism emerges as the unique predictor of the siege beliefs. ----------------------Insert Table 1 about here----------------In order to test the hypothesis that siege beliefs and the Jewish conspiracy stereotype independently mediate the relationship between collective narcissism and anti-Semitism we used the bootstrapping method suggested by Preacher and Hayes (2008). This approach allows us to assess the extent to which each of these variables mediates the effect of collective narcissism on anti-Semitism, conditional on the presence of the other variable in the model. Since siege beliefs and conspiracy stereotype were positively correlated, analyzing them simultaneously in a multiple mediator model teased apart their individual mediating abilities that could be attributed to their content overlap (Preacher & Hayes, 2008). To assess the independent indirect effects of collective narcissism on anti-Semitism via the siege beliefs and the conspiracy stereotype of Jews, we used bootstrapping to obtain the bias corrected 95 % confidence intervals for the total indirect effect and the specific indirect effects of each mediator analyzed together. The analyses controlled for age and gender. 3 The coefficients of the model are presented in Figure 2 and the bootstrapping confidence intervals are presented in Table 2. ----------------------Insert Figure 2 and Table 2 about here- ---------------The total effect of collective narcissism on anti-Semitism amounts to B =.23, SE =.11, t = 1.99 p =.05, while its direct effect is B = -.21, SE =.14, t = -1

Two studies examined the relationship between collective narcissism -an emotional investment in an unrealistic belief about unparalleled greatness of an in-group (Golec de Zavala et al., 2009) -and anti-Semitism in Poland. The results indicate that this relationship is simultaneously mediated by (a) a belief that the in-group is constantly threatened by hostile intentions of other groups (Polish siege beliefs; Bar-Tal & Antebi, 1992) and (b) a belief that the Jewish out-group is particularly threatening because its members secretly aim to dominate the world (the conspiracy stereotype of Jews; Bergmann, 2008;Kofta & Sędek, 2005). The results confirm earlier findings that collective narcissism is linked to increased sensitivity to intergroup threat which drives its association with intergroup hostility. The sensitivity to intergroup threat is composed of beliefs about vulnerability of the in-group and hostility of the out-group.

of the other variable in the model. Since siege beliefs and conspiracy stereotype were positively correlated, analyzing them simultaneously in a multiple mediator model teased apart their individual mediating abilities that could be attributed to their content overlap (Preacher & Hayes, 2008). To assess the independent indirect effects of collective narcissism on anti-Semitism via the siege beliefs and the conspiracy stereotype of Jews, we used bootstrapping to obtain the bias corrected 95 % confidence intervals for the total indirect effect and the specific indirect effects of each mediator analyzed together. The analyses controlled for age and gender. 3 The coefficients of the model are presented in Figure 2 and the bootstrapping confidence intervals are presented in Table 2. ----------------------Insert Figure 2 and Table 2 about here- ---------------The total effect of collective narcissism on anti-Semitism amounts to B =.23, SE =.11, t = 1.99 p =.05, while its direct effect is B = -.21, SE =.14, t = -1.53, p =.13. The difference between the total and direct effects is the total indirect effect via the two mediators. It had a 95% bootstrap confidence interval of.25 to.69, i.e. the total indirect effect of both mediators is significant. The path from collective narcissism to siege mentality had a coefficient of B =.69, SE =.10, t = 7.28, p <unk>.001 and the path from collective narcissism to conspiracy stereotype had a coefficient of B = 1.18, SE =.23, t = 5.11, p <unk> 001. The direct effects of mediators on anti-Semitism were also significant: B =.27, SE =.12, t = 2.34, p =.02 for siege mentality and B =.21, SE =.05, t = 4.33, p <unk>.001 for conspiracy stereotype. The predictors included in the full model accounted for a significant portion of variance in anti-Semitism, R 2 =.30, F (5,81) = 7.04, p <unk>.001. An examination of the specific indirect effects indicated that both siege beliefs and the conspiracy stereotype of Jews were statistically significant and independent mediators of the effect of collective narcissism on anti-Semitism. The specific indirect effect via the siege beliefs had a confidence interval of.05 to.40 and the specific indirect effect via the conspiracy stereotype had a confidence interval of.13 to.41. In order to establish whether the effects of mediators differ significantly in magnitude we conducted a pairwise contrast of the two indirect effects. Since the siege mentality minus conspiracy stereotype contrast had a confidence interval of -.18 to.26, we cannot infer that the two effects differed in magnitude. Finally, we used the Sobel test (1982) to confirm the significance of the indirect effects of collective narcissism on anti-Semitism via siege beliefs and via the conspiracy stereotype. Collective narcissism had significant indirect effects on negativity towards Jews both via siege mentality (z = 2.14, p =.03) and via the conspiracy stereotype of Jews (z = 3.25, p =.001). Since siege mentality and the conspiracy stereotype of Jews were positively correlated, it is also plausible that they influenced each other in driving anti-Semitic sentiments. For example, one could argue that collective narcissism is related to siege mentality, which in turn leads to conspiracy stereotyping and, thus, to hostility towards Jews. To verify this possibility we tested the multiple-step multiple mediator model proposed by Hayes, Preacher & Myersa (in press). We used the MODTHREE macro that allows for a simultaneous test of specific indirect effects of each mediator alone and a specific indirect effect through both mediators. In the first analysis we tested the collective narcissism the siege beliefs conspiracy stereotyping anti-Semitism multi-step mediation. While the independent specific indirect effects of siege mentality and the conspiracy stereotype remained significant, the indirect effect of both mediators became insignificant (95% bootstrap confidence intervals ranged from -.01 to.16). Similar results were obtained when we tested a model with collective narcissism conspiracy stereotyping siege beliefs anti-Semitism multi-step mediation. Both indirect effects of single mediators were significant and the indirect effect of the two mediators was not significant (its 95% bootstrap confidence intervals ranged from -.01 to.05). These analyses provide further support for our hypothesis that the siege mentality and conspiracy stereotype are two independent links between narcissistic attachment to the Polish national group and anti-Semitism. --- Discussion of Study 2 The results of Study 2 replicate the findings of Study 1. In addition, they confirm the hypothesis that siege beliefs and the belief that Jews secretly conspire to achieve economic and political power and dominate the world independently mediate the relationship between Running head: COLLECTIVE NARCISSISM AND ANTI-SEMITISM 20 Polish collective narcissism and anti-Semitism. Collective narcissism is associated with a tendency to perceive the in-group as threatened and the out-group as particularly threatening. Both tendencies independently predict out-group negativity. The strength of in-group identification is not related to prejudice against Jews. It is related to the Polish siege beliefs but this relationship is driven by the overlap between the strength of in-group identification and collective narcissism. Once the narcissistic aspect of positive in-group identification is teased out, the relationship between the strength of in-group identification and siege beliefs disappears. These results support our claim that it is not the strength of in-group identification in general, but rather the individual level of specific, narcissistic identification that is related to the perception of intergroup threat. Consequently, it is not the strength of ingroup identification but collective narcissism that is related to out-group negativity. --- General Discussion According to social identity theory (e.g. Tajfel & Turner, 1986), a tendency to derogate out-groups intensifies under perceived threat to the social identity and increased salience of group membership. An intergroup threat may be realistic and concern group interests (e.g. Brown, Maras, Masser, Vivian & Hewstone, 2001) or symbolic and concern difference in worldviews, values and beliefs (e.g. Esses, Dovidio, Jackson & Armstrong, 2001;Greenberg, et al, 1990;McGregor et al., 1998;McLaren, 2003;Pyszczynski, et al, 2006;Sears, 1988). It can come from out-group aggression, embarrassing rejection, unjustified discrimination (Britt, Boniecki, Vescio, Biernat & Brown, 1996;Stephan & Stephan, 1985;Corenblum & Stephan 2001) or uncertainty and awkwardness in the presence of out-group members (e.g. Islam & Hewstone, 1993;Stephan & Stephan, 2000). Importantly, intergroup threat can be also embedded in a negative stereotype of an out-group that attributes threatening features and intentions to the group and its members (Stephan & Stephan 2000;Stephan, et al., 2002). Running head: COLLECTIVE NARCISSISM AND ANTI-SEMITISM 21 It has been suggested that the strength of in-group identification is one of the important antecedents of perceived intergroup threat. The more people identify with their group, the more they are chronically aware of their group membership (Branscombe & Wann, 1994;Tausch, Hewstone, Kenworthy, Cairns & Christ, 2007) and the more sensitive they are to anything that can harm the in-group (e.g. Corenblum & Stephan, 2001). In addition, high identifiers are more likely to see the threats to the in-group as personally threatening (Bizman & Yinon, 2001). However, the meta-analytic review indicates that the relationship between the strength of positive in-group identification and perceived intergroup threat, although statistically significant, is inconsistent and on average rather weak (Riek at al, 2006). Moreover, the relationship between positive group identification and prejudice is not consistent and across numerous studies averages close to zero (e.g. Hinkle & Brown, 1990; see also Jackson, Brown, Brown & Marks, 2001;Pehrson et al., 2009) We propose that it may be not the strength of in-group identification that reliably predicts the perception of inter-group threat and prejudice, but rather its specific, narcissistic form. The results of two studies presented here confirm that people who narcissistically identify with their national in-group perceive more intergroup threat, stereotype a stigmatized out-group as particularly threatening and report more prejudice. Results of Study 1 show that Polish collective narcissism is related to the Polish siege beliefs pertaining that the national group is threatened by aggressive intentions of other groups and stands along against the hostile world. This relationship mediates the link between collective narcissism and anti-Semitism. Results of Study 2 replicate these findings. In addition they indicate that the relationship between collective narcissism and anti-Semitism is independently mediated by increased narcissistic sensitivity to the intergroup threat embedded in the negative stereotype of Jews as a particularly threatening out-group that conspires to dominate and rule the world. The perceptions of the in-group as vulnerable to intergroup threat and the out-group as Running head: COLLECTIVE NARCISSISM AND ANTI-SEMITISM 22 threatening independently drive the relationship between collective narcissism and anti-Semitic prejudice. The present results complement earlier findings indicating that collective narcissism predicts retaliatory hostility in response to the threat from out-group aggressiveness, outgroup distancing and rejection of the in-group (Golec de Zavala et al, 2009) or out-group criticism of the in-group (Golec de Zavala & Cichocka, 2009a). The present results go beyond the earlier findings revealing an intriguing novel aspect of the relationship between collective narcissism and out-group negativity. They indicate that collective narcissism predicts not only retaliatory hostility in response to momentary intergroup threat but is also associated with enduring prejudice towards an out-group stereotyped as threatening. The present results confirm suggestions that anti-Semitism is related to threat and narcissistic national pride (e.g. Bergmann, 2008;Krzemi<unk>ski, 2004). They indicate that anti-Semitism is grounded in beliefs in national superiority that are insecure and narcissistic and fuel the sense of the in-group's vulnerability in an intergroup context and fear of the hostile intentions of the Jewish out-group. We suggest that a similar mechanism is likely to drive the relationship between collective narcissism and prejudice against other out-groups with whom the in-group shares competitive or conflictual relations. In such relations the opportunities to injure the collective pride are plenty. Collective narcissists neither forget nor forgive wrongs done to the in-group by out-groups (Golec de Zavala et al, 2009). Thus, they are likely to see the out-groups that in the past transgressed against the in-group as potential threats, even if the transgression was mostly in 'the eye of the beholder'. Thus, the more frequent and less unequivocally positive the intergroup relations with a given group, the greater the chance is that this group will be targeted by prejudice. The examination of the generalizability of the present mediation model beyond anti-Semitic prejudice would be an important direction for further research. It is, however, noteworthy that the previous and present results indicate that collective narcissism is not universally associated with prejudice. Instead, collective narcissism seems to describe a combination of group-based feelings that can fuel hostility in certain intergroup situations and in response to certain out-groups. We propose that the relationship between collective narcissism and the sensitivity to intergroup threat can be explained by the nature of narcissistic in-group evaluation. Collective narcissism is related to high regard for the in-group combined with a belief that the in-group is not sufficiently recognized by others and the lack of the positive evaluation of the in-group on the implicit level. Thus, collective narcissists are easily convinced that the image of their in-group is being undermined. Intergroup hostility in response to the perceived threat serve as means of protecting the in-group's image and maintaining the in-group's positive esteem (see Golec de Zavala et al, 2009). In addition, our results reveal that collective narcissism is a better predictor of the perception of intergroup threat embedded in stereotype and prejudice than the mere strength of identification with the in-group. The positive relationship between the strength of in-group identification and perceived intergroup threat was reduced and became non-significant after collective narcissism was taken into account. --- Limitations The present studies provide strong support for the hypotheses derived from the concept of collective narcissism. However, they have several shortcomings that should be considered. Firstly, in both samples, there is a disproportionate number of women among the participants. However, in all analyses we included gender as a control variable and found no significant effect of gender. In addition, we do not have any theoretical reasons to assume that men and women differ with respect to their individual levels of collective narcissism. Secondly, the present findings are based on university student samples, which may not be representative of the population as a whole (Sears, 1986). Future studies should extend the investigation of collective narcissism and its correlates and effects to different populations. However, it is worth noting that we found remarkably consistent patterns of relationships across both presented studies. In addition, although the present studies provide support for the mediational hypotheses derived from the concept of collective narcissism, they are based on correlational data and do not allow for unequivocal conclusions about causality and order of the variables. In the present paper, we provide sound theoretical reasons to justify our assumption that collective narcissism is related to siege beliefs and the conspiracy stereotype of Jews and through these beliefs to anti-Semitism. In addition, most research in the social sciences confirm the direction of causality assumed in the proposed model, suggesting that broader ideological orientations and basic in-group identification constrain specific attitudes and beliefs, such as the siege belief or the conspiracy stereotype of Jews, and out-group hostility (rather than vice versa; see e.g. Cohrs, Moschner, Maes, & Kielmann, 2005;Duckitt, 2006;Duckitt & Sibley, 2006;Feshbach, 1994;de Figueiredo & Elkins, 2003;Sidanius, Feschbach, Levin & Pratto, 1997). However, further experimental studies are needed in order to examine the role of collective narcissism in eliciting increased sensitivity to intergroup threat and prejudice. Such studies will deepen our understanding of individual difference variables and situational conditions, increasing the likelihood of out-group negativity and intergroup aggression. Footnotes 1 Collective and individual narcissism are positively associated, although this relationship is rather weak. Importantly, collective narcissism is related to intergroup hostility, whereas individual narcissism is associated with interpersonal aggressiveness, especially in the context of ego threat (Golec de Zavala & Cichocka, 2009a; see also Golec de Zavala, et al, 2009; for results on individual narcissism see Baumeister & Bushman, 1998). 2 Collective narcissism is also distinct from social dominance orientation (Pratto et al., 1994;Sidanius & Pratto, 1999) or authoritarianism (e.g. Altemeyer, 1998) and independently predicts out-group negativity. We discuss these differences in more detail in a separate paper (Golec de Zavala et al, 2009). Summing up, collective narcissism and social dominance orientation overlap in the preoccupation with the in-group's greatness. However, for collective narcissists, any excuse, not only power, social status or economic dominance, is sufficient to support the belief in the uniqueness and greatness of the in-group. Collective narcissism is not related to opposition to equality, an important aspect of the social dominance orientation. Collective narcissism and authoritarianism are related because of the concern with the coherence and homogeneity of the in-group. For authoritarians, cohesiveness secures a predictable social environment and reduced cognitive uncertainty (e.g. Duckitt, 2006;Jost, Glaser, Kruglanski & Sulloway, 2003;Kruglanski & Webster, 1996). For collective narcissists, it confirms the assumed, unanimously accepted greatness of the in-group. Authoritarians aggress against others to protect the group as a predictable social environment, collective narcissists -to protect the in-group's positive image. 3 For example, our data indicate that only collective narcissism, but not blind patriotism or high national in-group identification, is predicted by the interaction of high private (positive opinion about one's national group) and low public (a belief that others do not hold a positive Running head: COLLECTIVE NARCISSISM AND ANTI-SEMITISM 39 opinion about one's national group) collective self-esteem measured with reference to one's national group (Golec de Zavala, 2007;Golec de Zavala et al., 2009). 4 Analysis controlling for group identification was also conducted. The pattern of results remained the same. Note. BC=bias corrected, 10.000 bootstrap samples. Figure Captions

Two studies examined the relationship between collective narcissism -an emotional investment in an unrealistic belief about unparalleled greatness of an in-group (Golec de Zavala et al., 2009) -and anti-Semitism in Poland. The results indicate that this relationship is simultaneously mediated by (a) a belief that the in-group is constantly threatened by hostile intentions of other groups (Polish siege beliefs; Bar-Tal & Antebi, 1992) and (b) a belief that the Jewish out-group is particularly threatening because its members secretly aim to dominate the world (the conspiracy stereotype of Jews; Bergmann, 2008;Kofta & Sędek, 2005). The results confirm earlier findings that collective narcissism is linked to increased sensitivity to intergroup threat which drives its association with intergroup hostility. The sensitivity to intergroup threat is composed of beliefs about vulnerability of the in-group and hostility of the out-group.

Introduction --- S ocioeconomic inequalities exist in paediatric metabolic syndrome (MetS), [1][2][3] and parental health literacy may partially explain these inequalities. MetS is a cluster of cardiometabolic risk factors (i.e. waist circumference, mean arterial pressure, insulin resistance, triglycerides, and high-density lipoproteins), 4 which can result in increased cardiometabolic risk later in life. 5,6 Due to the long-term impact of MetS, prevention early in life would likely provide favourable consequences over the life course. In order to prevent paediatric MetS, we need a better understanding of the mechanisms linking socioeconomic status (SES) to MetS, such as the role of parental health literacy. Parental health literacy may be an important modifiable mechanism linking SES to paediatric MetS, but this pathway remains understudied in children. Health literacy can be defined as 'the degree to which individuals have the capacity to obtain, process, and understand basic health information and services needed to make appropriate health decisions'. 7 In adults, there is evidence that health literacy mediates the relationship between own SES and health, and that the strength of these relationships vary depending on which proxy of SES is used 8 ; education has the strongest relationship with health literacy. Additionally, there is a well-known relationship between parental health literacy and various aspects of children's health. 9 Parents with low health literacy are more likely to experience difficulties managing their child's chronic diseases, and are more likely to incorrectly use medications. 9 Additionally, parents with low health literacy are more likely to have less optimal health behaviours, e.g. smoking, alcohol, nutrition and physical activity. 9 These factors contribute to the association between low parental health literacy and children's obesity, which is a major contributor to MetS, 10 in children. 9 Given these relationships, it seems that parental health literacy may mediate the relationship between SES and paediatric MetS. Evidence on the mediating role of parental health literacy is relevant because it is a modifiable target for preventive interventions to reduce socioeconomic health inequalities in paediatric MetS. Therefore, we aimed to assess the extent to which parental health literacy mediates the relationships between different measures of SES and MetS. --- Methods --- Setting and population Data were used from the Lifelines Cohort Study, a multi-disciplinary prospective population-based cohort study examining in a unique three-generation design the health and health-related behaviours of 167 729 persons living in the North of the Netherlands. 11,12 It employs a broad range of investigative procedures in assessing the biomedical, socio-demographic, behavioural, physical and psychological factors which contribute to the health and disease of the general population, with a special focus on multi-morbidity and complex genetics. A detailed description of the recruitment strategy and data collection can be found elsewhere. 11 Briefly, Dutchspeaking individuals aged 25-49 years were asked to participate by their physicians. Those who accepted were subsequently asked to invite their family members. Individuals could also self-register through the Lifelines website. The first measurement wave took place between 2007 and 2014 and 2010 and 2014 in adults and children, respectively. During the first (T1) and fourth (T4) measurement wave, participants were asked to fill out questionnaires and, if aged 8 years or older, they also underwent physical exams. For the purpose of this paper, Lifelines provided the data of 15 016 children aged 0-17 years at baseline whose data could be linked to at least one parent. For this study, we included all 6683 children aged 8-17 years during T4, and they had a mean follow-up duration of 36.2 (SD: 9.3) months. Children under the age of 8 years at T4 (n 1<unk>4 2167) were excluded because the components of MetS were not assessed in those children, and children who turned 18 years old during the follow-up (n 1<unk>4 1332) were excluded as this study focuses on paediatric MetS. In addition to the 6683 children included in our analysis, 3086 eligible children were lost to follow-up. An additional 1748 children were lost to follow-up but would have likely been excluded due to being too old or too young during T4, because they were aged 0-4 or 16-17 years during T1. Written informed consent was obtained for each participant prior to participating in the cohort. The Lifelines Cohort study is conducted according to the conventions set forth in the Declaration of Helsinki, and it has received approval from the Medical Ethics Committee of the University Medical Center Groningen (METc approval number: 2007/152). --- Procedures Participants completed questionnaires, physical exams and venous blood draws during both the first and fourth assessment. During the third (T3) assessment, participants only completed questionnaires. Questionnaire data were self-reported and covered various topics including demographics. Physical exams and venous blood draws were conducted by trained research nurses using a standardized protocol. 11 --- Measures MetS was measured by a continuous MetS (cMetS) score, which was defined using measures of waist circumference, mean arterial pressure, fasting glucose, triglycerides and high-density lipoproteins from T4. Following an approach similar to Eisenmann et al., 4 each component of MetS was regressed on age and sex, and their residuals were standardized. Then, a cMetS score was built by summing the standardized residuals of all components. The standardized highdensity lipoprotein residuals were reversed due to their inverse relationship with health. The cMetS score indicates how an individual's cardiometabolic health compares to the rest of our sample. 4 Parental health literacy was measured at T3 using self-reported answers to the three validated questions from Chew et al. 13 (1) How often do you have trouble understanding your medical situation because you have difficulty with the written information? (2) How sure are you of yourself when you fill out medical forms? (3) How often does someone help you with reading information materials from the hospital or another healthcare provider? The parents of our participants indicate how often or to what extent these items apply to them on a scale ranging from 'never/not at all' (1) to 'always/very' (5). We reversed the scores of the first and third questions and then added up the scores of all questions, resulting in a health literacy scale ranging from 3 to 15. This score was then categorized into low (3-12) and adequate (13-15) health literacy, as was done in previous studies using data from Lifelines 14,15 and leads to percentages of low and adequate health literacy comparable to those from large-scale health literacy surveys in the Netherlands. 16 Because Dutch mothers usually spend more time parenting, 17 we constructed the parental health literacy variable using the mother's health literacy. If this was missing, the father's score was used. Additionally, we created an average measure of parental health literacy. If data from both parents were available, we took the mean of their health literacy scores and then dichotomized this into low and adequate health literacy. If data were only available for one parent, that parent's health literacy was used. SES was measured using three separate indicators: parental education, occupation and income. These data were obtained from both parents during T1. Education was assessed by asking parents about the highest educational level they attained, with eight potential responses ranging from 'no education' to 'university'. In an approach similar to De Graaf et al., 18 these categories were recoded into years of education using the number of years it would take to complete each category by the fastest route possible; e.g. no education, primary school, secondary school and university were coded as 5, 6, 12 and 16 years, respectively. Occupation was coded using the International Standard Classification of Occupations 2008. 19 This was recoded into Treiman's Standard International Occupational Prestige Scale, which is a continuous measure of occupational prestige. 20,21 For both education and occupation, the highest level from either parent was used. If only one parent was registered in Lifelines, then only data from that parent were used. To construct the measure of equivalized household income (income), net household income was divided by the square root of the number of people living on this income. 22 Similar to how health literacy was constructed, the mother's response was used, but the father's response was used if this was missing. Parental age was defined as the mean parental age from T1. If we only had records for one parent, then that parent's age was used. This approach was used because both parental SES and health literacy come from a mixture of maternal and paternal data. --- Statistical analysis First, we described the characteristics of our sample. We compared the characteristics of the eligible children who participated in T4 and the eligible children that were lost to follow-up. For this comparison, we only included children from the lost to follow-up group who most likely would have been eligible to participate in our study (baseline age 5-15 years). Before conducting any further analyses, we first imputed missing values for our variables. As all missing variables were continuous, we used the predictive mean matching (20 imputations) method from mice (v3.13.0) 23 ; we used as predictors of the missing values low-density lipoprotein, total cholesterol, glycated haemoglobin, weight, height, hip circumference, body mass index, systolic blood pressure, diastolic blood pressure, age, sex and parental age. To assess the extent to which parental health literacy mediates the relationships between different measures of SES and MetS (figure 1), we conducted a causal mediation analysis which uses the potential outcomes framework. 24 This approach defines the causal effect of an exposure as the contrast of outcomes that would be observed under different exposure and mediator values. 24 To achieve this contrast of potential outcomes, we expanded our dataset using an imputation procedure, from the medflex (v0.6-7) package for R, which results in the same individuals being evaluated at different levels of our exposure. 25 After expanding the dataset, we fit natural effects models using the medflex package. 25 Both the imputation procedure used to expand the dataset and the natural effect models were adjusted for baseline parental age. We also conducted sensitivity analyses in which we used average parental health literacy instead of the mostly Figure 1 Model demonstrating the direct and indirect pathways connecting parental SES to paediatric MetS maternal measure of parental health literacy. All data preparation and analyses were conducted using R version 4.0.2. 26 --- Results --- Sample characteristics One-fifth of the children included in the sample came from households with low parental health literacy (table 1). When parental health literacy was defined as the average of both parents' health literacy scores, the proportion of children from households with low health literacy rose slightly to 21.2%. The prevalence of low parental health literacy was similar in the children lost to followup (21.2%) as in those retained (20.0%). However, the children lost to follow-up were older than the children included in the analysis, 10.4 and 9.8 years, respectively. The children lost to follow-up also had worse biomarker levels and came from households with slightly lower parental SES than the children included in our analysis. --- SES and cMetS: relationships and mediation by parental health literacy Higher levels of parental SES were related to lower cMetS scores, and this was partially mediated by parental health literacy (table 2). If parental education were 4 years longer, e.g. university instead of secondary school, cMetS scores were on average 0.499 [95% confidence interval (CI): 0.364-0.635] units lower, which is a small effect (d: 0.18). Similarly, if parental income and occupational level would be 1 SD higher, cMetS scores were on average 0.136 and 0.196 units lower, respectively. The effects of parental income and occupation on cMetS were small (d: 0.05 and 0.07, respectively). Parental health literacy mediated the relationship between the individual measures of SES (education, income and occupation) and MetS by 6.7%, 11.8% and 8.3%, respectively. The total effects remained the same in the sensitivity analyses. However, the percentages mediated by parental health literacy, which was based on the average of both parents health literacy, increased to 10.5% (education), 17.5% (income) and 12.5% (occupation). --- Discussion We assessed to what extent parental health literacy mediates the relationships between different measures of parental SES and paediatric MetS. We found that higher parental SES resulted in slightly lower cMetS scores, and this was partially mediated by parental health literacy. The proportion mediated by parental health literacy varied across indicator of SES; it was the largest for income. Additionally, the percentage mediated was higher when parental health literacy was defined using the average of both parents' health literacy, but the pattern of our findings remained the same. The extent to which health literacy mediated the relationship between SES and MetS was limited; the proportion mediated was largest for income (11.8%). To the best of our knowledge, this is the first study to assess this mediating pathway. Importantly, we have done so in a robust manner by using a causal mediation analysis. Our findings are supported by a recent review which found that health literacy mediates socioeconomic inequalities in adults. 8 Given that education has the largest impact on health literacy, 8 it is not surprising that the indirect effect was largest for education. However, we also expected the proportion mediated by health literacy to be largest for education, instead of for income. Despite the strong relationship between education and parental health literacy, other factors appear to mediate this relationship. Health literacy may be an important attribute, especially amongst individuals with low income, as it could result in making better lifestyle choices given one's resources. Finally, it should be noted that the pattern of our findings was consistent in the sensitivity analysis, which used average parental health literacy instead of primarily using maternal health literacy. The proportion mediated increased across all indicators of SES when using average parental health literacy, and this may be explained by the fact that if only one parent has low health literacy, the other may be able to a: P-values from testing the difference between the two groups using Fisher's exact test for categorical variables or Kruskal-Wallis test for numerical variables. b: Equivalized household income: calculated as the net household income in Euros divided by the square root of the number of individuals who live off of the income. c: Occupation: measured using the standard international occupational prestige scale, which is a continuous measure of occupation. It focuses on the prestige an occupation gives its holder, not on the incomes associated with occupations. SES and pediatric MetS: mediation by health literacy 181 compensate. In short, parental health literacy explains a small portion of socioeconomic differences in paediatric MetS. We found that the effect of SES on MetS was rather limited, confirming findings of other studies that also used cMetS scores. 2,27 In the first study, it is difficult to gauge the size of the effect as the authors do not provide a standard deviation for their cMetS score. 2 However, they found that cMetS scores at baseline were on average 0.28 units higher for children whose parents had low educational levels than for children whose parents had high educational levels. The other study, on a sub-sample (n 1<unk>4 1217) of the children included in the present study, only found a small effect of parental education on cMetS. 27 The larger sample used in the present study may have provided additional power to detect the relationship between both parental income and occupation and MetS compared with that study. --- Strengths and limitations This study's strengths lie in the quality of its data and the robustness of our findings. Due to the community-based nature, standardized protocols and longitudinal design of Lifelines, we were able to obtain high-quality data for a large sample of children that are generally representative of this region of the Netherlands. 11,12 Using multiple indicators of SES to account for their unique relationships with health and the use of a causal mediation analysis added to the robustness of our findings. Furthermore, our sensitivity analysis, which had the same pattern of results as our primary analysis, demonstrated the robustness of our results. This study also has some limitations. First, the children lost to follow-up had slightly worse biomarkers and came from households with slightly lower SES at baseline than the children included in our study. This loss to follow-up may have introduced some bias, but it is unlikely to have had a major impact given the differences between the two groups were rather small. Additionally, we may have underestimated the full effect of health literacy as our measure of health literacy focused primarily on functional health literacy, which is a specific component of health literacy. More comprehensive measures of health literacy that include other components of health literacy, e.g. critical health literacy and communicative health literacy, may have led to a stronger mediating effect. Nonetheless, we used a validated questionnaire for health literacy 13 that has been used in other studies. 14,15 --- Implications This study indicates that interventions which aim to improve parental health literacy may partially reduce inequalities in paediatric MetS. Health literacy can be targeted at various levels. For example, a recent study found that using a health literacy intervention to target both the children's parents and their health care providers resulted in less weight gain during the first 18 months of life. 28 Additionally, the children themselves can be taught about health literacy during their schooling, and this has the potential for many benefits throughout the life course. 29 However, further research is needed into the potential impact of improving health literacy in children. Additionally, our results demonstrate that the effect of health literacy differs when defined using mostly maternal data vs. the average of both parents. Future studies should take this into consideration, and we also suggest studies should investigate the role of paternal health literacy. Finally, previous research has shown that health literacy impacts various aspects of children's health, 9 so it is likely that interventions targeting parental health literacy will influence various socioeconomic health inequalities. --- Conclusion Socioeconomic inequalities in paediatric MetS are relatively small, being largest by parental education. Targeting parental health literacy may not have a major impact on socioeconomic inequalities in paediatric MetS. Additional research is needed into the mediating role of parental health literacy on other socioeconomic health inequalities in children. a: Education: measured using the minimum number of years of education required to attain their highest level of education. The reference level is 12 years of education (secondary school), and the treatment level is 16 years of education (university education). b: Due to rounding, the proportion mediated may differ from the value calculated using the indirect and total effect presented in the table. c: Equivalized household income: calculated as the net household income in Euros divided by the square root of the number of individuals who live off the income. The reference level is the mean income level, and the treatment level is 1 SD above the mean. d: Occupation: measured using the standard international occupational prestige scale, which is a continuous measure of occupation. It focuses on the prestige an occupation gives its holder, not on the incomes associated with occupations. The reference level is the mean occupation level, and the treatment level is 1 SD above the mean. e: Parental health literacy was primarily measured using maternal data in the primary analysis, and it was measured using an average of both parent's data in the sensitivity analysis. --- Data availability Researchers can apply to use the Lifelines data used in this study. More information about how to request Lifelines data and the conditions of use can be found on their website (https://www.lifelines.nl/ researcher/how-to-apply). --- Welfare and Sport, the Dutch Ministry of Economic Affairs, the University Medical Center Groningen, University of Groningen and the Northern Provinces of the Netherlands. The authors are also grateful to Josue <unk>Almansa, PhD (Department of Health Sciences, University Medical Center Groningen, University of Groningen) who assisted with any statistical questions. Conflicts of interest: None declared.

Background: Parental health literacy may explain the relationship between parental socioeconomic status (SES) and paediatric metabolic syndrome (MetS). For this reason, we assessed to what extent parental health literacy mediates the relationships between parental SES and paediatric MetS. Methods: We used data from the prospective multigenerational Dutch Lifelines Cohort Study. Our sample consisted of 6683 children with an average follow-up of 36.2 months (SD 9.3) and a mean baseline age of 12.8 years (SD 2.6). We used natural effects models to assess the natural direct, natural indirect and total effects of parental SES on MetS. Results: On average, an additional 4 years of parental education, e.g. university instead of secondary school, would lead to continuous MetS (cMetS) scores that were 0.499 (95% confidence interval (CI): 0.364-0.635) units lower, which is a small effect (d: 0.18). If parental income and occupational level were 1 SD higher, on average cMetS scores were 0.136 (95% CI: 0.052-0.219) and 0.196 (95% CI: 0.108-0.284) units lower, respectively; these are both small effects (d: 0.05 and 0.07, respectively). Parental health literacy partially mediated these pathways; it accounted for 6.7% (education), 11.8% (income) and 8.3% (occupation) of the total effect of parental SES on paediatric MetS. Conclusions: Socioeconomic differences in paediatric MetS are relatively small, the largest being by parental education. Improving parental health literacy may reduce these inequalities. Further research is needed into the mediating role of parental health literacy on other socioeconomic health inequalities in children.

Introduction Names are often one of the first things we learn about people with whom we interact. A vast body of literature highlights the impact of first impressions, in part because they shape the way we process further information (Nisbett and Wilson, 1977;Nisbett and Ross, 1980). Names may thus influence perceptions and decisions, especially when other information is limited. One important example comes from the hiring process. Bertrand and Mullainathan (2004) find that resumes with distinctively Black names, such as Lakisha, receive 50% fewer callbacks from employers compared to resumes with distinctly white names like Emily. The authors postulate that this disparity may be due to employers' use of "quick heuristics" in screening applications. This explanation is supported by Barto<unk> et al. (2016)'s finding that employers spend less time reviewing resumes of minority job candidates. 1In this study we elicit people's beliefs regarding demographic characteristics, noncognitive skills, and productivity levels associated with different worker names. Beliefs are measured probabilistically, which provides precise variation in race associations both within and across respondents for given names. This novel data allows us to estimate the role of race perceptions on hiring decisions more accurately, without relying on distinctly Black names or making assumptions about how people perceive names. We can also estimate how much of hiring discrimination is accounted for by other beliefs than race, including unobserved beliefs about names that are shared across employers. By randomizing the hiring decision time, we further gain insights into race-based heuristics employers may use. The research design features two stages, similar to those of Barron et al. (2020) and Bohren et al. (2023). In the first stage, we recruit Black and white workers and collect data on their first names and productivity, measured as the share of financial receipts that they correctly transcribe (Abel, 2022). In the second stage, we recruit a nationally representative sample of 1,500 individuals from the U.S., over half of whom have real-life experience making hiring decisions, to serve as employers in our experiment. We first elicit their beliefs about demographic characteristics (race, age, education), noncognitive skills (e.g. responsibility and motivation), and productivity levels for a set of names from our worker sample. 2 For the second task, employers make hiring decisions for ten worker pairs with names drawn from our worker sample. They receive a bonus each time they select the more productive worker. We present four main results, following a registered pre-analysis plan. Firstly, the data show substantial variation in beliefs about characteristics across worker names and that these associations are strongly correlated with perceptions of race. Names commonly associated with being Black are associated with lower levels of education and noncognitive skills. The magnitude of the racial disparity is very large. For instance, perceiving a name to be 100% Black compared to 100% white is associated with a 16 pp (47%) decrease in the likelihood of holding a master's degree and a 29 pp (42%) and 26 pp (46%) decrease in the likelihood of being perceived as responsible and trustworthy, respectively -traits that play an increasingly important role in labor markets but are hard to observe from a resume (Deming, 2017). Our findings reveal that these biased perceptions lead to disparities in worker assessments, resulting in a 25% gap in productivity beliefs based on perceiving someone as Black vs. white. However, this race penalty is not primarily driven by differences between distinctly Black names (e.g. Shanice) and distinctly white names (e.g. Heather). Controlling for name fixed effects, which accounts for commonly shared associations of names, only reduces the productivity race penalty by 10 pp or 40%. It remains large at 15%, highlighting the importance of variance in race perceptions between employers for the same name. This within-name variation in beliefs is important as it may help explain heterogeneity in discrimination observed in recent studies (e.g. Kline et al. (2021)). The second key finding is that participants' racialized perceptions are highly predictive of their hiring choices. On average, employers are 29.6 pp less likely to hire a worker they perceive as Black compared to white. Controlling for productivity and noncognitive skill beliefs reduces this gap to 21.3 pp and 20.3 pp, respectively. Accounting for age, education, productivity and noncognitive skill beliefs simultaneously reduces the race gap to 13.1 pp. These results are robust to excluding distinctly Black names from our sample. Results are also unchanged when we exclude respondents who suspect that the research is about discrimination, which assuages social desirability concerns. We find that discrimination varies substantially based on certain employer characteristics. The overall 29.6 pp race gap narrows for female (by 7.5 pp), younger (6.3 pp), Black (12.1 pp), and liberal (11.7 pp) employers as well as those who support race-based affirmative action (13.7 pp). Despite these reductions, the race gap remains substantial and statistically significant (at the 0.1% level) across each of these groups. By contrast, no significant differences in the race gap are observed based on employers' educational attainment, the level of racial diversity in their zip code, or whether they had previous experience with hiring. Third, we investigate mechanisms underlying discrimination in hiring decisions. Across various tests, we find evidence suggesting that employers use race as a heuristic in their decision-making process. Differences in race perceptions between candidates increase employers' confidence in their hiring decision and reduce the decision time, even after controlling for differences in perceived productivity. 3 However, in cases where a candidate is perceived as both more productive and more likely to be Black, the decision-making time increases, indicating that decision heuristics based on race and productivity beliefs are in conflict. 4 To further investigate the role of heuristic-based decision-making, we randomize one-third of employers to make hiring decisions within a two-second time frame, reflecting the fact that recruiters often spend very little time reviewing resumes at early screening stages (Cole et al., 2007). We find that time pressure exacerbates hiring discrimination by around 25%, with the most adverse effects observed for Black candidates perceived to be highly productive. Importantly, there is substantial heterogeneity in who is affected by rushed decisions: discrimination increases by almost 70% among white employers who support race-based affirmative action (AA) but has no effect on the hiring decisions of Black employers or white employers who oppose AA. Fourth, to parse and quantify the role of heuristic biases and analytical belief-based evaluations, we use data on hiring decisions, response times as well as race and productivity beliefs, to estimate the parameters of a structural drift-diffusion model (DDM). The DDM, which is widely used in neuro-psychology, posits that decision-makers sequentially sample information about choice options from memory and make a decision once an evidence threshold has been reached. 5 This process causes decision values to drift gradually in the direction of preferred candidates. Motivated by reduced form evidence, we model this drift to be affected by employers' race and productivity beliefs. Our structural estimates reveal that both productivity and race beliefs affect the decision drift, and that their relative importance in the decision process varies over time. In quick decisions, there is a significant drift favouring workers with white names, suggesting the use of (race-based) heuristics in hiring decisions, especially when people are under time pressure. Interestingly, despite observing similar levels of discrimination in long hiring decisions between Black employers and white employers who support AA, the underlying mechanisms behind this outcome are very different. Black employers exhibit a relatively weak heuristic response to Black names, whereas white employers who support AA initially display a strong negative heuristic response, which is subsequently overridden in longer decisions. White employers who oppose AA also exhibit a strong heuristic response, but this judgement aligns more closely with that of their slower, analytical cognitive process. Counterfactual analyses based on our structural estimates indicate that, on average, a Black worker would need to be 20.3 pp (1.2 sd) more productive to have an even chance of being hired over a white worker in slow decisions. This figure increases to 39 pp (2.4 sd) for very quick decisions. 6 and Siegelman, 1993). We also find that the race gap does not differ along employers' level of risk aversion. 4 This aligns with the concept of implicit association tests (IAT), which measure the unconscious associations of, for example, race (Black vs. white) and evaluations (good vs. bad) by comparing the speed of sorting decisions when concepts and evaluations are aligned versus misaligned (Greenwald et al., 2003). 5 The DDM relates to a recent set of models. E.g. in Bordalo et al. (2020), memories about past consumption experiences and prices both serve as anchors for evaluating choices. Other models, explore the role of memory in forming beliefs (e.g. Mullainathan (2002); Bordalo et al. (2019); Enke et al. (2020)). 6 These results align with previous research that highlight the need for members of discriminated groups Our study relates to an extensive body of literature that uses audit studies to detect discrimination in dimensions such as race, gender, or nationality (for reviews see Baert (2018); Bertrand and Duflo (2017); Neumark (2018)). In these studies the characteristic of interest (e.g. race) is typically signaled through individuals' first names. However, a well-known challenge in interpreting audit study results is that employers may associate candidates with distinctly Black names also with other characteristics important for the hiring decision such as lower educational attainment (Fryer Jr and Levitt, 2004). While other studies have collected name associations (e.g. Baert et al. (2022); Crabtree et al. (2023); Gaddis (2019)), our study is to our knowledge the first to incentivize truthful belief elicitation for demographic characteristics and to validate that reported noncognitive skill associations are predictive of incentivized decisions (in a trust game). Another key innovation is the collection of data on people's distribution of beliefs about names. 7 Simulations based on our data show that if we had followed existing studies and collected binary race associations, we would have missed about half of the race penalty in hiring and beliefs. This suggests that a significant share of discrimination occurs between candidates for whom employers agree on the most likely race category but differ in their level of certainty. For example, our data indicate that Colin is perceived as 77% likely to be white, whereas Robert is seen as 60% likely to be white. Yet, for both these names, around 90% of employers believe that the single most likely race category is white. Binary race perception data would treat these two names as equally white and thus fail to attribute employer preferences for Colin to race beliefs, leading to an underestimation of discrimination. A second key benefit of using probabilistic data is the ability to more precisely estimate the variance in race beliefs induced by studies that randomize names. This, in turn, allows calculating local average treatment estimates (LATE) from intent-to-treat estimates reported in audit studies, such as differences in callback rates. The LATE is important as it provides a more accurate measure of the impact of perceiving a candidate as Black. Our study also contributes to the growing body of literature examining the role of employers' beliefs about workers. Traditional models distinguish between taste-based discrimination (Becker, 1957) and statistical discrimination (Arrow, 1973;Phelps, 1972). Recent research has added more nuance to this categorization by demonstrating that beliefs can often be inaccurate due to biases, stereotypes, or lack of information (e.g. Barron et al. (2020); Bohren et al. (2023); Bordalo et al. (2016Bordalo et al. (, 2019)); Coffman et al. (2021); Chakraborty and Serra (2022)). Bohren et al. (2023) develop a model of "inaccurate statistical discrimination" and show that the accuracy of beliefs has crucial implications for identifying underlying mechanisms and for designing policies to reduce discrimination. Comparing employers' productivity beliefs to actual productivity data of 2,400 workers to outperform others in order to overcome discrimination (Bohren et al., 2019). 7 Reporting the (between-subject) variance of the binary race belief measure (such as Crabtree et al. ( 2023)) does not capture the within-person distribution of beliefs. Abel and Burger (2023) describe in more detail how researchers can access and use our data. performing the same task, shows that employers' perceived race gap of 25% greatly exceeds the actual productivity difference of 9%. Our paper quantifies the effect of such biased beliefs on hiring decisions and employs a structural model to investigate how these beliefs are incorporated into the decision-making process. We are, to our knowledge, the first to apply the DDM to study hiring decisions. Methodologically, we are also the first to extend the DDM to allow decisions to be determined by elicited beliefs rather than relying on stated preferences and to allow the drift to be affected by perceptions of both race and productivity. In addition, randomizing the decision time helps identify the extent to which discrimination operates through heuristic and analytical cognitive processes. These findings relate to a growing literature showing how (biased) beliefs affect how much attention people pay to information, which may inhibit learning (Schwartzstein, 2014;Esponda et al., 2020;Gagnon-Bartsch et al., 2021). Last, our paper helps explore the reasons behind the large heterogeneity in discrimination documented across jobs and industries in recent studies. For example, Kline and Walters (2021) conclude that "while most jobs barely discriminate, a few discriminate heavily." These findings are inconsistent with traditional models of statistical discrimination, but could be explained by heterogeneity in taste-based discrimination (Charles and Guryan, 2008). Our findings suggest that variation in (race) beliefs across hiring managers is one reason for this type of heterogeneity. This may help explain why variance in discrimination is more common for firms without centralized HR systems in which individual hiring managers have more decision power (Kline et al., 2021). Our results are also in line with studies identifying variation in the decision-making process including the time dedicated to screening of applicants as an important source of heterogeneity in discrimination (Bohnet, 2016). Our findings also highlight limitations of interventions to reduce discrimination, e.g. by increasing attention. The paper proceeds as follows: Section 2 describes the research design and Section 3 introduces the empirical strategy. Section 4 discusses results on name associations and Section 5 investigates the role of these associations on hiring decisions. Section 6 explores underlying mechanisms and reports results from our structural estimation. Section 7 concludes. --- Study Design --- Worker Recruitment and Name Selection A key feature of our study design is that it leverages real-life worker productivity data obtained from a previous field experiment. Specifically, we collect data on how people performed in transcribing financial receipts (see Abel (2022) for a detailed description of the study.) For the current study, we obtain permission from 150 of the initial sample of workers to use their first name in our research. We also collect data on incentivized choices in a trust game, as described in more detail below. We over-sample Black workers to obtain a racially balanced sample and categorize worker names as predominately Black, predominately white, or of ambiguous race based on pilot surveys that elicit race perceptions of names.8 From each of these three categories, we sample five female and five male names. These 30 names are used as our worker sample in the hiring experiment. --- Employer Recruitment and Sample Characteristics The research design for our hiring experiment, conducted in the second stage, is summarized in Figure 1. We recruit a new sample of 1,500 people based in the U.S. to act as employers in our experiment. These participants are recruited through Prolific, which allows us to recruit a sample that is nationally representative in terms of key demographic characteristics such as age, race, and gender. This represents a significant advantage over other studies that recruit subjects online, which tend to be disproportionately young and white. The characteristics of our sample can be found in Column 3 of Table A1. The median (mean) income of participants is approximately USD 45,000 (USD 55,700), which is close to the national median of USD 41,500. Despite being representative on key demographic characteristics, our sample is more educated than the national average with 47% holding a college degree. 56% of our sample has previously been involved in hiring, which is important since it allows us to test whether hiring behavior in our experiment differs for those with actual hiring experience. Our data is also notable for its broad geographical representation, with participants from all 50 states and 1,373 zip code-level locations. By linking the zip code level data to census data, we are able to learn characteristics of the participants' geographic locations, including levels of racial diversity. --- Belief Elicitation We inform employers that the study aims to learn about the associations people have with names (see Appendix Figure B4 for details). We first present employers with six names (in random order). The names consist of one female and one male from each of the three categories: distinctly Black, distinctly white, and ambiguous race associations. For each name we collect belief associations from approximately 300 respondents. We elicit beliefs about race/ethnicity, age, and education levels by asking participants the following question for each name "Out of 10 people named... how many people are...? ", followed by the categories of race/ethnicity, age, and education described in more detail below. 9 This method has several key advantages over previous research that asked about a single race category. Most importantly, it provides more nuanced data of the distribution of race beliefs and captures people's level of certainty over name associations. 10 Another advantage is that we incentivize participants to provide accurate answers by informing them that their responses will be compared to official national statistics, and that they will receive a bonus of USD 2.00 if they correctly order the characteristics of five name comparisons. 11 In addition to eliciting beliefs about race, age, and education, we also collect data on noncognitive skill associations. Specifically, we ask participants to select from a list of seven traits (responsible, trustworthy, cooperative, assertive, self-motivated, perfectionistic, decisive) that they believe are characteristic of someone with a particular name. 12 We chose noncognitive skills that are widely regarded as important in the workplace (Heller and Kessler, 2022). While there is no objective data on how noncognitive skills vary by name (preventing us from incentivizing truthful responses), we can still compare people's responses to incentivized choices described in more detail below. Last, to elicit beliefs about the productivity of workers, we present participants with six names in random order and ask them to estimate the productivity of workers with each 9 We mix concepts of race (e.g. Black) and ethnicity (e.g. Hispanic) as preliminary data collections showed that participants are confused by questions that elicit data separately (as e.g. done by the U.S. census). 10 Conceptually, a larger variance of beliefs may either mean that a person knows that the distribution is more disbursed or that the respondent is more uncertain about the name distribution. Both can be sources of uncertainty. In pilot data collection, we confirmed that reported certainty about associations are negatively correlated with a person's variance in race beliefs. 11 To further explain this scheme, we present people with a concrete example of worker names and beliefs about age that determine the bonus payout. We do not specify which characteristics we use to determine the payout. In practice, we have administrative data on race and age. 12 We ask: "What characteristics do you think does a person with the name NAME have? (Please select all that apply)". The seven traits are presented in random order. name. We provide context by explaining that we have collected performance data in a financial receipt transcribing task (see Appendix B.3 for details). We further follow Bohren et al. (2023) and inform participants that a "typical worker correctly transcribed 65% of receipts" and incentivize accurate responses by offering a bonus of USD 0.50 for each correct productivity guess they make. --- Hiring Decision and Randomization After eliciting beliefs, we conduct the hiring experiment in which employers make sequential binary hiring choice for ten pairs of worker names. The ten pairs, presented in random order, are selected from the six names for which beliefs were elicited, and include combinations of workers with both the same and different race categories. Similar to Barron et al. (2020), each time they select the more productive worker, the employer receives a bonus of 10 cents. We also record the time taken to make each hiring decision. The average (median) decision time is 2.1 (1.8) seconds. This means that making correct choices could result in an additional payout of up to USD 1.00 over less than half a minute. Employers do not learn about how many correct choices they made until the end of the experiment. The hiring experiment features a cross-randomization design that introduces exogenous variation in decision time and worker bonus (Figure 1). For the time variation, one-third of employers is informed that they must make a hiring decision within two seconds in order to be eligible for a bonus in a given hiring round. For the worker bonus variation, one-third of employers are informed before making their choices that the workers have the opportunity to earn an additional payout each time they are selected. 13 This "worker bonus" treatment mimics the benefit workers receive from being selected in real-world hiring decisions. Importantly, the worker bonus does not affect the monetary incentives that employers face and thus only affect hiring choices if people differentially care about the payout of white vs. Black workers, which could be one reason for why employers hire white workers. After completing the hiring task, we offer participants the opportunity to earn an additional bonus through a trust game.14 Specifically, we explain that we collected data on how much money each worker in our sample returned in the role of the recipient in the trust game. Participants can then choose a partner from a list of six names (presented in random order) and are paid a bonus according to how much the selected worker decision. Finally, we administer a short survey that collects data on employers' socio-demographic characteristics including age, education, income, location, and race as well as their political leaning, experience in hiring, level of risk aversion, and performance in a cognitive reflection test. --- Empirical Strategy --- Beliefs of Worker Characteristics and Productivity Suppose an employer believes the productivity of a worker, y, is determined as y = <unk> + r<unk> + x<unk> + u where x is a vector of observable worker attributes, <unk> is a vector containing the perceived productivity effects of these attributes, r is binary vector containing a single 1 corresponding to the worker's race, <unk> reflects perceived race productivity effects, and u is a scalar summarizing the perceived effect of unobservable worker attributes on productivity. In our experiment, employers are provided with a name, n, and are then asked to state their beliefs regarding the race, observable attributes and productivity of the worker. We denote these as rn <unk> E(r|n), xn <unk> E(x|n) and <unk>n <unk> E(y|n). Since race beliefs are elicited as probabilities, this directly measures rn = (r W n, rBn, rHn, rAn, rOn ) where each element represents the employer's reported probability that a worker with name n is of race white, Black, Hispanic, Asian, or other, respectively. If the employer believes, consciously or subconsciously, that worker attributes are correlated to race, then this would induce a correlation between race associations for a worker with name n, rn and beliefs about worker attribute k for a worker with name n, xkn : xkn = <unk> k + rn <unk> k + E(e|n)(1) This relationship can be estimated by regressing employer beliefs regarding worker attributes on employer beliefs regarding race across all names, as we do in section 4 below. If employers form productivity beliefs according to Bayes' rule, productivity predictions for a worker with name n are: <unk>n = <unk> + rn <unk> + xn <unk> + E(u|n)(2) The relationship between the employer's race beliefs and productivity predictions can be assessed by regressing predicted productivity on race beliefs: <unk>ni = <unk> + <unk> B rBni + <unk> H rHni + <unk> A rAni + <unk> O rOni + xni <unk> + <unk> n + u ni(3) where subscript i refers to the beliefs held by employer i. We choose the perceived probability of being white, rWn, as the reference category in our empirical analysis, so the race coefficients for race q now represent <unk> q = <unk> q -<unk> W, i.e., the perceived productivity effect of belonging to race q compared to white. We cluster standard errors at the employer level to account for correlation in outcomes between choices for the same employer. When not controlling for employer beliefs regarding worker attributes, the estimated race coefficients reflect both the direct perceived race effect on productivity, as well as the indirect affect that operate through perceived correlations between race and other worker attributes that are believed to determine productivity (Bohren et al., 2022). We can move closer to estimating this perceived race productivity effect by controlling for beliefs regarding commonly observable human capital measures (education and age) and noncognitive skills, xni. Our experimental design also allows us to control for name fixed effects, <unk> n, which can capture any variation between names and unobservable productivity determinants, like socio-economic status, that is shared between employers. In contrast to other studies in this literature, we directly measure employer beliefs regarding worker attributes that are difficult to observe from resumes. This means we can more comprehensively control for confounding effects that may inflate the race association coefficients. --- Hiring Decisions When making a hiring decision between two workers with names n 1 and n 2, the employer may compare their predicted productivities and select worker n 2 if E(y|n 2 ) > E(y|n 1 ), and candidate n 1 otherwise. Following equation 3.1, the expected difference in productivities can be expressed as: E(y|n 2 ) -E(y|n 1 ) = (E(r|n 2 ) -E(r|n 1 ))<unk> -(E(x|n 2 ) -E(x|n 1 )<unk> + E(u|n 2 ) -E(u|n 1 ) (4) Assuming that E(y|n 2 ) -E(y|n 1 ) is normally distributed allows the estimation of <unk>/<unk> and <unk>/<unk> with a probit estimator. A linear approximation of this process can be expressed in terms of observable sample data: h ji = <unk> B <unk>r Bji + <unk> H <unk>r Hji + <unk> A <unk>r Aji + <unk> O <unk>r Oji + <unk>x ji <unk> + v ij (5) where j denotes a specific name pair (n 1, n 2 ), <unk>r qji <unk> rqn 2 i -rqn 1 i is the difference in the perceived probabilities that worker n 1 and n 2 are of race q, and <unk>x ji <unk> xn 2 i -xn 1 i is the perceived difference in worker attributes. The estimated coefficient <unk> q = <unk> q -<unk> W (now implicitly scaled by <unk>) measures the effect of being perceived to belong to race q rather than being white. As before, we cluster standard errors at the employer level. --- Race Figure 2 shows the distribution of race associations for the 30 worker names in our sample. The names are ordered based on the perceived likelihood of being Black. Although many names in the sample are clearly categorized as either predominantly Black or white, some names, such as Jackson and Darryl, are perceived to be Black and white in roughly equal frequency. In addition to differences in the aggregate perceptions of names, there is also considerable variation in beliefs for a given name across employers. For example, while Charisma and Alijah share similar aggregate race associations, Figure A3 shows that there is a higher variance in beliefs for Charisma. (It is worth noting that due to our focus on the Black-white race gap in the selection of names, there is limited variation in the perceived likelihood of being Hispanic or Asian.) Eliciting data on the distribution of race associations has several key implications. Conceptually, it provides a more nuanced understanding of individuals' levels of uncertainty about their beliefs. Appendix Figure A4 shows that had we asked for a single race categorization, we would miss much of the nuance in people's beliefs and discriminatory behavior (as dis-cussed below). 15 From a methodological perspective, the distribution of beliefs about race is crucial for interpreting the results of studies that use names to signal race. For example, differences in call-back rates from sending fictitious resumes present intent to treat estimates. To back out the arguably more meaningful treatment-on-the-treated (LATE) estimate, it is necessary to know the (first stage) variation in race perception that is induced by names. --- Education Figure A1 shows the distribution of beliefs about the highest level of educational attainment across names. While there are some outliers, there is a clear correlation between beliefs about race and education. Table 1 shows results from specification 1. Names perceived as Black (compared to white) are associated with having 0.73 years (0.73 s.d.) fewer years of education (Col. 1). Next, we compare beliefs about specific categories of educational attainments. Strikingly, perceiving a name as Black lowers the chance that the person is believed to have a master's degree by 16 pp (47%) (Col. 2). By contrast, we do not see a difference for having completed (exactly) a four year degree (Col. 3) and an increase of 9 pp for having completed a two year degree (Col. 4). People also associate Black names with a 10 pp (62%) lower likelihood of holding a high school diploma and 3.8 pp (30%) higher chance of having less than a high school degree (Col. 6). Panel B shows the corresponding results controlling for name fixed effects. Only exploiting within name variation in race beliefs reduces the race gap in the number of years of education by 0.12 (Col. 1). Coefficients for specific education categories also converge towards zero but remain highly significant. --- Age We elicit age associations by asking participants how many people with a given name fall in each of the following age bins: 18-29, 30-44, 45-59, 60+. Using the midpoint for each bin (and 70 for the 60+ category), we compute the average age for each name. Perceiving a name as Black is associated with being 6.5 years younger (Table A2). One explanation is that distinctly Black names were more commonly used in the 1970s and 1980s (Fryer Jr and Levitt, 2004). Controlling for name fixed effects, the age difference drops by almost 80% to 1.4 years suggesting that much of the age race gap is associated with name-specific associations. 16 15 To assist researchers who want to choose names for studies on race and gender discrimination, we are compiling a companion paper with associations for a comprehensive set of names and guidance on how to systematically select names (Abel and Burger, 2023). 16 To assess how accurate people's beliefs are, we use census data to estimate the actual average age in the U.S. population. Specifically, we look at the frequency of birth at the midpoint of the age bins used in our belief elicitation method, which we use to compute a weighted average as a proxy for the actual age. Figure A2 plots the perceived vs. actual age data for our names, with points on the 45 degree line presenting --- Noncognitive Skills Figure 3 shows the belief distribution for four noncognitive skills across names ordered by the perceived share of being Black. Visual inspection shows that perceiving a name as Black is negatively correlated with being trustworthy and responsible, positively correlated with being assertive and uncorrelated with being motivated. (Appendix Figure B1 illustrates similar patterns for the other noncognitive skills.) Table 2 reports the corresponding coefficients from specification 1 for each noncognitive skill. Perceiving a name as Black (compared to white) reduces associations with responsibility by 29.4 pp (42%), trustworthiness by 26.5 pp (46%), cooperativeness by 24.8 pp (40%), and perfectionism by 19.6 pp (65%). By contrast, it increases associations with assertiveness by 34.1 pp (101%) and decisiveness by 12.6 pp (35%), while motivation perceptions are not correlated with race associations. Including name fixed effects reduces most coefficients by around 30-50% but they remain large and significant (Table 2, Panel B). Noncognitive skill perception data also show clear differences at the intersection of gender and race. For example, gendered associations regarding women being less assertive, more trustworthy, and more cooperative only extend to female names perceived to be white (Table A3). While not the focus of this paper, this points to additional distinct challenges faced by Black women. accurate perceptions. It shows that people tend to overestimate the age of people with names of younger people (e.g. Isaiah) and underestimate the age of names of older people (e.g. Robert). However, there is a strong positive relationship between perceived and actual age with a correlation coefficient of 0.81. As previously explained, to test whether these associations affect (economic) behavior, respondents can choose who they want to partner with in a trust game and receive the amount that person returns as a bonus. Figure A5 plots the relative frequency with which names are perceived as trustworthy and how often workers with these names are picked as a partner in the trust game. While there is some divergence (esp. for white men), associations and incentivized choices are highly correlated. Stating that a name is trustworthy increases the probability that this person will choose a worker with that name in the trust game by almost 80%. Overall, these results suggest that the noncognitive skill associations we collect are predictors of incentivized choices. --- Perceived Productivity We first show the relationship between productivity beliefs and race beliefs graphically. Figure 4 plots the average perception of being Black and productivity for each worker name, averaged across 300 employers. While there are some positive (e.g. Jazmine) and negative (e.g. Kevin) outliers, we find a strong negative and relatively linear relationship. Table 3 shows results from specification 3 using individual perceptions of productivity, measured in log of correctly transcribed receipts as the dependent variable. Without control variables a Black name is associated with about 25% lower productivity compared to a white name (Col. 1). While our pre-analysis plan does not include tests of discrimination towards other race groups given the limited variation in perception for names included in our sample, it is noteworthy that the race gap is half the magnitude (12.9%

economists, whose research aims to provide answers to the global labor market challenges of our time. Our key objective is to build bridges between academic research, policymakers and society. IZA Discussion Papers often represent preliminary work and are circulated to encourage discussion. Citation of such a paper should account for its provisional character. A revised version may be available directly from the author.

white men), associations and incentivized choices are highly correlated. Stating that a name is trustworthy increases the probability that this person will choose a worker with that name in the trust game by almost 80%. Overall, these results suggest that the noncognitive skill associations we collect are predictors of incentivized choices. --- Perceived Productivity We first show the relationship between productivity beliefs and race beliefs graphically. Figure 4 plots the average perception of being Black and productivity for each worker name, averaged across 300 employers. While there are some positive (e.g. Jazmine) and negative (e.g. Kevin) outliers, we find a strong negative and relatively linear relationship. Table 3 shows results from specification 3 using individual perceptions of productivity, measured in log of correctly transcribed receipts as the dependent variable. Without control variables a Black name is associated with about 25% lower productivity compared to a white name (Col. 1). While our pre-analysis plan does not include tests of discrimination towards other race groups given the limited variation in perception for names included in our sample, it is noteworthy that the race gap is half the magnitude (12.9%) for Hispanic names and that there is no gap for Asian names (Col. 1). When we control for the number of years of education, the Black-white race gap slightly decreases to about 23%, reflecting that education is a positive determinant of productivity that is negatively correlated with perceptions of being Black (Col. 2). Every year of education is associated with a 3.1% increase in productivity, meaning that the raw penalty of a Black name is equivalent to about 8 years of education. Controlling for age beliefs only has a very small effect reducing the race productivity gap to 24.1% as age is only weakly correlated with productivity (Col. 3). Controlling for noncognitive skill perceptions reduces the race gap by almost a fifth to 21% (Col. 4). This reflects that most of the traits positively associated with productivity, such as trustworthiness and responsibility, are also more likely to be associated with white names. By contrast, while Black names are more likely to be perceived as assertive, this trait is not correlated with productivity perceptions. For our most flexible specification, we estimate regressions with name fixed effects, which effectively only uses between-subject variation in beliefs for the same name to estimate the race gap. This specification reduces the race gap from 25% to 15% (Col. 6) or to 12% when we control for other beliefs (Col. 7). The importance of variation in perceptions for the same name points to an important potential source of heterogeneity in discriminatory behavior observed in audit studies (e.g. Kline et al. (2021)). To assess the accuracy of employer beliefs, we use data on the actual productivity of 2,400 workers in this task (collected by Abel (2022)). Table A4 compares the productivity race gap as perceived by employers (Panel A) versus the actual data (Panel B). While the perceived race gap is between 22 and 25%, the actual gap is below 9%, implying that employers overes- Figure 5 shows how productivity beliefs vary across employer characteristics. While every subgroup overestimates the race gap, there is substantial heterogeneity along certain characteristics in how much groups do so. Most strikingly, those supporting race-based affirmative action (AA) think the productivity race gap is 16% while those opposed to AA believe it is more than 40%. Participants involved in real-world hiring also overestimate the race-gap more. By contrast, it does not vary along other characteristics including employers' race and level of education. The next Section will first test the extent to which these employer beliefs affect hiring decisions. Section 6 will then explore mechanisms of how beliefs factor into the decision-making process. --- Hiring Decisions --- Hiring Race Gap Figure 6 shows the relationship between race perceptions and hiring probability for each of the worker pairs included in our experiment.17 The x-axis shows the difference in the likelihood of being Black between the two candidates, averaged across roughly 300 employers. We include worker pairs of similar race, distinctly different race, and intermediate race differences, resulting in three clusters of pairs along the x-axis. The y-axis presents the average hiring probability for the person listed first in the pair. Notably, almost all pairs are above the 50% hiring probability line, indicating the strong negative relationship between perceiving a name as Black and hiring that candidate. While we are fitting a quadratic function, the relationship is almost exactly linear, suggesting that there is not a threshold at which the hiring penalty of being perceived as Black is changing. Pairs above (below) the fitted line present cases where the first worker is hired more (less) than predicted by race perceptions. Over-performing names include Sara, Josh and Isaiah, while under-performing names include Brandon, Krystal, and Charisma. These patterns point to other beliefs, e.g. regarding education, that may systematically vary across names. To gain precision and account for other employer beliefs, Table 4 presents results from specification 5, using the full set of more than 14,000 hiring decisions. Without controlling for other beliefs, we find that a worker perceived to be Black is 29.6 pp less likely to be hired than someone with a name perceived to be white (Col. 1). Col. 2 controls for employers' productivity perceptions of the two candidates. A one percentage point advantage in perceived productivity increases the likelihood of being hired by 0.73 pp. Controlling for productivity perceptions reduces the race gap in hiring by roughly 30% to 21.3 pp (Col. 2). As productivity perceptions may not affect hiring decisions linearly, we next include a binary variable indicating whether a worker is perceived to be more productive. This increases the likelihood of being selected by 27 pp but only narrows the race gap to 23.7 pp (Col. 3). Next we control for perceived differences in years of education (Col. 4) and age (Col. 5). A one year increase in (perceived) education is associated with a 6.3 pp increase in the likelihood of being hired. This implies that being perceived as Black has a similar impact on one's chances of being hired as having almost five fewer years of education. Accounting for differences in education perceptions narrows the hiring race gap by 5 pp (18%). Age has a smaller but still positive effect on hiring and reduces the racial gap in hiring by 2.1 pp (7%). Perceptions of most noncognitive skills are highly correlated with hiring decisions and explain as much of the variation in hiring as productivity perceptions (Col. 6). For example, perceived a worker as more responsible than their competitor, is associated with a 12.8 pp increases in the chances of being hired. By contrast, the trait most commonly linked to Black names, assertiveness, is not associated with hiring decisions. Overall, accounting for differences in noncognitive skills reduces the race gap by approximately 32% to 20.3 pp, highlighting the importance of employer beliefs about these traits. Notably, perceptions of noncognitive skills remain significant predictors of hiring even after accounting for education, age and productivity beliefs. This is important since many of these latter characteristics are observable to employers on a candidate's resume. This most flexible specification reduces the race gap by about 45% to 13.3 pp (Col. 7). 18Lastly, we can estimate the previous specifications with worker pair fixed effects. Akin to our name fixed effects analysis, this specification estimate coefficients only from variation in race beliefs between people for the same candidate pair. Table A5 shows that while this reduces the race coefficient by about 45%, the race gap remains large at 15.8 pp. Results (not reported) further show that the size of the race gap is robust to including employer fixed effects. --- Subgroups Figure 7 presents the findings of various subgroup analyses, as specified in our pre-analysis plan (corresponding regression results are in Table B4). We find that women exhibit less discriminatory behavior as the race gap is 26 pp compared to 34 pp for men. In contrast, the race gap is not affected by whether a person has a college degree or not. The hiring race gap among Black employers is 19 pp compared to 33 pp for white employers and 27 pp for other race groups (Col. 4). The fact that we observe a substantial race gap among Black employers provides evidence against racial animus as the main driver of our results. We also find a negative relationship between participants' age and their level of discrimination. The race gap is 50% higher for people over 55 compared to 18 to 35. With regard to people's political leaning, we find that the the race gap is 70% larger for conservative compared to liberal employers (39 pp versus 23 pp). Likewise, we observe a strong negative relationship between support for race-based AA and hiring discrimination: Among those opposed to it, the gap is 42 pp compared to 19 pp for those in favor. Next, we look at the role of participants' location. Figure A6 shows how the race gap varies across states. We do not observe a clear geographical pattern of race discrimination. We can also compute the racial diversity of participants' zip-code level area using census data. Dividing this measure into terciles, we find no variation in discriminatory behavior along this dimension, although results (not reported) show that white respondents display a slightly higher race gap if they live in the most diverse areas (35 pp vs. 31.3 pp). Last, hiring discrimination does not differ based on whether participants have (real-world) experience in hiring. This partly assuages some concerns about the external validity of our results. Figure A7 replicates the same subgroup analyses but controls for differences in productivity, education, age, and noncognitive skill perceptions between workers. While the race gap closes by around 50%, the overall pattern across subgroups persists. Notably, the race gap remains statistically significant for all subgroups at the 0.1% level. It is important to note, however, that these coefficients only provide a correlation between groups, such as measuring the behavior of people with or without a college degree, rather than a causal relationship between obtaining education and exhibiting discriminatory behavior. --- Robustness In research related to discrimination, participants may be hesitant to reveal their true attitudes and behaviors, which would compromise the internal validity of results. To address this concern, we do not reveal the exact research question at the outset of the study, elicit a range of associations beyond just race, and also included hiring choices for worker pairs without racially distinct names. Despite these efforts, about 33% of respondents suspected at the end of the experiment that the research was related to race. 19 This share increases to 49% when we include other forms of biases and discrimination (e.g. gender). Table B5 replicates our main results separately for those who suspected the research was related to race and those who did not. While we acknowledge endogeneity and misreporting concerns, it is reassuring that our results remain robust, both for the overall race gap and the effect of controlling for beliefs. Specifically, the race gap is only about 1.9 pp (6.5%) smaller for those suspecting the research is related to race (Col. 1-4). Using our broad measure, this difference decreases to less than 1 pp (Col. 5-8). A related concern is that people do not suspect the research question in the beginning, but may alter their behavior once they become aware of the study's focus on race. To investigate this possibility, we analyze how discrimination changed over time by examining the (randomized) order of the hiring pairs. Our results, as shown in Figure B2, indicate that the race gap does not vary systematically across the ten hiring decisions. One possible explanation for these findings is that incentivizing choices may have addressed surveyor demand effects. However, it remains possible that people hide discriminatory behavior. Our results should thus be regarded as a lower bound of the prevalence of discrimination. One potential sets of concerns about the external validity of our results is related to the choice of names. Would we obtain different results if we had chosen a different set of names and, specifically, if we had excluded distinctly Black names? We address these important questions in two steps. First, we run simulations using 5000 different subsets of the data, each time estimating the race gap for a random sample of 50% of worker pairs (i.e. 25 pairs instead of 50). The results are shown in Figure B3, with dashed lines indicating the bottom and top 5% of estimates. We find that the 5th percentile of estimated coefficients is around 24%, and only 1 percent are below a race gap 21.6% (top left panel). 20Second, we reestimate our hiring results and exclude any worker pairs with at least one distinctly Black name. These are defined as having an average perception of being Black above 70% and include Shanice, Tyrone, Deshwanta, Deshawn, and Terell. Table B3 shows that estimates of the race gap remain almost identical (31 pp vs. 30 pp). This reflects the fact that the relationship between perceived race differences and hiring probabilities is linear (as shown in Figure 6) implying that results are robust to excluding pairs with large race differences. --- Mechanisms --- Motivation: What Explains Decisions that Contradict Beliefs? The results from Section 5 demonstrate that although productivity beliefs play an important role in hiring decisions, other beliefs about worker attributes, such as race, education and noncognitive skills, also strongly impact employer choices. In fact, in roughly 30% of hiring decisions employers select the worker that they previously predicted to be less productive. This may seem surprising given that the hiring task explicitly instructed and incentivized employers to choose the most productive candidate. However, it is in line with a large literature (e.g., Fehr and Rangel (2011)) showing that humans are prone to making errors when faced with simple binary decisions, particularly when placed under time pressure. One potential explanation for this phenomenon is linked to the process of forming beliefs by retrieving memories about people with the same name. Employers may initially retrieve salient attributes such as race, which are perceived to be correlated with productivity. 21 As this process continues, less salient but more relevant information will be retrieved and beliefs may drift away from this initial race-anchored impression and towards their true productivity belief (which may still be correlated with race beliefs). This process relates to dual-process models of cognition, which posit that one set of processes (labeled System 1 by Stanovich and West (2000)) is fast, reflexive and unavailable for conscious introspection, while the other (System 2) is slow, analytical and able to process information sequentially (Epstein, 1994;Sloman, 1996;Kahneman and Frederick, 2002). Applied to our study, racial discrimination can either be instinctive and subconscious, or deliberate and self-aware. This cognitive process is both noisy and effortful, and employers may terminate it when the cost of searching for additional evidence exceeds the potential benefits of improving belief accuracy. Due to the inherent noise in this process, employers may make different predictions about people with the same name at different times, leading to inconsistencies between their productivity beliefs and hiring choices, particularly when placed under time pressure. To explore this prediction, we investigate how the probability of being hired varies based on employers' race and productivity beliefs, as well as the randomly assigned decision time. While we formalize these ideas in Section 6.2, Figure 8 presents a visual representation of our findings. Based on previously elicited employer beliefs, we divide candidate pairs into three race difference groups and four productivity difference groups. As expected, when two workers are predicted to be roughly equally productive and equally likely to be Black, the probability of being hired is close to 50%. This is true for both long (left panel) and short (right panel) decision times. 22Examining long decisions with workers of a similar race (the middle line in the panel on the left), we observe that the hiring probability gradually increases as a candidate's productivity advantage grows, up to 75% for an advantage exceeding 30 pp. Workers who are perceived as more white receive an additional advantage in the hiring process, which increases from 10 pp to 15 pp as the perceived the productivity advantage grows. Conversely, workers with more Black-sounding names face a hiring penalty of 5 to 10 pp. For decisions under time pressure, the benefits of (perceived) productivity advantages remain largely unchanged for more white-sounding names (Figure 8, right panel). By contrast, Figure 8: Hiring Probabilities, by Productivity / Race Perceptions and Decision Time Notes: "More Black" includes workers perceived to be more than 20pp more likely to be Black; "More white" includes all worker who are perceived to be more than 20pp less likely to be Black; and "Similar race" includes workers whose difference in perceived likelihood of being Black is 20pp or less in absolute value. there is a notable decrease in the productivity-slope for workers with more Black-sounding names, leading to an almost 50% increase in the hiring race gap for Black workers with large productivity advantages. In other words, Black workers who are perceived to be highly productive are the most adversely affected by rushed hiring decisions. These findings are in line with the framework presented below, which posits that employers may initially anchor their productivity beliefs to easily retrievable race associations. When they are forced to terminate their search for worker attributes prematurely due to time pressure, productivity beliefs have less influence on hiring decisions, leading to a larger race gap for highly productive workers. In such scenarios, the "error" rate, defined as the inconsistency between hiring choices and stated productivity beliefs, increases from 10 pp for white workers to 43 pp for Black workers. --- Framework Employers make a simple two-choice hiring decision in our experiment. While economists have focused mainly on observed choice outcomes, cognitive and mathematical psychologists have made significant progress in understanding the joint distribution of choices and response times by modelling the neurological mechanisms that underpin such decisions. Neurophysiological data have provided support for a class of models that posit that individuals make these decisions by sequentially sampling noisy information, allowing them to accumulate evidence favoring one choice over the other (Sewell and Smith, 2016). The most successful of these models is arguably the drift-diffusion model (DDM), which offers a highly parsimonious representation of the cognitive process underlying binary decisionmaking (Ratcliff, 1978;Ratcliff and McKoon, 2008). According to the DDM, decision-makers accumulate evidences regarding the relative merit of choosing option a over b through sequential sampling of relevant information stored in memory, such as past consumption decisions and associated pleasure levels, to predict the hedonic impact of each option denoted as v(a) and v(b) (Fehr and Rangel, 2011). These cognitive computations are subject to noise caused by the stochastic nature of neuron firing rates. The relative decision value at time t can then be expressed as z t = z t-1 + <unk>(v(a) -v(b)) + e t (6 ) where <unk> is a constant drift rate, and e t represents the noise in the mental computations, which is assumed to be identically and independently distributed according to a standard Gaussian distribution23. When the decision value z t surpasses a positive threshold B, option a is selected, while crossing a negative threshold -B leads to the choice of option b. This process entails a sequential likelihood ratio test, which has been demonstrated to be the optimal statistical solution to the decision problem (Bogacz et al., 2006;Gold et al., 2007). The model also distinguishes between non-decision time, which is required for stimulus encoding and motor response generation (Ratcliff and McKoon, 2008), and the decision time, which involves the accumulation of evidence. The drift rate <unk> determines the speed at which evidence is accumulated and can be interpreted as the quality of the information available for evaluating the choices. When <unk> has higher values, decisions are reached more rapidly and consistently. The decision thresholds, B and -B, represent the required level of confidence for making a decision. Due to the inherent noise in the evidence accumulation process, different realizations of the same process with identical drift rates may not terminate at the same time or at the same threshold. However, the decision-maker can influence the quality and speed of the decision by adjusting the value of B. Lower values of B correspond to quicker decisions that are more prone to errors, whereas higher values of B lead to slower but more accurate decisions. This illustrates the speed-accuracy trade-off inherent in this model. Experimental manipulations have consistently confirmed these model predictions (Ratcliff and McKoon, 2008), but it has not been tested in the context of hiring decisions.24 Figure 9 illustrates the application of the DDM to our experiment. Employers are presented with candidate names from which they infer that candidate a is white, while candidate b is Black. According to the DDM, employers initiate a cognitive process where they sequentially sample information from their memory, emotions, beliefs and expectations related to these names. This information is utilized to compute decision values for each candidate, which are then combined to generate a relative decision value z t following equation 6. We elicit these types of beliefs from the employers prior to the hiring task in a incentive-compatible way and without imposing any time constraints. 25Based on the descriptive results indicating that both race and productivity beliefs affect the decision and decision time, we specify the drift process as follows: <unk>(v(a) -v(b)) = <unk> 1 (r a -r b ) + <unk> 2 (y a -y b ). <unk> 1 and <unk> 2 measure the relative importance of race and productivity during the process of gathering evidence for the hiring decision. Krajbich (2022) proposes an extension to this model that incorporates heuristic and subconscious racial biases by allowing race to influence the starting value of the evidence gathering process. 26 We specify this as <unk> = <unk> 1 (r a -r b ). This leads to the following estimable equation for the hiring decision: z t = <unk> 1 (r a -r b ) + (<unk> 1 (r a -r b ) + <unk> 2 (y a -y b ))t + E t (7) where E t = t <unk> =1 e t The next section will present reduced form analyses that test the model's predictions regarding the impact of race and productivity beliefs on the hiring choice and decision time. We then estimate a structural model to jointly estimate key parameters of this model including decision thresholds, initial bias, as well as productivity and race drift. --- Reduced Form Analysis To examine the predictions of the DDM, we begin by analyzing how employers' beliefs impact their decision-making time. According to our model, response times will be shorter when one worker is perceived to be more productive or more likely to be white, and especially when both of these perceptions align. Conversely, if the more productive worker is perceived to have a lower probability of being white, the race and productivity drift will have opposing influences on the decision value, resulting in slower decision-making. Table 5 provides evidence in support of these predictions. A greater (absolute) difference in race beliefs is associated with faster decisions (Col. 1). Choosing between one candidate perceived as Black and one perceived as white shortens decision-making time by 0.19 seconds, equivalent to 0.18 sd, compared to choosing between workers of the same race. While controlling for the (absolute) difference in perceived productivity between candidates reduces this estimate to 0.15 seconds, it remains statistically significant at the one percent level (Col. 2). Additionally, larger differences in perceived productivity also shorten decisions times (Col. 2). These correlations are driven by decisions with unlimited time, which are less affected by noise (Col. 3-4). Overall, these findings confirm our model assumption that both race and productivity beliefs are important determinants of the drift. 26 Biased starting values capture an immediate bias in favor of white workers that will be particularly influential in short decisions, but that dissipate over longer decisions. This may allow us to distinguish between racial biases that emanate from heuristics (System 1) and more analytical cognitive evaluations (System 2). However, since the estimated starting values are extrapolations of the linear drift function to decision times of zero, which is unobserved in the data, the validity of this interpretation hinges crucially on the assumption of constant drift rates. If System 1 and System 2 processes are more accurately depicted as a gradual shift in what employers attend to, then this process may induce non-constant drift rates which deviate from the DDM assumptions, and which would invalidate the interpretation of starting value estimates as the effect of heuristic biases on employment decisions. One way to test the validity of this assumption is to estimate the DDM separately for hiring decisions made under the randomized short and long decision treatments, and to test whether the drift rates are equal. These tests are performed in Section 6.4 below. Notes: The dependent variable in the time people take to make a hiring decision (winsorized at the 5% level). "Prod (Dif)" measures to the difference in productivity between worker 1 and 2 scaled to be between 0 and 1. "Black (Dif)" measures the difference in the share associated as Black between worker 1 and 2 scaled to be between 0 and 1. Specifications in Columns 1-4 use the absolute values of these differences. The interaction term uses the perceived productivity and race of worker 1 minus worker 2. All estimations are OLS. Robust standard errors clustered at individual level are in parentheses. * p <unk> 0.10, * * p <unk> 0.05, * * * p <unk> 0.01 Next, we investigate the interaction of beliefs. While perceptions of productivity and being Black are negatively correlated, there are many instances where a candidate is perceived as more productive and more likely to be Black. To explore these scenarios, we estimate specifications with an interaction term of differences in perception of being Black and productivity between candidates, without converting belief differences into absolute values. A positive interaction term thus indicates that a candidate is perceived as more productive and more likely to be Black. As predicted by the DDM, the coefficient on this interaction term is positive when decision-makers have unlimited time (Col. 6). The DDM also offers predictions regarding the outcomes of hiring decisions and the effect of time pressure. In particular, it suggests that rushed decisions can exacerbate the race gap in hiring, as employers may place greater emphasis on race during the initial stages of the decision-making process. Confirming this prediction, Column 1 of Table 6 shows that, in the overall sample, the race gap widens by 6.7 pp, or 24.2%, when employers are required to make fast decisions. Kahneman and Frederick (2002) conceptualize System 2 as a analytical, slow-thinking process that acts as a supervisor that can "endorse, correct, or override" the quick, heuristic judgments of System 1. In the context of our experiment, we hypothesize that System 2 will only override these System 1 judgments under two conditions: First, if System 1 exhibits bias against Black workers, and second, if these biases are not aligned with the analytical Notes: The dependent variable in a binary variable for choosing worker 1. Independent variables are coded to show the difference in perception of worker 1 minus that of worker 2. "Short" is an indicator variable for the randomly assigned fast decision time. Col. 5-6 and 7-8 divide the sample according to whether respondent are in support of race-based Affirmative Action, measured by whether they rate their support above 50 on a 0-100 scale. All estimations are OLS. Robust standard errors clustered at individual level are in parentheses. * p <unk> 0.10, * * p <unk> 0.05, * * * p <unk> 0.01 evaluation of System 2. To test this hypothesis, we compare the impact of time pressure 27 on the race gap across three distinct groups: Black employers, white employers in support of race-based Affirmative Action (AA) and white employers more opposed to AA. 28 We focus on support for AA for two reasons. Firstly, it is the single most important predictor of the race gap in hiring (Figure 7) and was specified in our PAP. Secondly, while opposition to race-based AA can stem from various factors, studies indicate that racial resentment is one 27 To test whether time pressure increases reliance on System 1, we divide our sample based on participants' performance in the cognitive reflection test (CRT) (using Frederick's (2005) three-question survey), which measures people's inclination towards analytical versus heuristic decision-making (Frederick, 2005). In line with predictions from dual process models, we find that the increase in discrimination in short decisions is driven by employers scoring above the median CRT who tend to rely more on analytical decision-making in the absence of time pressure (Table B7, Col. 7). Table B7, Col. 5 shows a similar pattern for productivity beliefs. While perceiving a candidate to be 10 pp more productive increases the chances of hiring that person by 7.8 pp when employers have unlimited time, this effect drops to 6 pp when employers are under time pressure (Table 6,Col. 2). This decrease is also driven by employers who score high on the CRT. These findings are in line with other studies showing that those with higher levels of cognitive reflection are less prone to heuristic biases (Toplak et al., 2011;Hoppe and Kusterer, 2011;Abel et al., 2021). 28 We split the sample by whether they rate their support for race-based Affirmative Action above 50 on a 0-100 scale. Figure A8 shows the effect of time for each decile. of the key predictors of such opposition (Feldman and Huddy, 2005;Mangum and Block Jr, 2022). Support for AA can thus act as a proxy for whether people deem it desirable to override the biases of System 1.29 Our findings indicate that rushed decision-making does not impact the race gap in hiring among Black employers. This suggests that this group does not exhibit a strong negative heuristic response towards candidates with Black names (Table 6,. 30 Time pressure also does not have a discernible effect on the race gap among white employers opposed to AA, who exhibit high levels of discrimination across decision times (Col. 7). Conversely, for white employers who support AA, rushed decisions increase the race gap by 70% from 20 pp to 33 pp (Col. 5). Controlling for productivity beliefs further increases this difference (Col. 6,8). In sum, white employers who support AA make decisions that align more closely with those of Black employers when given sufficient time. However, when put under time pressure, their decisions align more with those of white employers who do not support AA. These findings suggest that both System 1 and System 2 processes play distinct roles in influencing the race gap in hiring and that their relative importance varies across employers. To further explore this heterogeneity, we present the hiring results across different subgroups of employers and decision time in Table 7. Panel A shows that among white employers opposed to AA, both race and productivity beliefs have similar effects on decisions in short and long time frames. The race penalty of around 40% decreases by 7 pp (17%) when controlling for productivity beliefs (Col. 2, 5) and an additional 8 to 12 pp when we control for other beliefs (Col. 3,6). White employers supporting AA exhibit a similar pattern for short decisions: the large race gap of 32.5% closes by similar magnitudes when controlling for productivity and other beliefs (Panel B,. However, when these employers have unlimited time to make decisions, their pattern diverges: the race gap becomes 40% smaller compared to short decisions (Col. 1), and the race penalty decreases by nearly 50% when controlling for productivity beliefs (Col. 2) and becomes indistinguishable from zero when controlling for other beliefs (Col. 3). Notably, the magnitude of productivity coefficients is approximately 80% larger for this group compared to employers opposing AA, suggesting that productivity beliefs play a more important role when sufficient time is available. In Panel C, we find that the hiring pattern for Black employers closely resembles that of white employers supporting AA when making decisions without time pressure (Col. 1-3). However, changes in coefficients are smaller when Black employers face time constraints (Col. 4-6). --- Structural Estimation of Drift-Diffusion Model Given the reduced-form analysis' support for the prediction of the the DDM, we proceed to estimate the parameters of the structural model using our experimental data. We start by estimating a single DDM model on the observed response times and hiring decisions pooling data across both short and long decision frames. This model assumes that only the decision thresholds are affected by the time constraint, whereas the drift rates and biased starting values are constant. (Details of the estimation strategy can be found in Appendix A.1.) Estimates for the pooled employer sample reveal a large and significant effect of race beliefs on the decision drift, indicating a strong bias against hiring workers with Black-sounding names that accumulate with decision time (Table 8, Panel A, Column 1). As expected, productivity beliefs also affect the decisions through a strong positive drift. In line with DDM predictions, the information threshold of 1.42 for decisions not subjected to time constraints is reduced by 0.42 (30%) when employers are under time pressure. 31 We also estimate that employers require about 0.43 seconds on non-decision time for our task. Estimating the DDM separately by employer subgroups reveals substantial heterogeneity in how beliefs influence decision drift rates. For Black employers and white employers supporting AA, productivity drift parameters are six to seven times larger than the Black drift parameters (Panel A, Col. 3-4), indicating that decisions are mainly determined by their productivity beliefs. By contrast, the productivity drift is only twice the size of the Black drift for White employers opposed to AA, suggesting that they are more influenced by race beliefs. We also find that race beliefs do not have a significant impact on the starting values in the pooled sample or for any of the subgroups. We next relax the assumption of constant drift rates across experimental treatments by estimating the DDM separately for short and long decisions. Results reported in Panel B and C of Table 8 show that productivity drift parameters remain relatively constant across decisions times and employer groups. By contrast, the race drift parameter is more than three times larger in the short compared to long decision time in the pooled sample (Col. 1). This confirms the hypothesis that employers focus more on race early in the hiring decision and then gradually shift their attention to other worker attributes that are less salient but more relevant for hiring decisions. We again find large heterogeneity in this pattern across employers. Going from short to long decisions, the ratio of productivity to race drift changes from 2.2 to 12 for white employers supporting AA and from 2.5 to 6.8 for Black employers. By contrast, this ratio of only increases from 1.4 to 2.3 for White employers opposing AA. These estimates help explain the large variation of the effect of time pressure across employers (Table 6). 32 Figure A9 visualizes the changing role of the race drift across decision time and employers by plotting how expected relative decision values evolve. We can use these structural estimates for counterfactual analyses and calculate hiring probabilities for candidate pairs with different combinations of race and productivity differences. Figure 10 plots the hiring probabilities calculated separately across decision times and employer subgroups. Differences in productivity beliefs are plotted on the x-axis, hiring probabilities on the y-axis, and differences in beliefs that the worker is Black (normalized to be non-negative) are presented by different lines. In the pooled sample (top panel), lines have a steeper slope for long than short decisions, reflecting that productivity beliefs have a larger effect on hiring probabilities in decisions without time pressure. The bottom three panels visualize heterogeneity in responses to time pressure across employers. For Black employers and those favoring AA, the productivity profile becomes much steeper for long 31 A threshold of 1.42 should be interpreted as 1.42 standard deviations of the accumulated noise of mental computations after a second of decision time. 32 We also find that the starting point is significantly biased

economists, whose research aims to provide answers to the global labor market challenges of our time. Our key objective is to build bridges between academic research, policymakers and society. IZA Discussion Papers often represent preliminary work and are circulated to encourage discussion. Citation of such a paper should account for its provisional character. A revised version may be available directly from the author.

time pressure across employers (Table 6). 32 Figure A9 visualizes the changing role of the race drift across decision time and employers by plotting how expected relative decision values evolve. We can use these structural estimates for counterfactual analyses and calculate hiring probabilities for candidate pairs with different combinations of race and productivity differences. Figure 10 plots the hiring probabilities calculated separately across decision times and employer subgroups. Differences in productivity beliefs are plotted on the x-axis, hiring probabilities on the y-axis, and differences in beliefs that the worker is Black (normalized to be non-negative) are presented by different lines. In the pooled sample (top panel), lines have a steeper slope for long than short decisions, reflecting that productivity beliefs have a larger effect on hiring probabilities in decisions without time pressure. The bottom three panels visualize heterogeneity in responses to time pressure across employers. For Black employers and those favoring AA, the productivity profile becomes much steeper for long 31 A threshold of 1.42 should be interpreted as 1.42 standard deviations of the accumulated noise of mental computations after a second of decision time. 32 We also find that the starting point is significantly biased against Black workers for some employer subgroups in the long decision frame and that this bias appears to favour Black workers in the short decision frame for all employer subgroups. The fact that our estimates do not support the assumption of constant drift rates across experimental treatments combined with a lack of observations near the intercept means that we should guard against interpreting the starting point estimates as heuristic biases. decisions, capturing that these employers are more likely to override their heuristic bias and hire more productive (Black) workers when given time. --- Figure 10: Simulated Hiring probabilities The vertical difference between the highest and lowest lines in the graphs indicates the hiring penalty experienced by Black workers at varying levels of productivity differences. Notably, this penalty is larger for short than long decisions and peaks for Black workers with moderate to large productivity advantages (with a maximum around a 25 pp advantage for both decision times). 33 These estimates also provide insights into the necessary (perceived) productivity advantage required by a Black candidate to have an equal chance of being hired, as indicated by the horizontal difference between races at the 50% hiring probability threshold. In the pooled sample, this premium amounts to 20.3 pp for long and 39.0 pp for short decisions. For white employers who oppose AA these figures increase to 52.8 for long and 44.2 pp for short decisions. For white employers favoring AA the productivity premium is sizable at 34.1 pp for fast decisions but decreases to 10.1 pp for long decisions. Black employers require a more moderate productivity premium of 25.8 pp for short decisions, which decreases to 13.0 pp for fast decisions. The results in this subsection confirm that the DDM provides a useful framework for understanding the observed relationship between response times and hiring decisions, and helps explain seemingly contradictory responses to productivity beliefs and hiring decisions. Although variations in hiring decisions across short and long experimental treatments can not be completely accounted for in a single DDM with constant drift rates, the structural estimates from separately estimated models indicate that race beliefs have a strong effect on short hiring decisions, but becomes much less important when hiring decisions are longer. We also document important heterogeneity across employer subgroups in how attention to race and productivity beliefs shifts between short and long decisions. The results suggest that while all employer subgroups exhibit some degree of racial discrimination in their hiring decisions, the cognitive processes driving these decisions may vary significantly. --- Alternative Mechanisms --- Uncertainty and Risk Aversion An alternative explanation for the race gap in hiring is that firms may be more hesitant to hire Black applicants due to uncertainty regarding their productivity levels (Lazear, 1998). 34 That is, in addition to differences in expected productivity, differences in the variance of productivity beliefs may also influence decision-making (Aigner and Cain, 1977). 35 To test this hypothesis, we ask employers to rate their certainty about picking the more productive worker in each choice after they completed the ten hiring decisions. 36 In line with the DDM's prediction that people facing time pressure reduce the information threshold, we find that employers in the short time frame report being 3.2 pp (0.2 sd) less certain about their choices. If employers are indeed less certain about productivity of Black workers, race differences between candidates would also reduce certainty. However, our findings suggest the opposite. Perceiving one candidate as Black and one as white increases levels of certainty by 5.3 pp (0.3 sd) (Table A6, Col. 1). Even after controlling for productivity and other beliefs, we observe a similar pattern where perceived race differences increase levels of certainty by 0.2 34 Race differences in the variance of productivity beliefs can create spurious evidence from audit studies (Heckman and Siegelman, 1993;Neumark, 2012). 35 One theory proposed in the literature for why firms may be more uncertain about Black candidates is that the majority of hiring managers are white, and people tend to be more effective at evaluating applicants from their own race group (Cornell and Welch, 1996). 36 We ask: "You hired NAME over NAME. What do you think is the probability that you picked the more productive person? " to 0.3 s.d.. This suggests that race differences reduce uncertainty in the hiring decision, which is consistent with findings from the DDM that employers use race beliefs to reach an information threshold. We conduct subgroup analysis levels of risk aversion to further test whether uncertainty is a driver of the race gap in hiring. Risk-averse employers may be more hesitant to hire candidates with unfamiliar names, which could disproportionately harm Black candidates. Contrary to this hypothesis, we find that the race gap in hiring does not widen among more risk-averse managers (Table B4,Col. 8). This result provides further evidence against uncertainty as a driver of the race gap in hiring.37 --- Concerns about Worker Payouts Last, we explore the variation in whether workers obtain a chance of winning a large bonus for being hired. This offers a clean test of how much employers care about worker welfare in our setting as it does not affect their own payouts. The directional effect on the race gap is unclear ex-ante. If employers prioritize the well-being of white workers over Black workers, it may increase the race gap. On the other hand, if employers care about equity or are averse to inequality, it may reduce the race gap. We find that the bonus does not change the race gap in hiring (Table A7, Col. 1). One explanation is that the bonus has heterogeneous effects that offset each other. Some employers may prioritize the welfare of white workers, while others may be concerned about reducing inequality and thus favor Black candidates. To investigate heterogeneity of the bonus effect, we analyze the effect on different subgroups of employers. Interestingly, we find that the effect of the bonus does not vary depending on employers' support for race-based affirmative action, political views, or race (Col. 2-7). We interpret this as evidence that concerns about worker welfare, including explanations related to taste-based discrimination, are less relevant in our setting. This may be unsurprising since workers and employers do not interact in our study context. However, preference-based discrimination may still factor into belief formation and help explain why employers hold incorrect beliefs about the productivity of Black workers. --- Conclusion We explore the influence of names on hiring decisions, drawing on data from both actual workers and an incentivized hiring experiment in a nationally representative sample. Our analysis reveals racial disparities in beliefs based on workers' names and that these beliefs contribute to a race gap in hiring. These results shed light on one of the fundamental challenges in studying the effect of names, namely, that names are not randomly assigned. People hold many other associations that may be correlated with race and it is unclear to what extent these racial disparities reflect discrimination. We find that other beliefs explain some of the variation in the hiring gap, but that much remains unexplained, especially in rushed decisions. In addition, most of the race penalty remains when we look only at the variation in beliefs for the same name and when we drop the most distinctly Black names. These results address the criticism that discrimination detected through audit studies mainly reflect associations other than race or are driven by the choice of certain names. However, our findings also highlights the importance of measuring (probabilistic) race beliefs to correctly interpret differences in employer behavior measured by audit studies. An important question is how much name-based discrimination documented by our and previous studies matter for situations outside of experimental settings, especially as other studies using large administrative data sets find that Black names are not correlated with educational (Kreisman and Smith, 2023) and other life outcomes (Fryer Jr and Levitt, 2004) after controlling for other factors, leading Fryer Jr and Levitt (2004) to conclude that "carrying a Black name is primarily the consequence rather than a cause of poverty". One explanation for this disparity in findings is that once people have more information, e.g. through frequent interactions in the classroom, they are less likely to judge others by their name (Kreisman and Smith, 2023). Yet, there are many situations where name-based discrimination can prevent people from collecting additional information about a person as they may decide not to interview a candidate, accept as a renter, or offer credit (Barto<unk> et al., 2016). Esponda et al. (2020) show how biased beliefs reduces people's attentiveness to new information and prevents correcting beliefs. Learning requires that people are "willing and able to adjust their behavior" in response to information (Esponda et al., 2020). This can refer to new information and feedback or it may refer to thinking harder about a decision. Our findings show that perceiving someone as Black substantially shortens the time people dedicate to considering other qualities of that candidate. Biased beliefs thus prevents learning from ones own memory. Being "willing and able" to learn also requires that one considers other characteristics than race that may be less salient and harder to retrieve when spending more time on a decision. We find substantial heterogeneity in what beliefs people use in their decision. For rushed decisions, race beliefs are a key determinant of their decisions across all (white) employers. When given unlimited time, some employers put more weight on their productivity beliefs while others are guided by the same beliefs as in rushed decisions. In our data, we identify support for AA as a predictor of who is considering other beliefs than race. Support for AA is also the single biggest predictor in who is holding biased race beliefs, which is consistent with Esponda et al. (2020)'s conclusion that inattention can prevent people from correcting beliefs. One widely proposed strategy to reduce discrimination is to force people to slow down decision-making and thus reduce the role of heuristic biases associated with System 1 (Krajbich, 2022). Our results suggest that for some employers, this strategy is highly effective in reducing discrimination. However, our results highlight two reasons that limit the effectiveness of these policies: employers may be unwilling to consider other factors than race in their decisions and they may hold biased beliefs about these other characteristics. In our sample, these two tendencies are correlated, which makes it particularly challenging to reduce discriminatory behavior for these employers. Echoing recent papers documenting substantial variation in discriminatory behavior across employers (Kline and Walters, 2021), policies need to address the underlying source of this heterogeneity to be effective. f (t|v, c, w) = <unk> c exp -vcw - v 2 t 2 <unk> <unk> n=1 k exp - k 2 <unk> 2 t 2c 2 sin(k<unk>w) (8) This is a defective probability, in as far as it integrates to the conditional probability of crossing the lower threshold, rather than one. The probability density for the upper boundary can be obtained by setting v equal to -v and w equal to 1 -w. Calculating these density values is complicated by the presence of an infinite sum which needs to be approximated, as discussed in (Ratcliff and Tuerlinckx, 2002). However, recent contributions (Blurton et al., 2017;Navarro and Fuss, 2009;Gondan et al., 2014) have improved the efficiency and speed with which these calculations can be performed. We use the fddm package in R (Foster, 2023) to calculate the density values and the partial derivatives of the individual densities with respect to the model parameters. Our implementation of this model allows both race and productivity beliefs to determine the drift v = <unk> 1 (r a -r b ) + <unk> 2 (y a -y b ). In addition, the starting value is allowed to depend on the difference in race beliefs across workers: <unk> = <unk> 1 (r a -r b ). This requires adding another parameter to the model, so that the parameter vector is effectively (<unk> 1, <unk> 2, B, <unk> 1, t N ). This does not add any complexity to the estimation process beyond the application of the chain rule to calculate the relevant partial derivatives. We start the estimation process by dropping all decisions with missing choices, or productivity or race predictions. This is represents about 9.8% of the total sample. As is typical with such models, we also omit a small share of observations with implausibly quick or slow response times. Specifically, we drop 394 decisions with response times shorter than 0.5 seconds or longer than 20 seconds. These decisions comprise 2.3% of the sample, which is consistent with the the "2% to 3% of responses" (Ratcliff and McKoon, 2008, p. 885) that is common for such models. The maximum likelihood estimates are obtained by using a sequence of numerical optimization routines to identify the parameter values that maximize the likelihood value. This happens in three steps. First, we use the Bayesian global optimization routine implemented in R through the mlrMBO package (Bischl et al., n.d.). The optimization is "warm-started" by pre-evaluating the model over a 200 initial parameter values drawn using a space-filling Latin Hypercube design. These estimates are then used in a sequential optimization procedure across 100 additional iterations with Kriging models as surrogate learners and the expected improvement infill criterion. The terminal point of this global optimizer serves as the starting point for our second numerical optimization step, which uses the Nelder-Mead method (Nelder and Mead, 1965) on both the calculated likelihood and gradient values. This method was found to be more efficient and robust in approaching the same terminal point across different starting values than other methods we experimented with. Finally, we use the Berndt-Hall-Hall-Hausman algorithm (Berndt et al., 1974) with starting values determined by the previous estimation step. Both the individual log-likelihood vector as well as the gradient matrix are used, which allow us to cluster standard errors by employer using. These methods are implemented in the maxLik package (Henningsen and Toomet, 2011). --- A.2 Tables --- Figure A6: Geographic Variation in Race Gap Notes: The map shows how the race gap varies geographically. We estimate our main specification separately for each state. We limit the analysis to states for which we observe at least 100 hiring decisions. Affirmative action is measuring level of support from 0-100 scale coded as Con (0-33), Mid (34-66) and Pro (67-100). Racial diversity refers to zip code level racial fractionalization, divided into terciles. Hiring experience refers to whether the respondent has real-life experience in hiring. --- A Appendix --- A.1 Estimation In the DDM a decision is reached when the relative decision value, z t, crosses either of two thresholds, B or -B. In the most common version of this model, this process evolves with constant drift v, and is subject to random perturbations, e t : z t = z t-1 + v + e t. This process is sometimes called a Wiener diffusion process, due to the common normality assumption of the error terms, e t <unk> N ormal(0, <unk> 2 ). We also can allow this process to have a nonzero starting point, <unk>. The total response time to cross a decision threshold, t, consists of both decision time t D and non-decision time, t N. This process produces pairs of observable decision outcomes and response times that are random variables, whose distributions depend on four model parameters (v, B, <unk>, t N ). It is often more convenient to reparameterize this model so that c = 2B and w = <unk>+B 2B and to work with the parameter vector (v, c, w, t N ). The probability density of the response time for decisions that cross the lower threshold can be expressed as (Feller, 1968):

economists, whose research aims to provide answers to the global labor market challenges of our time. Our key objective is to build bridges between academic research, policymakers and society. IZA Discussion Papers often represent preliminary work and are circulated to encourage discussion. Citation of such a paper should account for its provisional character. A revised version may be available directly from the author.

within local government to support their ability to create and synthesis the evidence needed for preventative and public health interventions. Most research systems exist outside of the local authority and are based within health, community and academic partnerships [1]. In England, local government has become a key constituent for addressing health inequalities and influencing the health of individuals and communities [2]. While this has been lauded as an effective approach to tackle the multiple determinants of health, there are concerns that generating and utilising research evidence to inform decision-making and action is a challenge [3,4]. This situation is not isolated to England and international reviews have shown various ways in which local government access and acquire evidence for decision-making -one review suggesting six models and approaches between local government and research systems [1]. Indeed, evidence-informed decision-making is complicated and involves integrating the best available research evidence with contextual factors including community preferences, local issues, political preferences and public health resources [5]. With this backdrop, this paper reports research which sought to understand the capacity to collaborate, deliver and utilise research across one metropolitan district council. The research explored current assets within local government in relation to research development and evidence implementation and how these could be further harnessed. Moreover, the research identified limitations and shortcomings which prevented research use and activity from flourishing. The paper draws out implications more widely for local government and how to reconfigure the relationship between research, evidence and decision-making in public health. The transfer of public health functions in England from the National Health Service (NHS) to local government in 2013 aimed to bring about improvements to population level health and to reduce health inequalities. While the delivery of public health can vary in local authorities [6], this reorganisation saw a change in culture from a narrow focus on health care pathways to one of a politically led environment with opportunity to influence the wider determinants of health and wellbeing. As part of this, Health and Wellbeing strategies are a vehicle for local governments to act on the wider determinants of health and wellbeing and provide an opportunity to adopt an evidence-based approach to local decision making and prioritisation of limited resources across local government. Nonetheless, the use of evidence and published research within these strategies is not common practice. Analysis of Health and Wellbeing strategies by Beenstock et al. [7] identified that only five out of 47 Health and Wellbeing strategies referred to published research evidence and only three cited National Institute for Clinical Excellence (NICE) guidance. Barriers to the use of research and evidence to guide decision making include the questioning of the credibility of the evidence [3,4] and the transferability of evidence that is out of context and not generated in the local setting [3]. Studies exploring how evidence in local authority public health practice is used have highlighted the disconnect in understanding between policy makers and academics, especially in regard to what constitutes robust and useful knowledge [8]. Indeed, locally generated data are viewed by decision makers as fitting the political context, having more transferability, and thus having a bigger influence on their local decision making [9]. In addition, a systematic scoping review exploring the use of evidence in local public health decision making concluded that researchers need to develop a deeper understanding of evidence requirements from the perspective of decisionmakers [10]. The Local Government Association (LGA) [11] further reiterated the value of research in local government settings. The LGA recently highlighted that 'local government needs practical research providing solutions that can be applied in real world situations. Councils can benefit from engaging in research partnerships' (p.8). The report suggests the need for increased capacity and development of the local authority research system. Other reports have also signalled the importance of taking a population level, non-clinical and transdisciplinary approach to public health interventions and research [12]. How that vision translates into practice and work 'on the ground' is relatively under-explored and understood. So, while the rhetoric is strong, it is clear that there are significant challenges based on the "daily rush to support frontline delivery of services with a lack of resources" (p.8) in local government. This means that time, expertise, and space to use or generate research is a struggle [13]. In the UK the National Institute of Health Research (NIHR) funds health and social care research that aims to improve people's health and wellbeing. The NIHR recognised the position that local government can have to improve population health and set out a funding call (in April 2020) to identify how local authorities could be developed into locally based research systems and to shape future investment. The research presented here was conducted following a successful application to the local authority research system funding call. The research was based in one local authority in the north of England where qualitative methodology was employed, operationalised through interviews, focus groups, meeting observations and documentary review. This paper focuses specifically on interviews and focus groups with a range of local authority personnel (described in more detail shortly) to enable greater understanding of the capacity of the local authority to collaborate and deliver research. --- Methods The research was undertaken between August-November 2020. The overarching aim was to explore the current research assets in the local authority and to determine how these could be nurtured and replicated within the organisation to foster a stronger research culture. In addition, the research sought to identify any perceived barriers that exist to the local authority working with academic partners. In particular, to establish research capacity and opportunities, and explore with key members of the organisation how a sustainable research system could be developed to impact on local resident's health, reduce health inequalities and identify the most important research outcomes. The theoretical underpinning of the research was the Research Capacity Development Framework [14]. The study adopted a collaborative approach throughout from the funding bid development to outputs and dissemination. A project steering group was established which included representation from: the local authority at strategic, operational and political levels, neighbouring local authorities who had also received NIHR funding, local academic intuitions, NHS research infrastructure support networks and the local NHS hospital trust. This steering group supported with the study design, recruitment and data analysis and, following the study, knowledge transfer and dissemination. The study was chaired by an elected member. In the UK an elected member is chosen to represent their local area and inform and influence the decisions and running of the local authority. Elected members may have key responsibility for different portfolios such as health, children's services, planning and transport. Data collection was undertaken by an Embedded Researcher (ER) who was based within the local authority for the study period. The ER model is becoming increasingly highlighted as allowing a joined-up approach to creating and using knowledge by placing a researcher in a non-academic organisation to better link research and practice [15]. The decision to use an ER in this study was so that it could potentially provide greater depth and insight within an organisation through having a researcher integrated within the culture and environment. However, this was compromised during the Covid-19 pandemic and the ER became digitally, rather than physically, embedded in the local authority. As part of the ER process, a co-applicant of the study facilitated access for the ER to attend to attend online team meetings at the operational and strategic level with various departments across the local authority in order to meet employees, develop a rapport with teams and raise awareness of the study. This included attending team meetings and formal committees. The ER was introduced to strategic directors by another co-applicant of the study who was also a member of the Local Authority leadership team. Prior to the study starting the strategic leadership were informed and supportive of the study, this helped with rapport building in preparation of the interviews. While the research team conceded that the original intention was for the ER to be co-located in situ with staff in the local authority, there was still methodological learning and value from a digitally ER working within the organisation. This is reflected upon later. --- Setting The research focused on a single local authority in the north of England. The area is one of the largest Metropolitan Districts in the country and is one of the largest cities in the UK, without its own university, with levels of educational attainment below average. The area is in the top twenty-percent of the most deprived districts in England and on average, people die younger than in other parts of England. Cardiovascular, cancer and respiratory illnesses are in high levels in the district resulting in people becoming ill at a younger age, having to live with their illnesses longer compared to most of the rest of the country. --- Sample Purposive sampling was used for both identifying individuals for the interviews and focus groups. The sampling was conducted with support of the project steering group in which a discussion was had to identify the key strategic roles and groups from across the authority that would need to be included. The steering group also identified groups of people who were research active (involved in delivering or commissioning research or who held a research related qualification), roles within public health where research was considered to be used in practice on a regular basis, and elected members who had responsibilities for different portfolios across the local authority. Participants were recruited via email invitation. All participants were provided with a briefing paper, written by members of the project team and co-applicants employed by the local authority, and a participant information sheet, prior to data collection to ensure informed consent was gained. Consultation with the study steering group informed the sampling of three focus groups which were conducted with: Focus Group 1 -Elected Members (n = 3), Focus Group 2 -Public Health Officers (n = 6) and Focus Group 3 -Officers with research interests across the local authority (n = 4). Interviews (n = 7) were conducted by the ER with Corporate Directors and Service Managers purposively sampled to enable the research questions to be explored fully. All data collection was undertaken online using Microsoft Teams due to social distancing restrictions of the Covid-19 pandemic. All aspects of the study received ethical approval from both Leeds Beckett University and Sheffield Hallam University and access permissions were gathered from the local authority via the strategic leadership team. Interviews and focus groups were conducted in parallel due to the short time frame in which to conduct the research and lasted between 30 and 60 minutes. Interviews and focus groups explored a range of issues which were informed through the Research Capacity Development (RCD) framework developed by Cooke [14,16]. The RCD works at individual, organisational and systems levels, with a purpose to develop research that is useful and impactful to society [17,18]. Assessing both the assets and potential for RCD of an organisation can help articulate what a partner may bring to a collaboration and can be considered an important aspect of winwin research partnerships. The RCD framework has been applied in a range of contexts and in developing of organisational research strategy [16]. Using the principles of the RCD framework [18] the interview schedules and focus group guides covered: linkages and partnership; skills and confidence in the workforce and wider community; infrastructure of the council and wider partnerships, research use and dissemination, experience and assets of coproduction in projects (including citizen and public engagement in projects); and ownership, leadership and sustainability of research activity (both by Officers and Elected Members). --- Data analysis Interview and focus group recordings were transcribed by an external transcription company, anonymised and shared as a secure online file which was accessible by three members of the research team. All transcripts were coded on NVivo 12 by the ER and two members of the research team cross checked a sample for coding accuracy. Data were analysed using framework analysis [19]. Framework analysis was used as an expedite method given the short timescale for the project funding and was deductively informed following the RCD framework [14]. Specific elements of the RCD framework were used in the development of the matrices -a core aspect of framework approach -this seemed pragmatic in deductively analysing the data set given the RCD framework was used to inform the data collection tools (as discussed earlier). Given the limited timeframe set by the funder for the research delivery, the data was analysed sequentially with interview analysis being completed first followed by focus groups. This was based on pragmatics, but also was beneficial in refining analytical categories and themes during the process and supported the triangulation of the two sets of data. Inductive coding and inductive thematic development was also part of the analytical process to enable specific 'local' issues within the local authority to be represented. --- Results The analysis revealed a range of thematic areas relating to the focus of the research. This section presents these to highlight the barriers and potential in local authorities for improved, research-led, decision making to address health inequalities. --- Barriers Respondents identified challenges to improving researchled decision making to address health inequalities. --- The political and cultural context Respondents described a duality in the use of research and evidence within local authority decision making, and how the essential, political, nature of a local authority led to unavoidable tensions: --- "Sometimes politics and research meet in a way that's positive and constructive, and sometimes it collides, and sometimes research and objective factual information is inevitably used politically or influenced by politics. " (Interviewee 2). Pressures arising from the four yearly election cycle were acknowledged. As election time draws closer, Elected Members may begin to look to research for insights into, or solutions for, complex problems, such as health inequalities, but the time required to complete the research process and a need for prompt answers is incompatible. The political landscape may have moved on before research can provide answers, or political priorities changed. Many respondents highlighted the challenge in balancing the need to undertake robust research and the need to complete it quickly, with a tension between 'academic rigour and the political need to get things done' (Focus Group 2 Participant 10). The constraints of the four yearly election cycle also meant that where research and evidence was used to inform decision making, it may be focused on popular, short-term solutions and 'immediate response' (Interviewee 4) rather than engaging with the root causes of health inequalities and a longer-term view. The political leaders within the local authority were also felt to be more reluctant to deal with complex problems, such as health inequalities, as they may be viewed as 'a signal that something isn't working' (Focus Group 2 Participant 4) rather than as an opportunity for identifying potential solutions. Respondents felt this could lead to decisions being made because they would be popular with voters, but that these decisions were made quickly and without establishing what the most appropriate course of action may be: "There's a lot going on that you have to do, bang, bang, bang. It's a bit like that political, you know, this is what we have to do, and we have to do it now. But when do we actually have time to step back and ascertain whether we've done the right thing and what have we learnt from it?" (Focus Group 2 Participant 5). Respondents also described how it was politically expedient to be seen to be operationally focused and pragmatic with a strong focus on day to day delivery of services. Research could, therefore, be something which was a distraction from 'business as usual' and be a less attractive option for the use of resources: "But historically I think there's a view that research is not doing. So, we've become a council that is overly focused on action rather than consideration and careful development of those actions. So, across the organisation I would say it's kind of frowned upon as being a little bit academic and a little bit of nondelivery. " (Interviewee 6). The expectation for the local authority to be seen to be focused on delivery also led to constraints on those who had taken on formal training or qualifications, such as an MSc or PhD. On their return to the workforce they are fully committed back into delivery and had little opportunity to use their newly acquired skills: --- "I think there's a lot of people within the [Named] department who are doing their Masters or they have done their Masters, but then it's incorporating that into the everyday job. And I think sometimes you just revert back to the day job rather than what you've actually learnt through doing that programme. " (Focus Group 3 Participant 8). Barriers arising in the wider political landscape were also identified by respondents. The impacts of austerity and the financial restrictions within which a local authority must operate were widely acknowledged. Time and resource for developing or using research skills and capabilities were limited: --- "But again, it's about how you actually make that happen in terms of resourcing because as the workforce has shrunk, we have less flexibility to enable that to happen without then having to backfill posts. " (Interviewee 7). Respondents also suggested that the policy and practice of the wider research system was felt to be set up to support academic and NHS organisations conduct research, rather than local authorities: "So, I'm caveating I suppose that I think academic researchers go through [Professional Network] nationally to then reach individual local authorities. What we don't do, and there isn't a system for, is us saying individually or collectively as local authorities here's an area that we think would benefit from some research and some research expertise, could we collectively put that out to see whether we might find an appropriate research partner to work with us on this? So, it's a one-way system. " (Interviewee 3). This potential lack of dialogue could then leave those within the local authority feeling that researchers collected data from the organisation or community and then 'disappear with it for a couple for years' (Focus Group 3 Participant 9) without useful outputs coming back into the organisation. Furthermore, the language used by academics and researchers was not always helpful, or useful, and the perceived 'elitist world' (Focus Group 3 Participant 6) of academic research was not considered accessible to the delivery focused local authority. --- Lack of organisational strategy Local authorities, as with any organisation, have flux in terms of leadership and strategic direction. Variability in the leadership around research-led decision making presented a number of challenges to tackling health inequalities. Respondents explained that where individual Officers within a service had a personal or professional background or interest in the use of research, then a research-led response to health inequalities may develop. But the use of research was not yet an overarching strategic vision of the organisation. This 'patchy and sporadic' (Interviewee 3) approach to research was problematic. Even where there was a growing interest in the use of research-led decision making amongst practitioners, senior management may not share this position. As senior managers control the service budgets and resources this could then preclude any further action being taken: --- "It's also then about getting buy-in from the highest level, because what's the point in even trying to look at solutions for a problem if you don't have buy-in from senior management?" (Focus Group 2 Participant 1). The lack of a co-ordinated, organisation wide approach to research-led decision making was seen to lead to a culture of research as'somebody else's responsibility' (Focus Group 1 Participant 2), with services within the local authority providing policy and intelligence functions seen as responsible for providing relevant updates and insights, rather than research-led decision-making being embedded within the organisation. --- Facilitators Respondents identified several opportunities for research led decision making within the local authority. --- Recognition of the value of evidence Respondents described the growing support for research already present within the organisation, and the recognition that the tighter financial constraints required more careful targeting of limited resources for the greatest returns. Research was seen as: "Spending a little more up front to make sure your finances are focused in the right area. " (Interviewee 6). In addition, there was a willingness across senior leadership to engage with the culture shift required to take on board the insights available from research, with a growing interest in 'a bit more thinking about how we could deliver it in practice' (Interviewee 1). External research findings were felt to bring the additional advantage of being both instructive for changes to policy and practice while remaining uninfluenced by the possible biases present within the local authority: --- "The advantage would be purely that independence, because I know very much, I'm sure, I'm definitely guilty of it, and I'm probably sure other people are, quite often we maybe have a solution in mind before we even start. So, we're trying to do research that will fit our solution. So, you've got that inbuilt bias in the research that you're doing, so how you ask the questions, who you ask them to, what the content is, you're almost trying to fit the solution that you've got in mind; whereas somebody completely external is probably starting more with a blank piece of paper and is just supplying the evidence that leads you then to a potential solution. " (Focus Group 2 Participant 5). For a local authority, with the requirement for public consultation and feedback, 'evidence' inherently incorporates the 'local voice' (Focus Group 3 Participant 6). The value of intelligence generated locally was in the immediate geographical or cultural relevance which fed more easily into any decision making process. As such, respondents reflected the value of co-production to inform decision making around health inequalities. Listening to the voices of the community, and understanding that the use of these insights could result in better service provision and a more efficient use of resources, was driving the focus on evidence based decision making higher up the agenda within the organisation: --- "I think there's also something about leaders understanding what the national agenda and national conversation is around that and engaging with people with lived experience and the value that that can bring to an organisation. " (Focus Group 3 Participant 3). While acknowledging the cultural differences between the local authority and academics, respondents highlighted the opportunity to drive the use of research when addressing health inequalities by tailoring research findings to the needs of both the Elected Members and Officers separately: "I think it would be to managers, to me, that it will aid decision making. That if you've got the right information, it's much easier to make decisions on policies. And for politicians as well, the way forward it would be, to me, about helping make decisions. " (Focus Group 1 Participant 7). Overall, respondents were clear that the challenges of using research in a political organisation were not insurmountable and any research into health inequalities that could 'bring that strategic and operational-ness together' (Focus Group 2 Participant 5) would be well received. --- Building existing networks Though a divide between the culture and practice of academic, NHS, and local authority organisations was described by respondents, it was also clear that this divide was already being bridged and with further work (on both sides) could be mitigated further. The perception of a divide was manifested in a belief that local authority employees simply did not do research. But some respondents suggested that this was not wholly the case: --- "And that was a comment that came back from one of my Service Managers was no we haven't done any academic research as such. And I said that wasn't the question that I asked...certainly I had to prompt them to sort of say actually you have done a lot of research and you've used that research to put options and recommendations to Elected Members to inform their decisions. " (Interviewee 4). The national move towards greater integration between local authorities and NHS organisations was also described by respondents as a facilitator of the move towards greater use of research: "So, I think we're working on it and we're trying... because there are two very, very different cultures. --- So, it's about understanding each other's worlds and how we can come together and what we could share, what research we can share that's applicable to both of us. " (Focus Group 1 Participant 2). Potential to improve research-led decision making was also felt to lie in the networks around the local authority. Membership of professional networks provided exposure to new ways of working and allowed for the dissemination of research findings: "I would to some extent try and find out that myself by attending some public lectures at places like the [University], who get a lot of guest speakers in from the Office for National Statistics and the like, to talk about some of the cutting edge stuff that they're doing. " (Focus Group 3 Participant 9). Respondents also identified how relationships and networks need to be built with the local voluntary and community sector groups, not just with professional or academic networks, for improved decision making. These groups understood the context and lives of the communities the local authority served and could therefore provide greater insight to help target resources more effectively. --- Championing a research infrastructure A champion for research at a senior level was felt to be an important actor to facilitate the growing momentum within the local authority for research and evidence-led decision making. A senior leader would be able to identify where challenges remain in addressing health inequalities, and how to develop services, often as a result of their own background, interest, or simply by 'being curious' (Interviewee 7). Senior leaders could potentially make decisions to fund and support more research. In addition they need to manage the tension between the timescales within which the local authority operates with the timescales of a research process which seeks to create new intelligence. Greater understanding and tolerance of any delays could ultimately lead to the organisation being better informed and able to make more effective decisions about action: "It's understanding the timescales, and it's sometimes you may be asked to look at a problem and they're expecting a solution very, very quickly, whereas for quality research it's going to take a prolonged period of time. Obviously within local authority we tend to work in four-year cycles really, if that, coming towards elections and things like that. So, it's understanding that things don't happen overnight and that if you want quality information, quality data, it's going to take time to collect before the solutions can even be dreamt up. " (Focus Group 2 Participant 1). The economic constraints within which the local authority must operate are unlikely to shift, and limitations on the formal, funded, routes to developing research skills within the workforce are likely to remain. However, respondents identified the informal pathways within the organisation, such as mentoring or secondment, that were available. These could be 'used more effectively as an organisation' (Interviewee 6), and, while acknowledging the impacts on resourcing, would bring the benefits to the organisation: "So likewise, again if a member of my team said do you know what I'd love to spend a day a week with an academic institution researching this, as long as we can make it work in terms of, you know, the pressures that we have, work pressures, then I would really support that. " (Interviewee 5). --- Discussion This paper sought to understand how research evidence could be more effectively used to inform decision-making in a local authority, focusing particularly on what strengths and assets are currently embedded in the organisational make-up and to identify any potential areas for development. Uniquely, the research design was underpinned by an ER model which has high utility in gaining depth of information and recognising contextual and local factors -we argue that such an innovative methodological approach offers a new contribution to understanding the use of research and evidence in local government. While this ER was largely 'digitally integrated', there were particular benefits with adopting a model whereby rapport could be developed with individuals within the local authority to foster rich data gathering. This is discussed again later in this section. Influenced heavily by evidence-based medicine, evidence-based public health is a long-standing principle of great importance in research and practice. This principle has been amplified by the movement of Public Health into local authorities, with the increasing emphasis on 'economic rationalism' and the need to justify expenditure, and ensure that funds are deployed to maximum returns [20]. With the political dimension that local authorities hold, economic rationalism and evidencebased decision making is crucial to ensure democratic legitimacy but, to date, little research exploration has focused on this matter. If local authority personnel are to successfully implement change, then they must draw on the evidence base to aid and support decision-making [21] by elected officials. Indeed, this rhetoric was well understood in this study, and the practical challenges were also recognised by participants. This study showed the significant challenge for local authority practitioners and policy-makers using evidence to good effect. Some of these issues are unsurprising and have been noted elsewhere [7], it is perhaps axiomatic that busy practitioners working in local authority do not have the space or time to engage in research, evidence generation or assessment and this study reenforced that this situation has not necessarily changed over time. While this is understandable, it can be a fundamental shortcoming for effective evidence-based decision-making. There is also a strong ethical imperative to adopt the principles of evidence-based practice to ensure that health promotion and public health activity does no harm, either directly or indirectly, by wasting limited funds on ineffective or inappropriate interventions, or by raising unrealistic expectations about what might be achieved. Similar to the findings of the study reported in this paper, in a study by Li et al. [20,p.196], health promotion practitioners stressed the value of evidence for this reason. One participant in their study noted: 'I do firmly believe that we need some evidence before we launch into things. I think the prospect of doing harm is too great to not have some inkling of where it is going to go'. The context of public health within a local authority, a political domain, is also interesting for research and evidence utilisation. Lifestyle drift is the inclination for policy that recognises the need to act on upstream social determinants only to drift downstream to focus on individual lifestyle factors [22]. In a culture where lifestyle interventions are significantly easier to evaluate, and are facilitative of political cycles, it is understandable why more entrenched determinants of health, which takes years to address (i.e. poverty), are often ignored [23]. This strikes to the epicentre of the tension between academic rigour and expedient decision-making and was highlighted here as a common issue in local government. Public health is a very evidence-focused arena, and some have suggested that English local authorities are not a natural home for traditional evidence-based practice. Local government systems are political systems with key decisions needing locally elected officials' approval [6]. This has direct relevance to research leadership in local authority and having individuals who subscribe to research and evidence-based principles at the pinnacle of local authority structures. The research demonstrated that where this was in place, it fostered stronger commitments to research and evidence-based decision-making within teams and services. It has been suggested that training for practitioners in interpreting research evidence is a necessary competency to aid professional judgements [24]. Both Li et al. [20] and Owusu-Addo et al. [24] have demonstrated that practitioners in health promotion value evidence from researchers that is context-bound, and relates directly to their own practice, rather than evidence which is more abstract or out-of-context. This was shown in this study where decision-makers had a preference for contextspecific evidence. Yet, in reality this can be difficult, and extracting useful evidence from various contexts is critical and does require advanced skills and understanding. The research showed a strong appetite for individuals and groups within the local authority to improve their research skills, and moreover suggested viable ways to do that through training and qualifications and strong connections with academic organisations and institutions. The need for research competency and capacity in local authority is something that is commonly known both nationally and also internationally [25]. Owusu-Addo et al. [24] highlighted that training programmes which build and maintain common skill sets and language among local public health practitioners in Ghana was necessary to accomplish evidence-based public health goals. The literature highlights the benefits and challenges associated with utilising an ER approach to gather data [15]. Our experience was overwhelmingly positive, in terms of accessing rich and detailed data for analysis and interpretation. The ER approach drew on ethnographic principles, including interviews and observations, but was fundamentally premised on being responsive and agile to opportunities that were presented within the local authority. While the ER was 'digitally' embedded and not 'physically' embedded as a result of the pandemic, this did not pose significant disadvantage. Indeed, as discussed earlier, in some cases it facilitated expedient access to key personnel who may have otherwise not have been made available. There were, however, some limitations with the study: access and rapport building with employees at the local authority was limited through attendance at pre-arranged meetings and the methods of data collection with limited opportunity for informal conversations, such as those that take place in an office environment; the short-time frame set by funders to set up, deliver and report on the research meant the study team and ER had to focus on ensuring that data collection was prioritised with less time to establish the ER into wider teams across the local authority. The skill-set of the ER was crucial in being able to navigate both the local authority processes and also the academic collaborators making up the study team. Where challenges arose, they were mitigated by strong partnerships between the research team and the local authority staff (especially those acting as research collaborators) as well as the project steering group. This collective partnership between all constituents worked exceptionally well and enabled data gathering on barriers and facilitators to be conducted relatively smoothly. The ER approach offered the opportunity to gather insight from within the organisation that we are confident would not have been uncovered using other approaches to data gathering. --- Conclusions The study, utilising a unique ER approach, has explored and shed further insight into the decision-making processes and evidence-based decision-making in local government. Public health practice and practitioners are accustomed to the use of evidence-based decision making, yet this study showed how the democratic and organisational structure of local government challenges how effectively evidence is used in practice. Furthermore, increasing demands, limited capacity and resources impact on even the most research engaged practitioners' ability to do research. The research highlighted the criticality of research leadership to challenge the status quo in the process of policy development and decision making in local government and move it to one that uses evidence-based principles and prioritises the use and development of research undertaken within local government organisations. The ER model has high utility in gaining depth of information and recognising contextual and local factors which would support research capacity development in local government. Local government, place based collaborations, and academic institutions should explore and develop opportunities for ERs to bridge the organisational divides, in doing so developing trusted relationships, continued staff development and research capacity. --- Availability of data and materials The datasets generated and/or analysed during the current study are not publicly available to protect the anonymity of the participants within the local authority but are available from the corresponding author on reasonable request. --- Abbreviations --- Authors' contributions CH was the Principal Investigator for the study and led the writing of the project report and the publication. JW was a co-applicant, academic advisor for the study and co wrote the publication and project report. CF collected the data and was the embedded researcher on the study. JS was a co-applicant and senior academic advisor for the study. JC facilitated the focus groups and provided methodological input to the study. JH was a co-applicant and supported the dissemination of the findings into the locality. AH was a coapplicant and facilitated the research in the locality. SM was a co-applicant and facilitated the research in the locality. The author(s) read and approved the final manuscript. --- Declarations Ethics approval and consent to participate All methods were carried out in accordance with relevant guidelines and regulations. As the local authority had no formal ethics or governance procedures in place therefore ethics approval was granted from Leeds Beckett University 73115 and Sheffield Hallam University ER26394523 prior to data collection taken place. All participants provided informed consent prior to data collection taking place. --- Consent for publication Not Applicable. --- Competing interests The authors declare that they have no competing interests. • fast, convenient online submission • thorough peer review by experienced researchers in your field --- • rapid publication on acceptance • support for research data, including large and complex data types • gold Open Access which fosters wider collaboration and increased citations maximum visibility for your research: over 100M website views per year --- • At BMC, research is always in progress. --- Learn more biomedcentral.com/submissions Ready to submit your research Ready to submit your research? Choose BMC and benefit from:? Choose BMC and benefit from:

Background: Local government has become a key constituent for addressing health inequalities and influencing the health of individuals and communities in England. Lauded as an effective approach to tackle the multiple determinants of health, there are concerns that generating and utilising research evidence to inform decision-making and action is a challenge. This research was conducted in a local authority situated in the north of England and addressed the research question -'What is the capacity to collaborate and deliver research?' . The study explored the assets that exist to foster a stronger research culture, identified barriers and opportunities for developing research capacity, and how a sustainable research system could be developed to impact on local residents' health and reduce health inequalities.This was a qualitative study utilising semi-structured interviews and focus groups. The study used an embedded researcher (ER) who was digitally embedded within the local authority for four months to conduct the data collection. Senior Managers were purposively sampled from across the local authority to take part in interviews. Three focus groups included representation from across the local authority. Framework analysis was conducted to develop the themes which were informed by the Research Capacity Development framework. Results: Tensions between research led decision making and the political and cultural context of local government were identified as a barrier to developing research which addressed health inequalities. Research was not prioritised through an organisational strategy and was led sporadically by research active employees. A recognition across leaders that a culture shift to an organisation which used research evidence to develop policy and commission services was needed. Building relationships and infrastructure across local government, place-based collaborators and academic institutions was required. The embedded researcher approach is one method of developing these relationships. The study identifies the strengths and assets that are embedded in the organisational make-up and the potential areas for development.Research leadership is required in local government to create a culture of evidence-based principles and policy. The embedded research model has high utility in gaining depth of information and recognising contextual and local factors which would support research capacity development.

SM was a co-applicant and facilitated the research in the locality. The author(s) read and approved the final manuscript. --- Declarations Ethics approval and consent to participate All methods were carried out in accordance with relevant guidelines and regulations. As the local authority had no formal ethics or governance procedures in place therefore ethics approval was granted from Leeds Beckett University 73115 and Sheffield Hallam University ER26394523 prior to data collection taken place. All participants provided informed consent prior to data collection taking place. --- Consent for publication Not Applicable. --- Competing interests The authors declare that they have no competing interests. • fast, convenient online submission • thorough peer review by experienced researchers in your field --- • rapid publication on acceptance • support for research data, including large and complex data types • gold Open Access which fosters wider collaboration and increased citations maximum visibility for your research: over 100M website views per year --- • At BMC, research is always in progress. --- Learn more biomedcentral.com/submissions Ready to submit your research Ready to submit your research? Choose BMC and benefit from:? Choose BMC and benefit from: --- Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Background: Local government has become a key constituent for addressing health inequalities and influencing the health of individuals and communities in England. Lauded as an effective approach to tackle the multiple determinants of health, there are concerns that generating and utilising research evidence to inform decision-making and action is a challenge. This research was conducted in a local authority situated in the north of England and addressed the research question -'What is the capacity to collaborate and deliver research?' . The study explored the assets that exist to foster a stronger research culture, identified barriers and opportunities for developing research capacity, and how a sustainable research system could be developed to impact on local residents' health and reduce health inequalities.This was a qualitative study utilising semi-structured interviews and focus groups. The study used an embedded researcher (ER) who was digitally embedded within the local authority for four months to conduct the data collection. Senior Managers were purposively sampled from across the local authority to take part in interviews. Three focus groups included representation from across the local authority. Framework analysis was conducted to develop the themes which were informed by the Research Capacity Development framework. Results: Tensions between research led decision making and the political and cultural context of local government were identified as a barrier to developing research which addressed health inequalities. Research was not prioritised through an organisational strategy and was led sporadically by research active employees. A recognition across leaders that a culture shift to an organisation which used research evidence to develop policy and commission services was needed. Building relationships and infrastructure across local government, place-based collaborators and academic institutions was required. The embedded researcher approach is one method of developing these relationships. The study identifies the strengths and assets that are embedded in the organisational make-up and the potential areas for development.Research leadership is required in local government to create a culture of evidence-based principles and policy. The embedded research model has high utility in gaining depth of information and recognising contextual and local factors which would support research capacity development.

The language used to discuss aging has been criticized for its "poverty" (Marshall, 2022;Small, 2007, p. 3), reflecting and reinforcing instrumentalist and reductive approaches to aging that do not convey the nuances and lived experiences of a complex process lived in time (Baars, 2014). We argue that the same could be said for language used to discuss intergenerational relationships. The term has become ubiquitous across academic, policy, and media discourses, and is often used as a shorthand to describe different generational cohorts. Finding no adequate synonym, we use the term ourselves, but we seek to broaden the vocabulary and deepen understandings of what we mean by intergenerational relationships. We do so using the concept of lived ambivalence (Baars, 2014) and a creative qualitative methodology of analyzing participants' responses to novels. We argue that novels that depict characters whose experiences and relationships are portrayed in a narrative form, over time, can be a useful way of opening up discourses of lived ambivalence. Rather than dichotomous discourses that imply a "cross-sectional" approach to intergenerational relationships, in which each generation is fixed in time, fiction can facilitate more nuanced, processual, human, and fundamentally more meaningful understandings of how intergenerational relationships are lived in time and what they mean to people. --- Intergenerational Dichotomies We argue that there are two main discourses of intergenerational relationships, both of which reinforce a sense of rigid generational boundaries, instrumentalism, and reductivism, and neither of which are helpful in facilitating a more nuanced understanding of relationships between people of different ages. The first discourse can be understood as "generationalism," a mindset that "instigates artificial confrontations between the 'generations'" (Purhonen, 2015, p. 102) in a simplified and exaggerated way, perpetuating a narrative of generational injustice and ongoing crisis that divides generations (White, 2013, p. 217). This discourse relies on a conflict/ solidarity binary that frames intergenerational relationships as sites of antagonistic, zero-sum battles between different generational cohorts, most currently typified as "boomers" and "millennials." It is often found in the media and social media, where "the assumption of an economic and political conflict between the two generations has become an established trope" (Bristow, 2021, p. 766). These conflict narratives often invoke the future, with calls for "intergenerational justice" (Tremmel, 2006) to address what are framed as future threats caused to today's children and younger people by today's "older people." In response to this, there have been calls for a greater "intergenerational solidarity" that puts generational divisions aside and comes together for the sake of a more broadly imagined humanity (UN Secretary-General, 2013). The conflict/solidarity discourse is related to a second dichotomous discourse that identifies "intergenerational segregation" as a problem that needs to be addressed through increased intergenerational integration (Kingman, 2016). Examples of this include interventions and calls for policies intended to facilitate greater intergenerational interaction in housing (Hoolachan & McKee, 2018;Kingman, 2016), educational settings (Boström & Schmidt-Hertha, 2017;George et al., 2011), workplaces (Lagacé et al., 2019), and communities in general (Zhong et al., 2020). A desire for greater intergenerational solidarity and integration is understandable, particularly as intergenerational relationships have been centered on discourses concerning several recent political and environmental issues (Bristow, 2021;Elliott, 2021;Rios et al., 2021). However, seeking solidarity in intergenerational relationships can be limiting as it requires a consensus that minimizes the diversity of different experiences (Bengtson & Oyama, 2010;Connidis & McMullin, 2002) and risks ignoring how age intersects with other social divisions such as class, gender, race, and ethnicity (Elliott, 2021;White, 2013). An uncritical emphasis on solidarity can paper over the tensions and contradictions inherent in personal and social relationships (Davies, 2021) and render the term "intergenerational relationships" anodyne. To gain a more meaningful understanding of why and how intergenerational relationships matter to us, we must go beyond dichotomous discourses that mask the messiness of lived relationships, to find a way to expand the discourses of how people of different ages relate to each other. We argue that the concept of ambivalence can be helpful in this regard. --- Ambivalence and Intergenerational Relationships Ambivalence has been used in aging studies to cut through simplistic dichotomies, for example, in understanding conflicting feelings toward the future in people living with dementia (Thuessen & Graff, 2022). It has also been used in research on intergenerational relationships as a "bridging concept between social structure and individual action" (Connidis & McMullin, 2002, p. 559), that also has a "dynamic, transformative and temporal dimension" (Hillcoat-Nallétamby & Phillips, 2011, p. 214). As such, ambivalence allows for differences of opinion and reflects the way personal relationships can often be "sticky" and difficult to "shake free from... at an emotional level" (Smart, 2007, p. 45). Even when conflict is present around topics where values may differ (e.g., environmental concerns, care, policies, etc.) relationships still survive (Davies, 2021). Defining ambivalence as the coexistence of mixed or contradictory emotions, attitudes, or ideas about the same person, object, or situation by an individual subject, it can offer "a mature step toward acknowledging a more complex world of multiple perspectives and emotional resilience" (Biggs, 2007, p. 706). Lüscher (2011) conceptualizes "ambivalence" as a useful counterpoint to a desire for solidarity which, while appealing, can imply an idealized and static worldview that can ignore function and process (p. 194). The concept of ambivalence allows us to get closer to how intergenerational relationships are lived within kinship and social networks. The significance of aging as a process and of how relationships are lived coheres with calls from scholars of aging studies to draw on approaches from the humanities to better understand these "deeply humane processes and experiences" (Baars, 2014, p. 46). Baars argues that our experiences of living and aging through time cannot be reduced to the "either-or dichotomies" (p. 45) he identifies as being prevalent in academic and public discourses on aging, and so advocates for a "lived ambivalence" that can cut through these. One way to better understand the complexities of aging is through cultural and literary texts. Free from the conventions of policy-making and scientific research that often reduce human experience to empirical data (p. 45), fiction is better equipped to explore the ambiguities, contradictions, and richness of the relational and processual nature of aging. There is a substantial body of work by humanities scholars who draw on fictional texts to better understand aging as a process and experience (see e.g., Small, 2007;Woodward, 2019) demonstrating how aging is a lifelong process informed by cultural, social, and material forces (Kriebernegg et al., 2014). Fiction has also been used by health care professionals in geriatric medicine to communicate and explore the emotional aspects of aging with patients (Roitto & Rognstad Mellingsaeter, 2019). Sociologists have used fiction as an alternative source of knowledge to provoke new perspectives on the social world and as a "spark for theoretical rumination" (Beer, 2016, p. 410). Within literary studies, Felski (2019) has argued that identification with fictional characters can provide access to shared experience without negating difference and enhance a reader's sense of self in society. Alongside character, narrative emotion encourages readers to become empathetic co-creators of fictional worlds by "respond [ing] to the techniques of storytelling with curiosity, suspense and surprise" (Keen, 2015, p. 152). A growing body of work in the social sciences and literary studies demonstrates the value of using literary criticism as an empirical device to investigate social issues and ambivalences (Suckert, 2021) and to elicit lay reflections on cultural narratives of aging, addressing a perceived hierarchy and disconnect between professional and "ordinary" readers (Swinnen, 2023). The relationship between novelists and aging scholars also works the other way; the British author Margaret Drabble acknowledges that her novel The Dark Flood Rises was partly inspired by Helen Small's The Long Life (Drabble, 2017, p. 326). There is less of a tradition of using literary texts to understand the diversity and complexity of intergenerational relationships in gerontology, yet there is scope to use fictional representations of intergenerational relationships to relate and understand their social impacts. Literary texts have been used to facilitate and study intergenerational relationships, be that exploring the nature of play and connectivity (Deszcz-Tryhubczak & Kalla, 2021) and highlighting the mutual learning benefits that emerge in intergenerational reading groups (Lohman et al., 2003). Intergenerational reading groups offer a fruitful ground for analyzing complex attitudes to relationships between people of different ages, as together readers make meaning both individually and as a collaborative group, drawing on experiences across life spans. --- Methodology The findings presented here are part of an Economic and Social Research Council-funded project, Reimagining the Future in Older Age, which aimed to examine the relationship between older age and future time. The project received ethical approval from the University of Stirling. We undertook a qualitative study that brought together adults of different ages to analyze and reflect on dominant narratives of aging and intergenerational relationships by discussing novels that depicted these themes. We chose novels as a literary form because of their accessibility to a range of readers who might differ in confidence, and because we judged their typical length and narrative structure conducive to portraying aging and relationships as processes, in keeping with the project's focus on temporality. The project team drew up a long list of novels, which participants were then invited to choose from and add to. Only one novel was selected solely by participants. The novels were primarily chosen because they depicted themes of aging, time, and intergenerational relationships that were central to the project. We purposively sought novels from a range of genres and eras in order to represent different intergenerational relationships-from the historical to the contemporary and from the realist to the speculative-to give readers a range of imagined societies, life spans, and fictional world-building scenarios to respond to, sparking imaginative conversations about readers' own experiences. For more information on the novels and the intergenerational relationships foregrounded, please see Supplementary Material. Reading groups and book clubs have been found to encourage rapport and relational links, particularly in an intergenerational context where adults of all ages can develop a "mutually advantageous" understanding of one another's situation (Lohman et al., 2003, p. 104). Using literature as a creative method provided a way to engage with participants' lived experiences as they compared their social worlds with those in the novels. Specific genres, such as speculative fiction, are driven by a goal that returns us "to our own [world] enlivened with deeper understanding and insight... and fresh perspectives on how to build our age world in meaningful and just ways" (Woodward, 2019, p. 367). As such, fiction can be used as a provocation to prompt thoughts, ideas, and feelings on real-life topics as well as a springboard to imagine different solutions and scenarios in the real world. Although originally planned as face-to-face meetings, due to coronavirus disease 2019 (COVID-19), the reading groups were facilitated online by M. Lovatt and V. Wright, and this digital element allowed us to recruit participants from across central Scotland. Participants were recruited via social media and the project website and through snowball sampling. In an attempt to recruit a diverse sample of participants across different social characteristics, we also sent targeted invitations to a range of community and third-sector organizations representing minority groups in Scotland. Despite this, the majority of participants were white women-a bias that reflects the typical composition of reading groups (Tepper, 2000). Reflecting on our recruitment strategy in meetings with our project advisory group, we wondered if we might have been more successful had we been able to make in-person visits to organizations representing minority groups in order to explain the project in more detail, build relationships and elicit suggestions for novels; however, these opportunities were denied us due to social distancing regulations in place at the time. In the end, 28 participants were recruited to four reading groups, although not everyone attended all of the discussions due to a range of reasons including changing commitments and personal circumstances related to the pandemic. We tried to ensure age diversity in each group (see Table 1). The groups met monthly from June to October 2020. Each group was made up of adults of different ages to allow for reflections on the novels from a range of perspectives. The reading group sessions lasted approximately 90 min and usually began with the participants and researchers checking in with each other, providing a type of informal emotional support network as we shared how we were feeling under lockdown conditions and as things began to open again. We then invited participants to give their initial reflections on the novel, followed by a semistructured conversation based on prompting questions circulated in advance. The questions were based on M. Lovatt's and V. Wright's readings of the novels and related to the research interests of aging, time, and intergenerational relationships. Although the questions were useful as a starting point, discussions often spiraled out to a range of related topics raised by participants, and they shared their own stories and experiences outside of the reading group. All discussions were video-and audio-recorded and transcribed verbatim. Participants were further invited to keep diaries in which they could record their thoughts on the novels outside of the discussions and situate the readings in the contexts of their everyday lives and any other reflections they thought were relevant. The data comprised transcripts from 19 reading groups and 43 diary entries (1 group only met four times but discussed five novels). In the first stage of analysis, V. Wright and M. Lovatt read the transcripts, uploaded them to NVivo 13, and conducted open coding, which resulted in several themes of relevance to the project. J. E. French then led a further round of analysis which focused on the theme of intergenerational relationships and which was guided by the question "How can fiction expand our understanding of intergenerational relationships?." As J. E. French was not present for the data collection, she supplemented her analysis of the transcripts with watching video footage of the groups and sharing interpretations of the data with the other authors. J. E. French refined the themes using a mind-mapping approach to create a thematic map of the data and to further search for relevant patterns using an inductive analysis approach. This flexible approach was taken due to the breadth and depth of the discussions, in order to capture as much richness of the conversations as possible. That the analysis was conducted by researchers who were present during data collection and a researcher who was not, increased our confidence that our claims to knowledge were justified, and yielded insights that we may not have otherwise had, in a manner similar to what James (2013) calls "the analytic imagination." Not all novels are featured in this paper, reflecting this paper's focus on intergenerational relationships rather than aging and time more broadly, which are covered in the other novels. --- Findings Our analysis identified three key ways participants' discussions opened up new understandings of intergenerational relationships: (1) a greater understanding of what intergenerational relationships are, who they involve and what they ought to be; (2) a more nuanced understanding of emotional connection between generations; and (3) a desire for more "authenticity" in fostering intergenerational interaction. --- A Greater Understanding of What Intergenerational Relationships Are While the novels featured relationships between characters of different ages, none of these were explicitly defined as "intergenerational relationships." The fictional representations opened up lines of communication with the "generational other" (Biggs, 2007, p. 708) and the narration of complex characters entangled in relationships in a range of scenarios perhaps helped participants to talk about relationships in ways that transcended dichotomies. This was revealed in the ambivalence that participants expressed about their own relationships with people of different ages (particularly in the context of care), how they felt intergenerational relationships might change with increased life expectancies, and the relationships between ancestors and descendants. Intergenerational caring relationships within families were discussed in relation to The Last Children of Tokyo, a dystopian speculative fiction novel depicting the relationship between 108-year-old Yoshiro, who has an increased vitality and life expectancy, and his ailing great-grandson, Mumei, who will die young because of a variety of environmental and social disasters. Participants reflected on the extended life expectancy of the older characters in the novel and the implications of this for the nature of intergenerational relationships. Jen (40, G1) said "I think it... challenges us to think about what we mean by grandparents, great-grandparents... I think sometimes when you use the words like grandparents and great-grandparents we all have an image in our mind of what we assume that person to be, that age to be, but there's such a variety in society nowadays." Participants felt that the greater involvement of great-grandparents in caring for younger relatives was not unlikely, with Kitty (70, G4) suggesting: this might be the first generation where it's not unreasonable to assume that great-grandparents will have a caring role with their great-grandchildren... my grandchildren range from 18 months to 30 years old... I think for me the kinda realisation that great-grandparents now can be looking forward to actually having active roles... I hadn't thought about that before. When asked about what she felt about this extension of caring responsibilities, she expressed ambivalence, saying, "I have to be honest... deep down you want to help your children but there is a bit where you think 'I've already done this'... obviously there's lovely bits to it but physically I don't feel equipped really." She later added that while she was still working and active when her first grandchildren were born, "by the time the next tranche came along, the middle ones and the wee ones... I've been doing it just enough to really enjoy them, but I am exhausted with it, I really am." This shows the ambivalence of caring for (great)grandchildren but also the processual nature of the relationship as one that is lived through time as part of the aging process. Participants also expanded understandings of intergenerational relationships to include ancestors and descendants. Duine (71, G1) interpreted the decreased life expectancy of children in Last Children of Tokyo as "a message to society that when you live for today and to hell with consequences for tomorrow, it is building up untold harm for your children, grandchildren and following generations." Jamie (29, G3) echoed this, arguing that "the author has created this scenario as a kind of parable about each generation's responsibility over the generations to come. [It's] not as simple as living innocently and then being cared for in your old age but living with a radical responsibility." Frankie (60, G1) also felt that "maybe we should be looking back, it'd be interesting to look back seven generations and say, 'right that's what I need to be making amends for' or you know 'what happened seven generations back from me, what horrors were carried out' that might give a different perspective about it." The novel Sixteen Trees of the Somme uses the genre of a family saga to grapple with this exact question, as Edvard searches for meaning in the past 100 years of family history to decide the future of his farm. The novel prompted differing responses in our readers. Jamie (29, G3) reflected, "personally, I think that we owe more to future generations than our ancestors, because they are the ones who are going to inherit the world." Sanny (65, G3) added that: we have a responsibility to our ancestors as well as to future generations. Our ancestry is a major player in who we are. Values are passed down from generation to generation. It can sometimes take a considerable time before we realise this but that does not detract from it being a truth. This implies an awareness of change and continuity in intergenerational relationships, where it is possible to identify different generations while acknowledging how each is shaped by, and shapes, others. For participants, intergenerational relationships included familial, social, and public relationships happening now but also placed generations in dialogue with ancestors in the distant past and descendants in the near future. This perception of intergenerational relationships as one of mutual concern and responsibility occurring over several human lifetimes diverges from Scheffler's argument of a contemporary "temporal parochialism" (Scheffler, 2018). The participants' temporal framing of intergenerational relationships also runs contrary to the binary, antagonistic generationalism represented in contemporary dominant discourses, and is instead more akin to the longer-term, reciprocal relationships rooted in "environmental time" as articulated in Indigenous discussions of aging (Chazan & Whetung, 2021). Commitments to future generations, in particular, were evident in the emotional responses to the novels, which we discuss next. --- Ambivalent Emotions That Complicate the "Conflict/Solidarity" Binary Several novels prompted discussions of the environment, and participants of all ages acknowledged emotions of guilt, responsibility, and blame, relating them to intergenerational relationships. This is unsurprising considering that the climate crisis is typically framed as an issue of "intergenerational justice" (Diprose & Valentine, 2019). However, participants articulated these emotions in more complicated and ambivalent ways than attributing blame and victimhood to discrete generations in a zero-sum fashion. First, many of the older participants felt annoyed with what they perceived as being personally blamed for environmental crises. For example, Trouble with Lichen prompted discussions of the environmental impact of extended life expectancies. Christine (53, G3) felt she was being judged by her family on the topic: "I mean, even my daughter was saying 'oh it's you that's ruined everything' personally!... so already there's a bit of blame going on." Discussing Last Children of Tokyo Rebecca (48, G2) similarly expressed that her children were also blaming her for "ruinin[g] the planet." Although some of the older participants expressed some guilt and responsibility, they contextualized this within the socioeconomic, political, and technological practices and innovations experienced during their life course. Christine (53, G3) said "I'm far more aware of the climate emergency, I am more keen to do things... I utilise things, trying to stop using plastic... I don't think that [my] generation is responsible for all the plastic waste per se... but then that was the scientists at the time invented that and we all used it thinking that it was okay." Greta (70, G3) agreed with this, saying "I always have this argument with people when they start talking about plastic and I'll go "excuse me, not my generation... we get the sort of label, you're the previous generation so you must be responsible for all of that." Participants used the discourse of generations, but stressed the necessity to understand individual action within wider social practices. Second, participants of different ages invoked a temporal aspect to guilt; rather than seeing it as the preserve of particular age groups or generational cohorts, they projected themselves into the past or future, to consider what else they could have done-or could do-differently to take more responsibility. One example of this is a discussion about "generational guilt" in Group 3, which was prompted by the environmental dystopian setting of The Last Children of Tokyo. Clare (23, G3) began by saying "I think of when I'm older and there's younger generations, in my head I'm like 'they won't be resentful towards me, I've done my bit' but then... like, I am very much susceptible to causing climate change, but I do think that element of guilt will always be there." Responding to this, Jen (40, G1) said: when we were seeing all the climate change marches that were going on, cause they were kind of the next generation down from me and I feel really guilty that I didn't do enough when I was their age. When I was younger it was all about the... hole in the ozone layer... and stuff like that, and it kinda felt like that's the only thing we concentrated on and then once we'd solved that... we stopped doing anything else... So I think there's definitely a generational element to that guilt' Downloaded from https://academic.oup.com/gerontologist/article/63/10/1619/7145891 by Hochschule Luzern user on 19 February 2024 Kirsty (36,G1) agreed to some extent, although believed that: the generations that came before us that developed the technology to have everything disposable... did so with the right reasons... so I can see why there would be guilt but... it's up to us what we do with the technologies and the products that we have and I don't think we've made the right choices; I don't think that's my mum and my grandparents' fault. Discussions such as these were common across the groups. Elizabeth (60, G3) had a similar reaction to Jen and resisted the idea that "her generation" was ignorant of the environment: "in the circles I was in it was very much... about the environment, climate, Rachel Carson and Silent Spring... protests of all sorts." However, she also felt that "it's kinda like then we got to a certain level and everybody went 'ooh, we could have a holiday now, time for champagne and opera'." Jonathan (21, G2) thought that the characters in The Last Children of Tokyo were just resigned to a state of hopelessness [and...]it felt like there was a massive amount of guilt on the part of Yoshiro. Reflecting on the novel, he projected himself into the future, stating, "I'm going to be one of the older people so I'm going to be the one feeling guilty even though now I feel like I'm a younger person who gets to... be like 'well it wasn't my fault'... and I guess it just goes to prove a point that I don't think guilt is a very helpful reaction to it." A temporal dimension was also expressed in some participants' invocation of an understanding of generations in a broader sense, akin to that of the "ancestors and descendants" understanding discussed in the previous theme. Jess (30, G4) found The Last Children of Tokyo "shocking" for its clarity in describing a world in which a child might only know of animals through his great grandfather's memories. She said of Mumei: him drawing all of the animals and never getting the opportunity to see them, it just reallybrought home quite what we're doing to the planet and that that could be our future and I don't think we're quite ready in this generation to admit it. I don't think we've all, everyone that's alive now, has really accepted that climate change is a real thing. Here she implies a definition of generation as "everyone that's alive now" and in doing so, suggests a shared responsibility, regardless of age. This was also suggested by Jamie (29, G3) who argued "I think that if we are going to make any kind of advance on this then there needs to be a rallying together rather than throwing stones at each other." This can be interpreted as a desire for solidarity in the face of conflict, but we argue that the lengthy discussions devoted to generational guilt and other emotions complicate this and reveal underlying tensions that cannot necessarily be resolved in a desire to address environmental disaster. --- Authenticity and Ambivalence in Fostering Intergenerational Contact Participants welcomed relationships between people of different ages in the books and regarded intergenerational interaction as something to be valued and encouraged. However, they resisted what they saw as "contrived" attempts to bring people of different ages together. Instead, they called for "authentic" intergenerational engagement. This was prompted particularly when discussing The Summer Book and Young Art and Old Hector, both of which foreground the relationship between older and younger characters. Set in the Highlands of Scotland in the first half of the twentieth century, Young Art and Old Hector by Neil Gunn (1941) tells the story of 8-year-old Art and his older neighbor, Hector. Participants loved the depiction of the relationship for what they perceived as its warmth and empathy between people across age groups. Kitty (70, G4) praised the novel for showing "life through an older person's eyes and a young person, it was just that much clearer and I just felt like it just fitted somehow, you know, Hector's experience, as if he... really could connect with [Art],... it felt like he could see easily the issues there for the kid." Angela (39, G4) agreed: "I think it's just a lovely relationship, that both of them really valued." Participants across the groups reflected on the relationships in the novels in the context of their relationships with younger and older people in family and wider networks and spoke of the value of these experiences. Agnes (58, G4) recounted her own experiences of working with children and older adults in schools and the joy she took in this: I've got one photograph in particular of a lady who was in her eighties... talking to a child of... about eight or nine, and the look between the eyes of these individuals is a connection over all these years talking about school teachers, and to me that's the magic of it, it doesn't matter how old you are, you've got a shared experience. Although participants in all groups felt that these relationships were important for fostering empathy and learning across different stages of the life course, several participants felt that these interactions were rarer, for reasons including increased geographical mobility and a lack of time outside work and education. They called for new policies and practices that could encourage intergenerational relationships, focusing in particular on a desire for more "free time." However, participants also articulated an ambivalent desire for authentic relationships; they desired authentic community connection, free from political interference, at the same time as they articulated a need for new policies that would free up time and space to build such communities. Participants offered practical suggestions on how to use time more effectively. Kirsty (36, G1) felt having more time could encourage friendships across age groups and she wondered "if a younger retirement age might change that if there was a wider span of people enjoying the same activity together?." This type of shared activity across ages was a motif of The Summer Book, where many readers across different groups found the grandmother and Sophia's time together enviable, as they are free to stay curious and organize their own time on the island. The novel inspired them to imagine new possibilities for organizing time differently. Jonathan (21, G2) translated this desire for time into a potential policy solution: "we have the resources for, like, 3-or 3-day weeks for everyone and I think that would massively free up people to do what Sophia and her grandmother are doing which are, you know, exploring and playing and being creative [together]." Elizabeth (60, G3) suggested that a Universal Basic Income (UBI) would free people up to do more voluntary work or even change careers across their life course where "some might power on, some might want to change direction." Alongside desiring more time, a desire for more accessible "intergenerational" space was also important to participants. Responding to The Summer Book, Eleanor (62, G4) wrote in a diary entry: Do not keep elders and children apart, compartmentalising their experiences into age-appropriate ghettos, even if at times the relationship seems fraught or indifferent. Living and learning together improves our understanding of each other, builds stronger relationships, and teaches compassion. Participants were aware of existing projects that sought to bring different generations together, including intergenerational co-living spaces, nursery, and care home exchanges and integrated dementia villages. The space of the island in The Summer Book was seen to naturally encourage intergenerational connection, prompting Jonathan's (21, G2) idea to build a care home: "something that's kind of a lot more open to the community where people can just interact rather than kind of you only visit a home if you're visiting a relative," Participants imagined new spaces where intergenerational mixing could happen, with calls for community centers to be reopened and devolved to communities. Equally, participants said that they did not want these spaces to be "contrived" or "inauthentic." The relationships between characters such as the grandmother and Sophia or Art and Hector were praised for being "genuine." Reflecting on Young Art and Old Hector, Jane (69, G1) shared a story about a local man who was a stalwart in her community. Once he moved to a care home, he became more isolated and what would have been a natural, everyday occurrence of seeing his neighbors became reduced to visits that were "something you have to arrange." Katherine (76, G2) also craved more "real example[s] of directly interacting with the [older] generation" such as intergenerational house sharing, where people would be "not forced to but encouraged to mix and mingle." Angela (39, G4) cautioned against coercing participatory involvement, noting "you 'an't force someone to take part in something, you know, that takes away what 'ou're trying to do I think. Th're's only so much you can do, and it's about sometimes focusing on what groups are a success [...] Not everything's going to be a success and it hasn't been, but I think you've just got to try haven't you?" Participants hoped community would naturally occur and lead to active, intergenerational communities based around reciprocity, shared values, and sincere connections. --- Discussion and Conclusion In this paper, we have shown how fiction can be used to humanize and add nuance to the term "intergenerational relationships." Notably, the novels read by the groups did not explicitly mention intergenerational relationships and made no reference to cohorts such as "boomer" or "millennial." Instead, they presented readers with complex, emotional, and ambivalent examples of relationships between people of different ages that readers could relate to. The paper makes a number of key contributions to how intergenerational relationships can be engaged with more meaningfully, beyond simplistic and reductive dichotomies. First, it expands dominant definitions of intergenerational relationships beyond "fixed, cross-sectional cohorts" of people of different ages alive today, to include relationships that are lived and which transition over time. Second, it identifies greater complexity and ambivalence in the articulation of emotions relating to generations and generational identities relating to environmental crises than is depicted in "conflict/solidarity

Background and Objectives: The term "intergenerational relationships" is widely used in gerontological literature and age-related policies. However, discussions of the term often tell us surprisingly little about what it means or why it matters. We suggest that this is due to a reductivism and instrumentalism in 2 main discourses within which intergenerational relationships are usually discussed. First, intergenerational relationships are often conceptualized through a binary "conflict/solidarity" lens, reinforcing an entrenched "generationalism" . Second, they are predominantly constructed as a problem to be addressed within debates on how to tackle intergenerational segregation. Neither of these discourses provides much room for a more nuanced understanding of how intergenerational relationships are experienced or why they are meaningful. In this paper, we discuss how fictional narratives can introduce imagination and a richer vocabulary into discourses concerning how people of different ages relate to each other.We present findings from reading groups where adults discussed novels depicting themes of older age, intergenerational relationships, and time.In discussing the fictional narratives and characters, participants reflected on the significance and meaning of intergenerational relationships in ways that went beyond dichotomous and instrumentalist discourses. Drawing on the concept of lived ambivalence, we argue that fictional representations of intergenerational themes can elicit more meaningful reflections on the complexities and contradictions of relationships across age groups.We conclude that a more nuanced understanding of intergenerational interaction can inform gerontological discourses and policy, but also that gerontological awareness of social challenges concerning age relations can inform interpretations of fictional narratives.

you?" Participants hoped community would naturally occur and lead to active, intergenerational communities based around reciprocity, shared values, and sincere connections. --- Discussion and Conclusion In this paper, we have shown how fiction can be used to humanize and add nuance to the term "intergenerational relationships." Notably, the novels read by the groups did not explicitly mention intergenerational relationships and made no reference to cohorts such as "boomer" or "millennial." Instead, they presented readers with complex, emotional, and ambivalent examples of relationships between people of different ages that readers could relate to. The paper makes a number of key contributions to how intergenerational relationships can be engaged with more meaningfully, beyond simplistic and reductive dichotomies. First, it expands dominant definitions of intergenerational relationships beyond "fixed, cross-sectional cohorts" of people of different ages alive today, to include relationships that are lived and which transition over time. Second, it identifies greater complexity and ambivalence in the articulation of emotions relating to generations and generational identities relating to environmental crises than is depicted in "conflict/solidarity" discourses. Third, it presents ambivalent desires for, on the one hand, policies that foster greater "meaningful intergenerational interaction," while cautioning against interventions that are "contrived" and "inauthentic." The paper contributes to gerontological understandings of intergenerational relationships and ambivalence through its use of fiction and reading groups. Existing scholarship has identified ambivalence as a useful concept in "resolving rivalry and... moving beyond generational oppositions" (Biggs, 2007, p. 706) and to circumvent conflict/solidarity binaries (Lüscher & Pillemer, 1998). By using a methodology of readers' responses to novels, we show how fictional representations of intergenerational themes can elicit more meaningful reflections on the ambivalences and complexities of relationships across age groups. Notably, the ambivalence expressed by participants was rooted and revealed in reflections on their own lived experiences and their reactions to relatable fictional characters and scenarios. This is closer to the "lived ambivalence" that Baars (2014 invokes). Although Baars used the term more in relation to aging, rather than specifically intergenerational relationships, the experiential, temporal, and processual dimensions invoked by the word "lived" help us to make sense of the findings presented here. We endorse Baars' insistence that time be given more attention to understanding the aging process but extend this to specifically include intergenerational relationships. In reflecting on the novels, participants spoke about intergenerational relationships as ambivalent processes lived in time. This temporal dimension was evident in participants' discussions about past, existing and anticipated caring responsibilities, extended definitions of intergenerational relationships as encompassing "middle generations" and ancestors and descendants, personal regrets, and anticipated guilt contextualized within historical processes, and a desire for more time to allow for "authentic" intergenerational connection. In addition to showing how fiction can deepen understanding of gerontological issues, we suggest that our paper also offers insights into how gerontological concerns can inform interpretations of fiction. Although we argue that the very absence of terms such as "intergenerational relationships" or "intergenerational segregation" in the novels freed participants to express more nuanced and ambivalent thoughts than might otherwise have been the case, their awareness of these issues-in part, but not wholly prompted by our questions-guided their reading. Whether in speculative fiction such as Trouble with Lichen or The Last Children of Tokyo, or more realistic representations in Young Art and Old Hector and The Summer Book, participants responded enthusiastically and ably to our invitation to read them with a "gerontological lens" and were well versed in the vocabulary of dominant discourses of intergenerational conflict, inequality and justice in their discussions, even as they expressed more ambivalence. These interdisciplinary pathways between gerontological concerns and imaginative reader responses suggest that both approaches are needed to develop new ways of thinking through the complexities of intergenerational relationships and the transitional experience of living/aging through time. A number of policy implications arise from this study. Given the marked ambivalence expressed by participants across different subjects and their desire for "authenticity," we agree with Biggs's (2007) suggestion that policymakers would be wise to embrace ambivalence to "acknowledg[e] and negotiat[e] solutions promoted between generational groups" (pp. 706-707). Participants echoed concerns about age segregation (Hagestad & Uhlenberg, 2006) and, valuing what they perceived as the "genuine" intergenerational connections depicted particularly in the novels The Summer Book and Young Art and Old Hector, called for more time and space in being allowed to develop these relationships. The ambivalence expressed here was one where participants wanted policymakers to facilitate greater intergenerational connection, without necessarily labeling interventions as such. Accordingly, the policies they endorsed were not explicitly targeted at addressing age divisions. Therefore, organizations that work with intergenerational groups might consider how present policies currently being trialed such as the 4-day week (Cooper et al., 2021;Haraldsson & Kellam, 2021), UBI (Ugo et al., 2020), urban design (UN-Habitat, 2020), and climate action (UNEP, 2021) intersect with lived experiences of aging and what intergenerational communities can contribute to these ideas. This paper addresses calls to "myth-bust" generational conflict (Duffy, 2021) but also calls for nuance and complexity in invoking intergenerational solidarity. The findings presented in this paper are underpinned by a methodology that uses fiction to provide a platform for ambivalent discussions and feelings, where novels can open up new worlds and allowing for recognition to "forge connections across difference" (Felski, 2019) and offer new vocabularies for discussing intergenerational relationships. In doing so, we offer a way to avoid the pitfalls of a flattening "generationalism" (White, 2013) and bring more insight into what intergenerational relationships are and why they matter. --- Data Availability For the purpose of open access, the authors have applied a Creative Commons Attribution (CC BY) license to any Author Accepted Manuscript version arising. The data supporting the findings reported in this paper will be available via the UK Data Archive repository from April 2023. This study was not preregistered. --- Supplementary Material Supplementary data are available at The Gerontologist online. --- Conflict of Interest None declared.

Background and Objectives: The term "intergenerational relationships" is widely used in gerontological literature and age-related policies. However, discussions of the term often tell us surprisingly little about what it means or why it matters. We suggest that this is due to a reductivism and instrumentalism in 2 main discourses within which intergenerational relationships are usually discussed. First, intergenerational relationships are often conceptualized through a binary "conflict/solidarity" lens, reinforcing an entrenched "generationalism" . Second, they are predominantly constructed as a problem to be addressed within debates on how to tackle intergenerational segregation. Neither of these discourses provides much room for a more nuanced understanding of how intergenerational relationships are experienced or why they are meaningful. In this paper, we discuss how fictional narratives can introduce imagination and a richer vocabulary into discourses concerning how people of different ages relate to each other.We present findings from reading groups where adults discussed novels depicting themes of older age, intergenerational relationships, and time.In discussing the fictional narratives and characters, participants reflected on the significance and meaning of intergenerational relationships in ways that went beyond dichotomous and instrumentalist discourses. Drawing on the concept of lived ambivalence, we argue that fictional representations of intergenerational themes can elicit more meaningful reflections on the complexities and contradictions of relationships across age groups.We conclude that a more nuanced understanding of intergenerational interaction can inform gerontological discourses and policy, but also that gerontological awareness of social challenges concerning age relations can inform interpretations of fictional narratives.

Background In Mexico, 57.3% of the population do not have access to employment paid health care insurance [1]. This situation represents a scenario of vulnerability, where social capital could play an important role to solve health related issues. To understand the association between social capital and health, it is necessary to understand that the healthcare system conditions and access to healthcare services are marked by deep socio-economic, political, and cultural inequalities. It is also important to note that healthcare includes more than the possibility of getting health-related services. It incorporates access to knowledge about how to navigate the healthcare system, health education, efficient and high-quality medical care, and material and emotional support to deal with illness [2][3][4]. The National Mexican Council for the Evaluation of the Social Development Policy (CONEVAL) [5] defines the "lack of healthcare access" indicator as the population percentage that does not have healthcare insurance in public or private institutions. The indicator expresses the degree of compliance with the provision of the social right to health access listed in the Mexican Constitution. Although the gap in access to healthcare has decreased in recent years (from 38.4% of the population with no access in 2008 to 16.2% in 2018) [1], the prevalence of Mexicans without access to healthcare coverage is still relevant. In Mexico, health services are provided by differentiated subsystems [6]. Most of the medical services provided under a healthcare insurance coverage are delivered by the by the Popular Insurance (Seguro Popular), which covers 42.2% of the population who lack of medical insurance as a job benefit given that its beneficiaries do not have permanent job positions. The second most important healthcare provider is the Mexican Institute of Social Security (IMSS), which provides medical services to those employees working in private companies who cover the premiums as a benefit at the formal sector. The recipients of this coverage represent 36.4% of the population. Other healthcare providers are state owned as the Institute of Social Security and Services for State Workers (ISSSTE), which covers 5.5% of the population that receives this benefit as government officials. Other healthcare services are delivered through government managed hospitals who covers benefits to government employees at the National Oil Company (PEMEX), the Ministry of National Defense (SEDENA), the Ministry of Naval Forces (MARINA), state level ISSSTE. Also as healthcare special programs as the IMSS-PROSPERA who provides primary healthcare in low income communities. The remaining population receives medical under the private healthcare insurance, indirect insurance and other medical institutions [7]. Despite the positive trend in healthcare access, Mexico has some of the highest out-of-pocket expenses among all of the nations belonging to the Organization for Economic Co-operation and Development (OECD) [6]. The quality of service and the satisfaction level explain why individuals covered by public institutions prefer to access to private healthcare providers, where there is a larger amount of available facilities (11.4 public hospitals and 28.6 private hospitals for every million inhabitants). Some of the most common reasons for not demanding services at the public healthcare systems are the long waiting times, the lack of available medications, the distance from the household to health providers, the lack of money, and the poor service delivered by healthcare workers at the clinics [8]. People covered under the Seguro Popular medical insurance reported to have faced more difficulties to receive medical care at its facilities compared to have received assistance at other health systems [9]. This is a remarkable disadvantage considering that most of the beneficiaries of the Seguro Popular are in the first four income deciles and the most economically limited. Data concerning the conditions of Mexico's healthcare system suggest a scenario where social capital could play an essential role. Under this premise, we selected four states with differentiated levels of social welfare and healthcare access to conduct our research [10] which can see in Table 1. Mexico City (19.6%) and Oaxaca (15.9%) have the highest population percentage without access to healthcare, and are above the national average (15.5%). The State of Mexico has the same percentage than the national level (15.5%), while Tamaulipas has the lowest proportion of individuals without access to healthcare (12.8%) [11]. As it can be seen in Table 1, states with higher marginalization levels (i.e., Tamaulipas and Oaxaca) have more individuals covered by the Seguro Popular and less beneficiares of formal jobs insurance either from private companies or government agencies (i.e., IMSS and ISSSTE). Data suggest a potential association between formal employment, the marginalization index of the state, and the number of individuals covered by the Seguro Popular. Considering previous statements, we hypothesize that social capital is a relevant asset to manage health problems in Mexico. However, in the case of illness, support varies according to the conditions of marginalization. According to several scholars, families and friend's social capital can become a crucial option to deal with health issues in the least developed countries where healthcare systems are inadequate [12]. Networks provide assistance, trust, reciprocity and support in case of illness. Therefore, individuals with more social capital are more likely to be healthier [13,14]. --- Social capital and health Social capital is based on strong and weak ties, cultural dynamics of (re) production, and social articulations, such as a) norms, obligations, and commitments; b) trust; c) expectations of reciprocity; and c) support and cooperation practices [15][16][17][18][19]. Social capital operates at the micro, meso, and macro-social levels [19]. The micro-social level employs networks of relations that can be understood as bonds with relatives and friends and belonging to groups of similar interests [20]. According to Granovetter [17], these networks of relations (or strong bonds) play a decisive role in the solution of particular problems and are the main sources of social and emotional support. The meso-level refers to the strength of the relations within secondary groups. It comprises neighborhood and the sociocultural institutions of the community as a whole [19]. Therefore, social capital is part of each community's sociocultural system, its management, and its sanctioning structures. The macro-level includes a feeling of belonging to society and describes the degree of consensus and intensity of social bonds. Besides trust and institutional legitimacy, it includes participation in social, civil, or political organizations [20]. Previous research has reported an association between social capital and health self-perceptions [21][22][23]. There are also references to the influence of norms and attitudes on healthy behaviors and psychosocial networks that increases access to medical attention and mechanisms that improve self-esteem [24][25][26]. Social capital is also associated with community participation and mental health [27][28][29], physical and psychological health [30], and number of years lived [14,31,32]. Some authors have described a relationship between informal relations and mental health [33]. In this regard, social capital represents a strategy to mitigate the effects of social isolation and social disconnection on health and quality of life [34,35]. This is particularly relevant for older adults because it has been found that social capital has positive effects on slowing mental illnesses [36,37]. Indeed, regular family contacts have positive impacts on loneliness among older adults [38] due to the value of social influence, social support, norms, and the flow of information and resources [39,40]. In addition, social capital is considered a relevant concept in public health because it helps to identifying the importance of the links or bonds that individuals create throughout their lives [2]. Family, neighborhood, and identity relationships (sources of social capital) shape the source of social control, family support, and creation of benefits through extra-familial networks [41][42][43]. Several authors have suggested a positive association between social capital and health [44,45]. Links that individuals build throughout their lives are valuable assets that can be tapped into for personal or community profit [46]. Bonding-type social capital includes trust and cooperation relationships among members of a network that compare themselves as similar because they share a social identity [47]. This form of capital acts as a connector where trust and respect for norms are built, but where intolerance and mistrust toward members outside of the group can also grow. These circles include family members and very close friends [18,48]. As a consequence, networks are a support system for solving health-related issues [49]. However, behavior that is harmful to health might be reinforced or reproduced because they are ruled by habits, norms, and customs legitimized by the group [17]. Bridge-type social capital includes acquaintances in social groups or networks, where information is shared and exchanged [4,17,18]. Regarding access to health care, bridging-type social capital relationships create more opportunities to get information. Individuals can ask friends and other groups to get access to other networks and potentially earn valuable information [50]. Weak links turn into bridges among two or more networks that offer access to information, resources, and health services beyond the intimate circle [51]. The linking-social capital type includes norms concerning respect and trust. It also comprises networks among individuals who interact through different levels of explicit, formal, or institutionalized powers within the society [46,48]. Several studies show the association between social capital components (e.g., trust and reciprocity) and health depend on the context [4]. The consequences of such a relationship can be observed in populations that are socially excluded, to the detriment, or support to the wellbeing [43]. Social capital highlights the potential of social bonds and interactions, regarding disparities in access to health services [52,53], and other social, economic, and cultural inequalities [54]. Considering the above, some analyses have included income disparities [12,54,55] and the implications of these interactions [2,56]. However, the positive effects of such associations do not limit the state's responsibilities concerning public health and social welfare [57]. Social capital is used as an asset in the absence of an efficient health-care system. Nevertheless, the association is not homogeneous and can differ according to the characteristics of individuals and the context [4], mostly in populations that are socially excluded [43]. For this reason, our purpose is to answer, what role does social capital play when people with different levels of marginalization and access to health services get sick in Mexico? --- Methods --- Municipal marginalization index This study employs the Municipal Marginalization Index [58] as a measure of socio-economic comparison among contexts. The index includes four areas with nine types of exclusion, which are measured by percentages of the population that have no access to basic services. The four areas and nine types of exclusion are (1) education (i.e., illiteracy, population that has not finished elementary school); (2) housing (i.e., homes with no plumbing or sanitary services; homes with no electricity; homes with no water pipes; homes in any level of overcrowding; homes with dirt floors); (3) population distribution (i.e., towns with fewer than 5000 inhabitants); and (4) income (i.e., working population that earns up to two times the minimum average wage). Each of these categories has an indicator expressed as a type of deficiency, which is used to create an index that classifies marginalization in five levels: very high, high, medium, low, and very low [59]. In comparison to other similar measures as the Human Development Index (HDI), the Municipal Marginalization Index represents conditions of high marginalization as more vulnerable because they accumulate the highest percentages of indicators. Therefore, in low marginalization populations, there is a very low proportion of exclusion indicators. We labeled the marginalization values as follows: I (very low), II (low), III (medium), IV (high), and V (very high). --- Sample Data were collected through semi-structured interviews in four states with different degrees of social wellbeing [10]: Mexico City (very high level), Tamaulipas (high level), the State of Mexico (medium level), and Oaxaca (low level). Interviews were carried out in different municipalities according to the type of municipality (urban, semi-urban, or rural), Municipal Marginalization Index (IMM) [59], and population size (see Table 2). The total number of municipalities selected was 71 and distributed in the following way: 16 in Mexico City, 6 in Tamaulipas, 31 in the State of Mexico, and 18 in Oaxaca. A total of 247 semi-structured interviews were conducted: 78 in Mexico City, 44 in Tamaulipas, 53 in the State of Mexico, and 72 in Oaxaca. Interviewees were chosen through non-probabilistic sampling considering variables as socioeconomic level, educational level, age, and gender to achieve heterogeneous profiles (see Table 2). According to previous research [60], interviewees with different sociodemographic profiles support saturation of categories, which produces more sound and consistent results. Access to the communities was done with the help of several gatekeepers who were residents in the area. They were the starting point to the snowball technique to find individuals with the desired profiles. --- Information gathering Interviewers were trained in ethical research issues, as well as in the use of the protocol to handle respondents' information. Every interviewee was informed that all of their responses would be confidential and agreed to sign informed consent. The interviews lasted approximately thirty minutes major the social capital and health section. Answers were recorded with the interviewees' permission and later transcribed literally on a text processor. --- Analysis Data analysis was iteratively deductive and inductive, based on the grounded theory [61] that lies on fieldwork as the main knowledge source to come to conclusions based on empirical evidence [62]. Coding was concurrent with data analysis and interpretation. First, it helped to observe, to record, and to classify answers during fieldwork. Second, coding helped to disaggregate information and to group and to synthesize it from multiple associations and cognitive inferences [62]. This procedure was helpful to the interview analysis at different stages, such as understanding, synthesizing, and theorizing phases. The understanding stage enabled us to approach to the beliefs, values, and ways of life of the different contexts and to see the experience from the participants' perspective [63]. During the synthesis stage, we read the transcripts of the interviews. The purpose was to identify and codify the emerging trends in each category. To that end, we used the qualitative NVivo software, version 11. The following codes emerged from this process: In general, this codebook guided us to clarify the association between social capital and health. In specific, the codebook was also helpful to distinguish central discourses and meanings associated with the role of social capital in health, according to the informants' levels of marginalization and access to health services. 1. Finally, the theorizing stage allowed us to contextualize and to explain the relationship [63]. Authors discussed the findings in several meetings. Interpretations of categories in each level of marginalization and their link with the theoretical framework were examined during these meetings. --- Results --- Municipalities and interviewees In this study, municipalities were grouped into levels of marginalization and type (i.e., urban, semi-urban, or rural municipality), as shown in Table 2. Table 2 shows a potential relationship between the municipal marginalization level and the interviewees' socioeconomic profile. Data suggest most of the individuals with elementary school education attainment, in very low and low marginalization areas, live in rural and semi-urban areas. In contrast, those participants in areas with very high and high levels of marginalization with undergraduate and graduate studies attainment, live in urban areas. --- Social capital and health issues Regardless of the level of marginalization, social capital represents an essential tool to deal health issues for the interviewees. Low trust in health institutions turns social capital into a critical component to optimize the individual's economic resources. Bridge-type social capital contributes to the choice of medical service providers, by obtaining information from previous experiences and recommendations from neighbors, friends, and family. Moreover, interviewees indicated that belonging to networks of friends, relatives, or neighbors provides information that allows a better choice of hospitals, specialist doctors, and administrative procedures regarding medical insurers. Regardless of the marginalization level, findings show that the exchange and dissemination of health information are central resources for the closest networks. Emotional support is also fundamental to face and solve health-related problems. Still, there are some differences in how it works because it depends on the level of marginalization and access to health services. Participants living in very low or low marginalization areas, have health coverage in 96.5% of the cases. Such coverage includes access to private health care providers (e.g., health insurance with full coverage), as well as access to specialist doctors. At this level, participants have access to healthcare coverage due to their membership to a larger number of networks which include extended family, workgroups, or religious groups, besides strong links (bonding-type social capital). The interviewees' socio-economic level and their health coverage influence the function of social capital. In this case, it provides an emotional support system that leads to feelings of security and support. For instance, people accompany or visit a sick person: they give them fruit, they accompany them to the doctor, or they help with the housework. I feel, very, very supported. I mean I've known my friends for a long time, and I have no doubt that they would help because, at least with what we have shared, that's how they make you feel, don't they? And concerning my relatives, well, the time we spend together when I see them, um, let's say that they make me feel sure that, that they would give a hand whenever it might be needed, right? And... well, economically I'd do it because I have the insurance that would cover all their expenses (man, 36, IMM I). Individuals that live in high or very high marginalization areas are covered mainly by the Seguro Popular or lack of healthcare insurance. Seguro Popular has limited coverage of services, quality of the infrastructure and care, provision of medicines, and very high waiting times. For these reasons, 72.9% of interviewees mentioned that they had "insufficient infrastructure, for the number of individuals covered." Regarding funding for medicines, 54.5% of respondents stated they rely on generic medications, which affects the quality of the treatment. Meanwhile, 75.6% of respondents indicated to be unsatisfied with the quality and access to health services. Therefore, they would prefer to use private doctors' services (at low cost walk-in clinics where generic drugs are prescribed), home remedies, self-medication, traditional therapies, praying, or drinking. When they do not have enough money, respondents pray with the aim of tolerating the pain or healing. At these levels of marginalization, relatives and friends networks are valuable assets that individuals can use to transform into alternative and additional resources to solve health issues. Bonding-type social capital integrates the material and emotional support to deal with health situations. In the case of accidents or illnesses, these networks are useful to raise funds for the patient's treatment. Likewise, networks provide necessary emotional support to improve the health of a sick person. My family. Well, my brothers and my parents. I feel that in a strong need, yes, I have support from them. We have the support of the whole family (man, 49 IMM IV). These individuals' socio-economic situation and their vulnerable position regarding health coverage emphasize the essential role of social capital. According to our findings, support networks establish a critical resource for health emergencies, such as access to medicines, home remedies, or to transfer the patient to the nearest health care center. We have helped each other in the good and the bad times... if someone passes away or get ill, we collect money, and we help (woman, 47, IMM IV). To my friends and family, I would help them with actions... because I do not have much money (woman, 23, IMM V). Participants who have experienced these situations expressed they felt supported and grateful. Simultaneously, they felt guilty for being a burden to their loved ones with such moral and material responsibility. They also expressed constant feelings of vulnerability. Well, [I feel] very, very bad, because I can't even find a way to pay for it so that I can take care of it by myself or take care of my own accidents or illnesses, so you feel bad. I think anyone feels bad, don't they? That someone has to pay his/her expenses, and not out of pride, but because to have to say I can't pay it by myself, I mean it's hard, isn't it? (man, 31, IMM IV) The micro-level of social capital is essential in high or very high marginalization levels because networks provide the patient with information to navigate the healtcare system and to have accurate information about doctors and alternative health care centers. Likewise, people in networks recommend home remedies to the patient or donate medications that they did not consume in previous illnesses. To respondents, the support of these networks facilitates the transportation of the sick person to the health centers. In emergencies, private cars are a frequent alternative to the lack of ambulance and medical emergency services. In the middle level of marginalization, respondents present heterogeneous healthcare in coverage and how social capital works in different ways in case of illness. On the one hand, for individuals who have major medical health insurance, social capital is similar to very low and low levels of marginalization in situations of illness. On the other, for those who have the services of IMSS, ISSSTE, and Popular Insurance, and do not have major medical health insurance, social capital works similarly to those interviewed at high and very high levels of marginalization. Finally, 51% of respondents indicated that they have a low level of trust in health institutions. Although the main criticism targets at public institutions, individuals also mistrust the private ones. Results suggest that, regardless of the marginalization level, there are specific codes about the support levels, the type of support, and its length in Mexico. In the case of disease or long term health problems, the commitment level and the possibilities of material and emotional support are stronger in the family nucleus. Regarding friends' and neighbors' networks, commitment and support are usually temporary. --- Discussion The most vulnerable socio-economic levels are the most affected in the access to health services. Behind these issues, there are deficits in health sub-systems. For instance, there is a reduced number of doctors and nurses in the system (i.e., 2.2 doctors and 2.6 nurses for every 1000 inhabitants), including those that work in the private sector [6]. In addition, the public system usually provides less time for each patient's visit, which has negative consequences on the quality of medical service delivery. Also, only some of the Seguro Popular, IMSS, and ISSSTE medical centers have emergency rooms [6]. Finally, 33% of prescriptions in the Seguro Popular were not wholly fulfilled because the medicine was not in stock, while at the IMSS, 14% of the prescribed drugs were not available [6]. Regardless of the municipal marginalization level and the healthcare access, health system conditions could explain why most respondents mentioned that they do not use the basic or emergency services from public institutions. Most individuals indicated that these services are not efficient or effective. This perception is the result of (1) sick individuals that must travel long distances to reach the health care centers, (2) long waiting times, (3) perception of poor quality medical services delivery, and (4) there is a lack of funding for prescribed medications. Disparities between levels of marginalization suggest another source of inequality. Findings evidence that the participants at the poorest condition are in a very vulnerable situation regarding their right to health. Disparities in access to and quality of health services, explains the role of social capital to address health related problems. The evidence here supports that the micro-level of social capital has an essential role when individuals get sick. However, unlike other studies [63], the relevance of social capital can be explained by the precariousness of medical services, the poor health infrastructure, and the problems of access to medical care [6,8]. The low levels of trust in physicians and health institutions make bridging-type social capital an essential component to select a good quality medical service according to the individual's budget. However, the respondents' socioeconomic level and health coverage influence the role of social capital, or in other words, the type of support they provide and receive. Each social capital type has different functions. The bridging-type social capital helps to provide information about physicians or homemade remedies. Bonding-type supports emotional and instrumental provision to manage health problems [25,51]. Regardless of the level of marginalization and interviewees' socio-economic level, bridging-type social capital provides information based on direct experiences. These networks help in search and choice of healthcare insurance, economical and reliable hospitals, medical specialists, and advice for the follow-up and care of diseases [25]. Findings also evidence the value of emotional support in health situations [64,65]. In the low and very low marginalization levels, support networks are broad and include strong links (e.g., close friends and family), religious groups, and extended family. In the very high and high levels of marginalization, networks are limited to the nuclear family and closest friends. In specific, respondents in very high and high levels of marginalization indicated that family, close friends, and neighborhood are essential to face accidents or severe illnesses. These networks not only provide the necessary emotional support for the recovery of the patient, but they also articulate instrumental support [51,66], namely medicines, homemade remedies, taking the patient to the nearest hospital, and funding for health services. At medium marginalization level, social capital is broader: most of the participants mentioned they could count on extended family and friends. Networks also help in raising funds for more extensive medical services and better hospitals. For individuals in the low and very low marginalization levels, support networks include strong bonds that also extend to other groups and act as emotional support. In addition, the links are broader, and a positive relationship between social capital and income is developed [66]. Considering most of the participants at this last level have private healthcare coverage, social capital operates mostly as emotional support. --- Conclusions Mexican Constitution guarantees that every person hast the right to health protection. Health is a social right. According to interviewees, the delivery of the healthcare services is deficient, mainly at the primary care level as well as at the emergency room services provided by the state operated hospitals, as the Seguro Popular, IMSS, and ISSSTE. The access to this right should not be conditioned by any socioeconomic, political, and cultural inequalities. In Mexico, employment status (formal or informal employment) and socio-economic level define the access to the health care system and subsystems. The IMSS and ISSSTE relate to formal employees in the private sector and the federal government, respectively. Meanwhile, the Seguro Popular focuses on individuals in the informal sector not covered by the benefits of a formal job or permanent position medical insurance. Evidence suggests a positive association between informal work, socioeconomic vulnerability, and access to healthcare services. The population proportion covered by the Seguro Popular seems to be associated both to the municipality's informal employment level and its marginalization index. Participants indicated low levels of trust in healthcare institutions due to the poor infrastructure and quality of services. Although the main criticism is focused on public institutions, individuals also did not report to not trust the private ones either. These facts related to the access and quality of medical services and transform social capital into a significant asset. Besides, strong bonds or links are valuable resources that individuals can tap to solve health-related issues; however, the use of such resources is not homogenous and is modified by specific factors that were represented in this study through the municipal marginalization index. We can observe that at low and at very low levels of marginalization, the interviewees' socio-economic conditions and health care coverage (which in many cases includes major medical insurance) impacts on the role of social capital. Social capital acts as a type of emotional support and as a way of companionship that explains the positive feelings of security and support. At high and very high marginalization levels, networks are limited to the nuclear family and the closest friends. These individuals' access to healthcare defines social capital as an essential resource and an emotional support system when dealing with medical issues. For this reason, participants reported contradictory feelings. On the one hand, they feel supported and grateful, but on the other, they feel guilty about burdening their loved ones with such moral and material responsibility. One of the implications of this study is that, in Mexico, the use of social capital (especially bridging and linking types) could be a valuable asset in health education, access to information, and health resources and services. Besides, it could help to foster norms for respect and trust between medical healthcare providers and patients as a sine qua non for the quality and efficiency of health services. The second implication of this study is that, although we found that social capital is an essential resource for solving healthcare issues, the positive effects of the relationship do not exonerate the responsibility of the Mexican government to guarantee the quality of the healthcare services provided by the government. We should emphasize the importance of ensuring access to healthcare and quality medical services. It means to reduce waiting times, to increase prescription medicines availability, and having a more balanced ratio of medical staff in comparison to the number of individuals covered. Finally, there are some limitations to this study. First, the IMM is an indicator that provides measures at the context level conditions (neighborhood, municipality and, state) and not at the individual level (or by household). Second, the highest percentage of people interviewed were located in urban areas. Even in the locations under study, there are no rural areas at the very low level of marginalization (see Table 2). Third, the results are not generalizable to the entire Mexican population given that a qualitative analysis approach was used. Despite the sample cannot be considered as representative, the selection of the interviewees considered the heterogeneity and diversity of socio-economic, cultural, and geospatial aspects of the participants in order to improve learning possibilities and the reliability of the results. In this regard, the coding procedure used allowed the incorporation of emerging nuances and interpretations, which contributed to the consistency of the findings. Despite these limitations, the qualitative approach used in our research can be appropriate to any context (region or country) where health services have been stratified according to population socioeconomic conditions or employment status. It could also be useful in those countries where non-contributory health services cover individuals living in poverty because it shows how functional social capital can be in health-related situations. --- Availability of supporting data Datasets used and analyzed during this study are available from the corresponding author on reasonable request. --- Ethics approval and consent to participate This study was approved by the Ethics Committee at Iberoamericana University approved the study. Informed consent was obtained from all study participants. --- Consent for publication All authors consent for publication of this study. --- Competing interests The authors declare that they have no competing interests. --- Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Background: Social capital is employed as an asset when there is a lack of an efficient health-care system. However, this relationship is not homogeneous and can differ according to the characteristics of individuals and their context. In this paper, we aim to analyze the role of social capital in the solution of healthcare problems among individuals with different levels of marginalization and unequal access to health services. Methods: This qualitative study examines the role of social capital in the demand for healthcare among Mexican individuals with different levels of marginalization. The research draws data from semi-structured interviews (N = 247) that were collected in four Mexican states with different social welfare benefits: Mexico City, Tamaulipas, the State of Mexico, and Oaxaca. The interviewees were selected using the snowball method and other eligibility criteria such as education, age, and gender. Results: Findings suggest that social capital is a relevant factor in solving healthcare problems, depending on the level of marginalization. The role of social capital can be explained by the precariousness of medical service delivery, the poor health infrastructure, and the difficult access to health care in Mexico. Networks are the main resource to deal with health related issues, food, medicine, and out-of-the-pocket medical expenses in contexts of high levels of marginalization. In the middle level of marginalization, networks also help in raising funds for morespecialized medical services and higher quality hospitals. In the least-marginalized levels, social capital is used as companionship for sick individuals, while support networks act as emotional relief. At this level, most individuals have private health insurance, and many of them have major medical healthcare coverage. Conclusions: Participants reported low levels of trust in the health care system because of the poor infrastructure and quality of medical service delivery. Although the main criticism is focused on public healthcare institutions, there is a lack of trust in private medical services as well. These facts are related to the access and quality of medical service delivery and turn social capital into a significant asset. Despite that social bonds or links are valuable resources that individuals can use to solve healthcare related issues, the use of social capital is not homogenous. Indeed, it can be influenced by several factors that were represented in this study through the municipal marginalization index.

in low-income areas with limited access to healthful foods (termed "food deserts") are more likely to consume foods with poorer nutritional quality, when compared to those living in higher-income, higher-access areas 18. Studies show that living in a food desert is associated with risk for obesity 19 and nutritional deficiencies 20. Few studies have assessed the relationship between food deserts and metabolic health during pregnancy; however, one study showed that living in a food desert was associated with increased likelihood of morbidity during pregnancy 21, including metabolic disorders and another suggested decreased risk for gestational diabetes mellitus (GDM) in areas with more grocery stores 22. Food desert-mediated differences in dietary quality may be responsible for these associations. Among pregnant populations, some studies show that consumption of a more pro-inflammatory diet is associated with increased risk for GDM 23,24. However, another study observed that a more pro-inflammatory diet is associated with decreased risk for GDM 25. This heterogeneity may be related to differences in dietary data collection methods, with one study using self-administered food frequency questionnaires (FFQs) in the United States 25, while the others utilized FFQs completed with the assistance of trained dietitians/interviewers among people in Iran and China 23,24. While living in a food desert likely informs dietary quality, whether the inflammatory quality of the prenatal diet is related to living in a food desert during pregnancy has not, to our knowledge, been previously examined. The influence of socioeconomic status on metabolic health in pregnancy. Lower socioeconomic status (SES), an important correlate of health inequalities 26, is associated with poorer health outcomes for pregnant people. Compared to those with higher SES, pregnant people with lower SES show an increased risk for gestational diabetes 27, gestational weight gain 28, obesity 29, and metabolic syndrome 30. The mechanisms by which low SES influences metabolic health are complex and multi-faceted (see 31 ). Lower SES is associated with both reduced access to affordable, healthy foods 32 and more pro-inflammatory diets 33,34, which have been linked to metabolic health in non-obstetric populations, including increased risk for obesity 35, hyperglycemia 36, and metabolic syndrome 37. Low SES is also associated with inadequate prenatal care 38 and chronic stress 39, as well as insufficient access to specific nutritive foods 40 or food altogether 41. Whether living in a food desert mediates the association between SES and metabolic health for pregnant people has not yet been articulated. Socioeconomic status, food desert severity, and their interactive impact on metabolic health in pregnancy. Identifying contributing factors that help explain the relationship between SES and metabolic health during pregnancy is important for optimizing intervention strategies for pregnant people with lower SES, as well as for reducing the burden of poor metabolic health during pregnancy on offspring. Living in a food desert may be one mechanism underlying the relationship between variation in SES and metabolic health during pregnancy. We seek to extend studies showing that lower SES is associated with reduced access to affordable, healthy foods 18,32 and with increased pregnancy morbidity [27][28][29][30]42 by examining food desert severity as a potential mediator of the relationship between SES and metabolic health outcomes during pregnancy. For example, access to fewer financial resources may lead to reduced access to a healthful diet, which may, in turn, impact metabolic health by increasing adiposity, promoting peripheral inflammation, and/or disrupting glucose regulation. Current study. We propose that, in addition to being mechanistically linked, both SES and food desert severity may play a role in metabolic health and nutrition during pregnancy, an association that has not, to our knowledge, been formally tested. While many other studies of food deserts utilize a dichotomous assessment of food desert status (living in a food desert vs. not living in a food desert) 43, the use of a categorical variable does not capture the degree to which healthful and affordable foods are accessible and may limit the ability to detect meaningful differences. Thus, in the current study, we calculate the degree of food desert severity for each participant relative to others in their neighborhoods, allowing us to draw inferences about pregnancy metabolic health according to the spectrum of resource accessibility. We examine three outcomes related to metabolic health and nutritional quality in pregnancy: adiposity, glucose regulation, and the inflammatory nature of the prenatal diet. As these outcomes are linked to one another 44, and provide complimentary, but distinct information, unlike previous work, we control for their interdependence using structural equation modeling (SEM). Furthermore, as there is a graded, positive linear association between maternal hyperglycemia and risk for adverse perinatal outcomes 45, we examine a continuous measure of glucose regulation. As both low SES and living in a food desert are associated with consumption of a less healthful diet 46,47, we also examine the effects of low SES and food desert severity on nutrition during pregnancy. While the relationship between food desert severity and SES is complex and likely interrelated, elucidating the interacting role of each of these is critical for both practitioners and policy-makers as we work together to address disparate access to resources among pregnant individuals. Our analytic strategy is key to disentangling the unique role of SES and food desert severity on the metabolic and nutrition variables, as SEM allows us to simultaneously estimate the effects of SES and food desert severity on the metabolic and nutrition variables of interest, while controlling for their interdependence. Thus, we are able to examine the significance and strength of the different relationships in the context of the complete model (one in which SES and food desert severity are both present). Our use of SEM also limits the risk of multicollinearity impacting our results, as SEM is known to be robust to multicollinearity 48. SEM also allows us to reduce measurement error by explicitly modeling shared variance, thus more valid coefficients are obtained. Importantly, SEM also allows us to examine how SES may impact metabolic health and dietary intake during pregnancy by assessing the mediating and mechanistic role of food desert severity. In the current study, we examined whether food desert severity mediates the relationship between SES and metabolic health during pregnancy. This study had three hypotheses: 1. SES will have a negative relationship with metabolic health during pregnancy, with lower SES associated with increased adiposity, higher glucose concentrations during a routine glucose tolerance test (an indication of poorer glucose regulation), and consumption of a more pro-inflammatory diet during pregnancy. 2. Food desert severity will have a positive relationship with adverse metabolic health, with higher food desert severity associated with increased adiposity, higher glucose concentrations in response to a routine glucose tolerance testing, and consumption of a more pro-inflammatory diet during pregnancy. 3. Food desert severity will mediate the relationship between SES and metabolic health and diet during pregnancy, such that lower SES will lead to a higher degree of food desert severity, which in turn, leads to worse metabolic health during pregnancy. --- Methods Procedure. Data Measures. Glucose regulation. As part of their routine prenatal care, participants were screened for GDM via an oral glucose tolerance test (GTT). One-hour plasma glucose concentrations (mg/dL) were utilized in analyses as they are a common metric for predicting health outcomes 49, including whether further GTT testing is necessary. As participants were assessed at varying times during pregnancy (8.14-32.57 weeks), the number of weeks' gestation at the time of assessment was included as a covariate in all analyses that included GTT data. See Supplementary Materials for more information. Adiposity. To measure body composition, at 22 weeks' gestation (M = 22.53, SD = 1.41), air displacement plethysmography was utilized via the BOD POD <unk> Body Composition tracking system (Life Measurement, Inc.) 50,51. As an excess of body fluid is typically observed during pregnancy 52, estimates of fat and lean mass in the body were adjusted using previously-described equations 53 and second trimester pregnancy-adjusted percent fat mass was utilized in analyses. See Supplementary Materials for more information. Nutrition. Second trimester dietary intake was measured by three, non-consecutive 24-h diet recalls conducted by trained dietitians, using the multi-pass method 54 and the Nutrition Data System for Research software (versions 2018-2020; University of Minnesota, Minneapolis, Minnesota). See Supplementary Materials for more information. To capture the inflammatory nature of participants' diets, Dietary Inflammatory Index (DII) scores, a metric of dietary quality that may be particularly relevant for metabolic health 55, were calculated from the 24-h diet recalls. Positive values indicate a more pro-inflammatory diet 56. See Supplementary Materials for more information. Socioeconomic status. Self-reported information, including years of education completed, total annual combined household income adjusted for number of people in the household, and amount of savings after adjusting for debt was utilized to capture information about SES. See Supplementary Materials for more information. Food desert severity score. Geographic information system mapping was utilized to identify food deserts. This method offers advantages over others (e.g., those that enumerate the number of healthy vs. unhealthy food vendors in the neighborhood) because it provides a more comprehensive characterization of community resources when determining if an individual lives in a food desert. Participant addresses obtained at study enrollment were converted to latitude and longitude coordinates and mapped onto census tracts using data from the 2010 Census of the Population. Using the Food Access Research Atlas (FARA), published by the Economic Research Service of the United States Department of Agriculture (USDA 57 ) and census data, low-income (at least 20% of the population of the census tract has a median family income at or below 80% of the metropolitan area or state median income) and low-access (at least 33% of the census tract or at least 500 individuals reside far from a supermarket) census tracts were identified. Detailed information about the sources of the data used to characterize census tracts as food deserts can be found online (https:// www. ers. usda. gov/ data-produ cts/ food-access-resea rch-atlas/ docum entat ion/# data). Briefly, income data came from the 2014-2018 American Community Survey, urban or rural designation was from the 2019 urbanized area geographies, and two 2019 lists of stores (one including stores authorized to receive SNAP benefits and another list from Trade Dimensions TDLinx) were combined to produce a list of stores with affordable and nutritious food. Approximately one-quarter of participants in this study met criteria for living in a food desert (n = 70, 23.18%), a higher percentage than the US population (12.8%) 58. In order to capture the range of access to healthy and affordable foods, we calculated the degree of food desert severity for each participant relative to others in their census tract by dividing the number of individuals in a tract living more than 0.5 mile (urban areas) or 10 miles (rural areas) from the nearest supermarket, supercenter, or large grocery store) by the total number of individuals living within the tract. Food desert severity index was used as our primary analysis variable. Covariates. As advanced age of birthing parent (defined by age at last menstrual period), increasing parity, and minoritized race or ethnicity are risk factors for poor metabolic health during pregnancy 59,60, these variables were considered as covariates in analyses. Due to their relationships with SES and metabolic health [61][62][63], we also examined alcohol use during pregnancy, as well as the use of prenatal vitamins. See Supplementary Materials for more information. Analytic plan. Structural equation modeling (SEM) was conducted in Mplus (v. 8; Muthen & Muthen, 1998-2021). SEM was selected because it allows us to simultaneously model the complex associations between multiple predictors and multiple outcomes, while controlling for their interdependence. Furthermore, this approach allows us to test for mediation. All data utilized in analyses were first screened by qualified experts, including a trained nutritionist, to assess normality and plausibility. With the exception of percent adiposity during the second trimester, nearly all variables of interest were non-normally distributed (Kolmogorov-Smirnov tests; ps <unk> 0.05). Thus, a robust maximum likelihood estimator was utilized to accommodate for non-normal data distributions using a sandwich estimator. Mediation was tested using the model indirect command in Mplus. As the use of conventional tests of significance are unreliable when testing indirect effects, we utilized the recommended asymmetric confidence intervals based on bootstrapping methods 64 to incorporate the non-normality of the data into the model test statistics. Estimates and accompanying confidence intervals were calculated based on 10,000 bootstrapped samples. Missing data were handled using full information maximum likelihood 65. Model fit was assessed by examining the comparative fit index (CFI; adequate fit was considered at CFI <unk> 0.90), the Tucker Lewis index (TLI; adequate fit was considered at TLI <unk> 0.90), the standardized root mean square residual (SRMR; adequate fit was considered at SRMR <unk> 0.05), and the root mean squared error of the approximation (RMSEA; adequate fit was considered at RMSEA <unk> 0.08) 66,67. Non-independent observations (i.e., participants who were followed across multiple pregnancies; n = 5) were accounted for using the Mplus cluster command. See Supplementary Materials for more information. A stepwise approach to increasing model complexity was utilized: 1. Data reduction Prior to hypothesis testing, a confirmatory factor analysis was used to assess the appropriateness of considering our indicators of SES as a latent variable. 2. Covariate selection Covariates were examined and any that were significantly correlated (p <unk> 0.05) with the indicators of SES, maternal metabolic variables, or food desert severity were included in the more complex models, as described below. 3. Hypotheses 1 and 2: Regression Analyses. To test Hypotheses 1 and 2, a series of regression models were run where: (a) food desert severity was regressed on SES; (b) pregnancy adiposity, glucose concentrations, and DII values were regressed on SES (each pregnancy metabolic variable was considered in its own model); and (c) pregnancy adiposity, glucose concentrations, and DII values were regressed on food desert severity (each pregnancy metabolic variable was considered in its own model). 4. Hypothesis 3: Structural Equation Mediation Model. To test whether food desert severity significantly mediated the relationship between SES and each of the pregnancy metabolic variables, an SEM was estimated in which each of the pregnancy metabolic variables were simultaneously regressed on SES and on food desert severity. In the same model, food desert severity was also regressed on SES. Pregnancy metabolic variables were allowed to covary, allowing for control of their interdependence. Each metabolic variable and food desert severity were regressed on each covariate and SES and each covariate were allowed to covary. To preserve model parsimony, non-significant paths from covariates were removed from the final model. Beta weights and p-values are reported for each of the paths, as well as bootstrapped confidence intervals for the indirect paths. Institutional review board statement. Informed consent statement. Informed consent was obtained from all subjects involved in the study. --- Results --- Sample description. --- Bivariate correlations. Bivariate correlations among the focal variables (Table 2) suggest that higher food desert severity scores were associated with greater adiposity (r = 0.17; p = 0.02) and lower SES (years of education: r = -0.15, p = 0.01). Variables related to lower SES were associated with greater adiposity (total household income: r = -0.20, p = 0.01; reserve savings after debts are paid: r = -0.17, p = 0.04) and with pro-inflammatory dietary intake (total household income: r = -0.15, p = 0.02), but not with glucose concentrations. Adiposity, glucose concentrations, and DII scores were all positively correlated (rs > 0.21; ps <unk> 0.001). Age of birthing parent was positively correlated with variables associated with SES (total household income: r = 0.14; p = 0.02; years of education: r = 0.34; p <unk> 0.001; reserve savings after debts are paid: r = 0.25; p <unk> 0.001) and negatively correlated with DII scores (r = -0.16; p = 0.01). Minoritized racial/ethnicity status was positively correlated with glucose concentrations (r = 0.25; p <unk> 0.001). Multiparity was significantly correlated with total household income (r = -0.53, p <unk> 0.001), pro-inflammatory dietary intake (r = 0.13, p = 0.04), and age (r = 0.16, p = 0.01). Endorsement of alcohol use after pregnancy was known was positively correlated with food desert severity (r = 0.12, p = 0.04), and using prenatal vitamins was positively correlated with years of education (r =.13, p =.03). --- Table 1. Descriptive statistics for study variables. GTT Oral Glucose Tolerance Test (higher scores indicate poorer glucose regulation), DII Dietary Inflammatory Index (higher scores indicate consumption of a more pro-inflammatory diet), LMP last menstrual period. Higher food desert severity scores indicate that a greater percentage of individuals living in a census tract characterized by lower income and lower access to healthy and affordable foods. a Pregravid weight status is based on pregravid body mass index scores (Underweight: below 18.5, Healthy Weight: 18.5-24.9, Overweight: 25.0-29.9, Obese: 30 and greater). b Percent of census tract living in a food desert. c 0 = Primiparous and 1 = Multiparous. d 0 = White, non-Hispanic and 1 = Racial/ethnic minority. Participants self-identified as Asian (n = 24), Black (n = 4), Multiracial (n = 32), Native American (n = 2), Pacific Islander (n = 1), White (n = 234), Other (n = 3), or Prefer not to answer (n = 2). Participants also self-identified as Hispanic (n = 25), non-Hispanic (n = 276), or Prefer not to answer (n = 1). e 0 = Did not use alcohol after pregnancy was known and 1 = Used alcohol after pregnancy was known. f 0 = Did not use prenatal vitamins and 1 = Used prenatal vitamins. g 1 = Less than $5000, 2 = $5000 to $11,999, 3 = $12,000 to $15,999, 4 = $16,000 to $24,999, 5 = $25,000 to $34,999, 6 = $35,000 to $49,999, 7 = $50,000 to $74,999, 8 = $75,000 to $99,999, 9 = $100,000 to $199,999, 10 = $200,000 to $299,999, 11 = $300,000 and greater. h 1 = Less than $500, 2 = $500 to $4999, 3 = $5000 to $9999, 4 = $10,000 to $19,999, 5 = $20,000 to $49,999, 6 = $50,000 to $99,999, 7 = $100,000 to $199,999, 8 = $200,000 to $499,999, 9 = $500,000 and greater. 76, sf 77, and ggmaps 78. Table 2. Bivariate correlations for study variables. GTT Oral Glucose Tolerance Test (higher scores indicate poorer glucose regulation), DII Dietary Inflammatory Index (higher scores indicate consumption of a more pro-inflammatory diet). Higher food desert severity scores indicate that a greater percentage of individuals living in a census tract have lower income and lower access to healthy and affordable foods. a Total household income adjusted for household size. b 1 = Less than $500, 2 = $500 to $4999, 3 = $5000 to $9999, 4 = $10,000 to $19,999, 5 = $20,000 to $49,999, 6 = $50,000 to $99,999, 7 = $100,000 to $199,999, 8 = $200,000 to $499,999, 9 = $500,000 and greater. c Percent of census tract living in a food desert. d 0 = Primiparous and 1 = Multiparous. e 0 = White, non-Hispanic and 1 = Racial/ethnic minority. f 0 = Did not use alcohol after pregnancy was known and 1 = Used alcohol after pregnancy was known. g 0 = Did not use prenatal vitamins and 1 = Used prenatal vitamins. Bolded values indicate significance, p <unk> 0.05. --- SES measurement model. A latent variable for SES was created using the measures described above. All factor loadings were significant (p <unk> 0.0001) and greater than 0.40 in magnitude, confirming that this is an appropriate way to consider these data. The variance of the latent variable was significant (p = 0.02), suggesting inter-individual variability in SES scores. See Table 3 for detailed results of the measurement model. --- Regressions used to test hypotheses 1 and 2. Table 4 reports the results of the regressions used to test Hypotheses 1 and 2. Consistent with the results of the bivariate correlations, lower SES was associated with greater adiposity (<unk> = -0.27, p = 0.016), greater dietary inflammation (<unk> = -0.25, p = 0.005), and greater food desert severity scores (<unk> = -0.20, p = 0.008). Greater food desert severity scores were associated with greater adiposity (<unk> = 0.17, p = 0.013) and glucose concentrations (<unk> = 0.10, p = 0.045). SEM used to test hypothesis 3. Figure 2 displays the results of the SEM used to test Hypothesis 3. Lower SES was associated with higher food desert severity scores (<unk> = -0.20, p = 0.007), increased adiposity (<unk> = -0.23, p = 0.030), and higher DII scores (<unk> = -0.25, p = 0.003). SES was not associated with glucose concentrations (p = 0.765). Higher food desert severity scores were associated with greater adiposity (<unk> = 0.16, p = 0.036), but not with higher glucose concentrations (p = 0.090) or with DII scores (p = 0.848). Food desert severity scores significantly mediated the relationship between the latent SES variable and adiposity (<unk> indirect = -0.03, 95% CI [-0.079, -0.004]); however, they did not mediate the effect of SES on glucose concentrations (95% CI -0.053, 0.001) or DII scores (95% CI = -0.021, 0.037). This model simultaneously tested the relationships between SES and food desert severity and each metabolic variable, adjusting for their interdependence. SES was also associated with age of birthing parent (<unk> = 0.48, p = 0.0001), such that higher SES was associated with older ages. Using alcohol after pregnancy was known was positively associated with food desert severity (<unk> = 0.13, p = 0.026) and prenatal vitamin use predicted higher adiposity (<unk> = 0.15, p = 0.001) and higher SES (<unk> = 0.17, p = 0.048). Racial/ ethnic minority status significantly predicted glucose concentrations (<unk> = 0.22, p = 0.0001), such that individuals from minoritized backgrounds had higher glucose concentrations. Adiposity significantly covaried with glucose concentrations (<unk> = 0.45, p = 0.0001) and DII scores (<unk> = 0.25, p = 0.0001). Glucose concentrations significantly covaried with DII scores (<unk> = 0.22, p = 0.0001). Sensitivity analyses are described in Supplementary Materials. Table 3. Measurement model for SES latent variable. The measurement model for SES was just-identified, meaning that no fit indices were produced, but that model estimates are reliable. a Total household income adjusted for household size. b 1 = Less than $500, 2 = $500 to $4999, 3 = $5000 to $9999, 4 = $10,000 to $19,999, 5 = $20,000 to $49,999, 6 = $50,000 to $99,999, 7 = $100,000 to $199,999, 8 = $200,000 to $499,999, 9 = $500,000 and greater. --- Latent SES variable --- Indicator <unk> <unk> SE 95% CI p Total household income a 0. --- Discussion Metabolic health during pregnancy is an important determinant of health for pregnant people and their infants [1][2][3][4][5][6]8,[10][11][12][13][14][15][16][17]. Low SES is one risk factor for impaired metabolic health during pregnancy 27,28,30, but the mechanisms through which SES influences metabolic health and nutrition during pregnancy remain relatively underexplored. Here, we assessed whether food desert severity mediated the relationship between SES and adiposity, glucose regulation, and inflammatory level of diet during pregnancy. The findings of this study provide overall support for a relationship of both SES and food desert severity on metabolic health during pregnancy. Results suggest that food desert severity is a mechanism by which SES influences adiposity during pregnancy, such that individuals with lower SES tend to experience more food desert severity, which leads to increasing adiposity during pregnancy. Consistent with earlier work 28,[32][33][34], results from main effects models (Table 4) indicated that lower SES was significantly associated with greater adiposity, greater food desert severity, and consumption of a more pro-inflammatory diet. As adiposity, glucose concentrations, and the pro-inflammatory nature of participants' prenatal diets were all positively and significantly correlated, we account for their interdependence through our model design. It is interesting to note that the pro-inflammatory nature of participants' prenatal diets was significantly correlated with glucose concentrations. While dietary quality likely impacts glucose regulation and adiposity during pregnancy, it is also possible that overall metabolic state, informed by GTT results and body composition, may impact diet. For example, an individual at-risk for GDM may alter their diets in response to their GTT results or at the instruction of their physician, particularly if their physician recommends trying to limit pregnancy weight gain (as is often the case for individuals with overweight or obesity) 68. After adjusting for adiposity, glucose concentrations, and dietary inflammation, as well as the interdependence of these variables and for relevant covariates, the effect of SES on food desert severity survived. We also found that higher food desert severity was associated with higher adiposity and poorer glucose regulation during pregnancy (Table 4). The effect of higher food desert severity associated with increasing adiposity during pregnancy survived accounting for the effect of SES, the interdependence of metabolic health measures, and relevant covariates (Fig. 2). These findings are consistent with work in non-obstetric samples showing that living in a food desert is associated with obesity 19. While living in a food desert is certainly related to SES, our results indicate that they are not equivalent constructs or metrics. For example, it is possible that some individuals living in a food desert may have the financial resources necessary to access healthful foods or vice versa. Results from our study suggest that, after accounting for food desert severity, SES is not independently related Figure 2. Food Desert Significantly Mediates the Relationship between SES and Adiposity. Food desert severity scores significantly mediated the relationship between latent SES and adiposity (ß indirect = -0.03, 95% CI [-0.079, -0.004]); after adjusting for age of birthing parent, parity status, and racial/ethnic minority status, as well as for weeks' gestation at GTT, use of alcohol after pregnancy was known, and use of prenatal vitamins. Food desert severity did not mediate the effect of SES on glucose (95% CI -0.053, 0.001) or DII scores (95% CI -0.021, 0.037). *p <unk> 0.05, **p <unk> 0.01, ***p <unk> 0.001. CFI = 0.965; TLI = 0.947; SRMR = 0.039; RMSEA = 0.024, 90% CI [0.000, 0.049], p = 0.957 51. to glucose concentrations during pregnancy (Fig. 2). Our findings also suggest that living in a food desert is not associated with increased circulating glucose concentrations in response to the GTT, which is consistent with other work suggesting that living in a food desert is not associated with increased GDM risk 69, but in contrast to work suggesting reduced GDM risk in areas with more grocery stores 22. We posit that these discrepancies may be due, in part, because past work has not simultaneously accounted for the interdependent effects of SES and food desert severity. Food desert severity scores significantly mediated the relationship between lower SES and greater adiposity, even after accounting for the pro-inflammatory quality of the participants' diet, their glucose concentrations during pregnancy, and relevant covariates (Fig. 2). This important finding offers a mechanistic explanation of how SES impacts metabolic health during pregnancy. Consistent with our initial hypotheses, lower SES was associated with greater food desert severity scores (likely because SES may influence the types of neighborhoods in which one can afford to live), and food desert severity, in turn, was associated with increased adiposity during pregnancy. This is the first study to our knowledge to simultaneously assess the effects of food desert severity and SES on adiposity, glucose concentrations, and dietary inflammation during pregnancy. Notably, the effect of SES on adiposity via food desert severity was independent of the inflammatory nature of diet, which was not associated with food desert severity in this model. This is in contrast to other work in non-obstetric samples showing that living in a food desert is associated with an unhealthy diet (i.e., typically measured by lower fruit and vegetable consumption) 46. Future research should consider other mechanisms through which food desert severity may impact pregnancy adiposity. For example, individuals living in food deserts may experience chronic stress associated with poverty, which may be independently associated with dysregulated gestational weight gain 39. It is possible that measuring specific micro-or macro-nutrients in the dietary intake of pregnant individuals living food deserts, rather than the overall pro-inflammatory quality of their diets, may be more informative; however, we did not find that either total caloric intake or percent of total calories from fat during the second trimester had a significant relationship with SES or food desert severity (see Supplementary Figs. S5-S6). It is also possible that physical activity contributes to this relationship; however, we did not find that self-reported physical activity during pregnancy (as measured by the Active Living Index, calculated from the Kaiser Physical Activity questionnaire 70 ) explained our findings (See Supplementary Fig. S7). Thus, future work may benefit from targeting other differences in specific nutrient consumption by pregnant populations, as well as from exploring the influence of physical activity on pregnancy adiposity further. We note that our physical activity measure was self-reported physical activity during pregnancy, rather than a more robust objective measurement of physical activity during pregnancy, such as via accelerometry. This study had a number of strengths. To our knowledge, this is the first study to examine whether food desert severity mediates the relationship between low SES and poor metabolic health during pregnancy. Whereas most other studies of food deserts utilize a dichotomous assessment of food desert status 43, in this study, we utilize a continuous measure of food desert severity. The continuous measure of food desert severity allowed for quantification of the linear association between food desert severity and metabolic health and may be useful for drawing inferences relevant to individuals in low-income, low access areas, but who do not meet the threshold for living in a food desert, as well as for individuals who do not reside in food deserts, but have lower access to healthy food. This study was also strengthened by the detailed characterization of participants' diets via repeated 24-h recalls conducted by registered nutritionists (vs. participant report, as has been utilized in most previous research), which may have increased the reliability of the dietary information. Moreover, rather than using selfreported weight gain, we utilized air displacement plethysmography to assess body composition, allowing us to distinguish between the percent of fat and lean mass during pregnancy. Limitations and suggestions for future research. This study also suggests several directions for future study. Results suggest that factors beyond the pro-inflammatory quality of the diet may be responsible for the relationship between higher food desert severity and greater adiposity during pregnancy. Future studies should assess other factors relevant to the metabolic health of those living in food deserts. For example, neighborhoods with low access to healthful and affordable foods may also have poor incorporation of features that encourage physical activity such as sidewalks and greenspaces 71. Neighborhoods with higher crime rates may also be associated with lower physical activity due to safety concerns 72. Physical activity during pregnancy alone is unlikely to explain these findings, however, as analyses conducted with self-reported physical activity as an outcome measure did not show a relationship between physical activity during pregnancy and SES or food desert severity (See Supplementary Fig. S7). Furthermore, food deserts may have more pollution, which can contribute to poor cardiometabolic health 73. Due to our specific interest in mechanisms associated with metabolic health during pregnancy, we focused on adiposity during pregnancy in our primary models (rather than utilizing a measure of pre-gravid BMI, which may also be related to SES and food desert severity as well as health complications before, during, and after pregnancy 74 ). This focus allowed us to temporally align our various measures of metabolic health and to control for their interdependence, an approach that strengthened our ability to make inferences relevant to this unique developmental period. However, it is important to note that the effects of SES and food desert severity are not likely to be specific to pregnancy and may predate pregnancy and/or persist after birth. It remains clear that, while variation in SES and food desert severity may not be specific to understanding metabolic health only during pregnancy, SES and food desert severity contribute unique importance to identifying risk for poor metabolic health during pregnancy, a developmental period with long-term significance for the parent and child. Furthermore, while geospatial coding is a novel means of collecting objective data about the food desert severity, there are some limitations associated with this technique. For example, individuals within the same census tract may have different access to healthful foods, depending on where in the tract they reside. Although food desert severity provides a sense of proximity to healthful foods (and thus potential difficulty in www.nature.com/scientificreports/ obtaining these foods), this is, of course, not deterministic. Future research should explore other factors related to food access. For example, individuals may live in a food desert, but have access to healthful foods via grocery delivery or other means. Future studies may benefit from triangulating access to healthy and affordable foods via multi-modal data collection, including geospatial coding and surveys concerning accessibility to healthful and affordable foods.

Poor metabolic health during pregnancy is associated with health concerns for pregnant individuals and their offspring. Lower socioeconomic status (SES) is one risk factor for poor metabolic health, and may be related to limited access to healthful and affordable foods (e.g., living in a food desert). This study evaluates the respective contributions of SES and food desert severity on metabolic health during pregnancy. The food desert severity of 302 pregnant individuals was determined using the United States Department of Agriculture Food Access Research Atlas. SES was measured using total household income adjusted for household size, years of education, and amount of reserve savings. Information about participants' glucose concentrations one hour following an oral glucose tolerance test during the second trimester was extracted from medical records and percent adiposity during the second trimester was assessed using air displacement plethysmography. Information about participants' nutritional intake during the second trimester was obtained by trained nutritionists via three unannounced 24-h dietary recalls. Structural equation models showed that lower SES predicted higher food desert severity (β = -0.20, p = 0.008) and higher adiposity (β = -0.27, p = 0.016) and consumption of a more pro-inflammatory diet (β = -0.25, p = 0.003) during the second trimester of pregnancy. Higher food desert severity also predicted higher percent adiposity during the second trimester (β = 0.17, p = 0.013). Food desert severity significantly mediated the relationship between lower SES and higher percent adiposity during the second trimester (β indirect = -0.03, 95% CI [-0.079, -0.004]). These findings indicate that access to healthful and affordable foods is a mechanism by which SES contributes to adiposity during pregnancy and may inform interventions intended to improve metabolic health during pregnancy. Metabolic health in pregnancy. Metabolic disorders during pregnancy are associated with serious longterm health complications to the pregnant individual, including cardiometabolic disorders 1,2 , renal disease 3 , and urological conditions 4 . They also predict a host of birth and long-term complications for offspring, including preterm birth 5 , stillbirth and neonatal death 6 , altered birthweights (higher, in the case of gestational diabetes 7,8 and lower, in the case of pre-eclampsia 9 ), and macrosomia 10 , poor metabolic health at birth 11 , long-term obesity 12,13 , metabolic conditions 14,15 , and an increased risk for neurodevelopmental disorders 16,17 . Defining the determinants of metabolic health during pregnancy is critical to improving perinatal and long-term health for pregnant people and their children. The impact of nutrition and living in a food desert on metabolic health. Data from non-obstetric populations suggest that access to healthy and affordable foods is associated with metabolic health. Those living

predate pregnancy and/or persist after birth. It remains clear that, while variation in SES and food desert severity may not be specific to understanding metabolic health only during pregnancy, SES and food desert severity contribute unique importance to identifying risk for poor metabolic health during pregnancy, a developmental period with long-term significance for the parent and child. Furthermore, while geospatial coding is a novel means of collecting objective data about the food desert severity, there are some limitations associated with this technique. For example, individuals within the same census tract may have different access to healthful foods, depending on where in the tract they reside. Although food desert severity provides a sense of proximity to healthful foods (and thus potential difficulty in www.nature.com/scientificreports/ obtaining these foods), this is, of course, not deterministic. Future research should explore other factors related to food access. For example, individuals may live in a food desert, but have access to healthful foods via grocery delivery or other means. Future studies may benefit from triangulating access to healthy and affordable foods via multi-modal data collection, including geospatial coding and surveys concerning accessibility to healthful and affordable foods. Given the growing literature on "food swamps", areas where convenience stores and fastfood restaurants outnumber retailers with healthier options (e.g., grocery stores), the relationship between food swamp severity and pregnancy metabolic health should also be assessed. An additional and important limitation of this study is that the USDA Food Access Research Atlas data from which the food desert measures were calculated utilized data from the 2010 Census of the Population. The pregnancies in our study occurred between 2018 and 2021. It is possible that census tracts may have changed since 2010. Additionally, average incomes in particular census tracts may have changed between 2014 and 2019 and new supermarkets may have been constructed between 2018 and 2019. Thus, areas that were considered food deserts based off of these earlier data sources may no longer be considered food deserts. While these were the most up-to-date UDSA FARA data available at the time of this study, future studies should utilize updated USDA FARA data when it is released. Finally, while our sample characteristics are consistent with the demographics of the Portland, Oregon metropolis, the majority of our participants were white, non-Hispanic, and of higher SES. Future research should validate these findings in a larger more diverse sample that is more representative of the United States population. --- Conclusions The finding that food desert severity significantly mediates the association between low SES and adiposity during pregnancy is an important step toward identifying targets for interventions aimed at improving metabolic health during pregnancy, and, by extension, offspring birth outcomes and long-term health. Identifying food desert severity as a mechanism by which SES impacts adiposity during pregnancy lends important credence for supporting certain interventions aimed at improving metabolic health during pregnancy over others. For example, improving access to healthful and nutritious foods may be a step toward improving metabolic health during pregnancy for groups at increased risk for adiposity (i.e., low SES individuals), which may, in turn, improve offspring outcomes. As individuals with higher percent adiposity during pregnancy may not have exceeded clinical cutoffs for unhealthy pre-gravid BMIs 68, understanding the mechanisms responsible for increased body fat during pregnancy may be a useful means of identifying which individuals may be most liable for metabolic risk during pregnancy. Results indicate a critical need to focus on making healthful and affordable foods accessible in all communities. This can be implemented via micro-, mezzo-, and macro-interventions, which may include providing community gardens or encouraging healthful dietary habits via training/education on food preparation and nutrition (micro), enrolling eligible residents into federal supplemental nutrition assistance programs or encouraging participation in food system planning (mezzo), or enacting legislation and policies to incentivize the addition of full-service grocery stores and supermarkets to underserved communities and improving the quality of food available at existing convenience stores (macro). --- Data availability The data set described in this manuscript will be publicly available on the National Institute of Mental Health (NIMH) Data Archive (https:// nda. nih. gov/) under study #2069, DOI: https:// doi. org/ 10. 15154/ 15287 86. The study page provides instructions on locating the data. --- Author contributions E.K.W., G.S., H.C.G., and E.L.S. conceptualized the study. R.G. provided software expertise and data extraction. A.M.H. supervised the collection of dietary data. J.T.N., H.C.G., and E.L.S. supervised the collection of all other data. E.K.W., H.C.G., and E.L.S. conducted formal analyses. E.K.W. wrote the initial draft. G.S., A.J.M., R.G., A.M.H., O.N., A.Y., J.T.N., H.C.G., and E.L.S. provided edits to subsequent drafts. E.K.W. and A.J.M. created figures for data visualization. J.T.N., H.C.G., and E.L.S., provided project supervision and administration, as well as acquired funding for the project. --- Competing interests The authors declare no competing interests. --- Additional information --- Supplementary Information The online version contains supplementary material available at https:// doi. org/ 10. 1038/ s41598-023-32783-2. Correspondence and requests for materials should be addressed to E.L.S. Reprints and permissions information is available at www.nature.com/reprints. Publisher's note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Poor metabolic health during pregnancy is associated with health concerns for pregnant individuals and their offspring. Lower socioeconomic status (SES) is one risk factor for poor metabolic health, and may be related to limited access to healthful and affordable foods (e.g., living in a food desert). This study evaluates the respective contributions of SES and food desert severity on metabolic health during pregnancy. The food desert severity of 302 pregnant individuals was determined using the United States Department of Agriculture Food Access Research Atlas. SES was measured using total household income adjusted for household size, years of education, and amount of reserve savings. Information about participants' glucose concentrations one hour following an oral glucose tolerance test during the second trimester was extracted from medical records and percent adiposity during the second trimester was assessed using air displacement plethysmography. Information about participants' nutritional intake during the second trimester was obtained by trained nutritionists via three unannounced 24-h dietary recalls. Structural equation models showed that lower SES predicted higher food desert severity (β = -0.20, p = 0.008) and higher adiposity (β = -0.27, p = 0.016) and consumption of a more pro-inflammatory diet (β = -0.25, p = 0.003) during the second trimester of pregnancy. Higher food desert severity also predicted higher percent adiposity during the second trimester (β = 0.17, p = 0.013). Food desert severity significantly mediated the relationship between lower SES and higher percent adiposity during the second trimester (β indirect = -0.03, 95% CI [-0.079, -0.004]). These findings indicate that access to healthful and affordable foods is a mechanism by which SES contributes to adiposity during pregnancy and may inform interventions intended to improve metabolic health during pregnancy. Metabolic health in pregnancy. Metabolic disorders during pregnancy are associated with serious longterm health complications to the pregnant individual, including cardiometabolic disorders 1,2 , renal disease 3 , and urological conditions 4 . They also predict a host of birth and long-term complications for offspring, including preterm birth 5 , stillbirth and neonatal death 6 , altered birthweights (higher, in the case of gestational diabetes 7,8 and lower, in the case of pre-eclampsia 9 ), and macrosomia 10 , poor metabolic health at birth 11 , long-term obesity 12,13 , metabolic conditions 14,15 , and an increased risk for neurodevelopmental disorders 16,17 . Defining the determinants of metabolic health during pregnancy is critical to improving perinatal and long-term health for pregnant people and their children. The impact of nutrition and living in a food desert on metabolic health. Data from non-obstetric populations suggest that access to healthy and affordable foods is associated with metabolic health. Those living

Introduction Fake news has been defined as fabricated information that imitates news media content in form but not in organizational process or intent, which overlaps with other information disorders, such as misinformation-false or misleading information-and disinformation, which is false information that is deliberately disseminated to deceive people [1]. The impact of fake news in social media is a major concern in public health, as it can reduce or increase the effectiveness of programs, campaigns and initiatives aimed at citizens' health, awareness and well-being. The advancements in the methodologies related to social media analysis provide new insights to unveil how citizens share health information and the ways in which fake news influences public health. Social impact in social media (SISM) constitutes a novel methodology in both social media analytics and the evaluation of the social impact of research [2]. This article applies the SISM methodology to the specific case of fake news in health to identify the type of interactions related to the information shared in social media. --- Impact of Fake News on Public Health Fake news concerning health on social media represents a risk to global health. The WHO warned in February 2020 that the COVID-19 outbreak had been accompanied by a massive 'infodemic', or an overabundance of information-some of which was accurate and some of which was not-which made it difficult for people to find reliable sources and trustworthy information when they needed it [3]. The consequences of disinformation overload are the spread of uncertainty, fear, anxiety and racism on a scale not seen in previous epidemics, such as SARS, MERS and Zika. Therefore, the WHO is dedicating tremendous efforts aimed at providing evidence-based information and advice to the population through its social media channels, such as Weibo, Twitter, Facebook, Instagram, LinkedIn and Pinterest, as well as through its website. The MIT Technology Review highlights that social media are not only being used to spread false news and hate messages but are also being used to share important truthful data and solidarity with all those affected by the virus and hate messages [4]. We are in what some have called the second information revolution [5]. The first information revolution began with the spread of the written word through the press. Now, in this second information revolution, a digital transformation is shaping how citizens around the world interact with each other. We are facing an unprecedented global expansion in the ways we share, access and create information that is presented in many forms-one of which is social media. From diverse fields of knowledge linked to health issues, it can be stated that social media can have both a positive and a negative impact on public health [6,7]. On the one hand, the combination of artificial intelligence and big data can help public health providers identify pandemic diseases in real time, improving the coordination of the response of public health systems through information sharing and improving surveillance and diagnostics. Furthermore, research shows that social media information and social media responses are effective strategies to gain feedback on potential public health policy proposals. This positive impact of social media in health has been demonstrated in a recent study about heat protection policy for Australian schools, which, through the analysis of public comments posted on a national Australian Broadcasting Corporation (ABC) website, identifies the themes to support a national heat protection policy for schools [8]. There is evidence of the negative effects of using social media to spread misinformation, which produces harmful consequences to global health and well-being, becoming one of the greatest challenges for public health systems today [5,9]. The most extensively studied topics involving misinformation in health are vaccination, Ebola and the Zika virus, as well as nutrition, cancer, the fluoridation of water and smoking [5,[10][11][12]. Disinformation spread by the anti-vaccine movement has led to episodes regarding vaccination provoking easily preventable disasters, such as the measles epidemic in Washington state (January 2019). The spread of false information also explains a decrease in immunization behaviors with respect to measles-mumps-rubella (MMR) vaccinations, explaining the drop in the demand for this vaccine in the UK and the USA between 1999 and 2000 [13]. Furthermore, research related to the negative impacts linked to the authenticity of social media and identities has increased in recent years [14]. This includes the analysis of the problems surrounding social media messages/posts regarding privacy, posts ending with unintended users, concerns on how to use social media platforms, who to follow and how people portray themselves in an inauthentic manner [14,15]. --- Spreading Fake News on Health in Social Media --- The Context in Which Fake News Is Spread Never before in human history has the role of globalization processes had the impact that it currently has in decision-making processes and societies because of the speed of communication [5]. Globalization also plays a crucial role in the spread of health news, including social media, influencing the way users receive such news [16,17]. In this arena, it is important to highlight that in a globalized world, health content information can be perceived differently depending on the target group or context [18]. Therefore, fake news may reach citizens in different ways, depending on their age, culture, and other factors [19]. Moreover, research shows that social media and related global digital media content influence discourses about professions and how citizens perceive them, including public health professionals [20]. For instance, many public health programs aimed at children and youth have physical education at the core of their initiatives [21,22]. The teaching profession is often portrayed in digital media in relation to unhelpful physical crisis messages [23] or discourses related to bullying in movie scenes. In a similar vein, social media has also been demonstrated to influence the perceptions of adolescent students with regard to their sexual and reproductive health learning [24]. As a result, health professionals may recognize that social media channels, such as Facebook, offer possibilities to support their activities. Research on the role of mass media and messages and dominant discourses that are communicated to the public is an emerging topic of interest in scientific works that requires further investigation. The influence of social media discourses may differ depending on age, culture or gender [19]. For example, young people build their identities, construct knowledge and acquire information from digital media, including social media, beyond formal education and classroom learning, which is an approach resonating with "public pedagogies" [25]. Other authors such as Ulmer [26] argue that mass digital media provides the public an 'entry point' into the debates. The fact is that digital and social media contribute to the strength or undermine the diversity of points of view, influencing the development of specific health public health policies and interventions [27,28]. Such influence of the media has been defined by some authors as the "fourth state" [29]. Taking into account these contextual considerations, it is important to face fake news related to health in social media to support public health policies instead of trying to reverse them. --- Fake News, Health and Social Media In a globalized world, the spread of fake news content on health-related topics in social media and the ways in which it spreads have recently been discussed in depth [10,30]. Misinformation and disinformation-misinformation as inaccuracy and errors and disinformation as a falsehood created on purpose and the spread of it by malicious individuals (human or bots)-gain momentum from the desire to find a solution to a particular disease or illness by patients or their relatives, who inadvertently contribute to spreading misleading information. Globally, the narratives of misinformation are dominated by personal, negative, and opinionated tones, which often induce fear, anxiety, and distrust of institutions. Once misinformation gains acceptance in such circumstances, it is difficult to correct, and the effectiveness of interventions varies according to the personal involvement of each individual and his or her literacy and sociodemographic characteristics [10]. However, other studies have shown that ignorance rarely leads to strong support for a cause. For example, those who most strongly reject the scientific evidence of climate change are also those who believe that they are best informed about the issue. People's pre-existing attitudes often determine their level of belief in misinformation [31]. With respect to globalization processes, evidence suggests that false information spreads globally more pervasively and farther and faster than the truth spreads in social media. In examinations of possible explanations for this global phenomenon, it has been found that novelty is a pervasive component of false rumors, which are significantly more novel than the truth. However, data cannot support the contention that novelty is the only reason, or the main reason, for the spreading of falsehood [13,32]. However, other studies that have focused on the analysis of fake news in social media have reached different conclusions [33,34]. A recent study that analyzed the credibility of sources publishing articles online that may reach global audiences concludes that for the specific case of online health information and content on social media, people are more concerned about the veracity and credibility of the information source and tend to spread less misinformation about health. One possible explanation given by the authors is that people generally do not read health information for entertainment but rather search for information useful to their health or that of people close to them. Furthermore, in these cases, they are less likely to have a pre-existing opinion about health information than are those who share fake news stories about other topics [34]. A similar conclusion came from a fact-checking study of Twitter and Sina Weibo (the most-used social media platform in China), developed 24 hours after the WHO's declaration of the Ebola outbreak as a Public Health Emergency of International Concern in August 2014. In a globalized world, this declaration by the WHO had diverse impacts on the definition of private and public strategies to combat the virus. It contended that only 2% of the posts created on Twitter and Weibo were fake news or disinformation, while the rest were outbreak-related news and scientific health information, mostly coming from news agencies reporting information from public health agencies. This study was able to confirm that these two social media sources contributed to spreading the news of the Ebola outbreak, which was the key message of the WHO [33]. Research on fake news on health in social media covers a variety of channels, including Twitter, Facebook, Reddit, and Weibo [35][36][37]. The analysis of Twitter has gained special attention, and research shows the reach of Twitter codes and the manner in which information spreads on Twitter. This occurs in diverse fields related to public health, from physical education [38] to healthy eating habits or healthy lifestyles [39]. --- Identification of Social Media Interactions as Key to Spreading or Combating Health-Related Fake News Social interaction appears to be the main method of understanding how disinformation or fake news spreads over social media. Different studies have been conducted to identify by who and how health disinformation content is promoted in social media. In the case of Twitter, different types of malicious actors covering both automated accounts (including traditional spambots, social spambots, content polluters, and fake followers) and human users, mainly trolls, have been identified. It is very difficult to detect whether there is a human or a bot behind a profile. However, all of them produce distorting effects that may be critical to messages from public health systems [12]. One of the studies in the case of vaccines identified three types of profiles that had a special probability of spreading vaccine-related disinformation [12]. The first profile is trolls, or Twitter accounts with real people behind them, identified from lists compiled by U.S. authorities; these trolls use the hashtag #VaccinateUS and spread pro-and anti-vaccination messages, often with the apparent aim of encouraging people to believe that the medical community is divided. The second profile, called'sophisticated bots', is artificial intelligence that automatically spreads content via Twitter with the same objective of making people believe that the medical community is divided. The third group of profiles is comprised of content polluters, who use anti-vaccine messages to pique users' curiosity and lead them to click on links, such that every click leads to more income for those behind the website. Some studies have suggested the need to increase social media literacy, provide strategies and instruments to check the reputation, consistency, and evidence of any information, and avoid self-confirmation (based on assumptions or previous unchecked experiences) [40,41]. --- Combating Fake News on Social Media Several approaches have been proposed in recent years to automatically assess credibility in social media. Most of them are based on data-based models, i.e., they use automatic learning techniques to identify misinformation. Based on these techniques, different applications have been developed with different objectives and in different contexts, such as detecting opinion spam on review sites, detecting false news and spam in microblogging, and assessing the credibility of online health information [6,42,43]. These techniques include both human intervention and algorithms to verify the veracity of information across technologies, such as artificial intelligence (AI) and natural language processing (NLP) [44]. Other mechanisms developed as a remedy against fake news on social media are source ratings that can be applied to articles when they are initially published, such as expert ratings (in which expert reviewers fact check articles-the results of which are aggregated to provide a source rating), user article ratings (in which users rate articles-the results of which are aggregated to provide a source rating), and user source ratings (in which users rate the sources themselves) [34]. According to the literature, social media is an interaction context in which misinformation is spread faster, but at the same time, there are interactions focused on health that are evidence based. Furthermore, it is important to highlight that social media users share the social impact of health research. However, less is known about what these interactions are or what type of recommendations we can identify from them. The main aim of this work is to contribute to advancing methods of overcoming misinformation or fake news in health through social media data analysis. In this sense, we have applied the methodology of social impact in social media (SISM) [2] to identify which type of interactions spread misinformation and which type of interactions contribute to overcoming fake news or misinformation related to health. The main research questions are as follows: RQ1) How are social media messages focused on fake health information or misinformation?; RQ2) How are social media messages focused on health evidence with potential or real social impacts?; and RQ3) How do interactions based on health evidence with potential or real social impacts help overcome misinformation or fake health information? The results indicate that messages focused on fake health information are mostly aggressive, those based on evidence of social impact are respectful and transformative, and finally, deliberation contexts promoted in social media overcome false health-related information. --- Materials and Methods The method used and the sample selection are described in detail in the following sections. The method used is social impact in social media (SISM) methodology [2], which combines quantitative and qualitative content analysis of the sample selected considering the contributions of the social impact of the research [45]. According to Elo et al. [46], there are trustworthiness issues in the preparation phases of the data collection method, the sampling strategy and the selection of a suitable unit of analysis according to the way in which the research goals are defined. In this sense, the following sections develop in detail how the sample was selected and how the data collection and extraction were developed. The data analysis is explained in detail in the corresponding section, along with an explanation of the unit of the analysis used. Regarding ethical considerations, the present research adheres to international ethical criteria related to social media data collection and corresponding analyses; in particular, we have followed the ethical guidelines for social media research supported by the Economic and Social Research Council (UK) [47] and Ethics in Social and Humanities Sciences of the European Commission [48]. Furthermore, we have perceived the risk of harm to and conserved the anonymity of users. Additionally, we have read the terms, conditions and legalities of each of the social media channels, and we have used only public information without identifying any user. Likewise, the data were appropriately coded and anonymized to avoid the possibility of traceability. Sets of data have been secured, saved, and stored. The dataset analyzed and the calculations performed are available in the Supplementary Materials (dataset). We cannot share all raw data due to the current terms of the social media channels and the General Data Protection Regulation (GDPR). --- Sampling, Social Media Data Collection and Extraction The first step to develop this study was the selection of a suitable sample of social media channels to collect the data. The social media channels for the analysis are Facebook, Twitter and Reddit, and their selection corresponds to three criteria: (1) relevance of the number of active users in millions according to Statista 2019 data: Facebook (2414), Twitter (330), Reddit (330); (2) availability of public messages; and (3) suitability for online discussion. There are other social platforms with millions of users, but these three have been selected because they are more suitable for our research study. The chosen sample is exploratory and selective and is based on the following criteria. Criteria 1: Selection of suitable searchable keywords. We selected the word "health" as a general topic and the specific keywords "vaccines", "nutrition" and "Ebola". The use of these specific keywords is based on the findings by Wang et al. [10], in which the authors identify vaccines, nutrition, and Ebola as topics with more misinformation in social media. Specifically, we used the hashtags #vaccines, #nutrition and #Ebola to extract Twitter information. In relation to Facebook, we selected two public pages with more audiences in relation to the topic "health". With regard to Reddit, we selected the topic "vaccines" in a community focused on this topic and an "AskScience Ama Series" focused on vaccines. Criteria 2: Data extraction. The data extracted from Twitter contain tweets published under the hashtags selected in the last ten days. In the case of the Facebook page, the data are extracted from the last 100 posts published and the corresponding comments of the two Facebook pages selected. Finally, in Reddit, we selected the comments published in two conversations in two different communities (the AskScience Ama Series focused on vaccines and the vaccines subreddit). Table 1 shows the data collected. --- Data Analysis The strategy for data analysis aims to unveil the nature of interactions focused on misinformation or fake health information and the nature of interactions based on health evidence of potential or real social impacts. To do so, we have designed the following steps and strategy. The unit of the analysis is the full message published by the user (tweets, Facebook posts and Reddit messages (posts and comments)), which means that the information provided in the external links included in the messages is also analyzed. --- Steps Step 1: To identify which tweets (top 100 posts for each hashtag) and Facebook posts (top 20 posts) have received more attention. In the case of Twitter, this identification depends on likes and retweets. In the case of Facebook, this identification depends on likes on the posts of the Facebook pages selected and the public comments with more likes (top 20). In the case of Reddit, this identification depends on the total number of interactions of 4 conversations in the vaccines subreddit and the 100 most-valued comments by the community (the comments are sorted by reader preference), and each comment receives points given by different members of the community on the AskScience Ama Series. For the case of the AskScience Ama Series of one of the conversations, we selected the last 20 comments sorted by reader preference, such that each comment receives points according to the preferences of different members of the Reddit community. Table 2 shows the data selected. Step 2: Development of qualitative content analysis for each message selected (N = 453) (tweets, Facebook posts, Facebook and Reddit comments). Researchers apply a classification of messages according to the codebook (see Table 3) and interactions received. The social impact coverage ratio (SICOR) will be applied for each source of social media selected, which identifies "the percentage of tweets and Facebook posts providing information about potential or actual social impact in relation to the total amount of social media data found" [2]. The elaborated codebook has four categories defined a priori as a result of the literature review performed. The categories classify messages analyzed regarding evidence of social impact, false news, misinformation, opinion, and facts. While the research team performed the first analysis with these four categories, two new categories emerged from the analysis. These two categories were messages that ask for evidence of social impact and messages that contain misinformation but search for dialogue to contrast with the information-both messages search for deliberation. Step 3. In-depth analysis of interactions containing evidence of potential or real social impact. --- Classification Four categories defined a priori (deductive) Code Definition --- ESISM The message (tweet, Facebook posts or Reddit messages) is an example of evidence of the social impact shared in social media. This means that there is evidence of improvement in relation to the topic selected. This evidence could be a potential or real social impact of research results that is linked with societal goals, for instance, the UN Sustainable Development Goals, and that contributes to improving the specific health issue concerned. This information is useful in connecting citizens with trustworthy information. The message offers a link to evidence of contrasting information or sources of the evidence with possible contrasts. --- MISFA The message (tweet, Facebook posts or Reddit messages) is an example of misinformation or fake information in health. Both situations have negative consequences for public health and personal health. It does not offer a contribution in order to contrast but rather presents the message as evidence or trust information, or the source of the information offered is not scientific. --- OPINION The message (tweet, Facebook posts or Reddit messages) is an opinion, and the message is presented as opinion, not as evidence. --- INFO The message (tweet, Facebook posts or Reddit messages) is an event or fact, for instance, news. --- Two codes emerged from the analysis (inductive) --- ESISM D The message (tweet, Facebook posts or Reddit messages) is formulated as a question to ask for evidence of scientific results that ensure social impact or that are a starting point for deliberation. --- MISFA D The message (tweet, Facebook posts or Reddit messages) contains misinformation or fake information, but it contains questions that open dialogue in order to contrast one's own assumptions with the scientific evidence delivered by other persons or to begin a deliberation. --- Interrater Reliability (Kappa) The analysis of social media data collected for the second analysis was conducted following a qualitative content analysis method, in which reliability was based on a peer-reviewed process. The sample was composed of 453 messages. Each message was analyzed to identify whether it contains evidence of potential or real social impact (ESISM = 1), if it was a message of misinformation or fake information in health (MISFA = 3), if it was an opinion (OPINION = 4) or if it was information (INFO = 2). The researchers involved were experts in the social impact of research and fake news. Each researcher was provided with the codebook before starting to code the messages. Once the analysis was finalized, the messages were coded and compared. We used interrater reliability in examining the agreement between the two raters on the assignment of the categories defined using Cohen's kappa. The result obtained was 0.79; considering the interpretation of this number, our level of agreement was almost perfect and, thus, our analysis was reliable. In cases where no agreement was achieved, the raters decided to exclude the results (N = 450). --- Results Before answering the research questions, there are initial steps to determine whether, among the messages selected in the sample, there are more messages and interactions based on misinformation or, on the contrary, there is more evidence of potential or real social impact. For this purpose, we first classified the messages (see Table 4), and second, we calculated the SICOR. In relation to the Twitter sample analyzed, we found that tweets with a higher percentage of ESISM were those that were published under hashtag #Ebola (39%), followed by #nutrition (18%). In the case of #vaccines, a lower percentage of ESISM was found (only 4%), but MISFA had a higher percentage (32%). The higher percentage of three hashtags selected is the node of INFO, and OPINION is higher in the #vaccines hashtag with 29%. In relation to the data analyzed on Facebook, the lower percentage is under the code of MISFA (5% on Facebook page 1 and 3% on Facebook page 2), and the percentage of INFO is higher on Facebook page 1 than on Facebook page 2. However, OPINION is higher on Facebook page 2. The case of ESISM is only present in the case of Facebook page 1 with 15%. In the case of subreddits, we selected examples focused on vaccines because this topic was the most controversial on Twitter. One of the results indicates that the percentage of ESISM (27% and 14%) was higher than the MISFA code (6% and 10%), INFO had the highest percentages in the subreddits (37% and 62%), and OPINION had the second-highest percentages (31% and 14%). If we analyzed the total amount of data collected, we obtained the following SICOR for each social media channel selected, as shown in Table 5. The SICOR calculation [2] is a ratio that calculates the percentage of ESISM found in the full sample selected. In this case, the SICOR is the percentage of tweets with evidence of social impact in relation to all the tweets collected; the same applied to Facebook posts and comments and subreddit comments. As we can see, the selected subreddit comments have a higher percentage of SICOR (23%), followed by tweets (20%) and Facebook posts and comments (8%). In the case of MISFA, the percentage of total amount for each social media channel selected is shown in Table 6. According to the results, the MISFA percentage is higher in tweets (19%), followed by subreddit comments (7%) and Facebook posts and comments (4%). --- Fake Health Information Social Media Messages are Mostly Aggressive Regarding the research questions, the nature of the social media messages focused on false health information is that they are mostly aggressive in the sample analyzed. This result is mainly concerning the messages on vaccines in which the possibility of dialogue does not exist; there is no option, and the messages contained an affirmative closed sentence. First, we have detected hostility to arguments based on science and even defamation of scientists who have contributed to advances in the field of vaccines. One of the examples of fake news in vaccines said the following: "Vaccines have a long history of damaging the brain from day one", followed by "The-so-called Father of Vaccination left his first son brain-damaged by vaccinating him, Jenner was smart enough not to vaccinate his second". This is an example false of information that has negative impacts on the truth. First, Jenner devoted his life to overcoming smallpox. Jenner even freely treated poor people to save them from smallpox. His discovery had a substantial social impact, and his sons died due to tuberculosis, not due to smallpox vaccines. This scientific article explains in detail the contribution of Edward Jenner [49] and concludes how his discovery and the promotion of vaccination facilitated the eradication of smallpox. Thus, spreading false information about this crucial discovery with defamation of real history harms citizens' lives because this lie could damage their health. Another example found that parents should be encouraged to boycott doctors who recommend vaccinating their children. There is an active anti-vaccine movement that is continually sharing this type of message in social media. The negative consequences are that some children who are not vaccinated have contracted diseases that could be avoided, in addition to adults who have done the same. For instance, in 2019, there were 1282 cases of measles in an outbreak in 31 states [50]. This number was the highest number of cases reported in the U.S. since 1992. Most cases occurred among those who were not vaccinated against measles [50]. The negative impact of this type of interaction affects the public health of cities and villages where people decide to follow these anti-scientific arguments. For this reason, these types of messages are also aggressive, as they do not to follow scientific arguments related to health and cause physical damage and disease among children and those who share a common space. Regarding the examples of the type of MISFA, we found some messages that contain false information or misinformation but contain questions to open dialogue that may pose contrasts to one's own assumptions with scientific evidence delivered by other persons or the beginning of deliberation. For instance, one of the examples begins with "I would be interested in a healthy and respectful conversation", saying that he is a "vaccine agnostic", sharing his opinion that he believes there are more risks than benefits. However, he is open to dialogue. This message is based on opinion, not scientific evidence. However, he is honest, saying that it is an opinion, and he is not assuming that he knows the truth; this is a first step towards dialogue. This case reflects another type of message found, people who are influenced by false information but open to having a conversation. Nevertheless, this opinion also has negative consequences, as he stated that his children are not vaccinated and, thus, they are also at risk. --- Potential or Real Social Impact Social Media Messages Are Respectful and Transformative Messages that contain the potential or real social impact of health are respectful and transformative. They deliver quantitative or qualitative evidence of the social impact that contributes to knowing health is being improved. Some of the illustrative examples of this are those published under the Ebola topic. One of the examples analyzed was the impact that the Ebola vaccine was finally approved. Ebola is a health concern, especially in DRC, due to the number of people affected and who die due to this disease. One of the examples shares quantitative evidence of the social impact of this vaccine and congratulates the people that made this result possible "Merck's Ebola vaccine, which has been given to more than 258,000 people in the current outbreak in DRC". Another example of this type of message said that the Ebola vaccine is the best of 2019. This discovery offers hope and optimism for overcoming Ebola in DRC by offering qualitative evidence of the potential social impact of this vaccine. For example, a survivor of Ebola who took part in the vaccine trial was quoted as saying, "I can convince other people in my town that there is a treatment available for Ebola and that they can get better", and a link was added to the WHO article with the full testimony of this survivor [51]. Similarly, another example of qualitative evidence of social impact is delivered by a message that contains a documentary of Ebola through different testimonies, such as Jophet Kasere, who survived Ebola. However, his family did not; he works as a nurse, caring for children whose parents have been infected with the virus. The documentary recorded by Frontline shows how treatment delivered by WHO was improving the health of different members of the community, but at the same, shows how people who were against this international help tried to stop this improvement and leave people at risk [52]. --- In Deliberation Contexts, Messages with Evidence of Social Impact Overcome Fake Information in Health One of the results found is that deliberation contexts in social media promote the possibility of contrasting information and open dialogue based on valid claims. This example has specially been observed in the Reddit conversations analyzed. The social network allows conversations abiding by the rules of the communities. For instance, one of these rules states that conversations should be based on scientific information and not on false information. We have found examples of people with doubts or concerns regarding vaccines using Reddit to share their views and learn. In Reddit, we found MISFA D and ESISM D, because the common goal is to dialogue. For instance, one of the examples found was a conversation initiated by a girl who was not vaccinated. She said "my parents never vaccinated me," and she was concerned about this and her health. Her questions were addressed to the community, seeking help with regard to her situation. She received replies focused on helping her. For instance, she was told to visit her GP in order to receive an appropriate catch-up schedule, and the importance of talking to a doctor was stressed, "this is not something you should be deciding yourself or asking the Internet about; just ask your doctor". The second example selected was a conversation initiated by a person who holds anti-vaccine views but was searching for "some answers (if possible) to vaccine questions". This person affirms that "there has never been a vaccinated Vs. completely unvaccinated study" to extract reliable conclusions about whether it is better to vaccinate or not. This person received replies with evidence of social impact focused on comparative studies between people who were vaccinated and not vaccinated, where those who were vaccinated exhibited better health than those who were not. Direct links to these studies were also provided. Some of the information detailed the following, "German study on lower rates of asthma among the vaccinated", "comparing unvaccinated and vaccinated people who do catch the flu-vaccinated people are protected from the most serious effects, vaccinated versus unvaccinated children: how they fare in first five years of life, Nigerian study of 25 unvaccinated and 25 vaccinated children: one vaccinated child had a mild case of measles. Unvaccinated children: 3 dead, plus 11 non-fatal cases of measles". The second reply selected detailed cases of measles in the U.S. and explained how the number of cases increased due to unvaccinated children, and this was then compared with Romania, providing scientific sources where the data are published. The result of this conversation is that the person who began this conversation read extended replies that were well-argued based on evidence of social impact and official data, replying enthusiastically, "Thanks! I'll read these and think on the issue!" Thus a transformation was possible due to arguments based on evidence of social impact. Another example selected is from the ASK ME SCIENCE conversations; this is a conversation where scientists are available for dialogue with citizens about different topics. In this case, vaccines were used. One of the conversations selected was concern around Andy Wakefield and his research-that is, an ex-physician who became an anti-vaccine activist, among those responsible for purporting a link between vaccines and autism. The clear and overwhelming consensus among scientists is that "His malevolent influence on the vaccine world was terrible, and we have still not fully recovered even though his publications and ethics have been debunked. Because of his paper, millions of people were not vaccinated, and thousands have died. What a legacy to live with". Furthermore, the final example selected was regarding the negative impact of a community that opts to remain vaccinated. This dialogue was started by someone sharing the concept of "herd immunity". This concept details that in a community, there are people who cannot be vaccinated, such as due to allergies or those who are immunocompromised, and they depend on herd immunity to protect them. This was followed by highlighting that "If too many people who could get the vaccine but choose not to, it does not just affect the individual but can compromise others in the community as well". This person explained that in his/her county, a large population chose not to vaccinate and, consequently, there was a measles outbreak; further, an emergency was declared. He/she has a friend who is immunocompromised and needed to stay home for fear of contracting the measles, "it was terrifying and preventable if people who could get the vaccine would choose to do so. It is a choice that does have an impact on others". This conversation opens a dialogue on how our decisions based on false information can have a negative impact on the health of others. Thus, it is crucial to apply the evidence of social impact in collective matters to guarantee, in this case, successful public health. --- Discussion The previous studies reviewed have been useful in clarifying how health information is spread in social media, identifying the positive and negative impacts [6,7]. Regarding the adverse effects of using social media to spread misinformation, there is evidence of the harmful consequences to global health and well-being, becoming one of the

One of the challenges today is to face fake news (false information) in health due to its potential impact on people's lives. This article contributes to a new application of social impact in social media (SISM) methodology. This study focuses on the social impact of the research to identify what type of health information is false and what type of information is evidence of the social impact shared in social media. The analysis of social media includes Reddit, Facebook, and Twitter. This analysis contributes to identifying how interactions in these forms of social media depend on the type of information shared. The results indicate that messages focused on fake health information are mostly aggressive, those based on evidence of social impact are respectful and transformative, and finally, deliberation contexts promoted in social media overcome false information about health. These results contribute to advancing knowledge in overcoming fake health-related news shared in social media.

too many people who could get the vaccine but choose not to, it does not just affect the individual but can compromise others in the community as well". This person explained that in his/her county, a large population chose not to vaccinate and, consequently, there was a measles outbreak; further, an emergency was declared. He/she has a friend who is immunocompromised and needed to stay home for fear of contracting the measles, "it was terrifying and preventable if people who could get the vaccine would choose to do so. It is a choice that does have an impact on others". This conversation opens a dialogue on how our decisions based on false information can have a negative impact on the health of others. Thus, it is crucial to apply the evidence of social impact in collective matters to guarantee, in this case, successful public health. --- Discussion The previous studies reviewed have been useful in clarifying how health information is spread in social media, identifying the positive and negative impacts [6,7]. Regarding the adverse effects of using social media to spread misinformation, there is evidence of the harmful consequences to global health and well-being, becoming one of the most significant challenges for public health systems today [5,9]. Some of the studies have advanced the identification of the types of profiles that spread vaccine-related disinformation [12], and this helps to identify whether the profile that is posting could be a trusted source or not. Our study contributes to advances in the direction of overcoming false information in health through the analysis of how the messages and interactions are based on false health-related information and the transformative dimension of those messages based on evidence of social impact. This identification has made it possible to apply the SISM methodology, which is focused on evidence of social impact. The three social media channels show that there is a public online discussion regarding the object of study. The detailed analysis of the selected sample allowed us to identify deliberation contexts in the three social media channels; for instance, in Reddit, the open conversations encourage people to search for a dialogue based on valid claims. Moreover, in this context, messages based on evidence of social impact overcome false information, even among those with previous anti-vaccine ideas but with an open-minded attitude and respect. However, it is not possible to engage in dialogue with those who have an aggressive position against science. This finding is especially crucial because it allows us to identify whether citizens have access to evidence on social impacts and whether they can share this evidence in conversations in which false information is spread. The evidence of social impact is the vaccine against false health-related information. Future research lines could replicate this analysis in other topics in which false information is damaging. Nevertheless, civil rights movements could also promote these findings to quickly overcome false health-related information that is causing deaths in adverse but avoidable situations. On the basis of the research findings, there are several practical implications and recommendations for public health professionals. First, the results allow public health professionals to determine the type of health information with evidence of social impact that is most shared in social media. Second, the results also contribute to understanding the types of fake news with a stronger presence in social media that can reduce the effectiveness of public health social media campaigns. Third, this knowledge can be useful in the design of strategies in the public health sector to reverse fake news. Fourth, this knowledge can also be useful to narrow efforts to disseminate evidence of social impact in health to deactivate fake news. Finally, this study contributes to identifying discussion forums in which debates are occurring around health information to contribute to the dialogue providing health information with evidence of social impact. --- Conclusions This article demonstrates that SISM is a replicable methodology that has been successfully applied in social media analytics in relation to health and fake news, contributing to the further exploration of the possibilities of this methodology. This study offers the possibility to identify, on the one hand, evidence of social impact shared in social media and, on the other hand, misinformation or fake information related to health. Furthermore, the results show how the interactions in social media depend on the type of information shared or commented upon by diverse actors. The analysis of Twitter, Facebook and Reddit unveils the different types of interactions regarding evidence or fake news, but they all have the common pattern of showing more messages of events or fact-related information about Ebola, nutrition and vaccines. Furthermore, in most cases, the existence of interactions regarding evidence is higher than that of interactions regarding the misinformation of fake information, although the percentage is much higher for the misinformation of fake information than for evidence in the case of Twitter #vaccines. With regard to opinions, the results indicate that they are much more frequent on Facebook and on subreddits than on Twitter. Moreover, the percentage of tweets and Facebook posts providing information about potential or actual social impacts in relation to the total amount of social media data (SICOR) is higher in Tweets and subreddit comments than in Facebook posts and comments. Another relevant finding is that messages focused on false information regarding health are mostly aggressive, and messages based on evidence of social impact are respectful and transformative. Finally, deliberation contexts in social media allow for the transformation of even those who have false information but who are open to dialogue when they participate and access evidence of social impact. The findings provide insights into the way in which public health initiatives can support the presence and interactions of evidence as an effective strategy to combat fake news. Two main recommendations are suggested for public health professionals, among others. On the one hand, we narrow the dissemination strategies to reverse and deactivate fake news regarding health, considering that the percentage of misinformation on fake news is much higher than that observed for Twitter #vaccines. On the other hand, the design of concrete interventions for discussion forums in which health information is discussed (not only shared) can provide health information with evidence of social impact. This research contributes to including citizens' voices into research from a bottom-up approach, in line with the need to support science and social dialogue in relation to public health, including vulnerable groups [53] or the role of patients to overcome barriers to health access [54]. The possibilities of social media analysis have been widely explored in very diverse fields, from gender to digital protests [55,56], and this work contributes to advancing knowledge in social media analysis and fake news in public health. Future investigations can use SISM to analyze the interactions in social media regarding other public health issues to further explore how citizens use and share information. --- Supplementary Materials: The following are available online at http://www.mdpi.com/1660-4601/17/7/2430/s1, Table S1: "Dataset Kappa". --- Conflicts of Interest: The authors declare no conflict of interest.

One of the challenges today is to face fake news (false information) in health due to its potential impact on people's lives. This article contributes to a new application of social impact in social media (SISM) methodology. This study focuses on the social impact of the research to identify what type of health information is false and what type of information is evidence of the social impact shared in social media. The analysis of social media includes Reddit, Facebook, and Twitter. This analysis contributes to identifying how interactions in these forms of social media depend on the type of information shared. The results indicate that messages focused on fake health information are mostly aggressive, those based on evidence of social impact are respectful and transformative, and finally, deliberation contexts promoted in social media overcome false information about health. These results contribute to advancing knowledge in overcoming fake health-related news shared in social media.

Introduction In the last few decades, the rate of remarriage has grown rapidly in China. As illustrated in Figure 1, the annual number of people registering remarriage increased from fewer than 1 million in 1990 to more than 2.8 million in 2010. While remarriages accounted for approximately 8% of all marriage registrations in 1990, nearly 23% of marriage registrations in 2010 involved at least one remarrying spouse (Wang & Zhou, 2010). The importance of remarriage in China lies not only in its numerical increase. Against the backdrop of drastic socio-economic, cultural, and institutional transformation in modern China, the rise of remarriage serves as a kaleidoscope reflecting dynamic changes in the Chinese family and the shifting landscape of intimate relationships. Moreover, the institution and practice of remarriage-deeply embedded in complex family relations-raise a number of issues that have long interested social scientists: care provision, step-parenthood, inheritance, and family finances (Hans, 2008(Hans,, 2009;;Sweeney, 2010). Although a number of studies have been conducted on remarriage in the Chinese context, prior researchers have either focused on population-level trends (Wang & Zhou, 2010) or drawn on small-scale region-specific qualitative evidence (Huang, 2012). In this study, we aim to fill the gaps in existing research by providing an up-to-date, nationally representative individual-level analysis of remarriage in modern China. [Figure 1 about here] The study of remarriage requires the consideration of "linked lives" (Elder & Giele, 2009). The "incomplete" institutionalization of remarriage (cf. Cherlin, 1978) is vividly reflected in the state of existing research on remarriage, which relies heavily on the opportunity-preference framework developed for the study of first marriages (Graaf & Kalmijn, 2003). As individuals make decisions on remarriage in the wake of preceding lifecourse events (Cherlin, 2009), and previous marriages often entail "indisposable" ties with people such as children and in-laws (Koo & Suchindran, 1980), theories based on first marriage are unlikely to fully capture the distinctive features of remarriage. The need to consider family relations is particularly pertinent to the study of remarriage in China, where marriage is largely a family rather than an individual affair (Croll, 1981;Hu, 2016a). As a result, the presence of nuclear and extended family members may play crucial roles in configuring the probability of remarriage. Furthermore, the dynamic social interactions sustained by such family relations may also differ between remarried individuals and those who remain single after marital disruption (Palmer, 1995). In the Chinese family, women and men are also known to assume differentiated gender roles and relate to other family members in distinct ways (Chen & Li, 2014). Therefore, the first objective of this research is to explore the gendered relationship between family relations-both their structure and the dynamic exchange associated with such relations-and remarriage. It is also crucial to differentiate and compare distinctive life-course trajectories that lead to remarriage. Few previous researchers have distinguished between remarriage after divorce and remarriage after widowhood (Huang, 2012;Wang & Zhou, 2010). However, divorce and widowhood may entail drastically different life experiences, emotional contours, and circumstances. Although a small number of researchers have distinguished between remarriage post-divorce and post-widowhood, they have tended to focus on one or the other, not both (Graaf & Kalmijn, 2003;Sweeney, 2010). Only a few have compared post-divorce and post-widowhood remarriage in Western societies (Berntsen & Kravdal, 2012;Hans, Ganong, & Coleman, 2008). Therefore, it is our second objective to explore the similarities and differences in the relationship between family relations and remarriage after divorce and after widowhood in China. Analyzing data from the 2010 wave of the China Family Panel Studies, we examine the roles played by a diverse array of family relations in configuring the probability of remarriage after divorce and after widowhood, as well as how the dynamic care exchange sustained by such family relations differ between remarried and single divorcé(s) or widow(er)s. We further supplement and elaborate on our quantitative findings using qualitative evidence from 13 in-depth semi-structured interviews with respondents who have remained single and respondents who have remarried since divorce or widowhood. --- Background and Theoretical Considerations Changing Context of Remarriage in China Remarriage is not a new phenomenon in China. In the feudal and imperial eras, remarriage was common among widows from less well-off families. Despite the thendominant moral ideal of the "chastity widow," which obliged widows to remain faithful to their late husbands (Waltner, 1981), a lack of inherited assets made remarriage a viable strategy for subsistence for socio-economically disadvantaged women, especially given women's limited participation in the labor force at this time (Huang, 2012). Due to the rarity of divorce in feudal and imperial China, most remarriages followed widowhood rather than divorce (Waltner, 1981). The establishment of a socialist regime in 1949 and China's subsequent socioeconomic reforms brought about considerable and gendered changes in the context of and motivation for remarriage, as well as its sociocultural and symbolic meanings. First, the socialist revolution boosted women's labor-force participation to more than 90% in the socialist era (Hu, 2016a), enabling Chinese wives to enjoy unprecedented economic independence. Therefore, economic deprivation is less likely to encourage remarriage for the sake of subsistence, as it did in pre-socialist China (Holmgren, 1985). Second, the 1950 Marriage Law conferred on Chinese men and women equal rights to marry and divorce according to personal will (Chen & Li, 2014). Third, the "open-door" policy implemented in late 1978 exposed Chinese people to Western ideals of individualism. The resulting prevalence of divorce, particularly in big cities, has created a major path to remarriage, along with widowhood (Wang, 2001). Furthermore, as cultural and legal changes erode the ideal of the "chastity widow", traditional moral imperatives may no longer impede women from remarrying. Indeed, it is widely argued that the growth of individual autonomy in modern China has contributed greatly to the individualistic pursuit of intimate relationships and therefore remarriage (Huang, 2012;Yan, 2003Yan,, 2009)). --- "Linked Lives": Family Relations and Remarriage Given the ongoing influence of family relations on marital decisions, the "individualization" of remarriage remains an empirical question. Existing scholarship has consistently shown that nuclear and extended families continue to play an active role in events such as matchmaking, mate selection, and the sanctioning of marriage in China today (Croll, 1981;Hu, 2016a;To, 2015aTo,, 2015b)). Although prior research has focused on the relationships between individuals and particular family members such as children, parents, and relatives, investigating their respective influence on the probability of remarriage (Hans, 2008;Hans et al., 2008;Koo & Suchindran, 1980), research comparing distinct sets of family relations is scarce. It thus remains unclear whether and how different family members influence individuals' remarriage in distinct ways. Furthermore, family relations are experienced not only through the structural presence or absence of family members, but also through complex and dynamic care, resource and emotional exchange sustained by such structure (Ngan & Wong, 1996). Therefore, it is crucial to provide a systematic assessment of the ways in which the presence or absence of distinct family members may configure the probability of remarriage, and whether and how the patterns of care exchange differ between those who have remarried and those who have remained single since marital disruption. In China, family relations have two major dimensions: intergenerational, between parents and children; and horizontal, between extended family members such as siblings and relatives (Hu, 2016b). As gender roles are highly segmented between men and women in the Chinese family, both dimensions of family relations are gendered (Chen & Li, 2014). Intergenerationally, parents were traditionally considered to be a major impediment to remarriage (Huang, 2012), whose presence was believed to reinforce one's sense of filial obligation as well as the traditional ideal of chastity (Sommer, 1996;Waltner, 1981). This might not still be the case in modern China. In the wake of divorce or widowhood, parents may often serve as the first resort of help and support. In the modern era, the traditional moral ideal of "chastity wife" is being replaced by a rising sense of individualism that locates personal fulfilment in intimate relationships (Yan, 2003(Yan,, 2009)). This concurs with recent evidence that some parents are seen to encourage divorcé(e)s and widow(er)s to seek a fulfilled personal life and old-age security through remarriage (Stacey, 2011). The presence of children was considered as another major impediment to remarriage. Under the influence of patrilineal traditions (Hu, 2016b), minors were often considered to be undesirable "baggage" that may diminish one's chances in the remarriage market, particularly for women (Koo & Suchindran, 1980). In addition, the widespread vilification of stepfamilies for harming children's development has led to an image of remarriage as an option for "selfish" parents (Li, 2009). However, recent research suggests that the rising divorce rates may have contributed to a decline of the negative stigmas attached to stepfamilies in post-reform China, particularly in urban areas (Kuan, 2015). Moreover, single parents, especially those with young children, are increasingly challenged by the soaring costs associated with childcare and child-rearing to consider remarriage (Kuan, 2015). The legitimacy of marriage in modern China is sanctioned by law as well as by the recognition of a broader circle of extended family members (Croll, 1981;Hu, 2016a). Compared with those operating intergenerationally, care and resource exchanges between extended families and relatives are far less intense. However, social recognition from extended families, as a "generalized reinforcer" (Homans, 1958), may play an important symbolic role in legitimizing and reinforcing social norms pertaining to remarriage. Because marriage is considered an integral part of a fulfilled personal life-particularly for womenin modern China (To, 2015b), marital dissolution and widowhood may generate anxiety among extended family members, who may consequently pressure divorcé(e)s and widow(er)s to seek personal fulfilment through remarriage. Therefore, the presence of extended family members is likely to facilitate remarriage. --- Comparing Remarriage Post-Divorce and Post-Widowhood The need to differentiate post-divorce and post-widowhood remarriage is motivated by three major considerations. First, divorce and widowhood represent distinct life-course trajectories (Trivedi, Sareen, & Dhyani, 2009). A comparative analysis may yield important insights into the implications of the two major types of marital disruption for subsequent lifecourse developments. Second, divorce and widowhood tend to occur at distinct stages of the life course. Whereas divorce usually takes place in mid-life in China, widowhood is more likely to occur later in life (Wang & Zhou, 2010). This means divorcé(e)s and widow(er)s may have very different personal concerns over issues such as financial (in)dependence and old-age security and divergent life aspirations for intimacy and marriage. Third, divorcé(e)s and widow(er)s are also likely to have family members who are at different stages of their life course. In turn, this may demand divorcé(e)s and widow(er)s to respond to the needs that are specific to the life-course circumstances of their family members (Elder & Giele, 2009). For instance, due to the absence of state welfare for childcare, single parents with young rather than adult children may particularly view remarriage as a source of care and resources. Although we are prevented by the scarcity of prior research from deriving systematic hypotheses, our key aim is to explore whether and how the interrelations between family relations and remarriage differ between divorcé(e)s and widow(er)s. --- Method In this research, we adopted a two-stage mixed-methods exploratory design. We first analyzed nationally representative survey data to identify the relationship between the structure of family relations and the likelihood of remarriage after divorce and after widowhood, respectively; and to examine the patterns of association between remarriage and care-exchange activities. To further elaborate on our quantitative findings, we drew on indepth semi-structured interviews to explore how individuals make sense of their remarriage considerations. --- Quantitative Data and Sample Our quantitative data were drawn from the adult panel of the 2010 wave of the China Family Panel Studies (CFPS; Xie & Hu, 2014). The CFPS was conducted by the Institute of Social Science Survey at Peking University in collaboration with the Population Studies Center at the University of Michigan. Multistage probability-proportional-to-size sampling was used, with samples weighted in proportion to population structure at the administrative levels of county (or equivalent), village, and household. The survey covered 33,600 adults from 16,000 households from 25 provinces. The response rate was 81.28% at the household level. The 2010 CFPS data are suited to the current research for a number of reasons. First, the survey contained a large nationally representative probability sample of remarried individuals in China (N = 798). Second, the survey is representative of households from the eastern coast to the western hinterland, which is crucial given the considerable regional variation in remarriage rates across China (Wang & Zhou, 2010). Third, the CFPS collected event history data on both marriage trajectories and family relations. To construct our analytical sample, we first excluded 30,382 never-married respondents, those in first marriages and those who had experienced both divorce and widowhood. We then deleted 23 higher-order remarriages and 126 cases with missing values for our variables. We used the mcartest package in Stata to identify any potential biases resulting from our list-wise deletion of cases with missing values. As the Little's test was not statistically significant at the 10% level, the results confirmed that the list-deleted cases were missing completely at random (Li, 2013). Our final analytical sample comprised 3,069 individuals who had experienced either divorce or widowhood, of whom 722 had remarried for the first time. Based on the event history data from the CFPS, we restructured the cross-sectional (wide) dataset into the person-year format. Following the method recommended by Singer and Willet (2003), the data were structured such that the first person-year record represented the year in which the divorce or widowhood took place and the last person-year record represented the year of remarriage for those who had remarried by 2010 or the year of survey (i.e., 2010) for those who had not remarried as of 2010. This restructured (long) dataset contained a total of 33,964 person-year observations, i.e. each respondent was observed for an average of 11.07 years. Table 1 presents descriptive statistics for the analytical sample. [Table 1 about here] --- Measures --- Marital History Based on the respondents' current marital status and marital history, we distinguished between divorcé(e)s who had remained single (0) and remarried divorcé(e)s (1), and between single widow(er)s (0) and remarried widow(er)s (1). In our analytical sample, 830 respondents had previously divorced, of whom 49.2% had remarried; and 2,239 respondents had experienced widowhood, of whom around 14% had remarried. The proportions of divorcé(e)s and widow(er)s in our analytical sample are consistent with evidence from the divorcé(e)s accounted for around 1.4% of the Chinese population (Lu & Wang, 2013). We also derived a variable recording the year of divorce or widowhood. The respondents experienced divorce or widowhood between 1958 and 2010, with the average marital disruption occurring in 1997. On balance, divorce and widowhood occurred when the respondents were 33.37 and 53.67 years old, respectively. The previous marriage lasted an average of 9.02 years for divorcé(e)s and 32.81 years for widow(er)s. The average time between marital disruption and remarriage was 4.03 years for divorcé(e)s and 4.24 years for widow(er)s. --- Structure of Family Relations Based on the respondents' marital history and their children's demographic information (e.g., date of birth and decease), we derived a series of time-varying variables to capture the presence or absence of pre-school children aged 0-6, school children aged 7-17 and adult children, respectively, for each person-year observation. We coded the presence of children as binomial categorical variables, because only a small number of respondents had more than one child in each category at any given time. Our further checks indicated that no respondent had a child with their remarriage spouse prior to their remarriage. As no change was noted in the sibship status of the respondents in our observation window, we captured the presence of sibling(s) using a time-constant dummy variable (89%). Due to the lack of detailed life history data (e.g., time of decease) on parents and extended families, we coded the presence of living parent(s) (30%) and the size of the extended family (M = 3.29, SD = 2.23) as time-constant variables based on the information in 2010. Although the presence of parents and extended family members tends to predate the event of remarriage, we are careful to interpret the results for these variables in terms of association rather than causality due to the use of time-constant proxies. --- Care Exchange Activities We also measured the respondents' provision of care to their children and their reception of care from children in the six months prior to the survey. We defined "care" in a broad sense to include financial assistance, financial management, domesticity, and childcare. We coded the respondents as "active care providers" (23%) if they provided their children with one or more types of care. Similarly, we coded those who received one or more types of care as "active care recipients" (35%). Due to cell-size considerations, we coded care provision and reception as binomial categorical variables because the proportion of respondents who received or provided each specific type of care is relatively low. Unfortunately, the CFPS did not measure care exchange between the respondents and other family members, and only current but not event history data on care exchange were collected. This means our analysis only provides a correlational snapshot of whether and how the patterns of care-exchange activities differed between single and remarried respondents after marital disruption. Collecting and analyzing fine-grained longitudinal data on care exchange should be an important agenda for future research on remarriage. --- Covariates We controlled for the respondents' gender, and 62% of the respondents were female. We included respondents' age as a linear regressor, as preliminary work showed this to be the most parsimonious specification. The respondents ranged between 29 and 82 years old, with a mean age of 60.99 (SD = 14.80); and the divoré(e)s were considerably younger than the widow(er)s. Whereas the remarried divorcé(e)s and single divorcé(e)s were similar in age, the remarried widow(er)s were significantly younger than the single widow(er)s (t = 16.19, p <unk>.001). We controlled for the respondents' years of schooling (M = 3.83, SD = 4.73). As younger cohorts enjoy more educational resources than their predecessors in China and the divorcé(e)s were substantially younger than the widow(er)s, it is not surprising that the divorcé(e)s had received more education than the widow(er)s. We took account of whether a respondent had ever held a formal job for more than 6 months, and 25% of the respondents had never worked. In the absence of event history data on the respondents' income, we necessarily relied on the time-constant measure of individual income in 2010 (in the logarithm form due to its skewed distribution) as a proxy of one's socio-economic status. We also included a time-constant dummy variable distinguishing between residence in a rural area affiliated with a village committee (50%) and residence in an urban area under the jurisdiction of a neighborhood committee in 2010. Due to the lack of data on the respondents' migration history, this time-constant measure may not have captured the respondents' geographical mobility in our observation window. Nevertheless, we also controlled for one's hukou (household registration) origin, using the hukou status of the respondents' parents distinguishing between rural (83%) and urban types, which predates the event of remarriage and is unlikely to change over time. In China, hukou origin plays a prominent role in determining one's family values (Hu, 2016b) and marital mobility (Lui, 2017). Irrespective of one's location of residence in rural or urban areas, hukou origin is a major cause of segregation in the marriage market in contemporary China (Hu, 2016a;Lui, 2017). --- Statistical Analysis In our analytical sample, 19.3% of the cases included more than one respondent from the same household and the households were further nested within provinces. To account for the hierarchical nature of the data and unobserved heterogeneities at the household and province levels, we fitted all models using a multilevel framework with random intercepts at the household and province levels (StataCorp, 2015). First, to examine how the structure of family relations configures the probability of remarriage, we fitted separate multilevel Cox proportional hazards event history models for remarriage post-divorce (Model 1) and post-widowhood (Model 2) using the long person-year dataset. The Cox specification was used because (1) it makes no assumption about the underlying distribution of event time, (2) our key focus is on the predictors rather than the timing of remarriage, and (3) the specification provides a more parsimonious solution given our long window of observation (Singer & Willet, 2003). Our additional tests using the discrete-time specification yielded results consistent to those reported here. In Model A, we included the structure of family relations and all control variables. We then added the interactions between gender and the structure of family relations in Model B. We then nested Model 1 and Model 2 to conduct cross-sample comparison of the regression coefficients. Second, to compare the patterns of care exchange between the single and remarried respondents, we fitted multilevel binomial logistic regression models predicting the odds of care provision to children and care reception from children in 2010, using the wide dataset. The models were first fitted only with the main effects of all variables, and we then included the gender interactions for the care exchange measures. Notably, we used the respondents' remarriage status in 2009 rather than 2010 to ensure the key predictor (i.e., remarriage status) predated the dependent variables (i.e., care exchange). Given the dynamic and time-sensitive nature of care exchange, we are also careful to report and discuss the results in terms of association rather than causality. In both sets of models, the variance inflation factor (VIF) test was conducted and the VIF values were well below the conservative threshold of 2.5 for all our key variables (de Jongh et al., 2015). --- Qualitative Data Collection and Analysis To elaborate on our quantitative findings and explore how individuals make sense of remarriage considerations, we drew on qualitative data from in-depth semi-structured interviews. The interviews were conducted between 2015 and early 2016. Snowball sampling was used to recruit 13 single and remarried divorcé(e)s and widow(er)s with distinct experiences of remarriage from a number of cities in southern China. The interviewees ranged from 42 to 66 years old, with a mean age of 52. Of the 13 interviewees, 10 were female, 5 had experienced widowhood rather than divorce, and 5 had remarried since divorce or widowhood. With an overarching focus on people's considerations and decisions pertaining to remarriage, the interviews explored the respondents' family relations and biographical trajectories since their divorce or widowhood. Lasting for an average of 47 minutes (ranging from 30 to 65 minutes), the interviews were conducted in Chinese, audiorecorded where possible, and transcribed. A combination of selective and axial coding was used to analyze the qualitative data (Yin, 2010). As the main purpose of our qualitative procedure was to illustrate and enrich the interpretation of our quantitative findings, our initial coding of the qualitative data was necessarily guided by the patterns identified in our quantitative analysis. Here we paid close attention to the occurrence of themes that corresponded to our key variables. Axial coding was then used to identify and establish potential relationships between different themes (Yin, 2010). At this stage, we paid particular attention to how the interviewees subjectively experienced and made sense of their family relations when considering remarriage. --- Results Table 2 presents the results from the multilevel Cox proportional hazards regression models predicting the transition into remarriage after divorce and after widowhood, respectively. Coefficients are reported in Table 2. As we report the results, we also present the hazards ratios (HR) in the body of the text, where a value greater than 1 indicates a greater likelihood and a value less than 1 indicates a lower likelihood of having remarried rather than remaining single after divorce or widowhood. Figure 2 depicts the marginal differences in the odds of care provision to children and care reception from children between the single and remarried respondents. [Table 2 and Figure 2 about here] In Model 1A, the results show that there is a positive association between the presence of pre-school children and the probability of remarriage after divorce (HR = 1.88, p <unk>.01). In contrast, the presence of school-age (HR = 0.92, ns) and particularly adult children (HR = 0.47, p <unk>.05) are associated with a reduced likelihood of remarriage after divorce. In Model 1B, the results further reveal the gendered pattern of association between the presence of children and the likelihood of remarriage. Divorced men are particularly more likely to remarry in the presence of pre-school children (HR = 2.53, p <unk>.001; gender difference: p <unk>.10); and divorced women are particularly less likely to remarry in the presence of adult children (HR = 0.20, p <unk>.001; gender difference: p <unk>.10). The presence of school-age children has divergent influences on divorced women and men (gender difference: p <unk>.10): While divorced women are less likely to remarry in the presence of school-age children (HR = 0.55, p <unk>.05), divorced men are more likely to remarry in the presence of school-age children (HR = 1.42, ns). As depicted in Panel B of Figure 2, the association between care exchange and remarriage after divorce is also gendered, as remarried men (Odds Ratio [OR] = 0.93, p <unk>.001) but not remarried women are less likely to receive care from their children, compared with their single counterparts. The quantitative findings are further nuanced by our qualitative interviews, as evident from the following excerpts. I would be very concerned about my daughter if I were to remarry. I watch all these news reports of men harassing their stepdaughters. That's why very few women [with young daughters] remarry. It would be OK if I had a son. Look at Woody Allen and his adopted daughter. How can he touch her? She's his daughter! They married, even. His poor wife. (53, female, divorced, one 24 year old daughter) It's very hard to find a man who can take care of you and your children. You're old already. (45, female, divorced, no children) Citing media portrayals and popular stereotypes, the divorcé(e)s, particularly the females, almost unanimously expressed concern about the adverse effects of stepfamilies on their children's well-being. This echoes the widespread perception of stepfamilies as detrimental to the development of children and young adults (Huang, 2012). Meanwhile, (dependent) children from previous marriages were widely perceived as burdensome and unwelcome "baggage" in the remarriage market. Due to the persistence of patrilineal traditions in China, single mothers with children from previous marriages find it particularly difficult to remarry; in the words of one interviewee, "who wants to foster someone else's bloodline?!" (42, female, divorcée, one 12 year old son). Nevertheless, although our interviewees reiterated their preference as "good mothers" to stay single to protect their children, they also hinted at an inner desire for remarriage, as "life would be so much easier if there were someone to share the [parental] responsibilities" (47, female, divorced, one 21 year old son). However, despite the negative stigmas attached to single parenthood and stepfamilies, remarriage may function as a source of economic support that allows divorced mothers with young children to spend time on childcare rather than paid work, particularly in the absence of state welfare provision (Koo & Suchindran, 1980). Meanwhile, divorced men may view remarriage as a viable source of care for their young children and themselves, as women rather than men are expected to be the major care providers in China (Hu, 2016b). Similar to the results for Model 1A (predicting post-divorce remarriage), the results for Model 2A indicate that the probability of remarriage after widowhood is negatively associated with the presence of school-age (HR = 0.32, p <unk>.001) and adult children (HR = 0.13, p <unk>.001), similarly for women and men as shown in Model 2B. Furthermore, the comparison between Model 1A and Model 2A indicates that the negative association between the probability of remarriage and the presence of school-age and adult children are significantly stronger among widow(er)s than divorcé(e)s (between-model comparisons: p <unk>.05). Different from post-divorce remarriage (between-model comparison: p <unk>.10), the presence of pre-school children is negatively associated with the remarriage of widow(er)s (HR = 0.69, ns). The patterns of care exchange also differ considerably between single and remarried widow(er)s. As depicted in Panel C of Figure 2, remarried widows are less likely to provide care to their children, compared with their single counterparts (OR = 0.82, p <unk>.001). As depicted in Panel D, both remarried widows (OR = 0.77, p <unk>.001) and widowers (OR = 0.86, p <unk>.01) are less likely to receive care from their children, compared with their single counterparts. The findings may have two possible explanations. First, remarriage may be a cause of the observed differences in the patterns of care exchange. As remarried widow(er)s are closely involved in their conjugal lives, they may have less time and space for care exchange with their children. As one's spouse may serve as a source of care provision, the need for care provision from children may also be lower among remarried than single widow(er)s. Second, remarriage may be a consequence of pre-existing care exchange activities. In the absence of welfare legislation, the concern over old-age security constitutes a major driver for remarriage in China (Chen & Li, 2014). Therefore, pre-existing care with children may lessen widow(er)s' need of remarriage. Moreover, elderly widow(er)s in China often help their (working) children with daily chores and look after their grandchildren. The provision of care for family members may anchor care providers both functionally and emotionally in their family relations, and thus decreasing the perceived need of remarriage (Arber & Timonen, 2012). While we are not able to determine the causality between care exchange and remarriage using cross-sectional data, our qualitative evidence reveals the concurrent existence of both mechanisms. In the interviews, some of the widow(er)s expressed their hope to relieve their children of the burden of old-age care provision through remarriage. The other widow(er)s we interviewed were keen to enumerate their various forms of engagement with their offspring, and described a sense of security in the knowledge that they had families "to fall back on" in old age (61, male, widowed, 2 sons and 1 daughter). They said, "Grandma, you're so outdated, you don't know how to play video games." They helped me download apps. They gave me their stylus pen. I said, "Then you won't have one." They said that their dad had given them three [stylus pens], so they were able to give me one. My sons both come back for dinner every night. They live with me. I cook from 5 to 6, and then I am free to watch TV and listen to the radio, and read... I have no time to think about these things [dating]. (66, female, widowed, 3 sons) I follow my daughter wherever she goes. (65, female, widowed, 1 daughter) We observed a positive association between the presence of living parent(s) and the likelihood of remarriage for widow(er)s (HR = 1.97, p <unk>.01) but not divorcé(e)s (betweenmodel comparison: p <unk>.05). The positive association is stronger among widows (HR = 2.92, p <unk>.01) than widowers (HR = 1.06, ns; gender difference: p <unk>.05). Parents are often the first resort for support in the wake of widowhood. Different from Hans, Ganong, and Coleman's (2008) finding in the USA that remarriage is associated with a lower level of attachment to parents, our interview data vividly illustrate that parents' provision of such support often creates a strong sense of emotional attachment and fili

We are grateful for the helpful comments received at the conference. We thank the anonymous reviewers for their constructive suggestions.

cook from 5 to 6, and then I am free to watch TV and listen to the radio, and read... I have no time to think about these things [dating]. (66, female, widowed, 3 sons) I follow my daughter wherever she goes. (65, female, widowed, 1 daughter) We observed a positive association between the presence of living parent(s) and the likelihood of remarriage for widow(er)s (HR = 1.97, p <unk>.01) but not divorcé(e)s (betweenmodel comparison: p <unk>.05). The positive association is stronger among widows (HR = 2.92, p <unk>.01) than widowers (HR = 1.06, ns; gender difference: p <unk>.05). Parents are often the first resort for support in the wake of widowhood. Different from Hans, Ganong, and Coleman's (2008) finding in the USA that remarriage is associated with a lower level of attachment to parents, our interview data vividly illustrate that parents' provision of such support often creates a strong sense of emotional attachment and filial obligation in their children. Our interviewees were particularly vociferous in their gratitude to their parents for being a "reliable" cushion on which to fall back. The resulting feeling of "indebtedness" to one's parents was also a prominent theme in the narratives of the divorcé(e)s. Moving on from marital disruption, however, it is also evident from our interviews that the elderly parents are closely concerned about the old-age security of the widow(er)s (who are elderly themselves), and thus encourage their children to pursue a fulfilled personal life and old-age security through remarriage. We also found a significant positive association between the size of extended family and the likelihood of remarriage for both divorcé(e)s (HR = 1.05, p <unk>.10) and widow(er)s (HR = 1.13, p <unk>.01). These observations may be attributed to the widespread belief in modern China that marriage is crucial to a fulfilled personal life (To, 2015b)-a belief that may be reinforced by extended family members. Indeed, a number of interviewees reported that their singlehood was often questioned during gatherings with siblings and relatives. Extended family members customarily aired their concerns over the well-being of the singletons by encouraging them to "move on" to find happiness in a new relationship, as indicated in the following extract. Before I got together with my current wife, my family and particularly my brothers constantly prodded me to find someone with whom to start over. They said things like, "Look at the mess you are in, you need a woman to sort you out...not to mention that your son needs a mother." They told me to move on [from the divorce], and they went on and on... (41, male, remarried after divorce, one 17 year old son) The results reveal both similarities and differences in the patterns of association between family relations and remarriage after divorce and after widowhood. Particularly, the differences must be considered in relation to the distinct demographic profiles and marital histories of divorcé(e)s and widow(er)s. As shown in Table 2, compared with women, men are less likely to remarry after divorce (HR = 0.71, p <unk>.01), but they are more likely to remarry after widowhood (HR = 1.48, p <unk>.05). Notably, these gender differences are mediated and thus explained by the inclusion of gendered family relations in Models 1B and 2B. The likelihood of post-widowhood remarriage decreases with the age at divorce. A longer previous marriage is negatively associated with the probability of remarriage for widow(er)s (HR = 0.94, p <unk>.001) but not divorcé(e)s. We found that both divorcé(e)s (HR = 0.62, p <unk>.01) and widow(er)s (HR = 0.51, p <unk>.05) of urban hukou origin are less likely to remarry than their counterparts of rural hukou origin. Although socio-economically disadvantaged widows in feudal and imperial China relied on remarriage for subsistence (Holmgren, 1985), a lack of economic activity and income today does not seem to be associated with an increased probability of remarriage. We also found a positive association between one's years of schooling and the probability of remarriage after widowhood (HR = 1.07, p <unk>.001). --- Discussion and Conclusions The increasing rate of remarriage is often considered to reflect broader social changes in modern China. Prior research on the aggregate population trends of divorce, widowhood, and remarriage in Chinese settings has tended to attribute these phenomena to macro social trends such as socio-economic development and the massification of education (Wang, 2001;Wang & Zhou, 2010). Research drawing on nationally representative individual-level data is scarce; most individual-level analysis has focused on remarriage either after divorce or after widowhood, not both. Drawing on data from the 2010 China Family Panel Studies and 13 in-depth qualitative interviews, we address these gaps by exploring the gendered relationship between family relations and remarriage. In particular, we compare divorce and widowhood as two distinct life-course trajectories leading to remarriage. Our findings suggest that remarriage is far from an "individualized" affair. It is widely assumed that the prevalence of remarriage in modern China is caused by a trend of societal individualization (Yan, 2009). However, informed by the perspective of "linked lives" (Elder & Giele, 2009), we find that remarriage is closely associated with "indisposable" ties with children, siblings, parents and extended families. Indeed, we find that the probability of remarriage is closely shaped by the structure of family relations, and that the patterns of care exchange vary considerably by remarriage status. Above and beyond the structural presence or absence of family members, we find that the life-course circumstances of family members matter just as much in that children at distinct life-course stages were seen to configure the probability of remarriage in considerably different ways. In addition, distinct sets of family relations seem to play different roles in configuring the probability of remarriage, and these roles also differ between divorcé(e)s and widow(er)s. We argue and demonstrate that rather than considering remarriage as an (incomplete) institution (cf. Cherlin, 1978Cherlin,, 2009)), it would be useful to conceptualize remarriage as a social practice that responds flexibly to the diverse constellations of life-course circumstances. The relationship between family relations and remarriage may differ between widow(er)s and divorcé(e)s because they cluster at distinct stages of the life course and thus have very different life circumstances. Among divoré(e)s, we find a strong sense of "structural ambivalence" resulting from conflicting ideals and constraints in China's remarriage market. The divorcé(e)s we interviewed-particularly those with young children-expressed a desire to remarry. However, the structural presence of minor and young-adult children was perceived as a formidable constraint on divorcé(e)s' remarriage. This is not only due to the widespread cultural stigma attached to step-parenthood (Li, 2009), but also because the odds are stacked against those with children from previous marriages (i.e., "baggage") in the marriage market (Koo & Suchindran, 1980). Compared with divoré(e)s, the widow(er)s viewed the structure of family relations as an inhibiting "constraint" to a lesser extent; and the remarriage of widow(er)s seems to relate more closely to their "practice" of family relations. As vividly illustrated in our interviews, widow(er)s were seen to actively provide care for their offspring. Such care provision often generates a sense of emotional attachment and fulfillment that seems to lessen the perceived need for fulfillment through intimate relationships. Moreover, care provision might prevent widow(er)s from dating and getting remarried by consuming their available time and resources. The results should also be interpreted with reference to the distinctive institutional features of modern China. Due to the absence of childcare subsidies and social-service provision for single parents, the lack of state legislation for old-age welfare and care, and the persistence of patrilineality, Chinese families are often responsible for cushioning the repercussions of divorce and widowhood. This is consistent with Davis's (2014) finding that Chinese marriages are being privatized from the state's perspective. Paradoxically, however, due to the lack of alternative service and welfare provision, which has obliged individuals to fall back on their families in the wake of divorce and widowhood, remarriage and its considerations have become no less "private" from the individual's perspective. In other words, complex family relations continue to play a prominent role in shaping the institution of remarriage. Therefore, the state's privatization of the marital institution seems to reinforce the "familialization" of remarriages in China, which embeds remarriage practices further in the traditional familial system instead of de-anchoring it from traditional moorings as argued by individualization theorists (e.g., Yan, 2009). The limitations of this research suggest several important directions for future research. First, the quantitative results yielded from the analysis of cross-sectional data should be interpreted in terms of association rather than causality. Although the respondents are likely to provide accurate information on life events as important as marital disruption and remarriage, it is possible that the event history data may be susceptible to recalling errors. In future research, longitudinal analysis will be crucial to unpack the life-course dynamics of remarriage post-divorce and post-widowhood. Second, our qualitative procedure relied on convenience sampling. Therefore, we cannot claim to have obtained representative narratives, simply narratives that shed further light on our quantitative findings. Third, while our analysis took place at the individual level, it is important for future scholars to conduct dyadic analysis to address the tethered lives of remarried couples. Furthermore, given the rapid sociocultural and institutional changes ongoing in China, it will be crucial for future researchers to examine the over-time developments in the dynamics of family relations and remarriage. 6,638 27,326 Note: Reference category in parenthesis. a Female-male between-slope difference for remarriage after divorce statistically significant at the 10% level or below. b Difference between remarriage post-divorce and remarriage post-widowhood statistically significant at the 5% level or below, based on Model 1A and 2A for control variables. + p <unk>.10, * p <unk>.05, ** p <unk>.01, *** p <unk>.001. Note: Caculations based on three-level logistics regression models with random intercepts at the province and household levels. All models controlled for all variables listed in Table 1. Full regression results available upon request from the authors. Austerix star (*) indicates statistically significant difference beteween post-divorce and post-widowhood remarriage significant at the 5% level or below.

We are grateful for the helpful comments received at the conference. We thank the anonymous reviewers for their constructive suggestions.

Introduction The trafficking of human beings for the purpose of organ removal (THBOR) is not a new phenomenon [1]. With a shortage of legally sourced organs around the world, it is estimated that the illegal trade of human organs generates about 1.5 billion dollars each year from roughly 12,000 illegal transplants [2]. THBOR has serious consequences for human security, particularly for the most vulnerable populations, such as the unemployed, homeless people and migrants. For instance, in 2017, a growing number of organ trafficking cases was uncovered in Lebanon, as Syrian refugees were desperate to support themselves and their families [3]. Since the Declaration of Istanbul on Organ Trafficking and Transplant Tourism, governments and non-governmental organizations (NGOs) have begun to provide rough estimates about THBOR, and yet, little academic attention has been devoted to the study of such a global phenomenon [4]. In 2004, the WHO urged governments to take measures against organ trafficking and protect those who are the most vulnerable. Such efforts culminated with the Declaration of Istanbul on Organ Trafficking and Transplant Tourism, where more than 150 scientific researchers from 78 different countries concluded that commercialized transplants, organ trafficking, and transplant tourism should be prohibited [4]. In this paper, we aim to shed light upon the current state of the literature on: (1) THBOR and (2) its relationship to migrants. We employ a bibliometric analysis with a global sample for the 1990-2019 period, to find the current trends and shortcomings in the literature. In 2018, the International Labor Organization (ILO) asserted that about 40 million people were victims of Human Trafficking-roughly the population of Iraq today [5]. About 90 percent of all detected cases were for sexual exploitation or forced labour purposes [6]. The remaining 10 percent of cases are often lumped together in the "other forms" category-including organ removal. Organ trafficking is a 2 of 11 broad concept that includes several illegal activities, of which the main goal is to profit from human organs and tissue, for the sole purpose of transplantation. These activities include THBOR, transplant tourism and trafficking in organs and tissues [7][8][9]. Despite international and domestic efforts, about 10 percent of all transplants worldwide are believed to be illegal-approximately 12,000 organs per year [10]. While the number of reports on victims of trafficking in people has increased, only 700 victims of THBOR were detected from 25 countries for the 2006-2019 period [6]. Migrants are amongst one of the most vulnerable populations for organ trafficking [11,12]. Many migrants often face poor socio-economic and political conditions in their own countries. However, situations of vulnerability can occur while en route or in host areas, as migrants are exposed to abuse and exploitation by smugglers and opportunists [11]. For instance, in 2015, Pakistani refugees in Lebanon escaping the Syrian civil war were prevented from re-registering as refugees in a second country by the UN. In the words of an organ broker, "those who are not registered as refugees are struggling. What can they do? They are desperate and they have no other means to survive but to sell their organs" [11]. Moreover, a migrant's individual vulnerability can increase due to personal circumstances. Such situations arise when migrants travel with children, elderly, people with disabilities and the like [13]. The previously mentioned organ broker also acknowledged that one of his clients was a 17-year-old boy who left Syria after his father and brothers were killed. He had been in Lebanon for three years with no work and mounting debt, struggling to support his mother and five sisters. "He agreed to sell his right kidney for US$ 8000" [11]. To summarize, THBOR is a global problem with health and human rights consequences for millions of people, especially for those in vulnerable situations, such as migrants. The following research questions are relevant to our study: To what extent is scientific research addressing THBOR? Are migrants included in this research? Where is most of the published output from THBOR and migrants being produced? --- Materials and Methods --- Study Design We employed a bibliometric analysis to accomplish the goals of our study. A bibliometric analysis consists of the application of quantitative analysis techniques to data concerning scientific production, such as journal articles and their accompanying citation counts [14,15]. It is ideal to quantitatively measure the trends, volume, and scope of publications on THBOR. The use of a bibliometric analysis also allows us to find the geographical distribution of knowledge production about THBOR and migrants, as well as to explore the most frequently used methodological approaches in the study of organ trafficking. Therefore, bibliometric analysis constitutes a useful tool to quantitatively explore the evolution and situation of THBOR literature and particularly to find if, and to what extent, migrants, as one of the most vulnerable groups, are being included in publications dealing with such a relevant topic. In conclusion, it provides useful information for researchers and professionals to evaluate publication activity, offering a basis for future research. --- Search Strategy Table 1 presents our search strategy: iterations, keywords, and exclusions. Keyword selection is of great importance in bibliometric analysis, as keywords searched will determine the documents retrieved and, therefore, the findings and results. Search keywords were selected from previous published literature, relevant WHO documents such as resolution 44.25 [16], the WHO Global Glossary of Terms and Definitions on Donation and Transplantation [17], and the Declaration of Istanbul [4]. The second column in Table 1 presents the keywords searched among the topic field tags in Web of Science (WoS). Topic field tags include titles, abstracts, keywords and descriptors of every document. Although the search was not restricted to any language, most of the results come from English language publications. Iterations with specific organs were also used, as some publications do not address THBOR as a general topic, but focus on the trafficking or illegal transplantations of a specific organ or tissue. The specific organs utilized in our search are presented in the first column of Table 1. These keywords were selected for iterations because they are estimated to be the most frequently illegally removed [2]. Finally, excluded keywords were selected to minimize the number of false-positive results. As a result of this iterated search, we obtain a database of 482 publications, which will be used for the following bibliometric analysis. NOT TS = ("cell*" OR "gene" OR "genetic" OR "transport" OR "chemist*" OR "biochemist*" OR "internet" OR "molecule*" OR "DDoS" OR "halogens" OR "hardware" OR "aerosol" OR "single-shell") NOT PY = 2020 Note: symbol "*" includes all possible grammatical endings. --- Source of Information Data were obtained from the Web of Science (WoS) database for the 1990-2019 period-no publications relating to organ trafficking were found prior to 1990. WoS is a global citation search engine whose database concentrates "ideas across disciplines and time from over 1.7 billion cited references from over 159 million records" from leading academic, corporate, and government institutions over 115 years [18]. WoS includes the Science Citation Index Expanded (SCIE), Social Sciences Citation Index (SSCI), Arts and Humanities Citation Index (AHCI), Emerging Sources Citation Index (ESCI), Book Citation Index (BKCI), Conference Proceedings Citation Index (CPCI), and Current Chemical Reactions and Index Chemicus [19][20][21][22][23][24][25][26]. The database was created by Clarivate Analytics, an international, well-known enterprise dedicated to innovation and the delivery of critical data and information for, among others, scientific and academic research [27]. Furthermore, as Web of Science has been indexing every piece of content, and supplies a great variety of filtering options, it is the most reliable database for bibliometric analysis. --- Bibliometric Indicators Bibliometric indicators measure knowledge production about a certain topic (e.g., THBOR). The number of publications and citations are the most useful and widely employed metrics, as they quantitatively measure the volume of research output, and the impact and influence of it, respectively. Our bibliometric indicators are presented as a time-series to show a trend in THBOR research. We are particularly interested in the most productive journals and institutions, as well as their country of origin. This will allow us to analyse the spatial distribution of global output. In line with previous bibliometric analysis literature, we utilize a threshold of ten when presenting our results [28][29][30][31]. These indicators, namely the number of publications and citations, can be employed to further analyse research trends on the topic of THBOR, with the focus on which research approaches are being used, especially on how migrants' health has been included in organ trafficking research. --- Research Approach and Specificity The Web of Science categories show which research field a publication belongs to and which methodological approach was utilized. Bibliometric indicators are used to show the most common fields of study conducting research on THBOR. Previous studies on human trafficking have found health-related approaches to be underrepresented in the literature [25]. However, THBOR itself is a health-related issue. Therefore, we cluster all health-related publications for the following research areas (Web of Science categories): biomedical social sciences, cardiovascular system and cardiology, cell biology, gastroenterology and hepatology, general and internal medicine, genetics and heredity, health care sciences and services, haematology, immunology, infectious diseases, legal medicine, medical ethics, microbiology, neurosciences and neurology, nursing, obstetrics and gynaecology, pathology, pediatrics, pharmacology and pharmacy, psychiatry, psychology, reproductive biology, research and experimental medicine, respiratory system, surgery, transplantation, urology and nephrology. Publications that did not fall under these fields of study were classified as non-health related. An additional analysis of the database metrics was conducted to classify publications depending on their specificity. Some publications focus their THBOR research exclusively on the removal of a specific organ(s) or address the different dynamics between trafficking of different organs. Organ-specific documents are defined as those making the explicit mention of any organ in their title, abstract or keywords (see Table 1 for specific organs utilized in our search). We believe that this specificity will allow us to shed light on what kind of specific organs have researchers devoted most of their attention to, as well as the spatial distribution of where these research specific clusters are taking place across the world. Publications that mention a specific organ are henceforth referred to as "organ-specific publications." Given that one of the main purposes of our study is to analyse THBOR research that addresses migrants, publications were classified depending on their inclusion of migrants. We assume that when migrants are mentioned in the title or abstract of a document, it means that the article takes into account these relevant groups, as they may be the main research objects. Publications studying migrants are defined as those including the following words in the title or abstract: migrant/s, emigrant/s, immigrant/s, migration, emigration, immigration, refugee/s, asylum seeker. --- Results Our results suggest four key findings. First, a lack of research on organ trafficking. Second, the research that addresses the link between THBOR and migrants is staggeringly scarce. Third, most organ trafficking research originates from Western countries. Finally, the bulk of the literature comes from the medical sciences and there is a wide gap within the fields of social science and the humanities. --- Volume of Publications: Temporal Analysis In total, 482 documents about organ trafficking were used in our analysis. Figure 1 shows an increase in the total number of publications pertaining to organ trafficking from 1990 to 2019. Researchers have increasingly paid attention to this issue since the Istanbul Summit and the following publication of the Declaration of Istanbul on Organ Trafficking and Transplant Tourism in 2008. On average, before the Declaration of Istanbul, about 3.7 organ-trafficking-related documents were published per year. After the Declaration, the average increased to 34.7 documents per year. Therefore, we conclude that international attention to organ trafficking has led to a boost in publications about THBOR. The momentum created in 2008 can be also appreciated in the total number of citations received by organ trafficking research studies. Documents published in 2008 received more citations than those published in any other year and per publication (17.5) received double the citations than the overall average (8). The impact and influence of the Declaration of Istanbul of 2008 are clearly evident, but the momentum may have slightly decreased in the last couple of years, as suggested by the slight decrease in the slope of the curve in Figure 1. --- Spatial Distribution The retrieved documents were published in 219 different journals. As Table 2 shows, the most active journal in the field of organ trafficking was the American Journal of Transplantation, with a total number of 45 publications, closely followed by Transplantation, with 41 publications. Almost one in every five articles (18%) on organ trafficking was published in these two journals. However, documents published in The Lancet (30.8) and Current Opinion in Organ Transplantation (21.6) received the highest number of citations per article. As can be inferred from the journals' titles, most active journals belong to different medical research areas. --- Spatial Distribution The retrieved documents were published in 219 different journals. As Table 2 shows, the most active journal in the field of organ trafficking was the American Journal of Transplantation, with a total number of 45 publications, closely followed by Transplantation, with 41 publications. Almost one in every five articles (18%) on organ trafficking was published in these two journals. However, documents published in The Lancet (30.8) and Current Opinion in Organ Transplantation (21.6) received the highest number of citations per article. As can be inferred from the journals' titles, most active journals belong to different medical research areas. A similar pattern unfolds when we look at authors' institutional affiliations. Authors from the documents retrieved were affiliated to a total number of 859 institutions from 75 different countries. Table 2 also shows that Harvard University (52) and the University of California (51) are the leading institutions in THBOR research, when measured in terms of total publications. As was the case with most journals, the leading institutions on organ trafficking are concentrated in Anglo-Saxon countries: four of them in the USA, two of them in Australia, and one in Canada. N is the total number of publications. Figure 2 displays the most influential countries in THBOR research when using the total number of cited papers as a proxy. The geographical distribution illustrates that most of the literature is shaped and produced by Western countries. First, 85% of all cited organ trafficking-related articles are concentrated in the USA and UK. Second, while the rest of the leading countries are not Anglo Saxon, they tend to be economically affluent Western democracies-with the exception of South Africa and Turkey [32]. Although such a trend is not unexpected, we find important to point out that countries in Latin America, Africa and Asia that have the highest number of THBOR victims are underrepresented within the current literature [2]. Ironically, some rich countries that are amongst the primary organ-importers are also shaping the discourse on THBOR, while poor countries-often transit and/or destination areas-are largely underrepresented [2, [19][20][21]. A similar pattern unfolds when we look at authors' institutional affiliations. Authors from the documents retrieved were affiliated to a total number of 859 institutions from 75 different countries. Table 2 also shows that Harvard University (52) and the University of California (51) are the leading institutions in THBOR research, when measured in terms of total publications. As was the case with most journals, the leading institutions on organ trafficking are concentrated in Anglo-Saxon countries: four of them in the USA, two of them in Australia, and one in Canada. Int Figure 2 displays the most influential countries in THBOR research when using the total number of cited papers as a proxy. The geographical distribution illustrates that most of the literature is shaped and produced by Western countries. First, 85% of all cited organ trafficking-related articles are concentrated in the USA and UK. Second, while the rest of the leading countries are not Anglo Saxon, they tend to be economically affluent Western democracies-with the exception of South Africa and Turkey [32]. Although such a trend is not unexpected, we find important to point out that countries in Latin America, Africa and Asia that have the highest number of THBOR victims are underrepresented within the current literature [2]. Ironically, some rich countries that are amongst the primary organ-importers are also shaping the discourse on THBOR, while poor countries-often transit and/or destination areas-are largely underrepresented [2, [19][20][21]. --- Research Areas and Specific Organs Most of the publications in our sample address organ trafficking in broad terms and do not focus on the trafficking of a certain organ or tissue. Only 158 of all the organ trafficking studies retrieved (32.7%) mentioned a specific organ. Figure 3 illustrates the most common organs being addressed in the organ trafficking literature: almost 85% of the publications make references to kidneys, 16% mention liver, and about 6% refer to the heart. This ranking is consistent with the estimates of removals by organ, as most of the frequent illegal transplants involve kidney (67%), liver (22%) and --- Research Areas and Specific Organs Most of the publications in our sample address organ trafficking in broad terms and do not focus on the trafficking of a certain organ or tissue. Only 158 of all the organ trafficking studies retrieved (32.7%) mentioned a specific organ. Figure 3 illustrates the most common organs being addressed in the organ trafficking literature: almost 85% of the publications make references to kidneys, 16% mention liver, and about 6% refer to the heart. This ranking is consistent with the estimates of removals by organ, as most of the frequent illegal transplants involve kidney (67%), liver (22%) and heart (6%). Unexpectedly, publications that cover heart trafficking have a higher impact, receiving 21.2 citations per document, well above the average impact (eight citations per publication). heart (6%). Unexpectedly, publications that cover heart trafficking have a higher impact, receiving 21.2 citations per document, well above the average impact (eight citations per publication). When we consider that 76% of all THBOR research is done in the medical field, it comes as no surprise that eight of the top ten research areas within the THBOR literature belong to the medical field, as Figure 4 illustrates. This is a logical conclusion considering that organ removal/transplants require medical professionals. Nevertheless, it is surprising that such a transnational phenomenon and human rights issue is not being addressed by other fields such as international relations, which comprises only 2% of all publications. When we consider that 76% of all THBOR research is done in the medical field, it comes as no surprise that eight of the top ten research areas within the THBOR literature belong to the medical field, as Figure 4 illustrates. This is a logical conclusion considering that organ removal/transplants require medical professionals. Nevertheless, it is surprising that such a transnational phenomenon and human rights issue is not being addressed by other fields such as international relations, which comprises only 2% of all publications. heart (6%). Unexpectedly, publications that cover heart trafficking have a higher impact, receiving 21.2 citations per document, well above the average impact (eight citations per publication). When we consider that 76% of all THBOR research is done in the medical field, it comes as no surprise that eight of the top ten research areas within the THBOR literature belong to the medical field, as Figure 4 illustrates. This is a logical conclusion considering that organ removal/transplants require medical professionals. Nevertheless, it is surprising that such a transnational phenomenon and human rights issue is not being addressed by other fields such as international relations, which comprises only 2% of all publications. --- THBOR and Migration Is THBOR research addressing migrants? We find that migrants are not receiving proper attention within our main sample: only 13 documents, or 2.7% of 482 publications, address migrants. The omission of migrants from publications can be clearly seen in Figure 5. Migrants are not only neglected from research, but when they are acknowledged in the literature, the research focus is quite different. Publications that address migrants are less organ-specific, with just 21% of them mentioning any specific organ. This is a third less than non-migrant-related documents. Int. J. Environ. Res. Public Health 2020, 17, x 8 of 12 --- THBOR and Migration Is THBOR research addressing migrants? We find that migrants are not receiving proper attention within our main sample: only 13 documents, or 2.7% of 482 publications, address migrants. The omission of migrants from publications can be clearly seen in Figure 5. Migrants are not only neglected from research, but when they are acknowledged in the literature, the research focus is quite different. Publications that address migrants are less organ-specific, with just 21% of them mentioning any specific organ. This is a third less than non-migrant-related documents. The gap becomes wider when we narrow our focus to examine how many of the 13 documents specifically address the health of migrants. Previously, we established that the medical field was the most prolific on THBOR literature-arguably due to its natural relationship to the field of organ transplantation. However, such prolific production abruptly drops when we include migrants in our search: out of all 482 publications retrieved about organ trafficking, only 5 specifically address migrants' health. That is to say, about 1% of the entire THBOR literature has considered the relation between organ trafficking and migration from a medical field-a problem given that migrants are a particularly vulnerable group. --- Discussion The trafficking of human beings for the purpose of organ removal (THBOR) is a highly relevant health and human rights issue that disproportionately impacts migrants. The number of publications on THBOR has been on the rise since the Declaration of Istanbul on Organ Trafficking and Transplant Tourism of 2008, which puts the issue on the agenda of both International Organizations and researchers. However, the momentum created by the Declaration of Istanbul may have been slowing down in the last few years (as shown in Figure 1), and THBOR research has yet to catch up with the literature on other types of human trafficking. Despite the growth of scientific output on organ trafficking over the past decade, this literature comprises only 7% of the bulk of human trafficking literature [25]. The gap becomes wider when we narrow our focus to examine how many of the 13 documents specifically address the health of migrants. Previously, we established that the medical field was the most prolific on THBOR literature-arguably due to its natural relationship to the field of organ transplantation. However, such prolific production abruptly drops when we include migrants in our search: out of all 482 publications retrieved about organ trafficking, only 5 specifically address migrants' health. That is to say, about 1% of the entire THBOR literature has considered the relation between organ trafficking and migration from a medical field-a problem given that migrants are a particularly vulnerable group. --- Discussion The trafficking of human beings for the purpose of organ removal (THBOR) is a highly relevant health and human rights issue that disproportionately impacts migrants. The number of publications on THBOR has been on the rise since the Declaration of Istanbul on Organ Trafficking and Transplant Tourism of 2008, which puts the issue on the agenda of both International Organizations and researchers. However, the momentum created by the Declaration of Istanbul may have been slowing down in the last few years (as shown in Figure 1), and THBOR research has yet to catch up with the literature on other types of human trafficking. Despite the growth of scientific output on organ trafficking over the past decade, this literature comprises only 7% of the bulk of human trafficking literature [25]. Second, although THBOR disproportionately targets vulnerable populations such as migrants, the literature has largely excluded them from the issue. Only 2.7% of the entire THBOR literature in our sample acknowledges any direct or indirect relationship to migrants. This gap leaves the most vulnerable groups out of the academic and policy discourses, as well as the potential creation of any socio-political and healthcare policy agenda to tackle organ trafficking. Similarly, publications that do include migrants in their studies are less organ-specific, which is problematic, given that this is vital to understand the different dynamics of each organ being smuggled. This exclusion is even greater from the perspective of the migrant's health. Despite THBOR being a health issue since it entails the need for surgery to remove the organs, most publications about migrants in the THBOR literature do not approach it from a medical field. The health of migrants is currently being neglected by scientists of all stripes. Future research should address this blind spot in the literature. Such a limited number of publications can be attributed to a combination of factors. First, the differences in terms and definitions (before the Declaration of Istanbul and the WHO Glossary) make it difficult to conduct research consistently and in a homogenous manner. Additionally, the hidden and illegal nature of THBOR poses serious challenges to research on this topic, as it is complicated to gather reliable information-and often when the data is available, it is suppressed for political expediency. The lack of information is especially alarming in the case of quantitative data, imposing severe methodological constraints to organ trafficking research. Future research should further address THBOR in order to close the literature gap with other types of human trafficking research, focusing mostly on increasing reliable quantitative data on a global level. Given the global complexity of THBOR, it is surprising that in most areas of social science research (e.g., international relations), there is limited scientific knowledge pertaining to THBOR being generated. In the last 29 years, just 10 publications have examined organ trafficking from an international relations perspective. The phenomenon of organ trafficking and transplant tourism is a global problem, as trafficking networks are usually transnational-especially those targeting migrants-and often persist along the divide between rich (recipients) and poor (transit or origin) countries. Thus, research on THBOR should address this lack of attention from the angle of international relations and further analyse THBOR in the context of international migration dynamics. When mapping the country of origin of THBOR research, a remarkable bias is found. First, the skew highlights the presence of English-speaking countries, such as the USA, the UK, South Africa and Australia, which are among the most influential countries in terms of the cited work on this topic. Second, most of these countries are also net migration receiver countries. However, origin and transit countries for migration, which suffer more from THBOR, are not the ones publishing research on the topic (except Turkey and South Africa). To be able to address organ trafficking, research should also be promoted, incentivized and funded in developing and transit countries. Research made by these countries can best consider the local and regional dynamics of organ trafficking, as they are usually the victim's country of origin or where the organ removal surgery is performed. --- Limitations of Study While Web of Science is an internationally recognised group of databases which offers reliable information, it also has limitations. The main one is the language bias, which results in English-speaking countries being over-represented. This may explain why China, India, or Germany are not as present as they should be, regarding their number of publications. In addition, unlike many other bibliometric analyses, this paper does not include grey literature, as they are not included in the Web of Science. Thus, these limitations need to be considered when studying the findings of the investigation. Furthermore, the country of origin of the journal is not necessarily the same as the one of the editor or writers. Consequently, the results only show the countries in which more resources are being allocated. Moreover, we also acknowledge that it is possible that some of our retrieved literature could include the migrant population as well, but without mentioning this subgroup-perhaps due to legal/ethical constraints. Nonetheless, the available research so far does not allow for a robust examination of the impact of organ selling on migrants. --- Conclusions The current study is, to the best of our knowledge, the first to assess publication activity relating to the topic of trafficking of human beings for the purpose of organ removal. It has shown the huge quantitative impact the Declaration of Istanbul of 2008 had on stimulating research about organ trafficking. However, the literature on organ trafficking has yet to catch up with research on other typologies of human trafficking that are more established. We conclude that there is a clear need for improving and increasing quantitative data on THBOR, particularly focusing on migrants, who are usually the victims of organ trafficking and need to be included in the published articles. Currently, migrants are being ignored by research on THBOR, and medical research about organ trafficking and migrants is all but completely neglected. Furthermore, while academic publications mostly originate from rich countries, poor and middle-income countries are actually the most vulnerable, because they are either transit or origin areas for most international migrants. Future research should further address the relationship between migration, migrant's health and organ trafficking, in order to develop efficient public policies against organ trafficking and transplant tourism. This can be done by incorporating and advancing scientific output by developing countries. --- Author Contributions: J.G. and I.G. wrote the initial draft. J.G. also undertook data collection and analysis. A.S. contributed as supervisor to the project by undertaking research, data analysis and the revision of article content. All authors have read and agreed to the published version of the manuscript. Funding: This research received no external funding. --- Conflicts of Interest: The authors declare no conflict of interest.

The debate over trafficking of human beings for the purpose of organ removal (THBOR) remains largely absent from policy debates, as its crime is hardly detected, reported and sparsely researched. However, criminal networks continue to exploit vulnerable populations, particularly migrants. To help bridge this gap in knowledge, we employ a bibliometric analysis to examine whether the nexus between organ removal and migration is being addressed by the current academic literature. Our results indicate that (1) research exploring the link between THBOR and migrants is relatively scarce; (2) organ trafficking literature output is largely clustered in a couple of Western countries, and (3) despite the international nature of the topic, most empirical studies on organ trafficking and migration lack representation within the social sciences and humanities. Taken together, our results point to a huge gap on scientific publications between THBOR and migration. Quantitative data is required to lift the current knowledge constraints and better inform policymakers.

Introduction Quality sleep, marked by adequate continuity, duration, and frequency, is essential to human wellbeing. Sleep health is multidimensional, adapted to contextual considerations, and characterized by subjective satisfaction, appropriate timing, adequate duration, high efficiency, and sustained alertness during waking hours (Buysse, 2014). Sleep disturbance, defined as sleep that is inadequate in duration or quality, is correlated with impaired functioning of multiple physiological systems (Orr, Fass, Sundaram, & Scheimann, 2020;Chong et al., 2022). Sleep disturbance and related sleep disorders have been found to correlate with negative mental, behavioral, and physical health outcomes across all age groups (Ohayon, Carskadon, Guilleminault, & Vitiello, 2004). Investigators have identified many affective, cognitive, attentional, and motivational correlates of and sequelae to sleep disturbance including post-traumatic stress disorder (PTSD), depression and anxiety (Zhai, Zhang, & Zhang, 2015;Hertenstein et al., 2019;Freeman et al., 2020;Shalev, Liberzon, & Marmar, 2017). Sleep disturbance is considered a component feature of PTSD, anxiety, and depression (Shalev, Liberson, & Marmar, 2017;Chellappa & Aeschbach, 2022). Investigators have sought to disentangle the temporal relations between sleep disturbance and mental health challenges. Multiple prospective studies in higher-income countries have demonstrated that sleep disturbance, most commonly insomnia, may predict higher incidence and symptomology of subsequent depression, anxiety, and PTSD (Zhai, Zhang, & Zhang, 2015;Cox & Olatunji, 2016). Many cross-sectional studies have been analyzed in attempts to show causality between poor mental health and sleep disturbance (Oh et al., 2019;Du et al., 2020), potentially contributing to confusion regarding the temporal order of measured variables (MacKinnon, Krull, & Lockwood, 2000). A 2021 meta-analysis of randomized control trials in high-income countries show improved sleep quality causally leads to improve mental health -including symptoms of depression and anxiety, and PTSD (Scott et al., 2021). In addition to mental health sequalae of sleep disturbance, poor sleep correlates with impairment to many social processes (Gordon, Carrillo, & Barnes, 2021). Within mostly cross-sectional data, social correlates of poor sleep include increased physiological sensitivity to social-evaluative tasks, worse identification of social cues and social isolation, relationship strain, intimate partner violence victimization, and smaller social networks (Gordon, Carrillo, & Barnes, 2021). While there are few longitudinal studies assessing potential bidirectionality between social relationships and poor sleep quality, poor sleep has been shown to predict feelings of anger and worsened perceived relationship quality (Audigier, Glass, Slotter, & Pantesco, 2023). Nationally representative data from the United States demonstrate that sleep disturbance is more common among people who are socio-economically marginalized and engage less in healthy behaviors (e.g. unhealthy diet patterns, limited exercise, use of cigarettes and alcohol), raising concerns for health equity in promoting sleep quality (Hale, Troxel, & Buysse, 2020). Given the wide-ranging correlates and consequences of sleep disturbance and emerging concerns for health equity in promoting quality sleep, the global health agenda should include sleep quality as a potentially modifiable determinant of global population health (Chattu et al., 2019). Yet, the overwhelming evidence on sleep quality and health has originated from high-income countries and China (Chattu et al., 2019;Cosgrove et al., 2020). Extant research has identified associations between sleep quality and health correlates within low-or middle-income countries that parallel those found in high-income countries (e.g. Stranges et al., 2012). However, research that has been conducted on sleep in low-and middle-income countries has been almost entirely cross-sectional or limited to select clinic populations (Stranges et al., 2012;Simonelli et al., 2018;Wang, Cheng & Xu, 2019). The dearth of longitudinal data on sleep quality, mental health, and social relationships from within sub-Saharan Africa, and other low-or middle-income regions has unfortunately led to continual reliance on findings from western cultures to drive global mental health program and policy planning (Herman et al., 2022). Global health researchers, organizations, and activists have increasingly sounded alarm regarding the real global increase in common mental disorders (Herman et al., 2022). Social support is imperative for mental health, and promotes resilience to various threats to ecological, social, mental, and physical health -further underscoring the need to understand potential determinants of social support (Surkalim et al., 2022;Harandi, Taghinasab, & Nayeri, 2017). Clarifying relationships between sleep disturbance, social support, and mental well-being is essential to promoting global health and development. The estimated burden of mental health disorders within Kenya, and globally, is increasing (GBD 2019Mental Disorders Collaborators, 2022). There is a need for a "whole of society" approach to improving mental health across Kenya and other low-or middle-income countries (Herrman et al., 2022). Given the demonstrated impact of sleep disturbance on mental health within high-income countries, it is essential to understand how sleep disturbance may impact the population mental health within Kenya. --- Study Aim This study seeks to identify temporal relationships between sleep disturbance, social support and common mental disorder symptoms (i.e., symptoms of depression, generalized anxiety, and PTSD) among adults in semi-rural Kenya across two waves of an interventional cohort. Briefly, the intention behind the study intervention was to support the nurturing capacities of families and communities to which street-involved children are reintegrating and to support primary prevention of street-migration of children (Goodman et al., 2023). --- Methods Study design and participant selection. This study analyzes longitudinal interventional cohort data. All participants in the program are 18 years of age or older -though this is not an interventional requirement. The intervention is an active program with open enrollment, and no comparisons are made in this study between intervention participants and non-participants. Participants in this study were selected at random from among participants in a multi-sectoral, multilevel, community-based empowerment program in Meru County, Kenya. The intervention design has been described more completely previously (Goodman et al., 2021). Initial participants are recruited to the program through identification as a known family member of a child living on the streets (Goodman et al., 2023). Index families -the first within a village to join the program -invite neighbors to form an internal microlending group comprised of 25-30 participating families. Subsequent recruitment to the program within a village follows word-of-mouth and is open to anyone interested in joining. While all participants in the present study were intervention participants at the times of data collection, not all participants were "index families" -that is, most participants did not have a child living on the streets at the time of the study. potentially mediated by decreased loneliness and increased access to social resources (Goodman et al., 2021). Additionally, access to social resources available through the intervention moderates associations between prior adversity and future generalized anxiety among participants in the program (Goodman et al., 2022). Random selection process. With the aim of recruiting a representative sample of adult participants from participants, recruitment from the program to this interview questionnaire was conducted by offering each willing participant the chance to select a piece of paper from an opaque bag. The opaque bag contained folded pieces of paper, with "1" representing being selected to this study and "0" representing not being selected to this study. On the first interview day, seven pieces of paper with the number "1" were included in the opaque bag, and an additional "0" was added until there was an even number of willing participants and pieces of paper. Interviews were delivered in the location where the groups met, removed from the remaining participants. Selected survey participants joined the program in late 2022, and completed standardized survey questionnaires in February or March 2023 (T1), and again in April or May 2023 (T2). Inclusion criteria. Inclusion criteria required study participants to be active members of an internal lending group, present at the weekly session from which data were collected, 18 years or older, and willing to engage in the interview. Survey conduct. Each survey was delivered by a local, trained, paid language expert -most of whom were nursing students at a nearby university or recent graduates. Measures. Survey questionnaires utilized previously validated psychometric scales. Primary measures in this analysis included sleep disturbance, depression, anxiety, PTSD, and social support. All primary measures were analyzed as continuous variables. Local language experts worked in separate teams to translate questionnaires into the local language, Kimeru, and back to English. Comparisons between initial and back-translated items were conducted by these two teams, and an additional native English speaker (MG). Items in conflict were resolved by agreement between these teams, and administered to a small Kimeru-speaking sample (n=32) to assess psychometric properties. Items on all psychometric scales were averaged to contribute to a summative measure, following inspection of factor structure and internal reliability. Sleep disturbance was measured using the 8-item PROMIS sleep disturbance scale (Cronbach's alpha (<unk>)=0.84 at T1; <unk>=0.84 at T2; Yu et al., 2012). The scale asks respondents to recall their sleep over the past 7-days and respond to various statements about the quality and quantity of their sleep such as, "My sleep was restless" using a 5-point Likert-type response. The PROMIS sleep disturbance scale previously showed associations with higher prevalence of PTSD following road traffic accidents in an Ethiopia population (Fekadu et al., 2019). Depression was measured using the 21-item Beck's Depression Inventory-II (<unk>=0.81 at T1; <unk>=0.84 at T2; Beck, Steer & Brown, 1996). The BDI-II asks participants to reflect on their current thoughts and feelings covering multiple domains of depression -including negative or flattened affect, loss of appetite, and reduced motivation -on a 4-point response format. The BDI-II has been used and validated in diverse global populations (e.g. Win, Kawakami, & Htet Doe, 2019). Generalized anxiety disorder was measured using the GAD-7 (<unk>=0.86 at T1; <unk>=0.89 at T2; Spitzer, Kroenke, Williams, & Lowe, 2006). The GAD-7 is a 7-item anxiety scale using a 2-week recall period, prompting respondents to report the frequency of experiencing nervousness, worry, difficulty relaxing, and irritability. The GAD-7 has previously been validated in Kenya and other global populations (Goodman et al., 2022;Dhira, Rahman, Sarker, & Mehareen, 2021). PTSD was measured using the Primary Care PTSD Screen for DSM-5 (PC-PTSD-5) (Prins et al., 2016). While developed to screen for PTSD among veterans, the PC-PTSD-5 index has been shown to have excellent diagnostic accuracy among civilian populations (Williamson et al., 2021). Within the present sample, the PC-PTSD-5 showed high internal reliability (Kuder-Richardson (KR)-20=0.76 at T1; KR-20=0.8 at T2). The PC-PTSD-5 screens respondents for the presence of any life-threatening experience over the respondent's lifetime. Country-specific population-based estimates of lifetime exposure to such events range globally from 29% to 83% (Benjet et al., 2016). Respondents who indicate the presence of such an experience are then asked 5-items documenting the presence or absence of PTSD symptoms (e.g. experiential avoidance, related nightmares, being easily startled, experiencing guilt or self-blame, feeling numb or detached) over the previous month. The PC-PTSD-5 has previously been adapted for use in South Africa, Zimbabwe, and Uganda, and found to be associated with forced sex, and risk-taking in these contexts (Webb et al., 2022). Respondents who reported no prior life-threatening traumatic event, or who reported no symptoms from such an event, were considered "0" within this analysis. Social support was measured using the brief 2-way social support scale (SSS) (Obst, Shakespeare-Finch, Krosch, & Rogers, 2019). The brief 2-way SSS assesses the giving and receiving of emotional and instrumental support in the present, with 12 items such as "there is at least one person that I can share most things with" and "I give others a sense of comfort in times of need" (<unk>=0.86 at T1; <unk>=0.89 at T2). Within multiple cultural contexts, this relatively new measure of social support is associated with mental well-being as expected (Liu et al., 2022). Control variables. To control for within-subject variations potentially associated with sleep behaviors, we included measures of wealth, income, marital status, age, and years of formal schooling. Wealth index was recorded as household asset ownership of 12 common items (KR-20: 0.72). Income was recorded as a continuous variable reflecting estimated household monthly income. Marital status was recorded as a binary variable -married / living with someone as though married vs. not married / not living with someone. Age was recorded in years since birth. Formal schooling was measured as an ordinal variable between no school "0" and finishing a tertiary program "14." --- Statistical Analysis Descriptive analyses that included univariate mean (SD) and pairwise Spearman rank sum correlations were conducted of all primary and control variables. The correlation matrix included Bonferroniadjusted significance levels (<unk>=0.05). Wilcoxon matched-pairs signed-rank tests were used to assess the statistical equivalence of all continuous measures between T1 and T2. The primary analysis for this study utilized cross-lagged panel analysis, with a full set of potential pathways between each of the 5 primary, continuous variables (sleep disturbance, depressive symptoms, anxiety symptoms, PTSD symptoms, and social support) at T1 and T2. This model was created using structural equation modeling using a backward modeling process, removing correlations until only correlations under the alpha threshold (<unk>0.2) remained in the model. Control variables were added to all pathways, and further removed until only correlations under the alpha threshold (<unk>0.2) remained. Correlations between variables at T1 and T2 were included, and removed if under the alpha threshold (<unk>0.2). To be considered significant, path coefficients must be under a final alpha threshold (<unk>0.05). SEM correlation standard errors were calculated using robust variance estimators to account for variables with non-normal distributions (Mansournia et al., 2021). Additionally, we presented the standardized SEM coefficients between control and primary variables at T2 to enhance understanding of these relationships. All analyses were conducted using Stata v.16 (StataCorp, 2019). --- Ethical consideration Data were collected following ethical approval from the Institutional Review Boards at the Kenya Methodist University and the University of Texas Medical Branch in Galveston, TX. All participants provided informed written consent prior that emphasized the voluntary and confidential nature of the study to participating in the study. Compensation ($1) was given to participants' microfinance group on behalf of each study participant, following a practice designed by community members interested in ensuring equitable distribution of available resources following a random selection. --- Results The mean (SD) for each variable is shown in Table 1. Eighty-seven percent (87%) of respondents reported experiencing a life-threatening experience during their lifetime (T1). Symptoms of depression, anxiety, and PTSD significantly decreased from T1 to T2 (p<unk>0.01). Social support did not change significantly between the two waves. The mean (sd) age was 41.2 years (13 years). Respondents owned, on average, 5.4 items on the 12-item wealth index, which did not significantly change between the two time points. Mean monthly household income (USD) significantly increased from $34 (T1) to $45 (T2). Mean (sd) years of formal education was 5.2 (3.2), and 75% of respondents were married or living with a partner as though married. The mean (sd) number of weeks between two waves was 11 (0.9). <unk>Insert Table 1 here> Table 2 demonstrates the correlation matrix with significance between primary and control measures. Significant pairwise correlations exist between sleep disturbance (T1) and symptoms of depression (T1; r=0.35, p<unk>0.001; and T2, r=0.29, p<unk>0.001), and anxiety (T1, r=0.21, p<unk>0.01; and T2, r=0.24, p<unk>0.001). Sleep disturbance (T2) is significantly correlated with symptoms of depression (T2, r=0.38, p<unk>0.001), symptoms of anxiety (T2, r=0.29, p<unk>0.001), and social support (T2, r=-0.32, p<unk>0.001). <unk>Insert Table 2 --- here> The cross-lagged panel analysis, displayed in Figure 1 below, indicates that more sleep disturbance at T1 predicts subsequently worse symptoms of depression (T2; r=0.25; p<unk>0.001), anxiety (r = 0.19; p<unk>0.001), PTSD (T2; r=0.14; p=0.01), and less social support (T2; r= -0.14; p<unk>0.01) after controlling for withinvariable and within-time correlations between primary measures. More depressive symptoms (T2) were better predicted by prior symptoms of anxiety (T1; r =0.19; p<unk>0.001) and sleep disturbance (T1; r=0.25; p<unk>0.001) than by previous depressive symptoms (T1, r=0; p=0.9). Anxiety symptoms (T1) also predicted symptoms of PTSD (T2; r=0.16, p<unk>0.001) and later anxiety symptoms (T2, r=0.21; p<unk>0.001). Social support was significantly correlated with fewer symptoms of depression, anxiety and sleep disturbance at T1, and with fewer symptoms of depression, PTSD and less sleep disturbance at T2. Social support was not predictive of any included variable. <unk>Insert Figure 1 here> Table 3 shows the standardized correlation coefficients between included control variables and primary variables (T2). As shown, more years of schooling (reported at T1) predicted less sleep disturbance at T2 (r=-0.1; p<unk>0.05), but no other control variable was statistically significant. Married respondents reported significantly higher social support (T2; r=0.12; p<unk>0.05) and fewer symptoms of depression (T2; r=-0.12; p=0.01). Wealth predicted significantly fewer symptoms of PTSD (T2; r=-0.13; p<unk>0.05). --- Discussion This Unfortunately, we were not able to identify many variables that predicted sleep disturbance in this study, preventing meaningful nuance to subsequent investigations. Higher education attainment was found to predict less subsequent sleep disturbance, though income, wealth, age, marital status, and gender were not. Bivariate correlation analysis between control variables at T1 and sleep disturbance at T2 provides little information that might otherwise nuance these null findings from the full SEM model. Most bivariate relations between control variables (T1) and sleep disturbance (T2) were also nonsignificant. More basic behavioral research is required to understand the cultural and contextual factors contributing to, or potentially improving, sleep disturbance. Prior research in higher-income countries has found stress management, relaxation practice, stimulus control, sleep hygiene, and physical exercise may improve sleep distrubance (Murawski et al., 2018). Urban living has previously shown to be a greater risk for sedentary lifestyles -one common predictor of poor sleep within higher income countries (Martins et al., 2021;Yang, Shin, Li, & An, 2017). In the rural study context, where only 13% of respondents reporting owning motorized transportation (item #12 in the wealth index), sedentary behaviors are less likely to contribute to poor sleep in our study. Pain has been identified as contributing to poor sleep, and was unmeasured within the study data (Finan, Goodin, & Smith, 2013). Sleep disturbance as a predictor of reduced social support is an important finding. Social support and social functioning are essential to human flourishing and a central element within the Flourishing Community interventional model from which these data were collected (Goodman et al., 2023). This finding suggests the need for a deeper understanding of the processes by which social support is developed and maintained within the interventional context and the mechanisms by which sleep disturbance may both undermine, and be alleviated by, these processes. Economic modelling, comparing productive work time between individuals with different durations of average nightly sleep, finds that poor sleep contributes to a 1%-3% reduction in potential national economic productivity in five Organization for Economic Cooperation and Development (OECD) countries (Hafner, Stapanek, Taylor, Troxel, & Van Stolk, 2017). Whether sleep contributes to economic functioning merits research within low-and middle-income countries, and may be mediated by worse psychosocial functioning. While we expected temporal relationships between sleep disturbance and poor mental health, we did not anticipate depressive symptoms (T1) would be less predictive of (T2) depressive symptoms than were symptoms of anxiety, PTSD, or sleep disturbance. This finding suggests, perhaps, that persistent depressive symptoms are mediated by mental and behavioral processes better registered by the sleep-, anxiety-, and PTSD-related measures. Future research should add more waves of measurement, and qualitative investigation with people experiencing persistent depression to better understand operative mechanisms, and potential interventional points to reduce persistent depression. At minimum, these findings suggest improving sleep quality among participants may improve subsequent mental healthpotentially interrupting otherwise persistent mental health challenges. The percentage of respondents reporting any experience of life-threatening trauma over the course of their lives was slightly higher than national prevalence estimates in countries like the United States or South Africa (87% within sample vs. 83% within the United States; 74% in South Africa; Benjet et al., 2015). We find these differences credible, as the sample population was selected due to identification of street-involved children migrating from those villages -and socio-ecological risk factors of streetmigration of children may well increase the trauma risk for other individuals within the same villages. Future studies should adopt or adapt a life events checklist to improve the specificity of life-threatening traumatic histories (e.g. Kwobah et al., 2022). Trauma research may reveal population-level nuances regarding stress responses to traumatic events, the role of sleep in protecting against stress responses, and socio-cultural factors potentially providing resilience to diverse traumatic events. We interpreted responses to validated psychometric scales in many epidemiologic studies as reflecting symptomatic severity rather than diagnosed mental illness, as there are limited trained clinical diagnosticians within the study location -and across Kenya, sub-Saharan Africa and the world (Keynejad, Spagnolo, & Thornicroft, 2021). Thus, while this study is informed by practical constraints within the context, these constraints are found globally. To address the need for improved understanding and the development of public health approaches to community mental health therapeutics,the National Institutes of Health has led the development of an alternative to diagnosticbased approaches to mental health that primarily serve the clinician-patient relationship found within the Diagnostic and Statistical Manual (DSM-V) or ICD-10 diagnostic coding (Cuthbert, 2022). The NIH Research Domain of Criteria (RDoC) posits sleep quality as one essential component of mental and behavioral health (Kozak & Cuthbert, 2016). Our findings support the intent of the RDoC program by showing more sleep disturbance predict relatively worse future mental health symptomologies and less social support -and provide a route to exploring mechanisms to improve population-level mental and social well-being by reducing sleep disturbance. There is increasing recognition for the need to move beyond diagnostic criteria to support individual clinical encounters and towards a public health-enabled approach to mental health, which is consistent with the RDoC program and our aims in this study (Cosgrove et al., 2020). Finally, these data were not collected from a random sample of the population but rather randomly --- Limitations This analysis relies on interviewer-administered self-reported data; using interviewers was necessary to include participants with low or no functional literacy (15% of respondents reported never attending school). Self-reported data are subject to reporting biases such as social desirability bias and recall bias. Respondents, influenced by social desirability bias, would need to indicate worse than average sleep but average or better mental and social well-being at T1 and average sleep but worse than average mental and social well-being at T2 for social desirability bias to contribute to observed correlations. We think it is unlikely social desirability bias would lead to this differential reporting between time points. If social desirability bias leads to differential reporting between time points, it seems just as likely that misclassification could occur in the opposite direction (that is: over-reported sleep quality relative to mental and social health at T1 and under-reported sleep quality relative to mental and social health at T2). We believe findings are robust to influence by social desirability bias, though assert they are not evidence of causation -only of temporal relations controlling for other between-and within-panel correlations. Future research should investigate whether sleep quality is causally related to subsequent mental health as found in other contexts; research in any context is required to establish causal relationships between sleep and social variables. A review comparing reliability of objective measures of sleep to self-reported sleep quality found comparable reliability is lower than one may expect (Cudney et al., 2022)

Sleep quality is essential to biopsychosocial functioning, yet there remains limited longitudinal research on sleep and mental or social well-being within low-or middle-income countries. This study utilizes longitudinal cohort data from a community-based empowerment program in Meru County, Kenya to assess cross-lagged correlations between sleep disturbance, social support, symptoms of depression, anxiety, and posttraumatic stress, Participants (n=373; 92% women; age range 18-86 years) who reported more sleep disturbance at T1 reported significantly more symptoms of depression, anxiety and PTSD, and significantly less social support at T2 (average 11 weeks later), controlling for all within-time correlations across measures, within-measure correlations across time, and sociodemographic background characteristics. Findings are consistent with research across high-income countries, underscoring the need for more contextualized research into sleep behaviors across low-and middle-income countries. Findings may inform interventions to increase mental and social well-being within Kenya.

Drinkaware began as a website set up in 2004 by the Portman Group, an alcohol producer-funded organization which has attempted to influence the evidential content of policy debates through a range of tactics, including attempts to pay academics to write anonymous critiques of World Health Organization (WHO)sponsored evidence reviews [1][2][3]. The Portman Group featured prominently in the previous UK government's 2004 strategy for reducing alcohol-related harms, being responsible for the provision of information on alcohol to the public [4]. This was strongly criticized at the time [5,6]. In 2006, Drinkaware was established as a separate charity 'with the objective of positively changing public behaviour and the national drinking culture to help reduce alcohol misuse and minimise alcohol-related harm' following a memorandum of understanding between the Portman Group, the Department of Health, the Home Office and the devolved administrations for Scotland, Wales and Northern Ireland [7]. Globalization has concentrated alcohol production among a small number of large multi-national companies. The alcohol market was worth US$979 billion in 2007, 40% of which is controlled by just 10 producers [8]. Large corporations invest heavily in a range of activities to foster national and international policy environments which favour their interests [9]. Access to internal tobacco industry documents resulting from US litigation, including those concerning the jointly owned Miller Brewing Company and Phillip Morris [10,11], shows that companies in both industries use corporate social responsibility (CSR) activities to hone their reputations, which in turn helps them to access and influence policy makers [12]. --- SOCIAL ASPECTS/PUBLIC RELATIONS ORGANIZATIONS (SAPROS) Central to the alcohol industry's CSR activities are social aspects/public relations organizations (SAPROs), set up 'to manage issues that may be detrimental to its interests, particularly in areas that overlap with public health' [2]. SAPROs divert attention away from population-level strategies that limit the availability, price and promotion of alcohol, and thus threaten corporate profits, towards those focused on individual responsibility [13]. SAPROs operate in policy and research by disseminating consensus statements and codes of practice [14]. They have grown very rapidly over the last decade, and Drinkaware and the Portman Group are among more than 40 alcohol-related SAPROs now operating in at least 27 countries [1]. There has not been a systematic review of SAPRO activity [15], so we draw upon experience with the Australian SAPRO, Drinkwise, to compare its modus operandum with that of Drinkaware. Drinkwise was established in 2005 by the alcohol industry and funded later by the federal government of Australia in 2006. It describes itself as 'an independent, not-for-profit organisation focused on promoting change towards a healthier and safer drinking culture in Australia'. When given public funding, critics argued that it should advocate evidence-based public health policies [16]. Instead, Drinkwise lobbied the government for ineffectual information programmes (its tagline is 'Get the Facts') while opposing evidence-based policies not in industry interests [17]. In 2009, 57 health experts and scientists wrote to the Medical Journal of Australia opposing further public funding and declaring that they would not accept funding from Drinkwise [18]. Drinkwise responded by writing individually to selected signatories, including two of the present authors, suggesting that the letter was defamatory and implying possible litigation, in the manner of the tobacco industry [19]. --- WHAT DOES DRINKAWARE DO? Drinkaware is not publicly funded, although its activities are very similar to those of Drinkwise. It is 'the mechanism in England for government-industry partnership on public education campaigns' [7]. Its sophisticated multimedia website is promoted widely on alcohol packaging and marketing, although much less prominently than the core content (see Fig. 1). Drinkaware and Drinkwise have similar forms of governance, annual budgets and stated aims. Both have doctors and corporate members on their boards, and claim to provide independent, evidence-based advice to the public, particularly to help individuals make informed decisions about their drinking. Minimum unit pricing (MUP) is a key proposal in recent British alcohol policy that was opposed strongly by sectors of the industry [20]. Aside from the commitment to MUP, the UK Government strategy placed partnership with industry at the heart of policy [21]. The result has been an energetic public debate about the evidence supporting MUP. However, the Drinkaware website, despite being promoted as the place the public should go to 'for the facts' [22], did not acknowledge the existence of any evidence supporting MUP. During the debates which followed the release of the government alcohol strategy it suggested that 'Happy hours would also become slightly less cheerful' and refers wistfully to 'the days of the £10 crate of beer' among other negatively toned, out-dated content-see Box 1. The websitelinked tweets in Box 2 sent by Drinkaware include content that normalizes alcohol use and provides cues to drink on occasions when it may not be planned. For example, there is no British tradition of Halloween parties involving alcohol. Drinkaware claims that its founding memorandum of understanding [7] precludes its involvement in policy issues; yet when MUP was debated in Scotland [23], Drinkaware's written evidence to the 2009-10 Health Select Committee (HSC) alcohol enquiry argued that: 'Behavioural change is a process which cannot happen quickly. The UK drinking culture can be changed if educational initiatives receive sufficient investment over a long enough period' [24]. In the 2012 HSC alcohol enquiry, these claims of non-involvement in policy were repeated (e.g. 'we are proscribed from talking about policy or lobbying') in the face of several examples to the contrary, and incredulity among HSC members (see questions 86-109 in Ev16-18 in [25]). --- CONCERNS ABOUT DRINKAWARE The 2012 HSC [25] noted significant concerns about industry influence on Drinkaware and the content, purposes and value of its activities-see Box 3. A longdelayed review of Drinkaware's effectiveness, which the HSC hoped would address the 'perceived lack of independence' from industry influence, was published early in 2013. The review, undertaken by an 'integrated creative communications agency' [26], was overseen by a fivemember panel including Jeremy Beadles, Corporate Relations Director of Heineken UK and former Chief Executive of the Wine and Spirit Trade Association (WSTA), who oversaw a vigorous WSTA effort to dissuade the Scottish Government from introducing MUP [27,28]. The Drinkaware website also describes as 'independent research' an evaluation of the website by a 'brand and communications research company' who see their role as helping 'our clients build stronger brands through the use of better and more relevant communications' [29]. Both advertising agencies have histories of working with the alcohol industry. The review identifies 'a perception of industry influence resulting in a suspicion that Drinkaware is not truly Eight out of 10 people don't know the correct amount of units recommended in Government guidelines Sir Liam Donaldson proposed that the minimum price for a unit of alcohol should be 50p per unit to curb binge drinking. For example, a 13% bottle of wine containing nine units of alcohol could not be sold for less than £4.50. Cheap supermarket promotions on bulk quantities of alcohol would also get pricier-long gone would be the days of the £10 crate of beer. Happy hours would also become slightly less cheerful. Minimum pricing could bring an end to some drinks promotions in pubs and bars. For instance, a pint of lager with an alcohol content of 5% contains nearly 3 alcohol units, so with a minimum pricing of 50p per unit it couldn't be sold for less than £1.50. What exactly is a unit? So with all this talk about alcohol units, what exactly are they? Unfortunately it's not as simple as one drink equalling one unit. A unit is 10 mL or 8 g of pure alcohol. To put this into context, you would consume one unit of alcohol if you drank a 25 mL single measure of whisky (ABV 40%), or a third a pint of beer (ABV 5-6%) or half a standard (175 mL) glass of red wine (ABV 12%). Many people don't have a realistic idea of how much they're drinking. In fact, Drinkaware's research shows that eight out of 10 people don't know the correct amount of units that are recommended in Government guidelines. It's recommended that men should not regularly drink more than 3-4 units a day and that women should not regularly exceed 2-3 units a day. You can use our interactive Drinks Calculator tool to work out whether you might be drinking above the Government guidelines and also get tips on cutting down your daily intake if the results take you by surprise. --- So is minimum pricing far away? Whether or not the Chief Medical Officer's recommendations to put a minimum pricing of 50p on every unit of alcohol become reality remains to be seen. Gordon Brown's immediate reaction to the proposal was that putting a minimum price on alcohol would bring additional burdens on moderate drinkers. The minimum pricing headlines ensure that the debate around the UK's alcohol culture continues to be in the spotlight. At Drinkaware, our sole aim is to provide people with the information to decide for themselves about what role alcohol plays in their lives. Whether it be helping consumers understand units or allowing them to keep track of their alcohol consumption, we will continue to provide all the facts. *This page was created on 16/03/09, was last updated on 10/05/12 and was accessed on 21/11/12. It has since been removed as part of a major redesign of the website. independent of the alcohol industry' ( [2.8] in [30]) and criticizes a lack of clarity about the mission and purpose of Drinkaware. It finds no evidence of 'undue industry influence' ( [2.32] in [30]). Although recommending the building of an evidence base, it is not clear how industry actors can contribute to this when they promote information alone as sufficient for bringing about behaviour change ( [9.14] in [30]). Both industry actors and the authors of the review treat funding as 'investment' and see the Drinkaware brand as having explicit value for the companies associated with it. For example, complaining of industry bodies that do not provide funding for Drinkaware, one funder suggested: 'too many organisations that get to use the Drinkaware brand, get to benefit from what Drinkaware offers without actually having to put any cash up' ( [12.8] in [30]). --- Box 2 Tweets from Drinkaware --- WHY DOES DRINKAWARE MATTER? Alcohol problems in the United Kingdom continue to increase while they decline in most of Europe [31]. Drinkaware has hitherto avoided concerted public health attention as an alcohol industry body. Its centrality to policy has seemed unremarkable because it merely con-tinues in the public information role fulfilled previously by the Portman Group [4]. Despite the growing problems alcohol causes British health and society, lobbying has successfully positioned the alcohol industry very close to successive UK governments [32]. Henry Ashworth, the Chief Executive of the Portman Group, moved into that role directly from the UK Government Cabinet Office Behavioural Insights Team. Interestingly, a different situation pertains in Scotland, where the implementation of MUP is most advanced. There, the current government is not formed by a Westminster-led party and does not have close relationships with the alcohol industry [23]. It has been observed that the chapter on working with industry in the March 2012 UK Government's Alcohol Strategy 'eschews any enhanced regulatory stance; concrete recommendations are absent... corporate friendly content on the importance of alcohol to the economy, and the need to cut red tape, i.e. deregulation' [21] are emphasized instead. The intent of this material became apparent in late 2012 when the government launched its Consultation on the Implementation of the Strategy [33]. Impact assessments described various options for extending the availability of alcohol to places such as florists and motorway service areas. Early in 2013 it was reported that MUP would not be implemented and industry lobbying was the reason, according to a Conservative Member of Parliament with a health background [34]. --- WHAT TO DO ABOUT DRINKAWARE? Because of high levels of support from successive UK governments, Drinkaware has attracted medical and academic colleagues to support and work with it, and its materials are used in the National Health Service (NHS). In addition to examining Drinkaware's specific activities, the wider economic context needs to be considered. Both Drinkaware and the Portman Group belong to a global network of SAPROs. Corporations are legally required to protect shareholder interests, and any expenditureincluding that on SAPROs or any other CSR activitywill necessarily be servant to this obligation. The alcohol industry has to find a way to reconcile pursuit of profits through increased sales and thus consumption, with the needs of governments to act to reduce the attendant increases in health and societal costs. One possible strategic direction is an aspiration to exercise'soft power' in the form of'subtle forms of steering and control, constraining and limiting the options available for political choice' [35]. This requires a 'post-political' style of partnership in which SAPROs work with governments in order to draw attention away from fundamental conflicts between economic with social and health interests [35]. SAPROs are especially useful as they can claim not to have any economic interests themselves. --- Box 3 2012 Parliamentary Health Select Committee views on Drinkaware [25] 'Chris Sorek stressed that Drinkaware is an independent charity[80], but its role is seen by some as compromised because of its links with the alcohol industry. The British Medical Association told us that The involvement of the Drinkaware Trust in providing public health communications is a significant area of concern. This form of industry social marketing is counterproductive because industry responsibility campaigns are less effective than ones from other sources, keep messages in a commercial comfort zone, and distract attention away from more effective measures to regulate alcohol use. Industry-related messages about alcohol have been found to subtly enhance sales and company reputations. This is despite the fact that the public is cynical about the motives of corporate sponsors, and that nongovernmental organisations make a more effective and credible source.' (Paragraph 94) (page 32) The HSC recommended that: if Drinkaware is to make a significant contribution to education and awareness over the coming years its perceived lack of independence needs to be tackled, and as part of the review that is to be held this year the Committee recommends that further steps are taken to entrench that independence. (Paragraph 97) (page 33) We encourage NHS commissioners, public health practitioners and academic colleagues to reconsider their relationships with Drinkaware. Most would not engage with the Portman Group and we suggest that there is no obvious basis for viewing its offspring, Drinkaware, any differently. Importantly, the review of Drinkaware laments its 'isolation within the alcohol harm reduction community' ( [2.8] in [30]), and Drinkaware is actively seeking to recruit scientists to support it (including the first author). The British public health community should bear in mind WHO's recommendation that alcohol industry bodies only be engaged in their roles as producers, distributors and retailers [36]. Key corporate tactics in influencing policy include the manufacture of doubt about unfavourable evidence [37] and creating divisions among researchers [21,38]. The evolving international literature provides new frameworks for understanding SAPRO and other CSR activities [1]. These are needed to address the historically unparalleled levels of concern about the international activities of the global alcohol industry [39,40]. SAPROs work with, and learn from, each other. The Portman Group/Drinkaware operational model, whereby a public information role is assumed by the latter SAPRO, and the former more obviously promotes industry positions on alcohol-related issues, may well be replicated in other countries. National governments and their policy-making processes are key targets for the alcohol industry [41]. The addiction and public health research communities should examine industry influence on alcohol policies [42]. Drinkaware, like other SAPROs, appears to us to be an industry vehicle to subvert evidence-based public health policy. We propose that it is not worthy of any form of support. In the past the Portman Group divested itself of a public information function, so SAPROs may also be dispensable to the alcohol industry if they do not further its strategic objectives. We urge policy-makers to address industry influence on global and national alcohol policies [39,43,44] more assertively to reverse the mounting toll of alcohol on population health and social wellbeing. --- Declaration of interests The authors have no sources of funding, direct or indirect, nor any other connection with the tobacco, alcohol, pharmaceutical or gaming industries nor any body substantially funded by one of these organizations.

In 2006, Drinkaware was established as a charity in the United Kingdom following a memorandum of understanding between the Portman Group and various UK government agencies. This debate piece briefly reviews the international literature on industry social aspects organizations, examines the nature of Drinkaware's activities and considers how the public health community should respond. Although the British addiction field and the wider public health community have distanced themselves from the Portman Group, they have not done so from Drinkaware, even though Drinkaware was devised by the Portman Group to serve industry interests. Both long-standing and more recent developments indicate very high levels of industry influence on British alcohol policy, and Drinkaware provides one mechanism of influence. We suggest that working with, and for, industry bodies such as Drinkaware helps disguise fundamental conflicts of interest and serves only to legitimize corporate efforts to promote partnership as a means of averting evidence-based alcohol policies. We invite vigorous debate on these internationally significant issues and propose that similar industry bodies should be carefully studied in other countries.

Based on the identified risk factors, the highest area under the receiver operating curve (AUROC) and accuracy of the best classification model (weight overestimation in all participants) were 0.758 and 0.761, respectively. Adding various associated lifestyle factors to the baseline model resulted in an average increase of 0.159 and 0.135 in AUROC for classifying weight underestimation and overestimation, respectively. --- Introduction Obesity is a complex and multifactorial disease that is a major public health issue. The prevalence of obesity has been epidemically increasing worldwide; being overweight and obese affects over one-third of the world's population (1), and it is expected that 38% of the world's adult population will be overweight, and 20% will be obese by 2030 (2). Excess weight and obesity are well-established critical risk factors for comorbidities of chronic diseases such as type 2 diabetes, hypertension, dyslipidemia, cardiovascular events, or premature mortality (3,4). Obesity represents a challenge in public health given its high prevalence, increasing trend, and accompanying health risks (2). Issues of obesity and its related health problems have generated an increased attention to the perception of and satisfaction with body image and shape. Body image distortion can lead to inappropriate diet restriction, eating disorders, and excessive weight loss, which in turn, give rise to reproductive function issues, such as menstrual irregularity, hypothalamic dysfunction, infertility, and bone loss due to impaired sex hormone metabolism (5)(6)(7)(8)(9). Moreover, body dissatisfaction is linked to adverse psychological consequences such as poor selfesteem, depression, and anxiety (10)(11)(12). Existing studies report certain factors associated with the discrepancy between perceived and actual body shape and size (13-15). Significant gender and age differences have been reported for this discrepancy; males and younger individuals tend to underestimate their weight (and vice versa for females) (14)(15)(16)(17). Studies have explored comprehensive lifestyle factors, ranging from health-related behaviors to socioeconomic status; however, most of this research used cross-sectional designs, highly selected cohorts, or populations with specific age groups (e.g., high school or university students, and young adults) (18)(19)(20)(21)(22). In addition, there is growing attention on how to identify the target populations with weight misperception, and only a few studies have performed predictive analytics based on the associated behavioral and social factors that lead to body weight misperception. This study investigated the perceived body weight and/or its disturbance and trends across age groups in Korean women, using large population data from the national registry to determine the associated female-specific and age-specific risk factors for the discrepancy between actual body size and selfperceived body weight. One of the main purposes of the present study was to predict the future occurrence of body weight misperception with a machine-learning approach, by providing data that could be used by the public health branch management for preventive interventions. --- Materials and methods --- Study population --- Sociodemographic and lifestyle variables The KNHANES survey questionnaires assessed a wide range of lifestyle factors, ranging from health-related behaviors to socioeconomic status. Data on age, education level, income level, marital status, self-perceived health status, alcohol consumption, and employment status were obtained using structured survey items. Education level was categorized as elementary school or lower, middle school, high school, and college or higher. Income level was classified according to monthly income quartiles for each member in the household: low, middle-low, middle-high, and high. Participants consuming more than 12 alcohol drinks per year were classified as alcohol drinkers. The following clinical or health-related information were collected: past medical history of chronic diseases, experience of anxiety and depressive mood (none/mild/severe), degree of chronic pain (none/mild/severe), days of anaerobic workout per week, days of walking per week, exercise for weight loss, weightcontrol practice, gravidity, and menstruation condition. Past medical history of chronic diseases was defined as having been diagnosed from a physician, which was dichotomized into present and none for each disease, including diabetes, hypertension, hyperlipidemia, or depression. Weight-control practice was divided into the following categories: "weight loss management," "weight maintenance management," "weight gain management," and "never tried to control weight". Gravidity was defined as the number of pregnancies the woman has ever had, including all live births, stillbirths, miscarriage and abortions. The menstruation condition of the participants was assessed whether they were menstruating or were amenorrheic due to pregnancy, lactation, or menopause (menopause type). The full list of the variables used in this study is provided in Supplementary Table 1. --- Assessment of objective weight status and classification of body weight perception To assess the objective weight status, anthropometric indices, including height, weight, and waist circumference, were measured, and the body mass index (BMI) was calculated using the formula [bodyweight (kg)/(height x height) (m 2 )]. In addition, information on the self-perceived body image was collected based on the following questionnaire: "Do you consider yourself as (1) very thin, (2) a little thin, (3) average, (4) a little obese, or (5) very obese." Based on a comparison between the subjects' actual weight status (measured using BMI) by the World Health Organization Asian-Pacific criteria (23) and their self-perceived body image, two categories of mismatched body images were generated: (1) overestimation, defined as women who had normal weight or were underweight (BMI less than 23) who perceived their body as a little obese or very obese and (2) underestimation, defined as women with a BMI <unk> 23 (overweight or obese) who responded that their body size was average, a little thin, or very thin. The control group was defined as women with normal perception who were not belong to either of overestimation or underestimation categories. --- Statistical analysis First, Chi-square test was performed to examine linear trends in proportions of weight misperception among age groups. Then, we fitted multivariate logistic regression models to examine the associations between overall weight misperception and a range of lifestyle factors for all the participants. The main null hypothesis of our logistic regression model was that there is no association between specific lifestyle factors and weight misperception. Among the tested factors, gravidity was the only continuous variable and we replaced its extreme outliers (calculated as the 3rd quartile + 3 <unk> the interquartile range or 1st quartile-3 <unk> the interquartile range) with median values. To explore age-specific differences in lifestyle factors associated with weight misperception, we classified the participants into three age groups and repeated the analysis as follows: early adulthood (19-45 years), middle adulthood (46-59 years), and late adulthood (<unk> 60 years) (mean ages of 33.8, 52.6, and 69.8 years, respectively). For age-specific analysis, different subsets of health-related and lifestyle factors, including age, alcohol consumption, anxiety and depressive mood, chronic pain, days of anaerobic workout, days of walking per week, education level, exercise for weight loss, gravidity, income level, marital status, self-perceived health status, and management of weight (gain, loss or maintenance) were tested (Supplementary Table 2). The Bonferroni-adjusted significance level of 0.0055 for multiple comparisons (3 different age groups, each with overall misperception, overestimation, and underestimation patterns) was applied. All the analyses were performed in the R environment (version 4.1.1) using the svyglm function in the survey package, accounting for the KNHANES sampling weights. --- Predictive analysis We trained an ensemble of decision tree models to classify the participants' self-perceived body weight status based on the identified lifestyle factors from the aforementioned regression analysis. We hypothesized that several behavioral and social lifestyle factors would account for the misperception of body weight. For the model training and evaluation, the data were split into training (70%) and test (30%) sets, among which the normal controls were randomly undersampled. The baseline model was constructed only including BMI, and its result was compared with our combined model that additionally incorporated significantly identified lifestyle factors. The model performance was evaluated using the held-out test set by computing the values of the area under the receiver operating characteristic curve (AUROC), the 95% confidence interval (CI), and accuracy. --- Results --- Demographic characteristics and overall misperception patterns of body weight We analyzed multivariate data from 22,121 Korean women who participated in the KNHANES between 2010 and 2016. Table 1 presents the demographic characteristics of the participants. A wide range of socioeconomic, lifestyle, and health-related factors were included in the analysis, ranging from self-reported questionnaire data, reporting the individual's environment or daily behaviors, to objective health/weight measurements or clinical diagnosis history (Supplementary Table 1). The overall proportions for the "correct" and "mismatch" types of perceived body weight by age groups are shown in Figure 1A, where the age categories were 19-45 years (early adulthood), 46-59 years (middle adulthood), and <unk> 60 years (late adulthood). The proportion of participants who correctly perceived body weight was similar for early adulthood and middle adulthood (82.6% and 82.2%, respectively) but lower for the late adulthood group, as only 68.2% were correctly matched. Moreover, there was a trend towards an increase in the underestimation of weight with increasing age (P <unk> 0.001); in contrast, the proportion of overestimation decreased with age (P <unk> 0.001) (Figure 1B). We identified several lifestyle factors that were significantly associated with the overall misperception (overestimation and underestimation) of body weight (Figure 2). Age, alcohol consumption, days of walking, education level, marital status, self-perceived health status, effort of weight management (gain, loss, and maintenance) were negatively associated with misperception patterns, whereas, exercise behaviors for weight loss was positively associated with misperceived body weight. --- Age-specific factors associated with misperception of body weight status Based on the factors associated with misperception patterns, we analyzed the data using category-specific variables. In the early adulthood group (age 19 to 45), women who tended to consume less alcohol, had a lower education level, were married, and perceived themselves as unhealthy were significantly more likely to underestimate their body weight (Figure 3A, left). On the contrary, women who tended to consume more alcohol, had chronic pain, did fewer anaerobic workouts, perceived themselves as healthy, made less effort for weight gain and more effort for weight loss were more likely to overestimate their body weight (Figure 3A, right). For the middle adulthood group (age 45-59 years), women who did more anaerobic workouts, had fewer episodes of depression, had a lower education level, had lower gravidity, and were not menopausal state were significantly more likely to underestimate the body weight (Figure 3B, left). Women who did less anaerobic workouts, were not in menopause, made less effort for weight gain and more effort for weight loss, and who perceived themselves healthier were significantly more likely to overestimate their body weight (Figure 3B, right). In the late adulthood group (age <unk> 60 years), less alcohol consumption, fewer days of walking, lower education level, more exercise for weight loss, lower income, no menopause, and perceiving themselves as unhealthy were associated with an underestimation of body weight (Figure 3C, left). Among the variables of chronic medical diseases, the presence of diabetes and the absence of hyperlipidemia were significantly associated with an underestimation of body weight. Women who perceived themselves as healthier, with less effort for weight gain, and more effort for weight loss, were significantly associated with an overestimation of body weight (Figure 3C, right). --- Predictive performance The bagging tree ensemble model that included the previously identified lifestyle features showed a moderate overall performance for assessing weight overestimation (Figure 4A) and underestimation (Figure 4B). The feature included in each model were presented as footnote of Supplementary Table 2. Across the tested models, the highest AUROC and accuracy of the best classification model (weight overestimation in all participants across ages) were 0.758 and 0.761, respectively, with the 5-fold cross-validation (Supplementary Table 2A). Adding various lifestyle factors to the baseline model resulted in an average increase of 0.159 and 0.135 for AUROC when estimating weight-underestimating and weight-overestimating behavior, respectively. The classification models for the weightoverestimating behavior did not show any improvement compared to the baseline model, which only included BMI (Supplementary Table 2B). The feature important analysis revealed the most useful relevant variables for classifying weight underestimation, which were BMI, weight management status, and self-perceived health. --- Discussion In this study, we investigated the age-specific trend of selfperceived body weight in a large population of Korean women, and observed that the incidence of mismatch of body weight increases with age and is highest in late adulthood. Weight overestimation patterns were more commonly observed than weight underestimation in early adulthood (ages, and the pattern was reversed in middle adulthood (ages 45-59) and late adulthood (ages <unk> 60) in our study. Existing studies have reported that race, gender, age, and behavioral factors influence the misperception of one's body weight (16,(24)(25)(26)(27)(28)(29)(30). To date, most of these studies were conducted in the US or on Western populations, which have limited application to women in other countries (16). Unlike Western populations where an underestimation in overweight/ obese women is the main issue, women in Asian countries tend to overestimate their body weight and attempt weight control more than Western women (31,32). Our work represents one of the few studies examining the overall trend, prevalence, and agespecific patterns of body weight perception in Asian populations. Similar to the findings of other studies, we observed that young adults are likely to overestimate their body weight or image, and this phenomenon is frequently reported across countries and ethnicities (33)(34)(35). Young adults are often exposed to unrealistic body ideals portrayed in mass media that are regarded as the norm of societal standards of beauty (36). Accumulated studies have shown a high prevalence of misperception among the young population; however, data on older adults are currently limited. The results of the present study revealed that the proportion of individuals with a mismatch in body image was highest (31.8%) in the late adulthood group (<unk> 60 years) in our cohort. Notably, underestimation (29.6%) was more prevalent than overestimation (2.2%) in this age group. A possible explanation for this finding is that age-related changes in overall cognitive function or reduced interest in one's body image affect one's perceived body image. The importance attributed to body image and physical appearance decreases with age, and this reduced emphasis on appearance may be protective against negative self-evaluations associated with body image (37). It is also possible that becoming feeble or caredependent with age may cause the person to feel weak and misperceive themselves as underweight. More importantly, an awareness of this phenomenon of the misperception on body images is necessary because chronic diseases such as diabetes, hypertension, metabolic syndrome, and health issues related to obesity may be overlooked because older women do not feel the need for weight management (38). In this study, we identified several sociocultural, habitual, health-related, and female-specific factors that were significantly associated with misperceptions of body weight in Korean women. Our regression analysis revealed that age, education level, days of walking, and efforts for weight management were negatively associated with a misperception of body weight. Of note, self-perceived health status was negatively related to the misperception of body weight across all age groups, indicating that women who perceived themselves as unhealthy tended to underestimate their body weight. Also, making specific efforts toward weight loss was associated with weight-overestimation patterns in all age groups; these results support earlier observations that dieters are more likely to overestimate their weights even though their weights were within a healthy range, and those bias could be risky as weight loss effort are a wellknown risk factor for reduced nutritional state, mortality risk, or significant health problems (25,(39)(40)(41). It can also be suggested that dieters' particular attention to their body weight are associated with biased estimates of weight status and this is likely to influence their subjective perception of healthiness, or the evaluation of 'healthiness/unhealthiness' might be closely related to one's self-perceived weight status. Further work is required to clarify which of these two factors, weight loss efforts or self-perceived health status, is more important for the weight misperception behavior and if there is an interference or interaction between them. One of the strengths of our study is that a large nationally representative sample was collected from a centralized dataset derived from a national registry cohort. Existing studies have reported that women tend to be more sensitive and overestimate their body weight than men (42)(43)(44)(45). Our study aimed to demystify female-specific risk factors that might influence the weight misperception, including gravidity, menopausal status, or pregnancy history. In addition, we sought to identify women with weight misperceptions to quantify the extent of variance that the identified risk features could explain. To our knowledge, this is the first study with a machine learning approach to classify an individual's weight misperception based on their demographic and lifestyle data. Even though we showed a slightly improved prediction model compared to the baseline model, we believe that the inclusion of more lifestyle and behavioral features will identify social determinants for weight misperception and could address some crucial public health concerns. By estimating the relative feature importance within the prediction models, we confirmed that an individual's perceived health and weight management s t a t u s w e r e i m p o r t a n t a n d d i r e c t i n d i c a t o r s o f weight misperception. Our study had several limitations. First, most of our lifestyle features were derived from self-reported questionnaire from the KNHANES dataset. Even though self-reported features enable deep curation of the environment surrounding an individual, they are inherently biased by an individual's subjective feelings and do not comprehensively cover every part of the human lifestyle. Aside from the self-reported features, if other types and ranges of objective measurements were available to describe an individual's lifestyle better, the findings would have been strengthened by identifying more reliable and valid risk features associated with weight misperception. In addition, we observed a significant case-control imbalance in our dataset, especially for weight underestimation, hindering the performance of our weight misperception prediction models based on a machine-learning approach. We attempted to address these challenges by undersampling the training data; however, for future studies, the use of a large population will strengthen the classification models and unveil meaningful outcomes. Lastly, even though the BMI is the most common anthropometric index to estimate general adiposity, it does not provide accurate information for distinguishing between fat and lean mass or fat distribution (46,47). Thus, the use of BMI index alone might be insufficient to determine whether or not an individual is at her proper weights. Future studies should consider incorporating other anthropometric indices that can better identify the adiposity and fat distribution, to validate our findings further. --- Conclusion Our study identified several sociocultural, habitual, healthrelated, age-specific and female-specific factors that are significantly associated with misperceptions of self-perceived body weight in a large sample of Korean female adults. Our prediction models revealed that an individual's perceived health and weight management status are important and direct indicators of weight misperception. to the crucial discussion and manuscript revision and approved the submitted version. --- Data availability statement The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation. --- Ethics statement The studies involving human participants were reviewed and approved by Institutional Review Board, Korea University Ansan Hospital. Written informed consent for participation was not required for this study in accordance with the national legislation and the institutional requirements. --- Author contributions YYJ, GJC, and KWY contributed to the conceptualization, design of the study, acquisition of the data, and writing of the original draft. YYJ, JK, and KL organized the database and performed the statistical analysis. All authors have contributed --- Conflict of interest The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. --- Publisher's note All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher. --- Supplementary material The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/ fendo.2022.1007129/full#supplementary-material

Background: Misperception of body weight is associated with various psychological and health problems, including obesity, eating disorders, and mental problems. To date, female-specific risk factors, including socioeconomic or health-related lifestyle features, or their indicative performance for the misperception in Asian women according to age groups remain unknown. Objectives: To investigate the prevalence and associated risk factors for the mismatch in self-perceived body weight and evaluated the classification performance of the identified risk factors across age groups in female adults. Methods: We analyzed data of 22,121 women (age 19-97 years) from the 7year Korea National Health and Nutrition Examination Survey dataset (2010)(2011)(2012)(2013)(2014)(2015)(2016). We evaluated self-perceived body weight of the participants with their actual weight using the body mass index cut-off and grouped them by age: early adulthood (19-45), middle adulthood (46-59), and late adulthood (≥60). Logistic regression was conducted in each age group based on their weight misperception. The classification performance of the identified risk factors was evaluated with a bagging tree ensemble model with 5-fold cross-validation. Results: 22.2% (n=4,916) of the study participants incorrectly perceived their body weight, of which 14.1% (n=3,110) and 8.2% (n=1,806) were in the underestimated and overestimated groups. Among the age groups, the proportion of participants who misperceived their body weight was highest in late adulthood (31.8%) and the rate of overestimation was highest in early adulthood (14.1%). We found that a lower education level, absence of menopause, perception of themselves as unhealthy, and efforts for weight management were significantly associated with the overall misperception (overestimation or underestimation) of body weight across age groups.

| 61 de GRUCHY --- | BACKG ROU N D Recently, calls have been made at a global level for policies, at all levels of government, to be migration and health aware (Khan et al., 2016;Vearey, 2014;Wickramage & Annunziata, 2018). Research from the South African context has shown that local responses-or a lack thereof-can contribute to either exacerbating or mitigating the experiences that migrants have of accessing healthcare (Halogen & Vearey, 2010;Landau & Singh, 2008;Misago, 2016;Vearey, 2011). One kind of local, community-based response to issues of health is programmes that train and equip community members to facilitate access to healthcare within the community. This approach has received renewed attention in South Africa, as well as in other lowand middle-income countries (LMICs), since the mid-1990s, as the under resourced South African public healthcare system is faced with multiple challenges in addressing the 'four colliding epidemics' of 'HIV and tuberculosis; chronic illness and mental health; injury and violence; and maternal, neonatal, and child health' (Mayosi et al., 2012;Nxumalo, Goudge, & Manderson, 2016;Schneider, Hlophe, & van Rensburg, 2008). However, although the state increasingly recognises the role that these programmes can play in addressing unmet health needs at the local level, policy responses have largely been fragmented and highlight that while the state sees these programmes and their cadres of workforce as important, they remain peripheral to the health system itself and their training and management is by-and-large left up to non-governmental organisations (NGOs) (Daniels, Clarke, & Ringsberg, 2012;van Ginneken, Lewin, & Berridge, 2010;Schneider et al., 2008). This reliance on external non-state actors has implications both for the sustainability of these programmes and for the security and well-being of these workers (Clarke, Dick, & Lewin, 2008;Nxumalo et al., 2016;Suri, Gan, & Carpenter, 2007). Additionally, it has also meant that there is little uniformity in the structure of these programmes (Friedman, 2005;Schneider et al., 2008). For the purposes of this paper, these individuals will be referred to as community-based healthcare workers. However, within specific programmes they have different titles, varied responsibilities and levels of training, are sometimes volunteers and at other times are remunerated (Clarke et al., 2008;Friedman, 2005;Mwai et al., 2013). This paper explores two programmatic interventions that independently developed cadres of community-based healthcare workers in order to improve the access that migrant farm workers in the area around the South African town of Musina had to healthcare. Ten kilometres from the Zimbabwean border, Musina has always seen the coming and going of Zimbabwean nationals, some of whom have traditionally found work, both seasonally and more permanently, on the farms surrounding the town (Bolt, 2016;Rutherford, 2008;Rutherford & Addison, 2007). In 2007 and2008, in response to an increase in the number of Zimbabweans crossing the border to escape electoral violence and a cholera outbreak in Zimbabwe (Staff Reporter, 2008;Tran, 2008), several international organisations, including the International Organization for Migration (IOM) and Médecins sans Frontières (MSF), set up projects in the area. By 2009, this crisis had largely dissipated, and both organisations moved their focus to migrant farm workers on the commercial farms surrounding Musina, as vulnerable groups whose access to healthcare was limited and could be improved. Globally, while migrant farm workers are key to many commercial farming industries, more often than not structural barriers ensure that this workforce bears an undue burden of both communicable and non-communicable diseases (Arcury & Quandt, 2007;Rye & Andrzejewska, 2010). South Africa is no exception. While legislation 'covering rights to collective bargaining, basic conditions of employment, social security benefits and workplace health and safety' (London, 2003, p. 60) exist, the access that migrant farm workers have to these rights, including to healthcare, is limited. Subject to low wages and poor living and working conditions (Bolt, 2012;Jinnah, 2017;London, 2003), within a context in which the public health system is severely under resourced and programming is not migration-aware (Vearey, 2014(Vearey,, 2018;;Vearey, Modisenyane, & Hunter-Adams, 2017), migrant farm workers are known to have one of the highest HIV prevalence rates in the country (International Organization for Migration, 2010). In Musina, at the time that MSF and the IOM were developing programmes, the Department of Health had a mobile clinic programme that was meant to visit farm worker communities on rotation. However, it was severely under resourced and, importantly, unable to provide any form of HIV care or support. Furthermore, continuity of care for HIV-and other chronic conditions-has historically been very difficult in this area as patients move regularly and South African health systems are yet to respond to the realities of patient mobility, both cross-border and internal (Médecins --- What is known about this topic • Increasingly calls are being made for migration aware policies and programming at all levels of governance. • In South Africa, health policies do not consider the realities of migration, which affects the access that migrants, including migrant farm workers, have to care. • South Africa's approach to community-based interventions is fragmented and characterised by a reliance on non-state actors. --- What this paper adds • Original, empirical insights into community-based interventions, and their potential as migration-aware responses to health. • An analysis of why community-based interventions are unsustainable in South Africa. • Urgency to the call for sustainable migration aware policies and programmes. de GRUCHY Sans Frontières, 2012, 2013;Vearey, 2014;Vearey et al., 2017). To address these gaps, MSF implemented the Musina Model of Care, an initiative that included a mobile clinic programme that provided voluntary counselling and testing (VCT) for HIV, antiretroviral therapy (ART) and developed a cadre of Community Health Workers (CHWs) to work alongside the clinic. In 2012, 10 CHWs were trained across 10 farms to work both with the mobile clinic and independently to test for malaria, support HIV treatment, run support groups and provide basic medical care within the farm compounds. As MSF was developing and implementing its CHW programme, the IOM, working with the Centre for Positive Care (CPC) a local non-governmental organisation that acted as an implementing partner, trained on many of the same farms 103 farm workers as peer educators, referred to as Change Agents. This was part of the organisation's Ripfumelo project, a large-scale regional project which looked 'to reduce HIV and TB vulnerability amongst migrants and mobile populations and the communities affected by migration' across several areas in Southern Africa, including Musina. The primary difference between the two groups being that Change Agents educate and mobilise workers to seek care, while CHWs are able to provide some basic biomedical care. Both of these interventions were envisaged as supporting the provision of biomedical care through the mobile clinic. At the height of the interventions' success, CHWs and Change Agents were part of a robust migration-aware response to the intricacies of healthcare access for this group of workers. However, neither MSF nor IOM could commit to sustaining the programmes for more than a few years. MSF had hoped that when they left the CHWs would be incorporated into the Department of Health. However, at the time of MSF's exit, the Department argued that it was not able to take over the cost and the maintenance of the programme. As such, the programme was managed and funded by North Star Alliance, an international NGO that provides health services to mobile workers, for a year, after which it was handed over to the CPC, this time as an implementing partner of the Department of Health. CPC, at this time, remained an implementing partner of the IOM and involved in the training and management of Change Agents. At the end of 2017, the IOM brought an end to its migration and health related activities in the area, including funding for CPC. CPC have consequently left the area, handing the management of the CHW programme over to the local branch of a humanitarian organisation in the area, but leaving no provision for the Change Agents (see Figure 1 for a timeline of these two programmes). This paper is based on research conducted as part of a broader project that has examined the role that non-state actors have played in responding to migration and health in South Africa, and the longer term implications of their involvement with migrant farm workers around Musina (see Author and others, 2019). Key issues on which the research reflects have included the nature of responses by both state and non-state actors to migration and health. This paper uses the CHWs and Change Agents to illustrate the ways in which such workers can form part of a sustainable, migration aware response to health. But the paper also demonstrates how this is undermined by both the timebound nature of non-state interventions and the state's reluctance to incorporate community-based healthcare workers more formally within the health system. --- | MATERIAL S AND ME THODS Several qualitative methods were used in this research, including key informant interviews, an analysis of grey literature and observation of the mobile clinic programme. --- | Key informant interviews A total of 79 in-depth, key informant interviews were conducted, the specifics of which can be found in Table 1. To understand workers' experiences of the interventions, interviews were conducted across two farms on which the interventions had been implemented and where farm management were willing for researchers to conduct interviews during work hours. Most of these interviews were conducted by two research assistants who had received training on conducting interviews with farm workers, and the ethics thereof, and were able to conduct interviews in ChiShona, the language of choice for many of the farm workers. Audio recordings of interviews that were conducted in ChiShona were sent to a translator, who both translated and transcribed the interviews into English for analysis by the author. Individuals were approached as they stopped work for lunch, waited for the mobile clinic to set up or leave, or were introduced to the researchers by others who had already been interviewed. Informed consent was sought prior to interviews commencing. Interviewees were asked to provide either written or verbal consent Data from interviews was thematically analysed using Dedoose 8.2.14. 74 codes were identified through the initial analysis of the data, and these were then examined in relation to one another as themes. F I G U R E 1 --- | Observation It proved challenging to formally interview many of the Department of Health employed healthcare professionals who work on the mobile clinic given the constraints of their work. As such, observation of the mobile clinic was undertaken between May and July 2017. Time was spent in the mobile clinic offices (specifically in the morning as the nurses prepared to leave for the day), travelling with the mobile clinics and observing how the healthcare workers related to workers when they arrived on the farms. Observations and informal conversations were recorded as field notes, which were subsequently thematically analysed. No treatment or care itself was observed. --- | Grey literature In addition, to supplement a review of published literature, a thematic analysis of 76 documents-including policy directives, meeting minutes, memos, project reports, policy documents and emails relating to the projects-was undertaken. By-and-large, these documents were collected as interviews took place and participants indicated that a particular document might be of interest or use. Documents were thematically analysed in conjunction with interview data to triangulate information and provide details that may have been forgotten by participants. Use of these data were limited however, as key informants had to be relied upon to send the author documents that they had deemed sufficiently relevant to keep. The details of this study were reviewed and approved by the University of the Witwatersrand Non-Medical Research Ethics Committee (REC). Ethical clearance was given under protocol H16/08/10. --- | RE SULTS In this section, the two programmes are presented together to highlight their differences, as well as the ways in which the two workforces interacted and continue to interact against a fragmented policy and programmatic landscape. The Change Agents, on the other hand, were trained to act as community leaders, facilitate dialogue, educate farm workers on issues of healthcare, dispense condoms, and plan weekend activities so 'that farm workers use their free time in sports rather than engaging in unprotected sexual activities and abusing alcohol and drugs' (IOM_10). --- | Improving access to care Given this broader range of potential roles that Change Agents could and can play, these individuals are not exclusively regarded as healthcare workers by community members. Most farm workers interviewed reported positive experiences of both CHWs and Change Agents, and importantly that they know who these individuals are and the work that they do. However, across the two farms, three farm workers reported that they would only be able to identify the CHW by face, rather than by name, and three reported that they were unaware of the CHW. While these are not large numbers, they do highlight the limitations of community-based workers: even within a confined farm there are those farm workers who, for whatever reason, are beyond the CHW's reach. --- | Motivations and status While the experiences that farm workers had of the CHWs and Change Agents are important, the experiences that these individuals | 65 de GRUCHY Across the two farms there is however one salient difference in the CHW's role and status. On one of the farms, the CHW is very clearly regarded as an important part of farm life. She grew up on the farm and is married to the head Change Agent (and in some interviews quite explicitly referred to as 'The wife of Sizwe' (pseudonym)), who is a senior employee and has a closer relationship with farm management than many of the other workers. In one interview referred to by a farm worker as 'our leader' (MFW_ EG_103), she is very present on the farm and the lives of the workers, describing herself as motivated to fulfil her role as CHW, 'a full-time job'. On the second farm, however, the importance of the role of the CHW is acknowledged, but the CHW is regarded with some frustration and irritation, as she is often absent from the farm: The problem that we have is that, she cannot be found easily... If you are lucky to find her, she can help you. --- (MFW_EG_201) The role that this CHW plays on this farm is markedly different from the previous example. Here, the CHW indicates that while she has a room on the farm, this is not her home, and her role is to be present every second week-coinciding with the mobile clinic visits. When interviewed, this CHW often deferred to Change Agents and indicated that, in her opinion, the role of the CHW and Change Agents This is what I have seen happening in this compound. (MFW_LM_204) --- | Frustrations and insecurity The frustrations expressed in relation to the CHW on the second farm are part of a broader set of frustrations that both Change Agents and CHWs expressed during the research. Although the Change Agents continue to have some status within the community, the benefits that the Change Agents derive This frustration spills over into resentment over the lack of more formal recognition of their work. While Change Agents acknowledge that they signed up as volunteers, the apparent promise of compensation at some future point appears to have been a motivating factor. CPC, for their part, acknowledge this expectation, but express the view that Change Agents should have become formalised within and compensated through farm structures, although there is no evidence that this was ever discussed with farm management. However, while CHWs are remunerated for their work, they remain financially insecure. CHWs are paid their monthly stipend through whichever organisation is acting as the state's implementing partner at the time. However, contracts between the state and these NGOs need to be renewed annually. Regular delays in this process lead to delays in the payment of CHW stipends. For example, in 2017, the CHWs around Musina were not paid until July as CPC waited for their contract with the Department of Health to be renewed. In Regardless of this lack of support and the irregularities around remuneration, however, the fact that CHWs are remunerated while The structural differences in the programmes, specifically around remuneration, pose a barrier to the two cadre of workforce happily co-existing. When additional social factors exist, as on the first farm, this barrier is overcome. However, in the absence of these factors, as on the second farm, tension is a reality. (Vearey et al., 2017). In addition, regardless of their migrant status, workers enthusiastically took part in these programmes and, by-and-large, felt that the programmes had positively affected the health and well-being of their community. However, the sustainability of the programmes has been undermined as, reflecting much of the literature, the key'software' underpinning the efficacy of these cadre of workforce-motivation and supervision-as well as their integration within the broader health system have been neglected by the state (Gilson, 2006;Kok et al., 2017;Sheikh, George, & Gilson, 2014). --- | D ISCUSS I ON While the disparity in financial remuneration between the CHWs and the Change Agents emerged as a key tension in this case, both the research presented here, and the literature highlight that financial remuneration is not the sole motivation that individuals have for becoming community-based healthcare workers (Akintola, 2011;Greenspan et al., 2013;Kidman, Nice, Taylor, & Thurman, 2014;Mwai et al., 2013;Pallas et al., 2013). A desire to learn, as well as to improve the lives of loved ones and the general health and wellbeing of the community, are also key factors that motivate individuals to participate in such programmes. However, as Greenspan et al note (Greenspan et al., 2013), a'strong volunteer spirit... does not preclude a desire for financial rewards'. As many CHWs and Change Agents saw these programmes as an opportunity to up-skill and improve their status within the community, the lack of continued engagement by and support from external actors is keenly felt. This echoes findings by Akintola (2011) The lack of supervision and integration of the programmes within the broader state healthcare system are two additional factors that undermine the sustainability of these programmes. Supervision and integration have both been identified as key enabling factors or barriers to the sustainability of community-based programmes (Assegaai & Schneider, 2019;Mwai et al., 2013;Pallas et al., 2013). Kok et al. (2017) argue that community-based workers need to be seen as'social actors'; that trusting relationships with both the communities that they serve and the healthcare system are pivotal to their efficacy and sustainability. The role that the former plays is illustrated here by the different responses to the CHWs on the two farms, which also highlight the importance of 'community fit' for sustainability (Pallas et al., 2013) and perhaps the lack of care that was given to 'community fit' in the selection of the CHW on the second farm. Gilson and others highlight the importance of 'workplace trust' in health systems (Gilson, 2006;Gilson, Palmer, & Schneider, 2005). This research indicates that community-based healthcare workers can have a good working relationship and trust one part of the health system, in this case the mobile clinic staff and local emergency medical services, while simultaneously not being integrated into or trusting the broader structures of the health system. Research on the relationship between community-based healthcare workers who are able to provide some biomedical care and nurses indicates that nurses often feel threated by these workers and are prone to enforcing professional hierarchies in counter-productive ways when forced to work together (van Ginneken et al., 2010;Schneider et al., 2008). The good working relationship between the CHWs and nurses documented here contradicts much of this. This may be because in this case CHWs are isolated on the farms, and, as such, have a very clearly demarcated role-providing support and care to farm workers between mobile clinic visits. Consequently, they are of no direct threat to the nurses who leave the farm after each visit. Regardless, trust in the broader health system is limited, and may in fact be compounded by the fact that mobile clinic staff themselves indicate little trust in the broader Department of Health, the nuances of which are described in de Gruchy and Kapilashrami (2019). Finally, theories of sustainability point to the importance of integrating interventions into broader structures (Schell et al., 2013;Shigayeva & Coker, 2015). Here, implementing organisations were unable to secure this integration prior to their departure, due to the state's fragmented interest in and response to such interventions, characterised by a lack of sufficient attention to or support for this workforce (Daniels et al., 2012;van Ginneken et al., 2010;Schneider et al., 2008). | 67 de GRUCHY This paper shows how community-based healthcare workers can, and do, play an important role in health systems. However, as it makes clear, within the South African system, the fragmented approach to these workers undermines this potential. The findings from this research support the call for migration and health to be prioritised within policy and programmatic responses to well-being (Vearey, 2018;Vearey et al., 2017;Wickramage & Annunziata, 2018), and highlight the role that community-based healthcare workers can play within this. However, it raises important questions about the development of these cadre of workforce within a context in which factors enabling their sustainability are limited. As such, ensuring that the ways in which migration and health are included in policy making are sustainable emerges as a necessary element to be included in global calls. --- ACK N OWLED G EM ENTS The author thank all who agreed to participant in this research, --- CO N FLI C T O F I NTE R E S T There are no conflicts of interest to be declared. --- O RCI D Thea de Gruchy https://orcid.org/0000-0003-2589-5640 --- R E FE R E N C E S

Reflecting global trends, migrant farm workers in South Africa experience challenges in accessing healthcare. On the commercial farms in Musina, a sub-district bordering Zimbabwe, Medécins sans Frontières and the International Organization for Migration both implemented migration-aware community-based programmes that included the training of community-based healthcare workers, to address these challenges. Using qualitative data, this paper explores the experiences that migrant farm workers, specifically those involved in the programmes, had of these interventions. A total of 79 semi-structured interviews were completed with migrant farm workers, farm managers, NGO employees and civil servants between January 2017 and July 2018. These data were supplemented by a review of grey and published literature, as well as observation and field notes. Findings indicate that participants were primarily positive about the interventions. However, since the departure of both Medécins sans Frontières and the International Organization for Migration, community members have struggled to sustain the projects and the structural differences between the two programmes have created tensions. This paper highlights the ways in which local interventions that mobilise community members can improve the access that rural, migrant farming communities have to healthcare. However, it simultaneously points to the ways in which these interventions are unsustainable given the realities of non-state interventions and the fragmented state approach to community-based healthcare workers. The findings presented in this paper support global calls for the inclusion of migration and health in government policy making at all levels. However, findings also capture the limitations of community-based interventions that do not recognise community-based healthcare workers as social actors and fail to take into account their motivations, desires and need for continued supervision. As such, ensuring that the ways in which migration and health are included in policy making are sustainable emerges as a necessary element to be included in global calls. K E Y W O R D S community health workers, community-based responses, migrant farm workers, migration and health, South Africa, sustainability How to cite this article: de Gruchy T. Responding to the health needs of migrant farm workers in South Africa: Opportunities and challenges for sustainable communitybased responses.

Background In 2014, Zambia rolled out a new and ambitious framework for Comprehensive Sexuality Education (CSE) targeting children and adolescents enrolled in grades 5-12 in schools across the country [1][2][3][4]. In Zambia, sexual and reproductive health (SRH) knowledge is inadequate and unevenly distributed, leading to considerable SRH-related problems among Zambian adolescents [5][6][7]. Aimed to address such unequal access to knowledge about SRH, the development of a CSE programme was heavily supported by UNESCO [3,4]. In Zambia, as many as 25% of married girls aged 15-19 have an unmet need for family planning and about 30% of girls aged 15-19 have begun child bearing [5,7,8]. Moreover, Zambia has high rates of early marriage with as many as 31% of those aged 20-24 reporting to have married before the age of 18 [5,8,9]. While abortions in Zambia are allowed on the broad grounds spelled out in the Termination of Pregnancy Act of 1972, the same law also severely restricts access to safe and legal abortion services by demanding written consent of three medical doctors including a specialist for a legal abortion to take place [10,11]. This is problematic in a country with critical shortage of health workers. Data on abortion in Zambia is scarce, but recent policy documents from the Ministry of Health estimate that 30-50% of all acute gynecological admissions are caused by abortions and that as many as 6 per 1000 women in reproductive age die from abortion-related causes annually [12,13]. The problem affects teenage girls in particular; approximately 80% of women taken to health facilities for abortion-related complications are adolescents [9,14]. Studies in Botswana, Nigeria and South Africa have shown that sexuality education may contribute to overcoming the adolescents SRH challenges that Zambia and other countries face [15][16][17]. At the core of the Zambian sexuality education policy is the idea that there is a substantial need to support adolescents in delaying their sexual debut, to reduce the number of sexual partners and to increase safer sexual practices [15][16][17]. Backed by evidence on its positive effects on adolescents' level of knowledge, skills, attitudes and values related to sex and sexuality, CSE has been promoted in a series of global policy guidelines and recommended to be integrated into ordinary school curricula [3,[15][16][17]. It is anticipated that the positive effects on knowledge, skills, attitudes and values will empower adolescents to realize their health, well-being and dignity; to develop respectful and pleasurable social and sexual relationships; and to understand and ensure the protection of their rights throughout their lives [18]. Many low income countries have committed to international policies to roll out CSE in their schools [3,[15][16][17]. Together with 21 other countries, Zambia has signed 'The Eastern and Southern African Ministerial Commitment on CSE and SRH services for adolescents and young people' which has shaped expansion and implementation of CSE across the region [1]. This agreement was in turn informed by the International Technical Guidance on Sexuality Education published by UNESCO [18], a guideline that grounds sexuality education within a human rights framework springing out of the CEDAW and the ICPD programme of Action on sexual and reproductive health and rights [18]. The coordination of the development of CSE in Zambia was done by UNESCO [2,3], and the Zambian framework was developed with continuous reference to UNESCO's guidelines document. In retrospect, it has been documented that the process of developing and disseminating the content and format of the Zambian CSE was done in a way that left key stakeholders including religious leaders, civic leaders, parents groups and youth without sufficient representation [2,3]. This may have left the CSE policy without much needed public support. Zambia has had reproductive health education since the 1990s, but its original content was limited. It did not cover central SRH themes such as gender relations, sexual behavior, information on contraceptive methods as well as values, attitudes, and self-realization life skills which have now been included in the new CSE framework [4]. A key feature of the revised framework is that it is not supposed to be offered as a standalone subject, but is to be integrated in carrier subjects such as science and social studies [4]. Concerns that CSE is incompatible with the religious and cultural norms have been reported to affect acceptability [18]. In Zambia, this is commonly expressed as a conflict between CSE and a tradition of grandparents providing sexuality education along with cultural norms condemning discussions about sexuality between the sexes except for in grandparents-grandchild relations. It is also a common concept that providing sexuality information to young adolescents should be avoided since it will trigger sexual promiscuity. [19]. Similar difficulties in teaching sexuality education have been reported in other countries [20,21]. Conflicting inter-generational discourses on sexuality between teachers and community members as well as taboos associated with discussion of sexuality [22,23], and gender-related challenges [23][24][25], have been reported to affect the acceptability of sexuality education in studies from South Africa and Botswana. Closely linked to cultural norms and moralization over sexuality are religious values. Zambia was declared a Christian Nation in 1991, a declaration that was included in the preamble of the national constitution [26,27]. This declaration has given Christian morality a particularly prominent place in Zambian politics and society. It emerges in dominant discourses and weighs heavily in public health discussions about access to reproductive health services to homosexuals, or contraception and safe abortion services to adolescents. This contributes to the conditions causing unequal access to SRH knowledge and services among adolescents [28]. While quite a bit of documentation exists on the challenges of approaching sexuality education in schools in Zambia, there is inadequate knowledge about how teachers handle the task of teaching CSE in schools. This study aimed to investigate teachers' experiences with the implementation of the CSE curriculum in the Zambian context. We are particularly interested in teachers' interpretations of their roles in teaching about sexuality, love relations and contraception both vis a vis the pupils and their parents in the community. In examining the teaching process, we draw upon Lipsky's theory of'street-level bureaucracy' which relates to the role that frontline workers or'street-level bureaucrats' -such as teachers -play in concrete policy implementation [29]. Street-level bureaucrats are civil servants, or others tasked with the on-the-ground implementation of policies. They function as gate-keepers to services or reallife policy makers since any policy is dependent on health workers, teachers, social workers or others to convert the policy from paperwork to practice. Lipsky notes that, in order to gain enhanced understanding of public policy implementation, one needs to understand that the policy implementation process is dependent on the actions or discretion of those who carry out the policy in actual practice. Discretion, which is the central tenet of the theory, refers to the use of individual decisions or autonomy during policy implementation to vary the quantity and quality of services or information offered to citizens. Discretionary power can also take the form of inaction or resistance to delivering services or providing information [30]. This discretion may be influenced by many issues such as difficulties in making complex decisions [31], limited availability of information and resources as well as when policies are deemed not to be fully compatible with the local context [29,32]. We used this theory as it is one of the most comprehensive and widely-used theories in understanding bottom-up policy implementation process [32,33]. --- Methods This study is part of a comparative research project named "Competing discourses impacting girls' and women's rights: Fertility control and safe abortion in Ethiopia, Zambia and Tanzania" funded by the Norwegian Research Council and the University of Bergen, Norway [34]. We conducted the study in Nyimba district in Eastern Province of Zambia in 2017. The district was purposively selected as it is one of the provinces with the highest rate of early pregnancies and marriage in Zambia. Primary data were collected by the first author of this paper together with a research assistant at the district level. Designed as a case study of teacher's experiences of implementing CSE in schools, the study focused on the teachers of six schools, conceptualized as cases and combined in-depth-interviews of teachers with observation of the teaching process and classroom situations. The semi-structured interviews loosely followed an interview guide developed by the first author with input from co-authors. After the first phase of data collection, the results were discussed among all authors and the interview guide was further revised. A total of 18 teachers were interviewed from six schools in Nyimba district, reaching a level of saturation. We purposively selected the study participants to ensure inclusion of informants with diverse views and experiences about sexuality education. An attempt was made to include teachers across different grades and subjects. The average numbers of hours that the teachers teach varies from about 20 h per week in primary school to about 25 h per week in high school. Classes are made up of about 60 learners. Teacher expertise was largely grouped in two; those who taught basic sciences including mathematics and those who taught social science related subjects such as social studies and religious education. In conducting the recruitment process, we informed the head teachers in the six schools that we were interested in interviewing the head teacher and two other teachers per school (one from the social scienceand one from the basic science category). Based on this criteria, teachers discussed and agreed on who would be interviewed for the study. The sample was composed of seven female teachers and 11 male teachers. The male bias was caused by the deficiency of female teachers in some of the schools. The age range of the study participants was from 27 years to 48 years. The data did not suggest that gender, seniority or age had an effect on their experience or forms of engagement with the CSE. The semi-structured interviews varied in duration between 40 and 55 min and covered the teachers' experiences with teaching CSE and their thoughts and attitudes towards it. In addition to the interviews, we also reviewed the Zambian CSE curriculum and other relevant policy documents for documenting their content, framing and approaches used. We analyzed the material using thematic analysis [35], drawing upon Lipsky's perspectives on the use of discretion during policy implementation [29]. We focused on developing key themes in decisions about what teachers teach and what shapes their decisions regarding teaching of CSE in schools. The analysis process started with transcribing audio interviews and reviewing the full data set. After a thorough review of the interviews, the development of a code sheet and later coding of the interviews took place. Coding was done using NVIVO version 7 (QSR Australia) which is a qualitative software used to organize qualitative data. The codes were merged into categories, and then themes focusing on forms of discretion and drivers or sources of discretionary power were developed. This was an iterative analytical process which involved: moving between writing themes; reading and analyzing the data; and redrafting the analysis [36][37][38]. The quotes presented in this text are based on the interviews with the teachers, their experiences being the core focus of the paper. Ethical approval for the study was obtained from the ERES ethics committee in Zambia (Ref. No. 2017-Mar-003) and the Ministry of Education. Informed consent was given by all participants before being interviewed for the study. --- Results This section presents findings on the process of integrating CSE into the school curriculum for grades 5-12 in Nyimba district. While the first sub-section describes teachers' experiences and the dilemmas they encountered in teaching CSE, the second sub-section presents teachers' reflections on why their role in teaching CSE is problematic. --- Teachers negotiating the comprehensive sexuality education curriculum We found that the CSE curriculum was treated in an arbitrary manner, leaving much room for the teachers to decide how, when and what to teach as well as what to leave out. With very little guidance, these choices ultimately depended on the individual teacher's judgement on what would be appropriate to teach considering the time available, the age of the learners and the local norms about sexuality and sexuality education. Their decisions and how they reasoned around their choices is described below. --- This framework does not provide guidance Teachers were set to teach CSE in grades 5-12, and to integrate the subject into science, social studies, civic education, home economics and religious education. The teachers we interviewed struggled with how this could be done in an appropriate and natural manner and without compromising the attention to and the learning outcomes in the core subjects. Although their knowledge of the content of the CSE in general was incomplete, the teachers shared a feeling of being overwhelmed by the comprehensive list of topics they were expected to integrate into existing subjects: We are expected to teach relationships, values, attitudes and skills, culture, society and human rights, human development and sexual and reproductive health (IDI, Teacher 14). These topics are the same across all grades [5][6][7][8][9][10][11][12], but as the teachers explained they were expected to provide different levels of detail for the different grades: In the lower grades, for example Grade 5, on pregnancy prevention, the focus should be on identifying benefits of abstaining from sexual activities while as you go up let's say to Grades 8, 9 and 10, we are also expected to teach how hormonal contraceptives are used (IDI, Teacher 2). The CSE framework describes the expected learning outcomes for each grade, but according to our informants and to our review of the framework, it does not provide guidance on how teachers are supposed to integrate CSE into the specific subjects. This was experienced as both demanding and confusing: So when I am teaching home economics or religious education, when and how do I introduce sexuality issues in these subjects? This framework does not provide guidance on such issues. This makes teaching very difficult (IDI, Teacher 5). In this void between stated learning outcomes and lack of guidance on how to reach them, teachers were left to solve the problem themselves and make decisions on the integration process on an individual basis as clearly illustrated in this quote: I decide on my own on what to teach, and how to teach it (IDI, Teacher 1). The lack of direction in teaching and integrating CSE has thus implied dependence on individual teachers' priorities and judgements, and has entailed great variations in the content being taught, when it is taught and how it is taught both within and across schools. Teachers stories show how they moved to amend the curriculum to fit what they perceived to be appropriate through holding back information, emphasizing only some aspects of information or dropping classes on CSE. These tactics or strategies to cope with the problem are explored in detail below. --- Holding back information Teachers reported being selective about which CSE material they taught and what they left out. I do not teach them everything, I leave out some stuff (IDI, Teacher 14). According to our informants the concept that the school should teach pupils different ways of preventing pregnancy was highly controversial and teachers generally felt very uncomfortable about discussing the wide specter of contraceptive methods listed in CSE and the myths associated with them with learners: We are expected to discuss types of contraception such as oral contraceptive pill and depo-provera through injectable method, implant, but I do not teach them such things (IDI, Teacher 10). To develop the discussion in class into how the different types of contraceptives should be used, and where they could be accessed was perceived as even more awkward and unacceptable: Why should I teach about the steps to using contraception? It's like teaching someone how to fish, they end up always wanting to fish; so in the case of sex, what will happen when they do have contraception, they will still have sex (IDI, Teacher 8). The topic on developing good relationships and attributes of a good relationship was seen as difficult to integrate since they did not 'fit well in subjects such as science and home economics: When I am teaching science, how do I bring in issues relating to differences between love, affection and infatuation? It does not work for me (IDI, Teacher 10). Another reason for leaving this topic out was that teachers feared that it could be misinterpreted by learners as support from the school to engage in romantic relationships. --- Promoting abstinence only While some teachers would withhold a few selected pieces of the CSE curriculum, others would only agree to teach very limited fragments of it according to what they deemed to be appropriate for learners. Contrary to the philosophy of the CSE curriculum of making information available to adolescents in order to prevent pregnancy, some teachers believed that such information would be counterproductive and decided only to teach one method: In the school setting, when I teach about preventing pregnancy, the main message is only about abstinence (IDI, Teacher 16). Hence, most of the time available was dedicated to abstinence and the benefits of abstaining from sexual activities. The very strong moral message on abstinence was put across in several ways: We have a song about the importance of abstinence, we sing it before we start each session on CSE. I also tell them repeatedly that abstinence is the only method that can help them avoid teen pregnancies and STIs including HIV (IDI, Teacher 14). This group of teachers saw their role as much as one of preventing sexual activity among their pupils as one of preventing pregnancy. They reported to prioritize their time teaching about the importance of avoiding exposure to situations that could tempt pupils into sexual stimulations or encounters: I tell them to avoid intimacy, being with someone of the opposite sex in a secluded place or watching things that will make them think of the opposite sex (IDI, Teacher 3). --- Dropping topics Other teachers dropped topics or defined sessions in the CSE altogether. This took different forms within and between the schools. Interviews with teachers showed that some of them substituted the whole CSE topic, which they were not comfortable teaching, with other topics which they believed were more appropriate for learners. A teacher told us how he substituted a topic with another: I skipped the whole topic on pregnancy prevention. Instead of teaching about condom use, I moved to another topic. I repeated sessions which I thought were good for learners such as communication, assertiveness and decision-making skills (IDI, Teacher 9). In some cases, when teachers had already taught the topics in the CSE they were comfortable with or felt were appropriate, they turned to teaching completely different subjects with little relevance to the CSE curriculum: When I realized that I had taught all topics that I was comfortable with, instead of teaching topics on condoms or oral contraception use, I decided to only teach social studies (IDI, Teacher 9). Teachers that were most reluctant to teach CSE could even took a more radical step to avoid teaching. Some told us that when it was time for them to teach CSE, they sent learners to do outdoor activities which were not related to CSE: I opt to send the students out for sport activities, preventive maintenance work and other club activities instead of teaching CSE (IDI, Teacher 5). Interviews with teachers showed that teaching CSE was not done on a routine basis and in a standardized manner and that the CSE curriculum was treated haphazardly in the schools. Teachers could not state on average how often they taught CSE and a few teachers reported that they had stopped teaching CSE altogether and as one teacher put it; All I can say that it is something that happens by chance (IDI Teacher 7). --- Lack of local ownership of the CSE agenda The decision-making process among teachers regarding when, what and how to teach comprehensive sexuality education was informed by a number of factors. The reasons, which we outline in detail in this section, included perceived incompatibility of CSE with the local culture, teacher-parent role dilemma, concerns about the legitimacy of the CSE concept and practical challenges related to lack of training and access to manuals. --- Incompatibility with the local culture Comprehensive sexuality education was seen as incompatible with the local culture and religious values, as it confronts local ideas about sexual morality. There were concerns that some topics were too sensitive as they were believed to promote pre-marital and casual sex among learners. We are a Christian country, so the message for us is no sex before marriage (IDI, Teacher 9). Another teacher explained how provision of CSE information would promote casual sex: Many children avoid sex because of fear of pregnancy. No, if they know that they can prevent pregnancy by using contraception, children may get too excited and confident, and start getting involved in casual sex (IDI, Teacher 7). Teachers further reported that parents were against the teaching of some components of CSE in schools as they considered CSE topics to be sacred only to be taught by traditional counsellors at community level: The controversy is also about the place where such information is delivered from not being culturally appropriate, it's taboo to teach sexuality education in a school (IDI, Teacher 11). At one school, a teacher narrated an event which had caused uproar from the community. In an effort to implement the CSE curriculum, the teacher had asked learners to do an exercise at home on initiation ceremonies for girls when they reach puberty: Having seen the assignment, which I gave to the learners, parents came in numbers to the school in the morning and demanded to see the headmaster. I was called to attend the meeting. The parents then complained to the headmaster that the initiation ceremony is something special which should not be handled at the school (IDI, Teacher 6). The notion of sexuality education as sacred and belonging to arenas of learning very different from the school surfaced strongly and placed teachers in some squeeze vis a vis the parents. A complicating factor was gender mixed classes. It was highly uncustomary to discuss sexuality and reproductive health issues specific to female or male learners in the presence of the opposite sex. Adding to the problem was age. Customarily sexuality education was not supposed to be introduced to children in the lower grades. It should be introduced only during the initiation ceremony taking place later after girls attain puberty, and many teachers shared this understanding with the community and had difficulties discussing sexuality issues and using sexuality terminology particularly with the youngest learners. As the young students were not conversant in English, the classes on CSE had to be provided in the local language which was experienced by the teachers as more challenging since the local terms emerged as more insulting than the English ones. In order to cope with the embarrassment teachers used different strategies. As one of them explained: I close my eyes when I mention the sex organs (IDI, Teacher 1). The taboo related to mentioning sex organs in the local language in teacher-student discussions was clearly expressed in the practice of giving the teachers insulting nicknames. As a way of avoiding antagonism with the community, teachers reported leaving out or omitting issues that they perceived as inappropriate from the community perspective. --- Teacher-parent role dilemma The dissenting or opposing views from the community about teaching sexuality education in the school, coupled with cultural and religious values about morality presented a professional challenge for teachers. On the one hand, they were supposed to convey knowledge and stimulate reflections as described in the curriculum. On the other hand, teachers were expected to have a broader role vis a vis their pupils bringing them up according to social and cultural norms and values. Teachers reported that they struggled to strike a balance between teaching sexuality education to their pupils and maintaining the broader parental role of shaping them into responsible adults: It is very difficult for me. As a parent I need to promote abstinence, but as a teacher this curriculum wants me to talk about the importance of using condoms. One topic for example requires us to describe the steps that one has to follow when using a male or female condom. Now, how do I demonstrate such steps to learners who are almost the same age as my child? No, that's like teaching children to be'sex experts' (IDI, Teacher 7). This situation was even more challenging for teachers who had biological children in their class and bolstered the tendency to skip CSE sessions on sensitive topics: I think about my children, so when I know that the topic is not good for them, I skip the topic (IDI, Teacher 9). Because of this role dilemma, other teachers suggested the need to think about other approaches to delivering CSE. One recommendation was to engage other actors to deliver CSE: Some topics can be taught by teachers and other topics can be taught by people outside the school such as health workers or community health workers (IDI, Teacher 7). --- Concerns about the legitimacy of the CSE concept Concerns about the legitimacy of the CSE concept also emerged throughout the interviews. Many teachers reported not being comfortable teaching CSE as they considered it to be something that was externally driven with little relevance to local needs. When we asked one teacher why some teachers have stopped teaching CSE, he seemed to perceive it as a foreign agenda: You mean this donor funded program, some teachers have sidelined sexuality education, it's just extra work for us (IDI, Teacher 13). In addition to cultural incompatibility, the inadequate involvement of actors at the district level during the development, validation and dissemination process affected the legitimacy of the curriculum. Some teachers argued that instead of offering CSE, the community would have preferred more topics that directly address poverty-related issues: They (developers) should have known that this one is a hot issue. It is not as simple as introducing a new curriculum for social studies or science. This one (CSE) touches on what people believe in, people's culture and how people bring up children. To make it even more complicated, we did not cover it during the training process in college, thus as people that are supposed to implement it, we should have been consulted (IDI, Teacher 14). --- Limited prioritization of sexuality education Teachers reported that, compared to other subjects, CSE implementation was weak and was characterized by several severe gaps, including lack of adequate training of the teachers involved: The headmasters attended a 2 days training in CSE, and then they briefed teachers in schools on CSE for only one to two hours. So how do you expect us to effectively teach? (IDI, Teacher 2). Lack of teaching aids or images and reading materials in schools was another gap and was seen as particularly important for explaining complex and sensitive topics: We also need images to explain for example topics on unsafe abortion, cancer, STIs. For now, we have to borrow images from the health facility (IDI, Teacher 4). Teachers also had challenges accessing the manuals as only one manual was given to each school: The headmaster locks the only copy in his office. So how do we teach? (IDI Teacher 4). Furthermore, the topics in CSE were not reflected in the common scheme of work which all schools in the district were supposed to teach: After the schemes were completed, then we just realized that we accidently left out comprehensive sexuality education (IDI, Teacher 1). While some kind of teaching of CSE is going on in the schools in the district, many teachers were grappling with the puzzle of why CSE was introduced. This was primarily related to the weakness that has surrounded the implementation of CSE compared to other subjects: We have been teaching social studies for a long time and at no point did we see parents come and protest about the topics, so why should we continue teaching something (CSE) that the community has concerns over? (IDI, Teacher 5). The decision by the Government to implement CSE without providing adequate support in schools made some teachers question the timing of the implementation process: My question is about why they (Government) decided to implement CSE when they were not ready. I always wonder what caused this rush? (IDI, Teacher 14). The puzzlement among teachers about the rationale for introducing CSE made them question why they have to teach CSE. This lack of appreciation of teaching the new CSE framework by teachers is best illuminated in the following question raised by a teacher when we her asked why she had stopped teaching sexuality education: If I may ask, why do they want us to teach sexuality education? (IDI, Teacher 7). --- Discussion The study has examined teachers' interpretations of their role in teaching sexuality, love relations and contraception during the early phase of implementation of CSE in a rural district in Zambia. We have noted that what, when and how to teach is dependent on the individual teachers' decisions. In line with Lipsky's [29] call for the need to move beyond the top-down approach to policy analysis,and consider other contextual realities that shape policy implementation -our study strongly demonstrates how the settings within the schools in which the CSE framework was implemented influenced how teachers made decisions about the curriculum and subsequently the pattern and nature of the implementation of CSE. This study's findings revealed that the lack of clarity in the CSE framework, on how to integrate CSE teaching into existing subjects, coupled with contextual challenges, left teachers involved in CSE with a great room for discretion. In this context, extensive use of discretion resulted in arbitrary and unequal management of the CSE curriculum in the district. Lipsky notes that unclear or vague policy guidance as well as features of work settings or context in which street-level bureaucrats or workers act can make the bureaucrats interpret and implement the policy content in different ways [29]. According to the theory of street-level bureaucracy, the differential policy interpretation happens because the lack of clarity in the policy gives the implementers space and power to exercise individual discretion in interpreting the content and direction of the policy [33]. In this study, some of the features of work settings that shaped decision making among teachers were socio-cultural factors. These factors included incompatibility of CSE with the local culture and religious ideals. For example, while the CSE framework required teachers to discuss different ways of preventing pregnancy, religious and cultural values expected teachers only to focus on abstinence. Such incompatibility created teacher-parent role dilemmas in the classroom setting. Teachers tended to see themselves in a parental role with the obligation to shape their pupils into responsible or morally upright adults. Abstinence was a key message in this regard. The setting had inadequate support for CSE such as inadequate training, materials and tools for teaching as well as insufficient leadership and guidance in the implementation process. These gaps made some teachers question the extent to which CSE is prioritized within the education system and why they have to teach sexuality education. We note that such doubts among teachers potentially provided more space for discretion, and subsequently enhanced the power teachers had to skip some aspects of CSE or not teach it at all. These study findings, like other studies that have discussed the concept of discretion, agrees that the application of discretion or autonomy during policy implementation is potentially also motivated by the availability, or otherwise, of resources [29,32,39,40]. As a way of dealing with these dilemmas and the gaps in support, teachers modified their teaching of CSE, a practice that is articulated in street-level bureaucracy theory. According to this theory, when they are faced with challenging situations, bureaucrats use their discretion to modify how they understand and execute their tasks or responsibilities [29,33]. Lipsky [41] notes that this invention or modification of modes of decision is done in order to serve the "workers' agency or purposes"(p.xiv). This modification of policy content, which is also known as coping, can happen in three forms. The forms include: bureaucrats' adjusting or moving towards clients through bending policy options in order to meet the needs of clients; moving away from clients or rationing services; and moving against clients through rigid application of rules [42]. In our case study, the modification process adopted by teachers in relation to their clients (pupils and parents) was moving towards clients by bending the CSE policy. This study further showed that the use of discretion to modify what to teach was justified by teachers as the best way to protect the children from sexual harm. Teachers feared that some information would motivate learners to engage in sex as they would no longer have to worry about pregnancy. They argued that the situation had the potential of turning the learners into'sex experts,'putting them at risk of pregnancies in cases where there is no contraception or contracting an STI if condoms were not available. This process of exercising discretion among the teachers in the district was motivated by paternalism, as they viewed or defined learners as, "children in need of protection, rather than as young people who have the right to relevant information about their own bodies and their sexuality" [43], p. 36). As described above, teachers justified adopting paternalistic values as they perceived themselves as 'parents' of all children in the class. It is important to note that paternalism was further articulated through resistance towards teaching CSE. Teachers resisted teaching CSE as they viewed it as something that was externally driven with little relevance to local needs as well as incompatible with the cultural norms and values. The actual policy that is realized vis a vis clients depends more on those who carry out the policy than the policy makers [39]. In the context of this study, implementing CSE is a 'negotiative process' between the teachers and the contextual realities such as the broader educational system, socio-cultural and community dynamics, as well as the experiences and values of individuals. We note that the agency and power among the workers, in this case the teachers, coupled with interactions between teachers and the school environment influenced the implementation of the policy [44,45] which resulted in unequal access to CSE among learners. In our case, these powers included holding back some CSE information, teaching only abstinence and dropping classes. These scenarios, therefore, make the outcome of the policy implementation process a result of the complex interplay or interaction between the frontline workers and the contextual realities. Lipsky [29] refers to this phenomenoninteraction and negotiating process-as "a gap between policy as written, and policy as performed" (p. xvii). We further note like Lipsky's(29)words, that as teachers interact and negotiate during the implementation process of CSE, "the routines they establish, and the devices they invent to cope with uncertainties and work pressure, effectively become the public policies they carry out" (p. xii). We therefore agree, based on the findings on this study, with Gilson's [33] view that for "all bottom uppers, policy-making is still in progress at the moment of delivery" (p,9). Meanwhile Lipsky [41] cautions that negotiations during the policy implementation process and subsequent policy modifications "may widen the gap between policy as written and policy as performed" (p.xvii). Thus although "discretion" may promote teachers' freedom to tailor and adapt their teaching to the needs of their pupils, it may lead to widening the gap between policy as stated and practiced. This widening gap between policy and practice may distort service ideals [33]. For example, adopting paternalistic approaches in delivering CSE may affect the acquisition of skills about reproductive health among young people. Paternalism may affect learning: learners may not be or feel able to ask questions freely on sensitive topics such as contraception use because of the limited interaction and lack of frank discussion between teachers and learners and the moralizing context of sexuality education [46]. This may negate the very essence of establishing CSE in such communities and perpetuate the absence of critical knowledge and life skills to prevent early pregnancy. In a country with high pregnancy and early marriage rates, this lack of knowledge is problematic. Our findings resonate with other studies that have examined the use of discretion in delivering welfare and prison services, ie, that nonconformity to prescribed policies by street-level bureaucracy can lead to disparities in access to services for some populations [31,47]. Improving policy implementation requires paying attention to the contextual realities that reinforce discretion during this process [30]. As we have discussed above, policy as experienced by clients is a reflection or a product of the interplay between both the formal and informal practices of street level bureaucrats [48]. Enhancing the implementation of CSE may require increased involvement of stakeholders at local level in developing and implementing CSE policies and programs, as well as providing comprehensive training in CSE to teachers. As observed in this study, the limited involvement of local actors made teachers see CSE as a foreign agenda which was not compatible with their local context or their mandate to teach. Other authors on CSE in Nigeria and a recent publication on international cooperation in sex education have also cautioned that limited involvement of local actors has the potential of developing CSE which is insensitive to local collective concerns and networks [49,50]. We stress the need for giving stakeholders at the lower level (policy implementers) a much bigger role in developing the content and implementation strategy of CSE as they have better knowledge of the context, networks and local support which they can use to negotiate or navigate micro level politics.

Background: Reproductive health problems such as HIV, unwanted pregnancy and unsafe abortion among adolescents are closely linked to insufficient knowledge about sexuality and reproduction and lack of access to contraceptives. Supported by international agencies, Zambia has introduced an ambitious nation-wide program for comprehensive sexuality education (CSE) to be implemented into ordinary school activities by teachers. The curriculum is firmly based in a discourse of sexual and reproductive rights, not commonly found in the public debate on sexuality in Zambia. This paper explores how teachers perceive the curriculum and practice discretion when implementing the CSE in mid-level schools in Nyimba district in Zambia. Methods: Using a case study design, data were collected through in-depth interviews with 18 teachers and analyzed thematically drawing upon theories of discretion and policy implementation. Results: Individual teachers make decisions on their own regarding what and when to teach CSE. This discretion implies holding back information from the learners, teaching abstinence as the only way of preventing pregnancy or cancelling sexuality education sessions altogether. Teachers' choices about the CSE program were linked to lack of guidance on teaching of the curriculum, especially with regards to how to integrate sexuality education into existing subjects. Limited prioritization of CSE in the educational sector was observed. The incompatibility of CSE with local norms and understandings about adolescent sexuality combined with teacher-parent role dilemmas emerged as problematic in implementing the policy. Limited ownership of the new curriculum further undermined teachers' motivation to actively include CSE in daily teaching activities. Use of discretion has resulted in arbitrary teaching thus affecting the acquisition of comprehensive sexual and reproductive health knowledge among learners.The CSE had limited legitimacy in the community and was met with resistance from teachers tasked with its' implementation. In order to enhance ownership to the CSE program, local concerns about the contents of the curriculum and the parent-teacher role dilemma must be taken into consideration. Not addressing these challenges may undermine the policy's intention of increasing knowledge about sexuality and reproduction and empowering adolescents to access contraceptive services and avoid unwanted pregnancies.

to services for some populations [31,47]. Improving policy implementation requires paying attention to the contextual realities that reinforce discretion during this process [30]. As we have discussed above, policy as experienced by clients is a reflection or a product of the interplay between both the formal and informal practices of street level bureaucrats [48]. Enhancing the implementation of CSE may require increased involvement of stakeholders at local level in developing and implementing CSE policies and programs, as well as providing comprehensive training in CSE to teachers. As observed in this study, the limited involvement of local actors made teachers see CSE as a foreign agenda which was not compatible with their local context or their mandate to teach. Other authors on CSE in Nigeria and a recent publication on international cooperation in sex education have also cautioned that limited involvement of local actors has the potential of developing CSE which is insensitive to local collective concerns and networks [49,50]. We stress the need for giving stakeholders at the lower level (policy implementers) a much bigger role in developing the content and implementation strategy of CSE as they have better knowledge of the context, networks and local support which they can use to negotiate or navigate micro level politics. --- Concluding remarks We conclude that the implementation of the CSE curriculum in this setting was largely dependent on an individual teacher's decisions on what, how and when to teach. This was related to lack of guidance, lack of legitimacy of the curriculum, and lack of local ownership of the agenda. The big space left for teacher discretion in sexuality education resulted in arbitrary teaching of CSE and great disparities within and between schools. If the CSE program is to be successfully integrated and taught, there is a fundamental need to take local culture into account in terms of the curriculum content and teaching approaches, and to secure local ownership of the curriculum. The lack of such considerations can leave the learners at disadvantage. In Zambia, there is rapidly increasing prevalence of early pregnancy, which suggests limitations and failures in efforts aimed at addressing sexual and reproductive health challenges among adolescents. To address this problem, CSE is needed, but as this study has shown, it requires repackaging of both the content and mode of delivery with the support of teachers and other stakeholders at district level. --- Availability of data and materials The datasets for this study are available from the corresponding author on reasonable request. --- Authors' contributions JMZ, KMM and AB contributed to the design of the study. JMZ collected the data and did the analysis and drafted the manuscript. All the authors critically reviewed the manuscripts, provided contributions and approved its final version. --- Consent for publication Not applicable. --- Competing interests The authors declare that they have no competing interests. --- Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Background: Reproductive health problems such as HIV, unwanted pregnancy and unsafe abortion among adolescents are closely linked to insufficient knowledge about sexuality and reproduction and lack of access to contraceptives. Supported by international agencies, Zambia has introduced an ambitious nation-wide program for comprehensive sexuality education (CSE) to be implemented into ordinary school activities by teachers. The curriculum is firmly based in a discourse of sexual and reproductive rights, not commonly found in the public debate on sexuality in Zambia. This paper explores how teachers perceive the curriculum and practice discretion when implementing the CSE in mid-level schools in Nyimba district in Zambia. Methods: Using a case study design, data were collected through in-depth interviews with 18 teachers and analyzed thematically drawing upon theories of discretion and policy implementation. Results: Individual teachers make decisions on their own regarding what and when to teach CSE. This discretion implies holding back information from the learners, teaching abstinence as the only way of preventing pregnancy or cancelling sexuality education sessions altogether. Teachers' choices about the CSE program were linked to lack of guidance on teaching of the curriculum, especially with regards to how to integrate sexuality education into existing subjects. Limited prioritization of CSE in the educational sector was observed. The incompatibility of CSE with local norms and understandings about adolescent sexuality combined with teacher-parent role dilemmas emerged as problematic in implementing the policy. Limited ownership of the new curriculum further undermined teachers' motivation to actively include CSE in daily teaching activities. Use of discretion has resulted in arbitrary teaching thus affecting the acquisition of comprehensive sexual and reproductive health knowledge among learners.The CSE had limited legitimacy in the community and was met with resistance from teachers tasked with its' implementation. In order to enhance ownership to the CSE program, local concerns about the contents of the curriculum and the parent-teacher role dilemma must be taken into consideration. Not addressing these challenges may undermine the policy's intention of increasing knowledge about sexuality and reproduction and empowering adolescents to access contraceptive services and avoid unwanted pregnancies.

Introduction Dental caries is a diet-related condition and is one of the most common oral diseases in China, with almost half of the children and majority of the adults affected [1]. Despite the decline of dental caries in Western countries in the past decades, inequalities in oral health still exist in most of the countries [2]. The high-level of caries in China is mainly attributed to the increased consumption of sugars, lack of dental services, and socioeconomic factors, particularly income and parental education [3]. Several studies found significant associations between lower socioeconomic position and higher risk of dental cavities [4]. In China, there were significant inequalities in the sum of decayed, missing and filled teeth (DMFT) index by geographical location among 12-year-old children [3] and in DMFT index by income among 5-year-old children [5]. Furthermore, parental education was found to be inversely related to children's caries experiences [6]. Parental education and income appear to affect children's oral health through an impact on knowledge of oral health risk behaviors and attitudes towards prevention, such as purchasing dental hygiene products, toothbrushing habits, frequency and patterns of health service utilization, all linked to children's oral health [7]. Furthermore, parental level of education allows a better understanding of information that enables them to care for their children which may affect their children's oral health [8]. Higher mother education and income may also affect children's dietary habits, which could lead to a positive impact on dental caries. On the other hand, the inequality in socioeconomic position might affect the dietary habits and nutritional intake of Chinese children, which could also affect children's growth (weight and height). With the increase in income inequalities in China over the past 25 years, it was argued that health inequalities in China are attributed to demographic and socioeconomic factors, among which income, occupation and education are the most important ones [9]. Yao's (2019) study indicated that there is an association between income inequality and BMI index [10]. Other studies also showed strong a socioeconomic gradient in stunting [11] and thinness [12]. Children from families with lower income are more likely to buy low-cost food and have a poor dietary intake, thus decreasing essential nutrients intake such as protein and fat, which increase the risk of being underweight and stunting [13]. These apparent income inequalities indicate that a reduction in economic inequalities could reduce inequalities in undernutrition [14]. Given the common role of socioeconomic factors in both dental caries and undernutrition, this longitudinal study set out to examine whether the same socioeconomic indicators are linked to changes in dental caries and children's growth in the same population. The objectives of this study are to examine socioeconomic inequalities in increment of dental caries and growth among preschool Chinese children, and to assess the strongest predictor of socioeconomic inequality for each of these conditions. Children whose parents agreed to participant were included at baseline and follow-up in this study. The sample size calculation of 636 is based on a previous study among preschool children in Hong Kong [15]. The number was increased to 1000 to account for dropouts. --- Materials and Methods --- Ethical approval was obtained for --- Study Population This is a longitudinal population-based study, and data was collected at two points in time. The study recruited preschool children in the cities of Shenyang, Liaoyang and Fushun, Liaoning Province, China. At baseline, 15 kindergartens (eight kindergartens in rural area and seven kindergartens in urban area) with 1111 participants were included in this study. At follow-up, the number of participants was 772 with a response rate of 70% at follow-up. The participants who were lost at follow-up either left the kindergarten, did not complete clinical examination or did not return a signed consent form. Mean ages of participants at baseline and follow-up were 50.82 and 60.55 months, respectively. Median and mean follow-up times were 10.12 and 9.73 (SD: 1.22) months, respectively. --- Data Collection Dental examination and anthropometric measurement were assessed at baseline and follow-up. A structured questionnaire was used to collect information on demographic factors, family income, mother's education, and children's dietary habits. The questionnaire was based on the WHO questionnaire [16] and modified following a Chinese questionnaire used in national oral health surveys. In this analysis, baseline data from the questionnaire were in the analysis. Sex was coded as male and female, and area was coded as urban and rural area. Income was collected using five options: 0-3999 RMB, 4000-5999 RMB, 6000-9999 RMB, 10,000 RMB or above and undeclared. Income groups were based on the 4th Chinese Oral Health Survey [17]. Mother's education was grouped as primary and middle school, high school and junior college, bachelor or above, and others. The parents reported the intake frequency of five food or drinks of their children: biscuit/cake/bread, candy/chocolate, fizzy drinks and fruit juice on 6-point response scales (twice or above per day, once per day, 3-4 times per week, once per week, once per month, or seldom or never). Weighted scores were calculated to reflect the frequency of consumption in each response category. Weighted scores of four snacks and drinks were then summed up to produce an overall score [3] which was used to reflect daily frequency of sugar intake. Frequency of fresh fruit consumption was calculated in the same manner and was used as a separate variable. The survey team of oral examination included two persons, a dentist and a recorder. The dentist was trained on data collection using WHO standard criteria for assessing dental caries [16]. Eighty participants were re-examined in different days to calculate intra-examiner reliability. Agreement level (Kappa) was 0.72, indicating substantial agreement. Children with an urgent need for treatment were referred to a dentist. The sum of decayed, missing and filled teeth (DMFT) was used to examine the dental caries level of participants. One health worker was responsible for assessing weight and height and was blinded to the dental condition of the participants. Height of the child was assessed with a weight and height scale to the nearest 0.5 cm. Similarly, weights were assessed to the nearest 0.5 kg. Children wore light clothes and removed shoes before they stood on the scale. Age-and sex-specific percentile for weight and height was used to indicate body measurements of the children [18]. The WHO Growth Standard's for 2006 and 2007 were also used [19,20]. The WHO Growth Standards 2006 was utilized to convert weight and height measurements to weight-for-age and height-for-age z-scores among children aged 0-5 years old and WHO Growth Standards 2007 was used for children older than 5 years [19,20]. --- Statistical Analysis STATA (Stata Statistical Software: Release 15. TX: StataCorp LLC, College Station, TX, USA) was used to analyze the data. Multilevel analysis was used to assess the association between baseline socioeconomic position and dental caries, weight and height. Participants who had two clinical assessments were included in the analysis. Firstly, the distributions of baseline demographic (age, and sex), socioeconomic factors (income and mother's education), area (urban/rural) and dietary habits (mean fresh fruit and mean sugar consumption) were assessed. Then distributions of DMFT index, weight-for-age z-score (WAZ) and height-for-age z-score (HAZ) at baseline and follow-up were assessed. Secondly, multilevel analysis was used to explore the association between baseline socioeconomic position and changes in dental caries. Three models were constructed to assess the association between changes in DMFT and each of income, education and area, accounting for time and sex. The fully adjusted model included all socioeconomic indicators, demographic factors and diet. Similar sets of multilevel models were constructed for each of change in height and weight. --- Results Mean age of those lost at follow-up was 51.15 months at baseline, 51.10% were boys, mean DMFT was 3.17, and means of actual weight and height were 18.56 and 106.26, respectively. In these parameters, differences between those who were lost at follow-up and those included in the analysis were minimal. Details of the study is published elsewhere [21]. Table 1 shows mean age at baseline (50.82) and follow-up (60.55) months. The percentage of boys was 51.42%, and 65.54% of children lived in the rural area. Only 14% of the participants were in the highest income group. The highest percentage of mother's education was high school and junior college (30.70%). The mean DMFT index was 3.18 at baseline and 4.21 at follow-up. Mean weight z-score (WAZ) was 0.58 at baseline and 0.66 at follow-up. Mean height z-score (HAZ) was 0.49 at baseline and 0.69 at follow-up. Table 2 shows results from the multilevel analysis for changes in DMFT index. Area (rural) was positively and significantly associated with an increase in DMFT (coefficient: 0.86; 95% CI (confidence interval): 0.36, 1.37) in the model adjusting for sex and age. Lower mother education was significantly associated with an increase in DMFT score in the semi-adjusted and fully adjusted models with coefficient 0.96; 95% CI: 0.18, 1.74 and 0.86; 95% CI: 0.20, 1.51 for the lowest and second lowest education groups. Only the highest income group (10,000 RMB or above) was negatively and significantly associated with changes in DMFT index (-0.90, 95% CI: -1.77, -0.02) in the semi-adjusted model. However, the association was insignificant in the fully adjusted model. Table 3 shows results from the multilevel analysis for changes in weight-for-age z-score (WAZ). In the semi-adjusted model, higher income was associated with an increase in weight-for-age z-score. After adjusting for all socioeconomic and dietary factors, only the second lowest income group was significantly associated with weight (coefficient: 0.36; 95% CI: 0.11, 0.61). Table 4 shows the association between changes in height-for-age z-score (HAZ) and socioeconomic factors (income, mother's education and area). The relationship between income and changes in height-for-age z-score was positive and significant in the second highest income group (coefficient: 0.26; 95% CI: 0.03, 0.49) in the semi-adjusted model. --- Discussion In this study we assessed socioeconomic inequalities in both of children's growth and dental caries in the same population of Chinese preschool children. The main findings showed that there was negative and significant association between income, mother's education and changes in dental caries. Income, rather than mother's education, was positively and significantly associated with weight and height. Children from rural areas had a greater increase in dental caries, a smaller increase in weight and height compared to those from urban areas. The findings of the current analysis of dental caries inequalities among Chinese children are consistent with studies from other countries. Dental caries is related to lower socioeconomic conditions, such as parental education, family income and area [22]. Untreated decayed teeth increase consistently as the family income decreases [23], which is consistent with the finding of this study. The places where people live (urban or rural areas) have also been reported as a factor that influence the association between health outcomes and related factors [24]. Education was considered as the most important indicator of socioeconomic inequalities in children's caries [25], a finding consistent with this study. In this study, mother's education appeared to be a stronger predictor of children's dental caries than family income or area. This is probably because highly educated mothers are more likely to have better oral health knowledge and to provide better care for the children, particularly related to behaviors such as diet, oral hygiene and use of dental services. Furthermore, mother's education is also linked to income, which impacts on better living conditions and facilitates access to oral hygiene products, better diet and preventive services [26]. Several studies demonstrated that the level of caries is associated with deprivation [27]. Mothers in lower socioeconomic positions are more likely to allow their children to consume unhealthy food such as sugary, fatty and acidic food [28,29] and to select food that optimizes quantity rather than nutritional quality, compared to mothers in higher socioeconomic positions [30,31], which may result in dental caries. Moreover, children who belong to families at the bottom of the social hierarchy are more likely to have inadequate access to food, food shortages and inadequate eating patterns. These dietary patterns are likely to lead to an increased consumption of fermentable carbohydrate, a risk factor for dental caries [30,31]. These dietary habits are implicated in the mechanism linking socioeconomic factors and dental caries [30]. On the other hand, children living in countries and communities with higher socioeconomic conditions are more likely to consume food with high calories, snacks, soft drinks and even fermentable carbohydrates [32,33]. This could also contribute to an increased risk of dental caries and weight gain in developed countries [34]. Socioeconomic inequalities in dental caries are not only attributed to different patterns of food and drinks consumption, but also to other oral health behaviors such as toothbrushing habits [35], limited access to dental healthcare services, poor oral hygiene and poor living conditions, which could accelerate the development of decayed teeth [28]. Socioeconomic factors and material abilities are also strongly linked children's growth [36]. Poorer families are more likely to experience poor dietary intake due to reduced access to healthy food containing less sugar and fat, and rich in essential nutrients such as milk and protein [13]. Poor diet also impacts children's abilities to concentrate, sleep and attend schools, which could also contribute to undernutrition [28]. Low income is also associated with poor sanitation and poor hygiene that lead to increased infections in children, which detrimentally affect children's development [37]. In the current study, there was consistent income inequality in weight gain than in height gain. This was consistent with findings by Wagstaff and Watanabe [38] who concluded that inequalities in being underweight tend to be larger than inequalities in stunting [38]. Unlike in developed countries, in China, obesity and overweight are more common among those with a higher socioeconomic status. Factors such as overconsumption of food among children in higher socioeconomic families in developing countries leads to the risk of excessive weight gain [36]. Economic wealth and income inequalities also impact childhood development as they usually use private transportation rather than walking and engage in sedentary entertainment such as video games, which influence their energy balance [39]. Sugar consumption is a leading risk factor for both caries and being overweight due to a higher energy intake [40]. However, in this study, sugar consumption was significantly associated with dental caries, but not with children's growth. The lack of significant association between sugar consumption and anthropometric measures observed in this study could be because sugar is not the sole determinant of children's growth. The nutritional transition in China has been linked to an increase in consumption of animal proteins [41]. This increase of energy intake is considered as a key determinant of children's growth [42]. The strength of this study is in using longitudinal data to assess socioeconomic inequalities in both dental caries and children's growth in the same population of preschool children in China. There are some limitations worth mentioning. Approximately 30% of the participants were lost at follow-up, mainly because the children were recruited from kindergartens and they could have moved from kindergartens as their parents changed jobs. Furthermore, those who left the study were to a great extent similar to those included in the analysis in terms of clinical outcomes and sociodemographic characteristics. Secondly, a longer time interval between the two assessments could have affected the findings; this however was not possible given the short period children spend in the kindergarten. Thirdly, only children who attended selected kindergartens were included in this study. These children are usually from families with higher income who can afford kindergarten fees. It is possible that if the study was conducted among the general population, greater inequalities could have been observed. --- Conclusions The study demonstrated socioeconomic inequalities in changes in both dental caries and growth among preschool Chinese children. While mother's education appeared to be the strongest predictor of dental caries, income was the only factor significantly associated with weight and height gain. --- Author Contributions: A.S., W.S., conceived the research idea. E.B. advised on the study design. A.S. conducted the analysis and wrote the first draft. E.B. and W.S. advised on the statistical analysis and contributed to writing the manuscript. All authors have read and agreed to the published version of the manuscript. --- Conflicts of Interest: The authors declare no conflict of interest.

Background: This study aimed at assessing socioeconomic inequalities in the increment of dental caries and growth among preschool Chinese children, and to assess the best predictor of socioeconomic inequality for each of these conditions. Methods: This is a longitudinal population-based study. The sample data included preschool children living in three cities of the Liaoning Province, China. At baseline, 15 kindergartens with 1111 participants were included and dropped to 772 with a response rate of 70% at follow-up. Mean ages at baseline and follow-up were 50.82 and 60.55 months, respectively. Median follow-up time was 10.12 months. Data were collected through structured questionnaire, oral examination and anthropometric measurement. The questionnaire collected information on sex, age, family income, mother's education and children's dietary habits. The numbers of decayed, missing and filled teeth (DMFT) was used to indicate dental caries. Weightand height-for-age z-scores were calculated using the WHO Growth Standard. Multilevel analysis was used to assess the association between baseline socioeconomic position and each of dental caries and child's growth. Results: Mother's education was negatively associated with increments of DMFT. Family income was not significantly associated with DMFT in the fully adjusted model. The association persisted after accounting for other socioeconomic and dietary factors. Higher income was positively related to an increase in the weight-for-age z-score. The relationship between income and changes in the height-for-age z-score was positive and significant in the second highest income group. Conclusions: Mother's education appeared to be the strongest predictor of increments of dental caries. Only income was significantly associated with an increase in children's weight and height.

Introduction Disability is a general term used to denote impairments, activity limitations, and an individual's restriction in participating in various activities, which are the result of the interaction between human physical characteristics and the society in which adults with disabilities live [1]. Referring to the International Classification of Functioning, Disability, and Health (ICF), disability mainly consists of "functional impairments", "activity limitations", and "participation restriction" (WHO, Geneva, Switzerland, 2019). The data released by the World Health Survey (WHS) estimated that more than 1 billion people worldwide suffer from various disabilities, accounting for 15% of the world's population [2]. As a developing country, China has the largest disabled population in the world. It was estimated by the China Disabled Persons' Federation that the number of disabled people had reached 85.02 million in 2010, implying that 6.2% of the total population are suffering from various disabilities, and this proportion is gradually increasing (CDPF, 2021). Individuals with functional limitations and physical impairments are usually disadvantaged in their participation in society, which is contradictory to fundamental human rights and affects their health and well-being [3][4][5][6]. Compared to the non-disabled population, people with disabilities are at a higher risk for mental health issues due to the many challenges caused by poor health, lower education levels, employment difficulties, lower income levels, higher incidences of poverty, and limited social participation [7,8]. Previous studies have found that people with disabilities suffer from more serious anxiety and depression owing to physiological diseases, and nearly all disabled people have varying psychological problems. For instance, the quality of life of disabled people is significantly lower than that of healthy people [9], and they are more likely to experience symptoms of anxiety and depression [10,11], which lead to the deterioration of their overall mental health status. More importantly, in the context of accelerating global population aging, the number of disabled people in the older population will also rise, which will limit their social participation, and thus worsen their mental health level. The data showed that nearly one out of every five older adults with ADL disability is associated with depression in the United States, and that the prevalence of depression among the elderly with disabilities reached 20% [12]. Depression is more common among older people in Republic of Korea; nearly 20-70% of older people were affected by varying degrees of depression [13]. The depression seen amongst disabled people in China is also not encouraging. The detection rate of depression was 17.40% in 2021, and this rate was even higher among middle-aged and elderly adults with disabilities, reaching 54.5% [14]. Numerous studies have confirmed that depression was associated with increasing medical costs [15], mortality [16], a lower quality of life, and lower levels of well-being [17,18]. How to encourage middle-aged and older adults with disabilities to actively participate in social activities and improve their mental health status has become a significant issue that the Chinese government has to address. Fortunately, the development of the information society provides a convenient opportunity for individuals with disabilities to communicate and participate in social activities, the basic manifestation of which is the popularity and widespread use of the internet. According to the data issued by the China Internet Network Information Center (CINIC), the total number of internet users in China increased to 1.032 billion in 2021, and the internet penetration rate reached 73.0%. Among older adults, more than 119 million utilize the internet, with an internet penetration rate of 43.2% (CINIC, 2021). Middle-aged and elderly people can receive massive amounts of information, communicate instantly, and use a variety of convenient services without leaving their home via the use of the internet. Internet technology is reshaping the form by which people with disabilities can participate in social activities, and has excellent health spillover effects. Previous studies have showed that internet use has a positive impact on depression, mental health, social participation, and the subjective well-being of the elderly [19][20][21][22]. White (2002) mentioned that the use of the internet can improve the elderly's participation in online leisure activities and social activities, and this is beneficial to reducing loneliness and depression [23], thereby improving their life satisfaction and subjective well-being [24,25]. Cotten et al. (2012) found that internet use was positively related to the mental health of retired adults, which could reduce the risk of depression of retired elderly people by 20-28% [26]. Against this background, some scholars began to focus on the impact of internet use on the mental health of disabled adults, but existing studies have not reached a consensus on the health spillover effects of internet use. Most studies suggested that internet use can significantly reduce the risk of depression among older people with disabilities. Using a randomized controlled trial, Cotten et al. (2013) found that internet use could significantly reduce the risk of depression symptoms among elderly disabled people [27]. Duplaga and Szulc [28] held that internet use could decrease the loneliness experienced by individuals with disabilities by 40%. Birnie and Horvath [29] found that those who were not active in the physical world may utilize the internet more frequently and benefit from it. For those people with disabilities, online communication may be regarded as a compensation for the lack of social interaction. Lee and Cho [30] found that frequent internet use could significantly reduce the depression levels of disabled people, and the use of social media could help them establish social support networks and healthy psychological tendencies. A few scholars, however, have drawn different conclusions, believing that there is a huge digital divide for disabled people. The proportion of disabled people using the internet is significantly lower than that of the general population [20,31,32], and internet use was found to be positively associated with depression levels, loneliness, and stress [33]. Overall, the existing literature has some shortcomings. First of all, many scholars have discussed the effect of internet use on the mental health of the elderly in high-income countries, but more empirical evidence is needed from middle-and low-income countries. Moreover, although a few studies concentrated on the impact of internet use on the mental health and depression levels of disabled people, they ignored that the different purposes of internet use may have different effects on depression risk, and this has not been addressed in any previous studies. Third, the digital divide caused by disabilities varies significantly with the types of disabilities, so internet use may have different impacts on the risk of depression among older adults with different types of disabilities. However, few works in the literature have explored the possible differential effects of internet use on different types of disabilities. Finally, there is a digital gap between urban and rural areas, and the internet penetration rate in rural areas is only 57.6%, far below the average of 73.0%, which may have different impacts on the mental health of people with disabilities. This article aims to respond to the above questions, and the possible contributions are as follows: First, we used the nationally representative China Health and Retirement Longitudinal Survey (CHARLS) to measure the internet use of middle-aged and elderly people with disabilities from three aspects-internet use, frequency of internet use, and social activities related to internet use-and further systematically explored the association between internet use and the risk of depression among them in the information society. We aimed to provide instant and verified evidence to allow the government to actively address the problem of "ageing brought about by disability" and "disability caused by ageing". In addition, we classified the disabled people based on the types of their disabilities, and analyzed the impact of internet use on people with different types of disabilities so as to provide evidence regarding maximizing the positive spillover effects of internet use on health. Finally, we analyzed the possible age and urban-rural differences. --- Materials and Methods --- Data Source The data used in our study were obtained from the 2018 China Health and Retirement Longitudinal Study (CHARLS). CHARLS is a national and continuous largescale social survey project conducted by the Chinese Social Science Research Centre at Peking University, which aims to collect demographic, internet access, health status, and other information from middle-aged Chinese people (aged <unk> 45). The database consisted of 7 modules, including "Demographic Information", "Family", "Cognition and Depression", "Healthcare & Insurance", "Work, Retirement & Pension", "Household Income & Expenditures" and "House Property". According to the research purpose, this study mainly used the "Demographic Information", "Cognition and Depression" and "Healthcare & Insurance" modules of the 2018 CHARLS to analyze the impact of internet use on depression risk among middle-aged and older adults with disabilities. We cleaned the data as follows: firstly, we matched all modules and removed the duplicate samples to obtain a sample size of 19,816. Secondly, we only kept the samples with disabilities to obtain a sample size of 5329. Finally, after removing the samples with extreme values of household income and missing values of all controlled variables, a final sample size of 5220 was used to analyze the association between internet use and depression risk in middle aged and older adults with disabilities. Figure 1 presents the process of sample selection. --- Variables Dependent variables. We used depression to represent the mental health level of middle-aged and older adults with disabilities. Depression has been proven to be the most vital predicting indicator of mental health. Residents with depressive symptoms are more likely to suffer from serious psychological problems [34]. In our study, we adopted the 10item Center for Epidemiologic Studies Depression Scale (CES-D-10) to assess the depression status of individuals. The CESD-10 included 10 questions, and each question inquired as to the respondent's mood during the previous week. Each respondent was asked to answer each question with one of four options ("Rarely or none of the time" = 0, "Some or a little of the time" = 1, "Occasionally or a moderate amount of the time" = 2, and "Most or all the time" = 3). After positive emotional responses were reverse-scored, we finally obtained a summed score of 0-30. Referring to previous studies, the cut-off point of CES-D-10 was 10, and the respondents whose score exceeded 10 were deemed to be more likely to suffer from depression [35]. Therefore, we generated a dummy variable of "depression", and when a respondent's score exceeded 10 we assigned a value of 1, and 0 was valued when the score was less than 10. In this study, the Cronbach's alpha for CES-D-10 scale was 0.802, which demonstrated sufficient reliability and validity in CES-D-10. Table 1 shows that nearly half of the middle-aged and older adults with disabilities had depressive symptoms. Independent variables. According to the design of the 2018 CHARLS questionnaire, we selected three indicators, "internet use", "the frequency of internet use", and "social activities related to internet", to measure the internet use status of middle-aged and elderly adults with disabilities. First, we selected the question "Did you use the internet in the last month?" to measure the internet use of individuals, and we assigned a value of 1 if the respondents answered "yes"; otherwise, the value was 0. We selected the question "How often in the last month did you use the internet? Almost daily, almost every week, not regularly, or never?" to measure the respondent's internet use frequency. We gave a value of 0, 1, 2, or 3 if the respondent chosen "never", "not regularly", "almost every week", or "almost daily", respectively. Finally, we selected the question "What do you usually do on the internet?" to measure social activities related to internet use. There were four options for the response to this question in the questionnaire: "Chatting", "Watching --- Variables Dependent variables. We used depression to represent the mental health level of middle-aged and older adults with disabilities. Depression has been proven to be the most vital predicting indicator of mental health. Residents with depressive symptoms are more likely to suffer from serious psychological problems [34]. In our study, we adopted the 10-item Center for Epidemiologic Studies Depression Scale (CES-D-10) to assess the depression status of individuals. The CESD-10 included 10 questions, and each question inquired as to the respondent's mood during the previous week. Each respondent was asked to answer each question with one of four options ("Rarely or none of the time" = 0, "Some or a little of the time" = 1, "Occasionally or a moderate amount of the time" = 2, and "Most or all the time" = 3). After positive emotional responses were reverse-scored, we finally obtained a summed score of 0-30. Referring to previous studies, the cut-off point of CES-D-10 was 10, and the respondents whose score exceeded 10 were deemed to be more likely to suffer from depression [35]. Therefore, we generated a dummy variable of "depression", and when a respondent's score exceeded 10 we assigned a value of 1, and 0 was valued when the score was less than 10. In this study, the Cronbach's alpha for CES-D-10 scale was 0.802, which demonstrated sufficient reliability and validity in CES-D-10. Table 1 shows that nearly half of the middle-aged and older adults with disabilities had depressive symptoms. Independent variables. According to the design of the 2018 CHARLS questionnaire, we selected three indicators, "internet use", "the frequency of internet use", and "social activities related to internet", to measure the internet use status of middle-aged and elderly adults with disabilities. First, we selected the question "Did you use the internet in the last month?" to measure the internet use of individuals, and we assigned a value of 1 if the respondents answered "yes"; otherwise, the value was 0. We selected the question "How often in the last month did you use the internet? Almost daily, almost every week, not regularly, or never?" to measure the respondent's internet use frequency. We gave a value of 0, 1, 2, or 3 if the respondent chosen "never", "not regularly", "almost every week", or "almost daily", respectively. Finally, we selected the question "What do you usually do on the internet?" to measure social activities related to internet use. There were four options for the response to this question in the questionnaire: "Chatting", "Watching news", "Watching videos", and "Playing games". The specific definitions are presented in Table 1. Controlled variables. To reduce potential estimation deviation caused by omitted variables, four types of variables were controlled: (1) demographic characteristics variables, including age, gender, marital status, residence, and ethnicity; (2) variables related to health, including number of chronic diseases, self-rated health, smoking, and drinking; and (3) socioeconomic variables, including education condition, household income (logarithm), pensions, health insurance, and work status. Considering that the differences in internet use in different provinces may have an impact on depression levels, we controlled the provinces in which the middle-aged and older people were located. The definitions and descriptive results were presented in Table 1. --- Models Because the dependent variable, depression status, is a dichotomous variable with values of 0 to 1, the logit model was used to estimate the impact of internet use on the depression risk of middle-aged and older adults with disabilities. The basic model was set as: P(y = 1|x i ) = log p 1 -p = <unk> 0 + <unk> 1 Internet i + k <unk> i=1 <unk> k x ik + <unk> i P(y = 1|x i ) represents the probability of experiencing depression risk among middle- aged and elderly adults with disabilities; <unk> 0 is the intercept term; Internet i represents the internet use level of the i th individual; <unk> 1 is the correlation coefficient of the impact of internet use on the depression risk of middle-aged and elderly adults with disabilities; <unk> k i=1 <unk> k x ik represents a series of control variables; and <unk> is a random error term. --- Results --- Descriptive Results Table 2 shows the results of the Chi-square test for depression symptoms and internet use among middle-aged and older adults with disabilities. As shown in Table 2, there were significant differences between the symptoms of depression of disabled adults who used the internet and those who did not utilize the internet (p <unk> 0.001). --- Regression Results --- The Association between Internet Use and Depression Risk among Middle-Aged and Older Adults with Disabilities We used the maximum likelihood method (MLE) to estimate the association between internet use and depression status among individuals with disabilities. Column (1) of Table 3 showed that internet use was negatively associated with depression in people with disabilities when we did not include any control variables (marginal effects = -0.143, p <unk> 0.01). Column (2) indicated that the marginal effect decreased to -0.055 after controlling for other confounding variables (p <unk> 0.05). Columns (3) and (4) of Table 3 reported the relationship between the frequency of internet use and depression in disabled adults. The results suggested that compared to disabled people who never used internet, those who used the internet almost daily had a lower depression risk (p <unk> 0.05), which was consistent with the results of Columns (1) and (2). Among the controlled variables, increasing age, male gender, individuals married with a spouse, drinking wine, higher education levels, better self-rated health, and higher household incomes were negatively associated with depression risk (p <unk> 0.01). In contrast, living in rural areas and having chronic diseases were positively associated with depression risk among disabled adults (p <unk> 0.01). --- Robustness Test We used the substitute variable method to check the robustness of the regression results. Referring to the 2018 CHARLS questionnaire design, we selected the questions "Does your residence have a broadband internet connection?" and "Do you use WeChat?" as substitute variables for internet use. Table 4 reports the regression results for the impact of having a broadband internet connection and WeChat use on depression risk. The results in Column (1) showed that disabled people who had access to broadband internet in their residence had a lower probability of depression risks (p <unk> 0.01). Column (2) in Table 4 suggested that using WeChat was negatively associated with depression symptoms relative to those who did not use WeChat (p <unk> 0.1). The results in Table 4 are consistent with those in Table 3. In order to effectively reduce the potential selective bias between internet use and depression risk, the PSM method was used to estimate the association between internet use and depression risk. The results of Table 5 show that the ATT coefficients for K-nearest neighbor matching and kernel matching were -0.0838 and -0.0574, respectively, which means that internet use was negatively associated with depression risk. The ATT coefficients were consistent with the regression coefficient shown in Table 3, which indicates that internet use could significantly decrease the probability of depression risk after overcoming the potential selective deviation. --- The Association between Social Activities Related to Internet Use and Depression Risk among Adults with Disabilities Columns ( 1)-( 4) of Table 6 report the association between social activities related to internet use and depression risk among middle-aged and older adults with disabilities. The results showed that watching videos, reading the news, and chatting via the internet were found to be negatively associated with depression risk. However, playing games via the internet had no significant impact on depression risk. (3) Due to space limitations, the regression results of the controlled variables were not reported. --- The Association between Internet Use and Depression Risk among People with Different Types of Disabilities Table 7 reports the association between internet use and depression risk among people with different types of disabilities. The results suggest that internet use had different effects on depression risk among people with different types of disabilities. Internet use had a greater impact on the risk of depression in adults with physical disabilities (p <unk> 0.05). Conversely, in terms of hearing-related, vision-related, and other types of disabilities, internet use had no impact on depression risk. Table 7. The association between internet use and depression risk for different types of disabilities. ( (3) Due to space limitations, the regression results of the controlled variables were not reported. --- Heterogeneous Analysis Columns (1) and (2) of Table 8 report the age differences regarding the impact of internet use on the depression risk of disabled adults. The regression results showed that internet use had a greater effect on elderly people with disabilities than on non-elderly people. Columns (3) and (4) of Table 8 report the regional differences in the impact of internet use on depression risk. The results suggest that internet use could significantly decrease the probability of depression risk among urban adults with disabilities by 8.1%, but it had no impact on the rural group. --- Discussion Previous studies have mainly explored the impact of internet use on the health of the elderly; almost no studies have analyzed the potential association between internet use and depression risk among people with disabilities [26,36]. This study analyzed in detail the association between internet use and depression risk among Chinese middleaged and elderly adults with disabilities by using the 2018 CHARLS data. We found the following: (1) Internet use was negatively associated with depression risk, and the higher the frequency of internet use, the lower the probability of the depression risk. (2) Different social activities related to internet use had different impacts on the depression risk. The decline in depression risk was mainly related to watching videos, watching the news, and chatting via the internet, but playing video games had no impact. (3) Internet use was found to significantly reduce the risk of depression among middle-aged and elderly adults with physical disabilities, but had no impact on those with hearing-related, vision-related, and other types of disabilities. (4) The association between internet use and depression risk was mainly reflected in older adults with disabilities and those living in urban areas, but had no impact on middle-aged and rural disabled residents. Disability is an increasingly serious public health problem throughout the world, especially for elderly adults [37]. Individuals with functional limitations or physical impairments are usually at a disadvantage regarding their health and subjective well-being [9,10]. The advance of the information society has provided a chance for disabled people to participate more in social life. The internet integrates multiple functional advantages, such as human-computer communication, information sharing, and leisure entertainment [38]. People with disabilities can take part in social communication, carry out information searches, and enjoy leisure entertainment online through the internet, which can reduce their risk of loneliness and depression [27,29,30]. A study from Poland showed that internet use could reduce the loneliness of disabled people by 40% [28], and Cotten, et al. [39] drew a similar conclusion. Conversely, some studies report that internet use has a negative impact on the mental health of disabled people in the information society, and a significant positive correlation was found between internet use and depression, loneliness, and stress [31,33]. Our study supports the first view. We found that internet use could significantly reduce the risk of depression for middle-aged and elderly people with disabilities, and the higher the frequency of internet use, the greater the reduction in depression risk. A possible reason for this may be that, through the internet, people with disabilities can use social media to keep in touch with their relatives and friends, enjoy leisure and entertainment, search for health information, and improve their health awareness and health literacy, all of which can help them improve their overall health status [38,40]. For example, some scholars found that during the prevention and control of the COVID-19 pandemic, elderly patients with poor health received timely and effective treatment through online consultations. Moreover, they also learned relevant knowledge regarding disease prevention through the internet, and they decreased their depression risk through online leisure and entertainment during quarantine [41]. In addition, we found that the association between internet use and the risk of depression was constrained by different social activities related to internet use. Online chatting was found to help people overcome time and space barriers, maintain a stable interpersonal network, and obtain emotional support (e.g., by improving family members' communication and strengthening relationships with neighbors), thereby alleviating their sense of helplessness and loneliness [42]. Watching the news and videos was found to not only help increase their leisure and entertainment activities, but also helped them obtain health-related medical information, which is helpful in improving their health level [38]. One study found that the COVID-19 pandemic further increased the internet penetration rate, and a large number of middle-aged and elderly people acquired knowledge about infectious diseases and improved their health literacy through their use of the internet [41]. However, playing games was found to have a limited effect on reducing their depression risk because this would occupy the time that people should have spent on exercise or social participation. Some studies found that the frequency of playing games online was negatively correlated with one's level of happiness [28]. Compared with healthy people, people with disabilities were more likely to indulge in the entertainment functions of the internet [43], which is also related to certain unhealthy behaviors and can decrease the positive spillover effects of internet use relating to health. We also found that internet use had different effects on the depression risk in relation to different types of disabilities. Internet use had the greatest effect on reducing the depression risk of people with physical disabilities, but had no effect on the those with hearing-related, vision-related, and other types of disabilities. A possible reason for this may be that people with hearing-related, vision-related, and other disabilities experience a stronger digital divide, and they may be unable to effectively communicate and participate in social life through the internet; thus, internet use has no impact on their depression levels. In contrast, people with physical disabilities are merely limited in their daily activities; their disabilities do not affect their use of the internet. Moreover, a relatively higher education level is conducive to building online interactions, searching for information, and enjoying the leisure and entertainment functions of the internet, which can significantly reduce the risk of depression. Previous studies have shown that disabled people have strong heterogeneity, and the digital divide varies based on the type of the disability [31,44]. Our findings further confirm that the impact of internet use on the health of people with disabilities is affected by different types of disabilities. Finally, we found that there were significant age differences and urban-rural differences regarding the association between internet use and depression. Internet use mainly reduced the risk of depression among older adults and people living in urban areas, which is consistent with the findings of previous studies [36]. The internet penetration rate in urban areas was found to be significantly higher than in rural areas, so people with disabilities in urban areas have more convenient access to internet services. As a consequence, these groups are more likely to use the internet for social interaction, learning, participation in social activities, and other activities, contributing to a higher positive impact of internet use on their depression level [41]. Macro data also confirm this difference. At present, the internet penetration rate in rural China is only 57.6%, far lower than the average rate of 73.0% (CINIC, 2021). Therefore, strengthening the internet infrastructure, improving the penetration rate of the internet in rural areas, and achieving the positive spillover effect of internet use on individuals' health should be a core strategy of the Chinese government. --- Conclusions In conclusion, our study suggests that internet use can significantly reduce the risk of depression, and the reduction effect in depression risk is significantly related to the frequency of internet use. However, the association between the internet use and depression risk of disabled people was affected by different social activities related to internet use and the type of disabilities. Watching videos, chatting, and watching the news via the internet was found to decrease the depression risk among disabled people. Moreover, internet use mainly affected those with physical disabilities, and it was found to have no impact on people with other types of disabilities. Finally, we found that there were age differences and urban-rural differences in the association between internet use and depression risk. Internet use had a greater impact on the elderly and those who lived in urban areas. This study had several limitations. First of all, the sectional data of 2018 CHARLS were used to analyze the association between internet use and depression risk among middleaged and older adults with disabilities, so this study essentially intended to elucidate a correlation relationship. Although we used the PSM method to overcome the possible selective bias to some extent, we did not address the endogenous problem caused by missing variables and reverse causality. Thus, we failed to obtain a causal relationship. Moreover, we used the frequency and social activities close to the internet to measure internet use among middle-aged and elderly people with disabilities. Unfortunately, we lacked indicators such as the duration of internet use, which reflect the intensity of an individual's internet use, and which may also have an impact on their depression level. Finally, we explored the association between internet use and depression risk in people with different types of disabilities, but disability severity was not fully considered; it is clear that the impact of internet use on depression is affected by disability severity. Unfortunately, we failed to take this indicator into consideration due to data limitations, and we will continue to conduct related studies once the data are available. --- Data Availability Statement: The datasets generated for this study are available on request to the corresponding author. --- Author Contributions: Conceptualization, X.Z. and L.Z.; methodology, X.Z. and A.L.; software, X.Z.; validation, N.C., H.T.A.K. and B.G.; formal analysis, X.Z. and A.L.; investigation, B.G. and N.C.; data curation, X.Z.; writing-original draft preparation, X.Z.; writing-review and editing, H.T.A.K. and L.Z.; supervision, H.T.A.K. and L.Z.; funding acquisition, L.Z. All authors have read and agreed to the published version of the manuscript. --- Conflicts of Interest: The authors declare no conflict of interest.

Background: Globally, nearly 15% of people suffer from various kinds of disabilities, and China has the largest disabled population in the world. The poor mental health status of people with disabilities has become an essential issue in most countries. The main aim of this study was to explore the potential impact of internet use on depression risk among middle-aged and older adults with different types of disabilities. Methods: The data used in this study were obtained from the 2018 China Health and Retirement Longitudinal Study (CHARLS) collected by Peking University. A binary logit model was used to analyze the impact of internet use on the depression risk among adults with disabilities, and the substitute variable method and the propensity score matching method were used to examine the robustness of the results. Results: (1) Internet use was negatively associated with depression risk among disabled people, and the higher the frequency of their internet use, the lower the probability of their depression risk. (2) Different social activities related to the internet had different impacts on the depression risk, and the decline in depression risk was mainly related to watching videos, watching news, and chatting via the internet. (3) Internet use reduced the depression risk of adults with physical disabilities, but had no impact on those with other types of disabilities. Conclusions: Our study suggests that internet use may have a positive spillover effect on decreasing the depression risk of disabled people, but the reduction effect is significantly affected by the social activities related to the internet and the types of disabilities.

Open Access --- AbstrAct Outcome measures Both absolute and relative measures were calculated. First, age-standardised mortality rates have been calculated, directly standardised to the Belgian population. Second, mortality rate ratios were calculated using Poisson's regression, adjusted for education, housing conditions, attained age, region and migrant background. results This study highlights inequalities in sitespecific cancer mortality, both related to being employed or not and to the occupational group of the employed population. Unemployed men and women show consistently higher overall and site-specific cancer mortality compared with the employed group. Also within the employed group, inequalities are observed by occupational group. Generally manual workers and service and sales workers have higher site-specific cancer mortality rates compared with white-collar workers and agricultural and fishery workers. These inequalities are manifest for almost all preventable cancer sites, especially those cancer sites related to alcohol and smoking such as cancers of the lung, oesophagus and head and neck. Overall, occupational inequalities were less pronounced among women compared with men. conclusions Important SE inequalities in site-specific cancer mortality were observed by employment and occupational group. Ensuring financial security for the unemployed is a key issue in this regard. Future studies could also take a look at other working regimes, for instance temporary employment or part-time employment and their relation to health. --- IntrODuctIOn In industrialised countries, life expectancy increased during the epidemiological transition, with the most important causes of death (COD) shifting from infectious to chronic diseases such as cancer. 1 This favourable longevity trend went together with an increase in socioeconomic (SE) health inequalities in general 2 3 and with cancer inequalities in particular. 4 According to the fundamental cause theory (FCT), inequalities result from the differential distribution of valuable resources that can be used to avoid adverse health outcomes. 2 This implies that SE inequalities in health will be larger for cancers that are more preventable, either by avoiding risk behaviour or by access to medical interventions, 3 5 than for less or non-preventable cancers. SE inequalities in health are one of the most important challenges for public health policies. Therefore, it is crucial to thoroughly document these inequalities. Up till now, most --- Site-specific cancer mortality inequalities by employment and occupational groups: a cohort study among Belgian adults, 2001-2011 studies assessing health inequalities have used only one indicator of socioeconomic position (SEP), most often educational attainment or material wealth. However, detailed research on occupational inequalities in site-specific cancer mortality is scarce 6 7 and even non-existent in Belgium. Yet, many studies have shown that, although these SEP indicators are strongly related, they may represent different aspects of SEP. [8][9][10] Therefore, it is important to establish the role of every single dimension of SEP for every cancer site separately. 9 11 12 Hence, this paper probes into SE differences in site-specific cancer mortality, using employment and occupational group as measures of SEP, accounting for the other SEP indicators. By doing so, we aim to estimate the net association between employment and occupational group and site-specific cancer mortality in Belgium, which has not yet been studied. Our first research aim is to examine the magnitude of the association between site-specific cancer mortality and employment group, net of one's educational attainment and housing conditions. The second research aim is to assess whether among the employed group, occupational group is associated with site-specific cancer mortality, again net of education and housing conditions. Based on the FCT, we assume that we will observe inequalities by employment and occupational status, especially for the more preventable cancer sites. --- DAtA AnD methODs --- Design and study population Data are derived from a record linkage between the Belgian censuses of 1991 and 2001 and register data on emigration and cause-specific mortality for the follow-up period 01/10/2001 to 31/12/2011. This unique population-based dataset includes information on mortality, emigration, COD, sociodemographic (SD) and SE variables of the total de jure population of Belgium. All SD and SE variables are extracted from the 2001 census, apart from occupational status for which the 1991 census contains the most recent detailed information. All Belgian individuals who are alive at the 2001 census and who were within the economically active age range (25-65 years) at the census of 1991 are included in the study. For these individuals, we linked information on employment and occupational status, stemming from the 1991 census, irrespective of their employment status at the 2001 census. The study population includes 2 333 479 Belgian women of whom 49% are employed and 2 231 385 Belgian men of whom 80% are employed. Age is included as a time-varying covariate to account for age changes during the 11-year follow-up period. To do so, individual follow-up time is split into episodes of 5 years attained age groups using Lexis expansions. 13 Consequently, the age distribution of this population ranges from 35 to 85 years. --- Variables All cancer sites representing at least 1% of the total cancer mortality are included. Table 1 gives an overview of these cancer sites, the corresponding ICD-10 codes and the level of preventability. To classify the cancers by level of preventability, we apply the criteria used in Mackenbach's study: 14 amenability to behavioural change and to medical interventions. Cancer sites are amenable to behavioural change if the combined population attributable fraction of mortality for overweight and obesity, low fruit and vegetable intake, physical inactivity, unsafe sex, smoking and alcohol consumption was larger than 50% for the European population in the Global Burden of Disease and Risk Factors study. 15 Cancer sites are considered as amenable to medical interventions, if the 5-year relative survival rate for Belgian women and men in the EUROCARE project was higher than 70% between 2000 and 2007 16 and/or if effective screening is available in Belgium. 17 This study probes into cancer mortality inequality among the Belgian population within the economically active age range by using employment and occupational group as a measure of SEP. Both employment and occupational status are derived from the 1991 census and mortality is followed up for the period 2001-2011. Employment status is divided in four categories: employed, unemployed and looking for a job, unemployed and not looking for a job and disabled. Among the employed, we additionally defined their occupational group, based on the International Standard Classification of Occupations (see table 2): 18 managers and professionals, intermediate white-collar workers, service and sales workers, agricultural and fishery workers, skilled manual workers and unskilled manual workers. We choose the employed as a reference category and among the employed the managers and professionals. These choices have been made because of the size of these groups and because we assume lower cancer mortality for these groups, which facilitates the interpretation. --- Open Access --- statistical analyses To obtain a full picture of inequalities in cancer mortality, we calculate both absolute and relative measures, 19 using complete-case analysis. First, directly age-standardised site-specific mortality rates by employment and occupational group are calculated, using the Belgian population at the 2001 census as standard population. Second, mortality rate ratios (MRR) are calculated using Poisson's regression. We aim to assess the net effect of employment and occupational group on cancer mortality. Therefore, the Poisson's models are adjusted for educational attainment and housing conditions. Educational attainment is categorised using the International Standard Classification of Education: lower secondary education or less (ISCED 0-2, 'low'), higher secondary education (ISCED 3-4,'mid') and tertiary education (ISCED 5-6, 'high'). Housing conditions is the result of a combination of ownership (tenant or owner) and comfort of the house (low-comfort, mid-comfort and high-comfort), resulting in six categories. 20 Sensitivity analyses are additionally conducted without adjustment for education and housing conditions. These results are not shown but are available in online supplementary table 1 and 2. Important differences between the crude and net model are mentioned in the discussion section. As there is a strong association in Belgium between mortality and region 21 as well as migration history, 22 all Poisson's models are adjusted for region (Flanders, Wallonia and Brussels) and migrant background (native vs non-native). All analyses are stratified by sex and are performed using STATA 13.1. --- results --- Description of the study population The study population consists of almost 5 million Belgians within the economically active age range (table 2). Almost four out of five men are employed, whereas in women about one in two has a paid job. --- Absolute cancer mortality rates by employment and occupational status For almost all preventable cancer sites, men belonging to the employed category have systematically the lowest cancer mortality rates (table 3). Among the employed men, white-collar workers and agricultural and fishery workers have lower cancer mortality rates compared with manual workers (table 4). These results hold true for all preventable cancer sites, except for prostate cancer and malignant melanoma for which no inequality is observed. For the non-preventable cancer sites, less inequalities are observed, as expected based on the FCT. For the non-preventable cancer sites, inequalities generally represent differences between employed versus unemployed. For women, the absolute inequalities are less pronounced. Generally, mortality from preventable cancer sites is largest for women who are unemployed (table 3). Among the employed women there are no inequalities, except for lung cancer mortality which is somewhat more elevated for service and sales workers and manual workers compared with female managers (table 4). On the other hand, service and sales workers In this paragraph, we present the results of the net relative models, adjusted for education and housing conditions. Unemployed men and women have higher site-specific cancer mortality rates compared with employed people (table 5). This holds true for preventable and non-preventable cancers and are especially observed among the unemployed who are not looking for a job. To illustrate this with an example: unemployed men who are not looking for a job have a three times higher chance of dying from bladder cancer (MRR 3.37; 95% CI 3.05 to 3.73) compared with employed men. For most cancer sites (mainly the preventable cancers) particularly unemployed men and women who are either not looking for a job or disabled have consistently higher cancer mortality rates compared with employed men and women. Within the employed population, inequalities by occupation are also observed although more in men than in women (table 6). In men, managers appear to have higher colorectal and liver cancer mortality rates compared with men in other occupations. Furthermore, lower mortality rates in favour of manual workers are observed for several non-preventable cancer sites, among others cancer of the pancreas and kidney. Compared with male managers, skilled manual workers have a 7% higher lung cancer mortality rate. Another interesting observation is the mortality pattern of the agricultural and fishery workers. Compared with managers, they tend to die less from preventable cancers such as head and neck and oesophageal cancers. Their liver cancer mortality rate is even 76% lower compared with managers (MRR 0.24; 95% CI 0.15 to 0.39). In contrast, agricultural and fishery workers show higher mortality from some non-preventable cancer sites such as leukaemia and multiple myeloma. Female manual workers and service and sales workers have about 25% lower breast cancer mortality rates compared with female managers. --- Open Access --- DIscussIOn AnD cOnclusIOn methodological issues The findings are based on a high-quality and exhaustive dataset including the total Belgian population within the economically active age range. A numerator-denominator bias was eliminated through record linkage between census and register data. This dataset provides very rich information on SD as well as SE variables and mortality for a follow-up period of 11 years. This enables us to give precise estimates of site-specific cancer mortality inequalities at the individual level. However, these register data do not contain information on incidence or survival, nor on health behaviours, healthcare use or important aspects of the job (eg, occupational exposures, psychosocial factors), which are all likely to be associated with cancer outcomes. 23 Therefore, it is difficult to make solid conclusions on the relative importance of all these factors to explain the observed SE inequalities in cancer mortality. 23 24 We included the total Belgian population that belonged to the economically active age group (25-65 years) at the Reference category is managers and professionals. All analyses are adjusted for current age, region, migrant background, educational attainment and housing conditions. Table 6 Continued census of 1991, independently of their actual occupational status. We decided so because we were interested in the association between cancer mortality and both employment and occupational group. By doing so, we avoided a selection effect due to including only the healthy workers. 6 25-28 Our results indicate the importance of this issue, showing highest (site-specific) cancer mortality among the unemployed groups. Nevertheless, we cannot fully exclude a selection effect among the employed population because we can assume that unhealthy persons are less frequently employed in physically demanding jobs. 26 27 A healthy worker effect might then be more likely in the groups of manual workers, which could partly explain some of the observed mortality patterns in favour of manual workers. On the other hand, the whitecollar workers may have less physically demanding jobs, which can counter this effect. Occupational information was derived from the census of 1991, which is the most recent source of detailed information since the census of 2001 does not contain detailed occupational information. Because of the lag time between some (occupational) exposures and cancer mortality, we do not consider this as a problem. 29 Due to the cross-sectional nature of this information, the occupation is not necessarily the longest job respondents were involved in nor do we have information on the duration of this occupation. 25 However, occupation was grouped into broad groups, which leads us to assume that the bias due to transitions between occupational categories will be rather small. 30 We decided not to use the robust distinction between manual versus non-manual workers nor did we focus on one specific cancer site in relation to one or more job exposures. These choices have been made in order to gain insights into the overall association between occupational status and cancer mortality. 26 Many studies assessing SE inequalities in cancer mortality use only one SE indicator at a time although the importance of including multiple indicators has already been pointed out 9 11 12 because different SEP indicators tap into different pathways. [8][9][10] Housing conditions are related to material and financial resources and therefore to healthcare utilisation. 31 Education captures the human capital acquired early in life and may be related to the ability to adapt health education messages and hence to health behaviours. 6 Occupational status on the other hand reflects the ability to realise this human capital in the labour market and is rather a reflection of one's social class at older ages. Employment as well as occupation is likely to (partially) capture material resources, access to healthcare as well as social networks and work-related factors such as stress, autonomy and occupational hazards. Hence, we need to analyse all aspects of SEP in relation to health outcomes, although these indicators are closely related. 6 In this paper, we focus on the net effect of occupational status, thereby adjusting for educational attainment and housing conditions. The results indicate an association between occupational status and site-specific cancer mortality, independently of education and housing conditions. --- Open Access Finally, we classified the cancer sites by their level of preventability (see Methods section). We acknowledge that this classification does not exclude the fact that some 'non-preventable' cancers are also related to behavioural change and/or medical interventions nor does it eliminate some possible overlap between these two criteria of preventability. However, to enhance the comparability with other studies, we decided to adopt this often-used classification. theoretical considerations on the main findings This study reveals inequalities in site-specific cancer mortality by employment and occupational groups. Generally SE inequalities are less pronounced in women compared with men, which is consistent with the literature. 7 23 28 The unemployed group (and especially those not looking for a job) shows both for men and women higher cancer mortality rates compared with the employed, which is in line with previous studies in France and the UK. 23 32 Possible explanations are financial insecurity, which is related to a lower use of healthcare services and an unhealthy lifestyle. 32 Some studies suggest that part of this association between unemployment and health might be due to health selection, 23 although others observed an ongoing effect after adjustment for pre-existing morbidity. 32 However, it is very likely that for the unemployed that are not looking for a job, as well as for the disabled group, health selection will be at play. Also within the employed group, inequalities by occupational group are observed. In men, these findings are in line with the FCT which expects larger inequalities for more preventable cancer sites. 3 5 The absolute results show that male manual workers have the highest site-specific cancer mortality rates whereas white-collar workers and agricultural and fishery workers have the lowest cancer mortality rates. This discrepancy between manual and non-manual workers is in line with previous findings in Europe. 6 33 These absolute inequalities are manifest for all preventable cancer sites (except for non-malignant melanoma and prostate cancer), with alcohol-related and smoking-related cancers (eg, cancers of the lung, head and neck and bladder) being the main contributors to these inequalities, as reported in previous studies. 25 28 30 33 34 The excess mortality for cancers of the head and neck and lung are also observed for male service and sales workers, which can be explained by the higher likelihood of occupational exposure to tobacco and alcohol in bars and restaurants. 27 Another interesting finding is the favourable cancer mortality pattern for farmers, which might be related to their healthy life style with less tobacco and alcohol use and more physical activity. 35 36 According to the FCT, the availability of valuable resources such as knowledge, money, power, prestige and beneficial social connections are the social causes of health inequalities. 2 When there is sound knowledge of the causes and cures of cancers, those in high SEP contexts, with greater access to resources, will disproportionally benefit from this knowledge. Several mechanisms have been suggested to explain this association, such as a differential acquisition of knowledge on health-damaging behaviours (eg, smoking, bad diet, alcohol intake or a lack of exercise). 6 7 10 23 26 37 Also, material factors are important: having financial difficulties might be related to poor living conditions 7 23 30 or the inability to optimise the use of health services. 10 33 37 Both access to and quality of healthcare are crucial for health outcomes in all stages, from prevention (eg, through cancer screening) to treatment. Finally, the social aspect of SEP is associated with health, for example stress-related factors or the level of social prestige. 25 Since we cannot adjust for health behaviours or healthcare use, it remains difficult to decide on the extent of inequalities due to each of these mechanisms. 26 Yet, our findings showed that the inequalities were especially large for the preventable cancer sites related to health behaviours such as smoking and alcohol use. Previous studies indeed reported higher smoking prevalence among manual workers and people working in the catering industry. 38 However, studies mediating the observed gradients for smoking reported an attenuated but still significant relation between SEP and mortality, 7 24 30 which assumes that there might be other factors at play as well. Sensitivity analyses not adjusting for education and housing conditions showed both for men and women lowest lung cancer mortality among the managers compared with all other occupational groups (see online Supplementary table 2). However, in the models adjusted for education and housing conditions (presented in table 6), the association reversed for women, whereas for men lung cancer mortality only remains somewhat higher for male skilled manual workers, which could be related to occupational exposures. In contrary, in the adjusted model, higher breast cancer mortality rates were observed in female managers compared with all other occupations except sales and service workers. This can be explained by differences in reproductive behaviour, with less children and a later age at first birth for the whitecollar occupations. 24 28 In the net model, male managers had higher colorectal and liver cancer mortality compared with all other occupations. Yet, in the unadjusted model, the association between occupation and colorectal cancer mortality disappeared (see online supplementary table 2). Moreover, compared with the white-collar occupations, manual workers do have lower mortality for several other cancer sites (eg, malignant melanoma and kidney). These results are counterintuitive, suggesting that there are other factors at play, such as an unhealthy and sedentary lifestyle of managers. 39 40 We can assume that the results of the net model are more likely to be due to differences related to the job itself. Yet, a healthy lifestyle may be induced by the social environment (eg, the colleagues at work), and therefore might not be excluded as a potential mechanism. 6 Specific occupational exposures and hazards may explain part of the association between occupational status and health. 7 10 23 Another important aspect of work related to health is the psychosocial aspect of the job, such as the Open Access sense of control and autonomy, the level of job strain or long working hours. 7 23 25 30 41 This could be an explanation for the excessive colorectal cancer mortality among male managers. Colorectal cancer is associated with perceived stress and could therefore be related to the long working hours and work strain as perceived by managers. 39 41 Implications and conclusion This paper highlights important SE inequalities in site-specific cancer mortality by employment and occupational groups. Being unemployed, and among the employed, being employed as a manual worker or service and sales worker is associated with higher overall and site-specific cancer mortality. These unfavourable mortality patterns among the unemployed and manual and service and sales workers were especially observed for the more preventable cancers, as we assumed based on the FCT. These occupational inequalities change when other SEP indicators are adjusted for. Multiple SEP indicators should be taken into account when studying SE inequalities in health 9 11 12 because these allow for different causal pathways. [8][9][10] Yet, it is difficult to identify to what extent the potential explanatory factors attribute to these inequalities because we do not dispose of data on incidence or survival, nor on data on risk factors, healthcare use or job characteristics. Future studies having access to these kind of data could help to unravel the complex interplay between incidence, survival and mortality and to clarify at which steps the social differences operate. 23 24 This study can be helpful in providing evidence for policy makers in order to reduce SE inequalities in cancer mortality. 37 Our results prove that there is still a long way to go. We observed for example that the unemployed groups are at a much higher risk of dying from cancer compared with the employed population. Ensuring financial security for the unemployed is a key issue in this regard. 32 Future research could also study other working regimes, for example, temporary or part-time employment and their relation to health. Finally, there could be an important role for the general practitioner to make sure that the unemployed are getting the healthcare they need. 32 Also, for the high-risks jobs, regular health checks at the work floor are needed in order to detect cancers at an early stage. --- competing interests None declared. ethics approval This research as well as the data adhere to the ethical code of scientific research in Belgium, and all authors have signed the ethical code. Provenance and peer review Not commissioned; externally peer reviewed. Data sharing statement Analyses are based on administrative data from the Belgian census, the Belgian mortality register and death certificates provided by statistics Belgium. The availability of the data is restricted. Permission for analyses must be granted after verification of the research goals by the Belgian commission for the protection of privacy.

strengths and limitations of this study ► This paper probes into socioeconomic (SE) differences in site-specific cancer mortality, using employment and occupational group as measures of socioeconomic position (SEP), accounting for two other SEP indicators (education and housing conditions). ► The findings are based on an exhaustive dataset including the total Belgian population within the economically active age range, which provides very rich information on sociodemographic as well as SE variables and mortality for a follow-up period of 11 years. ► These administrative register data do not contain information on incidence or survival, nor on health behaviours, healthcare use or important aspects of the job (eg, occupational exposures, psychosocial factors).

recent editorial in the International Journal of Epidemiology interpreted 'the psychosocial' as a concept that operates at a'meso-level'; that is, between macro-(social-structural) and micro-(individual) levels. It described the meso-level as encompassing such social formations as,'religious institutions, the family, the firm and the club...' (Martikainen, Bartley andLahelma, 2002: 1092). The editorial located the roots of 'psychosocial' thinking in the influential definition of health originally recorded in the Constitution of the World Health Organisation in 1946Organisation in. (ibid.: 1091)). This defined health as, 'a state of complete physical, mental and social well-being, and not merely the absence of disease and infirmity'. But the psychosocial is not only a medical theory and practice. It also represents a form of politics. This article addresses some historical elements of the psychosocial as a politics through an examination of the International Congress on Mental Health held in London in 1948. The Congress comprised international conferences on child psychiatry, medical psychology, and mental hygiene. These were brought together under the organising umbrella of the mental hygiene movement with its conference playing the largest and most prominent role. Its theme was 'Mental Health and World Citizenship'. This emphasised that healthy personality development was related to stages of growth and development which took place primarily within the family, but that this process was mediated by wider cultural patterns and social institutions (Frank and Mead, 1948;International Preparatory Commission Statement, 1948, p. 82-3). In apparent conformity with the International Journal of Epidemiology's claim regarding the origins of psychosocial thinking, the 1948 Congress used the WHO's 1946 description of health as a guiding definition. However, as this article shows through its examination of the mental hygiene movement, the roots of 'the psychosocial' in fact stem from earlier than this. If the psychosocial is a form of politics, what form does it take? The sociologist Nikolas Rose has influentially described social psychology and the psychosocial as intimately connected with modern western liberal democracies. He has argued that, Whereas the ideals of democracy are abstract and general -respect for the individual, personal autonomy, social responsibility, the control by the people over those who govern -the power of social psychology is to enable these to be made congruent with specific programs for managing particular problematic areas of social life. (Rose, 1998: 118) For Rose,...social and dynamic psychologies were to offer the possibility of making democracy operable through procedures that could govern the citizen in ways consonant with the ideals of liberty, equality, and legitimate power. (ibid.: 118) It is certainly true that mid-twentieth-century social psychology and representations of 'the psychosocial' were replete with the terminology of 'democracy'. The 1948 Congress was no exception. However, Rose employs Foucauldian conceptualisations of 'power' and 'government' in which power has no'sovereign centre' in society and indeed power relations appear to take place only between power and knowledge. In Rose's hands these conceptualisations lead to the rather functionalist depictions of 'the psychosocial' and its interaction with liberal democracy, expressed in the above quotes. This article contests such a rendition. It suggests that pre-Foucauldian understandings of power, government and authority, common in the nineteenth and twentieth centuries (and indeed the twenty-first), need to be attended to in order to help appreciate the relationship of the psychosocial to twentieth-century liberal democracy. Since this article's analytical stance might seem unfashionable in the face of currently popular 'theory driven' historical work, it is appropriate that the sociologist Robert Nisbet's 1966 book The Sociological Tradition provides a schema for its analysis. Nisbet was a noted conservative and, according to one obituarist, 'So resolutely unfashionable that he regularly came back into fashion'. In The Sociological Tradition Nisbet set out a group of paired concepts that he believed epitomised the conflict between 'tradition' and'modernity'. These were, respectively, community, authority, status, the sacred, and alienation, on the one hand, and, society, power, class, the secular, and progress, on the other. Nisbet emphasised that the archetype of community 'both historically and symbolically' was the family, and that the nomenclature of the family was prominent in every expression of it (Nisbet, 1966: 48). In fact, although he did not draw it out fully, the family clearly played a fundamental role for his thesis in general. 'Traditional' social organisation was structured by authority relations that were deeply embedded in social institutions from the family through neighbourhood, parish and guild, and integrated throughout the social body. The form of this authority was primarily personal whether it be that of the patriarchal head of the family or the sovereign. It was characterised by 'personal intimacy', 'emotional depth','moral commitment', 'loyalty' and 'duty'. Nisbet described how on this view the rationalist image of society, along with associated philosophies of contractual relations and utilitarianism in the political and economic spheres, separated the individual from deeper ties of community which had provided social cohesion and secure personhood. Nisbet added that, when people are separated from long established social institutions, 'there arises, along with the spectre of the lost individual, the spectre of lost authority' (ibid.: 108). These perspectives were informed by four aspects of the sociology of power associated with concerns about the growth of the rational and contractual state (ibid.: 109-111). There were misgivings about the centralisation of power in the state. This was often related to concerns about the political legitimacy of power becoming derived from a mass popular base. A common view saw popular participation in power only leading to increased state centralisation through the dismantling of intermediate institutions of authority. There was an associated fear of the totalism of power; of its extension to the whole of people's life and being. Along with this there were concerns about rationalisation as a facet of power-the processes of measurement, ordering, and systematising, in the name of rationality and efficiency. Nisbet's rendition of each of his paired concepts as opposites is overdrawn (Lukes, 1979). It might be more accurate to view these pairs as coordinates that some social and political theorists attempted to place in new combinations in order to reconcile 'the traditional' and 'the modern'. In this sense Nisbet's schema offers, perhaps surprisingly, a useful means to unravel some of the complexities of 'the psychosocial' with reference to the mental hygiene movement and the 1948 Congress. It shows that, through the interwar and postwar years of the twentieth century, the psychosocial can be read as an uneasy and ambiguous attempt to reconcile 'the modern' and 'the traditional'. The International Congress represents one specific aspect of this engagement. It was a means to both further professional aspirations, and address a particular crisis of the relationship of individuals with the social order commonly perceived to have been revealed by the Second World War. --- Relevance of Nisbet's thesis to the mental hygiene movement in England Since the driving force behind the organisation of the International Congress was the movement for mental hygiene this article will first of all look at how Nisbet's five paired concepts might relate to the movement in its pre-war manifestation. This will be discussed by looking at the movement in England. Four organisations formed the movements' nucleus. These were the Central Association for Mental Welfare, the National Council for Mental Hygiene, the Child Guidance Clinic, and the Tavistock Clinic. Of these, the earliest was the Central Association for Mental Welfare. This had been set up, with official government backing, to assist in the operation of the 1913 Mental Deficiency Act. Under this Act all county and county boroughs in England and Wales were to ascertain the local population of people deemed'mentally defective', provide institutional provision, and arrange community supervision (Thomson, 1998: 217). So-called'mental deficiency' was associated with what was termed'social inefficiency' and, in turn, to a host of social problems. Partly because of these connections, the Central Association for Mental Welfare increasingly widened its scope to include people believed to be on the socalled 'borderline' between apparent 'pathology' and'mental health' (Jones, 1986: 27-8). Through this it combined with the other three interwar organisations in the formation of a movement for mental hygiene. Although these other organisations included mental deficiency as important, their concern with'social inefficiency' focused particularly on psychological disorders among individuals in the community. The main aim of the movement therefore became the prevention, detection and treatment of mental disorders in the population as a whole. The main targets were people in society whose behaviour or experience appeared to mental hygienists to represent a failure of mental adjustment. The central institution through which mental adjustment was to be achieved, and maladjustment to be understood, was the family. The National Council for Mental Hygiene and the Child Guidance Council in particular promoted the family as a crucial element in the achievement of mental health. This is important with regard to the movement for mental hygiene. Any reading of literature by the movement will show that the family was at the heart of its aim of preventing, or detecting and treating, mental maladjustment (for instance Stewart, 2006 and2007). But it is here, in fact, that the ambiguous position of the movement to the modern/anti-modern polarities set out by Nisbet is thrown into relief. The essential elements of this position can be summarised in the following way. For the mental hygiene movement, unlike the social authority of 'traditional' community characterised by Nisbet, the modern family's personal and hierarchical relations were not recapitulated across the social order from the realm of the family to that of the sovereign and all points in between. Instead, straddling the modern and the anti-modern, the mental hygiene movement made the personal authority relations of the modern nuclear family the foundation of the rationally marshalled and moralised individual. It is as if, under the mental hygiene movement, the authority relations of community, as described by Nisbet, have been thrown back and compressed into those of the nuclear family and turned into a developmental process to adult 'adjustment'. Authority relations, embedded in personal relationships organised throughout the community, which continually sustain personhood and social cohesion, appear replaced by the nuclear family as forcing house of the rationally organised, self-sustaining and moralised individual. The individual of durable rationality and stability remains, for mental hygienists, the epitome of 'civilization'. But it is formed through the family; it is not straightforwardly preordained as suggested in Enlightenment thought. How was this expressed by the mental hygiene movement? Its stance was of course structured by the theories that informed it. Two were prominent. One was a theory of instincts drawn from 'the new psychology' of the early twentieth century (Hall, 1885;Dewey, 1884;McDougall, 1908). On this view humans were dynamic. They inherited primitive instincts and these, with their accompanying emotions, constituted the basic impulses of human behaviour (Rees, 1929;Burt, 1952). The other theoretical strand was psychoanalysis and its various popular reformulations. With this the investigation of unconscious motivations became important. These lay in the emotional substrates of the mind. They were recalcitrant emotions but they could not just be suppressed by an act of 'will'. They required an expert knowledge of the development of personality and its emotional components. With this, emotionality could be nurtured and crafted to ensure mental adjustment. The essential claim was that rational thinking had finally grasped the fact that emotional experience underlay all growth and adjustment. The terminology of adjustment and maladjustment was important in mental hygiene theorising. The nineteenth-century concept of alienation suggested a failure of progress. The mental hygiene use of the term maladjustment echoed this. Its terminology derived from biology. In fact such terms had been applied to human psychology and social evolution since the mid-nineteenth century (Young, 1990;Keir, 1952: 6). Inherent in their use was the notion of progressive human and social history. By using it mental hygienists were able to situate their outline of human development and mental health, on pre-existing depictions of the past as a process of progressive change. Much nineteenth-century western thought had cast the past as a progression of stages in the development of the rational individual mind in association with an emerging 'civilisation' of society (Bowler, 2003). The mental hygiene movement echoed this view (e.g. Brown, 1935;Gordon, 1933). But mental hygienists described it in terms of emotionality as well as intellect. Emmanuel Miller, the director of the East London Child Guidance Clinic and a leading figure in child guidance, provides an example of how, in doing so, mental hygienists prioritised the family and its personal relationships of authority as crucial to 'progress', 'civilisation' and the emergence of the rational individual (Miller 1938). According to Miller, 'primitive societies' brought emotionality under rigid group control early on in a child's life. Consequently they thwarted individuality and freedom. But Miller described 'civilised' society as having superseded the 'group mind' with the individual mind. Under so-called civilised society the individual intellect now marshalled instincts and emotions. This was achieved through the temporary and less rigid authority of the modern family. J C Flugel, the psychoanalyst, and senior lecturer in the Department of Philosophy and Psychology at University College London, expressed the same theme in his 1921 book The Psycho-Analysis of the Family. The book was later described by H V Dicks, the prominent post-war mental hygienist at the Tavistock Clinic, as a work that had acted as a veritable 'Bible' (Dicks, 1954: 29). In it Flugel described a correlation between particular historical stages of development of the family and the relation of an individual to forms of social, political and religious authority (Flugel, 1921). Regarding the modern individual and family he wrote:... it is I think, abundantly clear that normal psychic development involves a gradual emergence from a state of dependence on parental authority and care to one in which the individual is dependent to a greater or less extent upon his own efforts as regards his livelihood, and upon his own judgement as regards his conduct. (Flugel, 1921: 41) The sense of a transformation of the kind of traditional order described by Nisbet into a developmental order within the modern family is summed up in a passage from a booklet published by the Association of Psychiatric Social Workers in 1932: The 'father image' in family life, in the community and in religion, seems to be at the present time divesting itself of authority, removing the protective guidance of a decalogue, and asking of individuals -instead of conformity and adjustment to a social norm -the harder but more adult task of adjustment to the selfachievement of inner balance. (Cosens, 1932: 12) It was the psychic integration of instincts and emotionality that the parent -child relationships of the modern family were understood to provide. This required, not unyielding domination, but shaping based upon expert knowledge of emotional development. The correct type of parental authority was crucial. The child was not a blank space that could be directed and imposed upon without concern. 'If we realize that the child is unique, an individual dynamic organism, then we are going to be a little more careful as to how we introduce authority', wrote E. Graham Howe, one of the founding members of the Tavistock (Howe, 1932: 39). Emmanuel Miller maintained that: Both mother and father... become part and parcel of a child's life not merely as persons from the outside world, but as part of one's own sentiments -of love, admiration, authority and self-control. (Miller, 1933: 69) This was why authority must not be overly dominant, distant or rigid, nor be overlyindulgent or utterly absent (Miller, 1933;Howe 1932;Rees 1929). Its aim must always be to assert authority in the interests of promoting the development of a self-governing mind. Parental authority must be wielded with the intention of making itself unnecessary. Flugel described a failure to become fully mentally adjusted within the relations of the family, like this:... one who has never progressed far from the infantile condition of dependence on the commands and opinions of others will be lacking in one of the character qualities which are essential for the attainment of any high degree of individuality or of social and economic responsibility.... At best his morality will be one of hard and fast rules, the dictates of parental, ecclesiastical, legal or social authority, incapable of enlightened growth or modification to suit the ever changing flow of circumstances and the widening experience of life. (Flugel, 1921: 44) But if the mental hygiene movement seems thoroughly'modern' so far, this impression must be tempered by mental hygienists' assertion that primitive emotionality continued to reside in individual minds and mediate their engagement with society (Rees, 1929;Crichton Miller, 1932;Miller, 1938: 179;Burt, 1933: 225-245). Emmanuel Miller put it like this:... while the looser texture of civilized society has allowed for individuality, it has also submerged those bonds which, at one time visible in group conventions, were strong enough to break the individual. If at one time the individual was moored by a visible rope to the social quayside, he is now anchored to the sea-bed by cables which, though hidden, are no less strong and binding. (Miller, 1938: 36) Hugh Crichton Miller maintained that, on becoming a hospital patient, the apparently 'civilised' individual,'very soon reveals the fact that he is little more than a grown-up child, and that his reactions are for the most part, those of primitive man' (Miller, 1932: 430). Once emerged in the circumstance of hospital, this 'primitive man' behaved much like Flugel's description of a person who had failed to become fully mentally adjusted; he was dependent on authority, craved security and relinquished his own initiative and responsibility. It was here that mental hygienists expressed concerns about the coordinates of modern power described by Nisbet. In 1921 Flugel wrote for example that, The world war has taught us the necessity of implicit obedience to the State and its representatives -military or civil; the right of independent thought, criticism and action being to a large extent suspended and the minute details of our lives being subject to order and inspection in much the same way as in our childhood they were subject to the supervision of our parents. Again, modern socialistic thoughtespecially in its cruder aspects -has produced a state of mind, as a result of which the individual becomes to a large extent absolved from the responsibility for his own education, progress and maintenance, and for those of his children. The adult individual is thus led to transfer onto the State that attitude of dependence which he originally adopted in relation to his parents... If our conclusions are correct, there is a danger in too wide a ramification of State provision and State control, inasmuch as it is liable to prevent that full development of individual power, initiative and self-reliance which can only be obtained by a high degree of emancipation from the primitive attitude of dependence on the parents. This is a classic example of concerns about forms of modern power. Centralised power has grown in the state and it is totalising and rationalising. It inspects and structures the minute details of people's lives in the name of rationality and efficiency. Fears about the legitimisation of state power through mass mandate are expressed in Flugel's relation of these negative aspects of modern power to the 'cruder' versions of socialistic thought. The mass political participation suggested by state socialism is depicted as in fact promoting more pervasive state control. And this power merely serves to infantilise the masses while eroding the only institution in society that can successfully moralise the individual -the family. A class bias is clear in much mental hygiene writing. Sir Maurice Craig, for example, claimed in a speech that correct parental authority offered the best hope for combatting mental disorder. But he made the general targets of this comment clear by adding that the'stability and happiness of the mass' depended on it. He added that current ideas about 'equal shares for all' might sound alright but they ignored differences in mental capacity. The majority of people must therefore be content to remain in the social station to which they were most suited (Craig, 1922). Attempts to justify such a hierarchy were made by other prominent mental hygienists (e.g. Gillespie, 1933;Burt, 1933). Mental hygienist's prioritization of the modern family as forcing-house of the adjusted individual supported such attitudes to status and authority in the community. Flugel concurred with Cyril Burt that'strikes and other crude forms of rebellion against authority in industry' were the result of a displaced father-hatred due to the primitive and repressive form of fatherhood common in the working classes (Flugel, 1921: 120). Similarly, J A Hadfield, a leading figure at the Tavistock Clinic, believed political disorders should be understood as'manifestations of mass neurosis'; a conflict between'spontaneous impulses' of the masses and an authority that had failed to restrict and direct them adequately (Hadfield, 1935: 14). This led some to question whether the idea of political democracy should not be amended. 'Men have lately begun to wonder whether the principle of "one man one vote" is working quite as well as they hoped', stated Burt in a 1933 radio broadcast. 'Is it fair to count the number of heads without stopping to consider their contents?' Hugh Crichton Miller, the director of the Tavistock Clinic, suggested that university graduates ought to be more independent thinkers and therefore more competent voters. They should therefore have more weight and responsibility in a 'democratic society'. He claimed that 'one man, one vote' was sham democracy, and that a community not educated in selfgovernment was'much more likely to profit from a dictatorship than from the semblance of democracy which permits the demagogue to achieve autocracy' (Hadfield, 1935: 150-151). --- The 1948 International Congress on Mental Health The ambiguous position held by a pre-war mental hygiene movement straddling the polarities of modernity and anti-modernity set out by Nisbet, is clearly apparent at the International Congress. Each of the conferences that comprised the Congress continued their pre-war mission of developing and publicising what they considered to be their expertise in understanding and preventing mental maladjustments. But now, faced with the catastrophe of the recent world war, this took on a new salience. With the onset of the Second World War many leading figures of the movement in Britain, the United States and Canada secured influential positions in the armed services. It was from these Anglo-American developments that ambitious plans for a post-war Congress on mental health were developed. J R Rees, a leading figure of the English mental hygiene movement, was one of the principal instigators. He had secured appointment as head of psychiatry in the British Army during the war. In the early 1940s he made two long study visits to the United States, making close contact with leaders of the mental hygiene movement as well as leading psychiatrists, including George Stevenson, Frank Fremont-Smith, Harry Stack Sullivan and the Canadian Charles Brock Chisolm (Manning, 1976: 268-9;Brock, 1998). After the war Rees collaborated with these people to plan an International Conference on Mental Hygiene. The ambitious aim was to promote the importance of applying knowledge on human personality and group relations for mental health and world citizenship through the coordinated work of the human and social science professions. The introduction to this article referred to a recent editorial in the International Journal of Epidemiology that considered the roots of 'the psychosocial' to lie in the WHO definition of health. We have seen that its origins lie much earlier than this. But, nevertheless, the World Health Organisation's definition of health was certainly used as a guiding definition for the 1948 Congress. The first of the four volumes that recorded the Congress displayed this extract from the WHO's constitution, on one of its opening pages. In fact, leading figures in psychiatry and the mental hygiene movement were closely involved with the creation of the WHO and the drafting of its Constitution (Brody, 1998). Several, such as George Brock Chisholm, the leading Canadian psychiatrist and mental hygienist, the U.S. psychiatrist Harry Stack Sullivan, and J. R. Rees, were, as I have noted, key instigators of the International Congress on Mental Health. However, the underlying orientation that had informed the interwar mental hygiene movement remains evident. As we have seen, the movement's primary attention was on the family and the role of personal authority relations in mental development. This concern with the role of human relationships in the creation of mentally 'adjusted' citizens encouraged increasing engagement with the sociological and anthropological disciplines. For example, several of the important contributors to the 1948 Congress, such as the psychoanalyst Franz Alexander, Harry Stack Sullivan and Lawrence K. Frank, the Director of the Caroline Zachry Institute of Human Development, had engaged with the anthropological and sociological professions between the wars. Additionally, the anthropologist Margaret Mead played an important role at the Congress. She had joined the U.S. National Committee for Mental Hygiene in 1945 and became a member of the International Preparatory Commission for the Congress. In one of the prominent papers to the Conference on Mental Hygiene Mead argued that the family was mediator of culture and the individual (Mead, 1948a). Franz Alexander concurred. For him, psychiatry had established the fact that, 'The factory in which human personality is moulded is the family' (Alexander, 1948: 149). He added that this knowledge had been extended by anthropologists such as Mead and her one-time tutor Ruth Benedict. They had shown that the personality traits characteristic of a nation were attributable to similarities in family influences prevailing in each culture (ibid.: 149). Mead's speech at the Congress on the theme of 'collective guilt' is instructive regarding the integration of these views with mental hygienist thinking. She argued that character structures were associated with different social arrangements, family structures and political forms (Mead, 1948b). Basing her analysis on contrasts in family socialization in various cultures, Mead argued that guilt was only one type of character structure used as a sanction in societies. Thus it was only one culturally specific mediator of the relationship between family training of the young and a nation's government (Mead, 1948b: 65). However, she also described guilt as 'perhaps one of the more special developments', and went on to say that:... the capacity to experience guilt, as a dominant psychological mechanism, is a human capacity which may be either developed or neglected by any given society, and cannot therefore be regarded as either universal or necessary, however desirable it may be found in terms of contemporary ethics. (Mead 1948b: 64) Despite her caveats, the idea that such mechanisms are either 'developed or neglected' surely implies that they are morally superior. The leading British mental hygienist, H.V. Dicks, was more forthright. He commented that'some of the societies which Dr Mead [had] quoted as having little guilt and much shame or pride or just sheer external compulsion for the coercion of their members, seem to stand on a psychologically less mature rung of the ladder' (Dicks, 1948: 84). Mead also distinguished between 'character structures' formed in relation to guilt. These could be differentiated between, guilt over the content of an act, guilt over the form of an act, and guilt over conflict between formally required acts. She contrasted Britain and the United States with Germany and Japan in this context. The latter developed character formations built around the second two forms of guilt. For instance, she maintained that investigations had shown German character structure to be based around an emphasis on 'the moral necessity to discharge one's duty, in the light of one's formal position in a family, the army, a government organization, or the state...' (Mead, 1948b: 63). This she contrasted with Britain and the United States where, she asserted, an emphasis was placed on the content of an act and whether it was to be judged good or bad, right or wrong. Thus the construction of an internal authority overrode any external human authority. According to Mead, this related to the fact that these countries allowed individuals free choice of those who govern them. Because of this freedom individuals therefore felt responsible for the content of the national behaviour. In the context of our schema derived from Nisbet, such views suggest an assertion of the'modern' over the 'traditional'. The interacting hierarchical authorities of 'traditional' community comprised of rigid status are rejected as inimical to mental health and social stability. Ives Hendrick, a Harvard psychiatrist, gave a similar rendition of German character to Mead, describing the culture as 'favouring feudal types of character' (Hendrick, 1948: 27). Dicks related character formation in Germany to the continued influence of Old Testament authority. According to him pre-war Germany had been, 'a kind of paragon of old testament virtues and values'. By which he meant it was more 'rigidly ordered and insistent on the "good behaviour" of its children' than most western countries (Dicks, 1948: 85). Toms However, these presumptions apparently in favour of democracy, were nevertheless mediated by a fundamental orienting principle of the Congress. This was the established mental hygienist claim that beneath the surface of apparently modern rational individuality lay 'primitive man' under the sway of irrational prejudices and fears (e.g. Crichton Miller, 1932). At the beginning of the six day mental hygiene conference, David Mitrany, the adviser to Lever Brothers and theoriser of functionalism in international politics, described human history as marked by the fact that humans had massively transformed their social environment and come to do so with increasing rapidity. But, understood biologically, human nature had changed very little. One consequence was that, 'The chains of primitive taboos have gone, but the complexity of modern life, after freeing the individual from formal chains, has wound round him a cocoon of distracting strains and conflicts' (Mitrany, 1948: 72). Others outlined associated features of modern societies. Mead summarised some of the common themes of the seventy-six Preparatory Commissions. She noted great concern about the insecurity, and thus anxiety and hostility, that was believed to be engendered by shifts from rural to city life, firm religious backgrounds to loss of belief, and from strong to loose family ties (Mead, 1948: 122-3). For Mitrany, this left individuals vulnerable to control and manipulation of their irrational impulses. He maintained that, whether one liked to admit it or not, it was an unquestioned fact that'most democratic devices, intended as they are to widen participation in politics and government, have added to the possibilities for appealing to mass emotions...'. Likewise, he continued, new methods of communication had vastly increased the opportunities for'mass suggestion' (Mitrany, 1948: 75). A prominent theme at the Congress was the extended power of the state and the general view of delegates about its deleterious effects on the formation of rational, moralised individuals (e.g. MacCalman, 1948;Segerstedt, 1948). Mitrany, for instance, described the modern state as an 'artificial organism' that maintained itself through an 'all-demanding' requirement of compulsive loyalty and unquestioning obedience (Mitrany, 1948: 78). It was a planned state, able to use social institutions and 'instruments of material and of cultural life to gain submission and conformity...' (ibid.: 72). He feared that the military style discipline imposed through economic planning for wartime total mobilization was creeping into wider aspects of social life (ibid.: 78-9). At the same time, he asserted that, however worthy the modern attempt to provide'social security', it had the effect of treating people abstractly and uniformly, removing initiative and promoting passivity and frustration. On Mitrany's view, the state had become too strong to allow liberty but too weak to provide security (ibid.: 81). A question raised remorselessly in the preparatory commissions' reports on the subject of 'Family Problems and Mental Disturbance', was: 'Is it likely that the family group is going to disappear completely, and that all its functions will be taken over by the state?' (Segerstedt, 1948: 159). The British mental hygienist, D.R. MacCalman (presumably referring to industrialized western nations) appeared to concur with Mitrany in his speech to the Conference on Child Psychiatry. He raised the 'apparent failure in function of the family'. 'Vast numbers of individuals would appear to be growing up with insufficiently mature independence', he claimed. And he asked, '[is] the family unit surrendering its functions to the wider unit of the state, which has not yet learned to exercise them adequately?' (MacCalman, 1948: 55). MacCalman's speech in fact provides a useful template to represent the general view. His opening remark rendered the aim of the Congress this way:... its goal is the more harmonious interpersonal relationships within the family group, which, in turn, will give a greater security to the individual child and allow him to mature with a minimum of uncontrolled aggression. If this takes place successfully he will be a more stable and efficient citizen of his community and the world. (MacCalman, 1948: 50) Displaying the modernist aspects of the mental hygiene movement, MacCalman maintained that too much aggression had been used against children in most cultures throughout most of history. He maintained that, 'the injunction "spare the rod and spoil the child" has been interpreted as an encouragement to lay the rod on lavishly'. This aggression had been accompanied, he added, by an interpretation of 'Honour thy father and thy mother' designed to 'crush any hint of retaliation'. MacCalman concluded, Thus protected by Biblical sanction parents and teachers have proceeded joyfully to 'knock hell out of the little blighters'. And what can

The Foucauldian sociologist Nikolas Rose has influentially argued that psychosocial technologies have offered means through which the ideals of democracy can be made congruent with the management of social life and the government of citizens in modern Western liberal democracies. This interpretation is contested here through an examination of the 1948 International Congress on Mental Health held in London and the mental hygiene movement that organised it. It is argued that, in Britain, this movement's theory and practice represents an uneasy and ambiguous attempt to reconcile visions of 'the modern' with 'the traditional'. The mental hygienist emphasis on the family is central. Here it appears as a forcing-house of the modern self-sustaining individual. Mental hygienists cast the social organisation of 'traditional' communities as static, with rigid authority frustrating both social progress and the full emergence of individual personality. Yet mental hygienists were also concerned about threats to social cohesion and secure personhood under modernity. If the social organisation of 'traditional' communities was patterned by the archetype of the family, with its personal relations of authority, mental hygienists compressed these relations into the 'private' family. Situated here they became part of a developmental process of mental adjustment through which 'mature', responsible citizens emerged. This reformulation of the family's centrality for the social order informed mental hygienist critiques of the growth of state power under existing forms of democracy, as well as suspicion of popular political participation or protest, and of movements towards greater egalitarianism.Democracy; Michel Foucault; mental health; mental hygiene; psychosocial The notion of 'the psychosocial' is highly influential in current mental health conceptualisation and practice. However, the term remains variously described and interpreted. In its most general sense 'the psychosocial' refers to a relationship of mind, body and social environment in terms of mental health and ill-health. This usually translates into assessments and interventions at various levels, such as individual experience and behaviour, the perceived dynamics of an immediate social context, or a wider social environment (Dagnan, 2007: 3). The psychosocial refers neither to a purely reductionist or isolated notion of the individual, nor to a primary focus on large-scale social structures. A

fact provides a useful template to represent the general view. His opening remark rendered the aim of the Congress this way:... its goal is the more harmonious interpersonal relationships within the family group, which, in turn, will give a greater security to the individual child and allow him to mature with a minimum of uncontrolled aggression. If this takes place successfully he will be a more stable and efficient citizen of his community and the world. (MacCalman, 1948: 50) Displaying the modernist aspects of the mental hygiene movement, MacCalman maintained that too much aggression had been used against children in most cultures throughout most of history. He maintained that, 'the injunction "spare the rod and spoil the child" has been interpreted as an encouragement to lay the rod on lavishly'. This aggression had been accompanied, he added, by an interpretation of 'Honour thy father and thy mother' designed to 'crush any hint of retaliation'. MacCalman concluded, Thus protected by Biblical sanction parents and teachers have proceeded joyfully to 'knock hell out of the little blighters'. And what can the little blighters do but wait till they are big and strong enough to do likewise. (MacCalman, 1948: 51) MacCalman made it clear in his speech that he saw the role of psychiatrists, psychologists and psychotherapists as to 'develop the power to influence mankind' (MacCalman, 1948: 55). But how might they do this? He drew the comparison between the influencing of children's behaviour that took place in the family unit and the influencing of adults in wider society. Social scientists, he argued, emphasised indirect 'rationalistic techniques' of influencing behaviour. MacCalman noted the antipathy of many psychotherapists to such large scale rationalised methods. He pointed out, however, that psychotherapists themselves had played their part in extending such rationalizing power and its penetration of family group relations in terms of calculation, objectivity, rationality and efficiency, in place of arbitrariness, domination and patronage. But MacCalman clearly lamented, along with the majority of Congress delegates, the perceived erosion in the role of the family and, 'the gradual transfer of influence from small primary groups, like the family, to larger and apparently artificial groups' (MacCalman, 1948: 53). MacCalman looked to the'small group' as the main vehicle of influence. In the introduction to this article we noted that the 2002 editorial in the International Journal of Epidemiology interpreted the psychosocial as a concept that operated at a'meso-level' between socio-structural and individual levels of human activity. It described this level as encompassing social formations such as,'religious institutions, the family, the firm and the club...' (Martikainen, Bartley andLahelma, 2002: 1092). MacCalman's explanation of 'the small group' relates directly to these social formations. The significance of the small group was that it operated on the basis of personal ties and Toms relations of authority. Echoing the 'traditonal' elements of Nisbet's schema, MacCalman noted that, 'It characterises small social units, such as the family, fraternities, clans, and communities' (MacCalman, 1948: 53). He made plain that the modern family was the archetypal root of these groups. It provided the primary medium through which 'intense inter-personal relationships' could be worked out (ibid.: 53). Other postwar mental hygienists did much the same. For example, the psychiatrist T A Ratcliffe was closely involved with the post war mental hygiene movement. He maintained that: The concept of mental health in a community would seem to demand two thingsthat individual members of that community should be themselves stable, secure and settled and that the community pattern itself should be a mentally healthy one. (Ratcliffe, 1951: 11-12) He argued that a community mental health service should be understood as a form of'relationship therapy'(ibid.: 14). The model for this kind of therapeutic relationship was the family:... just as the parent-child relationship should be the epitome of future relationships for the child and the path which leads him to adult maturity and independence of personality, so the client-Psychiatric Social Worker relationship should be an experience which leads the client on until he can form his own mature adult relationships in his environment. (ibid.: 14) But Ratcliffe also sought to promote small group work both for people considered maladjusted, and also as a means to bring a wide array of professional groups in society together to work for mental health. And his vision of such groups was that they would mirror the important elements of'relationship therapy'.: We should see an example of the importance of working out inter-personal tensions, of understanding and accepting our feelings of aggression, overdominance or over-dependence towards each other and of the significance of interpersonal relationships in the sense that I have described. (ibid.: 21) We have here, yet again, an uneasy straddling of the'modern' and 'traditional'. If, under the mental hygiene movement the authority relations of the 'traditional' social order have been compressed and turned into a developmental process, then the theme of development from childhood to adulthood lies at its heart. In part, this can be seen to uphold a form of liberal democratic political order. The family creates autonomous, responsible citizens capable of upholding a version of representative democracy. But this developmental family model also informs critiques of the growth of state power under existing forms of democracy, along with suspicion of popular political participation or protest, and of movements towards greater egalitarianism. They are either the causes or effects of dependence and infantilisation. In other words, in failing to echo the internal social order of the family, they become elements in society that impact negatively on it and thus on the creation of citizens. Class bias inhibits democratic values and this bias is readily apparent in frequent mental hygiene pronouncements. In general the vision of the mental hygiene movement was of a hierarchical ordering of society on the basis of measurements of intellectual capacity and the contents of personality. The need for leadership was a common Toms theme. Thus, in respect of the mental hygiene movement, the politics of the psychosocial might be less about how it acted as a conduit through which particular programmes for managing people could be made congruent with the values of democracy, and more about the ways in which the movement attempted to reconcile its vision with democratic values. --- Biographical note Jonathan Toms has previously been Postdoctoral Fellow at the Centre for the History of Medicine, University of Warwick.

The Foucauldian sociologist Nikolas Rose has influentially argued that psychosocial technologies have offered means through which the ideals of democracy can be made congruent with the management of social life and the government of citizens in modern Western liberal democracies. This interpretation is contested here through an examination of the 1948 International Congress on Mental Health held in London and the mental hygiene movement that organised it. It is argued that, in Britain, this movement's theory and practice represents an uneasy and ambiguous attempt to reconcile visions of 'the modern' with 'the traditional'. The mental hygienist emphasis on the family is central. Here it appears as a forcing-house of the modern self-sustaining individual. Mental hygienists cast the social organisation of 'traditional' communities as static, with rigid authority frustrating both social progress and the full emergence of individual personality. Yet mental hygienists were also concerned about threats to social cohesion and secure personhood under modernity. If the social organisation of 'traditional' communities was patterned by the archetype of the family, with its personal relations of authority, mental hygienists compressed these relations into the 'private' family. Situated here they became part of a developmental process of mental adjustment through which 'mature', responsible citizens emerged. This reformulation of the family's centrality for the social order informed mental hygienist critiques of the growth of state power under existing forms of democracy, as well as suspicion of popular political participation or protest, and of movements towards greater egalitarianism.Democracy; Michel Foucault; mental health; mental hygiene; psychosocial The notion of 'the psychosocial' is highly influential in current mental health conceptualisation and practice. However, the term remains variously described and interpreted. In its most general sense 'the psychosocial' refers to a relationship of mind, body and social environment in terms of mental health and ill-health. This usually translates into assessments and interventions at various levels, such as individual experience and behaviour, the perceived dynamics of an immediate social context, or a wider social environment (Dagnan, 2007: 3). The psychosocial refers neither to a purely reductionist or isolated notion of the individual, nor to a primary focus on large-scale social structures. A

Introduction Due to the internal turmoil in Syria, the influx of migration started in 2011, and then it became an international social problem affecting the neighboring and regional countries as well as the EU member states. Türkiye, the country with the longest border and historical ties with Syria, was most affected by this situation. Türkiye's open-door policy, adopting the temporary protection regime, which includes the principles of non-refoulement and meeting basic needs, has also been effective in this, which causes Syrian refugees to see Türkiye as a safe haven (T.C. Kalk<unk>nma Bakanl<unk>, 2018: 32). Today, Türkiye has become the country hosting the largest number of refugees in the world, hosting approximately 4 million Syrian citizens since 2015 (UNHCR The UN Refugee Agency, 2022). Hosting such a large population also necessitated a comprehensive organization. This organization requires the services of local government agencies as well as the support of international organizations and civil society. On the other hand, rights-based organizations, aid-based non-governmental organizations and academia are also working to inform the public about the refugee crisis originating from Syria, to identify the deficiencies in the implemented policies and to offer suggestions (T.C. Kalk<unk>nma Bakanl<unk>, 2018: 59). One of the most important of these policies has been the provision of education to children and young people in order to ensure compliance. With the fact that education policy is one of the most important tools in ensuring compliance, multidimensional planning has been made. The reason for this is the high young population and the unknown when the turmoil in Syria will end. This situation is a result of the fact that education is seen as a human right, as well as laying the foundations for a long-term coexistence between local people and immigrants. According to the December 2021 data of the Directorate of Migration Management, 1,259,435 of the 3,734,876 Syrians under Temporary Protection in Türkiye are school-age children between the ages of 5-18, and 503.128 are university-aged youth between the ages of 19-24. In this respect, it is seen that a total of 1,762,618 people are in the education age. (T.C. <unk>çişleri Bakanl<unk> Göç <unk>daresi Başkanl<unk>, 2021). It is seen with the data that the number of schooling among Syrian children in Türkiye has increased over the years. According to the data of the Ministry of National Education; 230 thousand Syrian children were enrolled in school in the 2014-2015 academic year, 311 thousand in 2015-2016, and 492 thousand in 2016-2017 (T.C. Milli E<unk>itim Bakanl<unk> Hayat Boyu <unk>renme Genel Müdürlü<unk>ü, 2019). According to the May 2018 data of the Migration Administration, 519,980 of the 1,150,791 school-age Syrians aged 5-18 in Türkiye attended Temporary Education Centers (TEC) and public schools. However, 630 thousand people in this age group did not go to school (T.C. <unk>çişleri Bakanl<unk> Göç <unk>daresi Başkanl<unk>, 2021). By 2019, while 680,000 children were enrolled in school, nearly 400 thousand children were still not enrolled in school (UNICEF Türkiye Milli Komitesi, 2022). According to the data of the Ministry of National Education, 339,541 out of 382,357 school-age Syrians under temporary protection in Türkiye in 2019 were enrolled in primary school. Therefore, approximately 89% of primary school children continue their education. Considering that according to UNHCR's 2019 data, the primary school enrollment rate of refugees is around 63%, it can be said. 2460 At<unk>f / Citation: <unk>KTEN, Serkan. (2023). "Educational Policies and Practices Regarding the Adaptation of Syrians Under Temporary Protection to Türkiye" ODÜSOB<unk>AD 13 (2), 2457-2482, Doi: 10.48146/odusobiad.1232477, Doi: 10.48146/odusobiad.1232477 that Türkiye is successful in this regard. However, at the high school level, the rate of enrollment and attendance is low. This rate is 32.55%. The main reason for this is that boys have to work to provide economic support to the family due to economic conditions and girls are married at an early age (Aksoy, 2020: 37). Despite all the arrangements and activities regarding the education of Syrian refugee children, the desired success has not been achieved in the participation of school-age Syrians in high school and higher education in Türkiye. In addition, there has been a significant increase in the number of Syrian students enrolled in higher education. While the number of Syrian students in the Turkish Higher Education system was 608 in 2011, it increased to 27,606 in the 2018-2019 academic year (Erdo<unk>an, 2019: 11). Considering the Syrians who were born, married, and set up a business in Türkiye, it is possible that they will be permanent in Türkiye in the long run. Therefore, it is important to integrate children into society and ensure their integration into society. In order to achieve this, an accurate and effective education policy must be established and implemented (Akgül et al., 2015: 12). In this study, the education policies of Türkiye, which is the country that hosts the most refugees with approximately 4 million people in the world, since 2011, in order to ensure the integration of Syrian children and youth, will be discussed. In this context, the current situation regarding the legislative arrangements made, the roles of local and international institutions and organizations in these policies and the results obtained after the researches on this subject will be revealed by document analysis method. In this respect, it is aimed to determine how education and integration policies develop in the case of Türkiye, which is the country hosting the most refugees in the world. --- Background --- Legislation Arrangements Türkiye has published many legislation related to the education of Syrian children and youth. In addition, various projects have been implemented in cooperation with the Ministry of National Education, the European Union and Non-Governmental Organizations in order to ensure the adaptation of Syrian children to the Turkish education system and their access to education (Canpolat, 2020: 116). Disaster and Emergency Management Presidency (AFAD), Ministry of National Education (MEB) and Higher Education Institution (Y<unk>K) have established a cooperation regarding the determination and implementation of education policies for Syrians under temporary protection in Türkiye (<unk>ahin, 2020: 388-389). After the refugee influx that started in 2011, the policies followed on education were prepared only for the camp, based on the assumption that the Syrians would return first (Aksoy, 2020: 37). The first legislation on providing education to Syrians was the circular titled "Measures Regarding Syrian Citizens Hosted Outside of Camps in Our Country" dated April 26, 2013 by the Ministry of ODÜ Sosyal Bilimler Araşt<unk>rmalar<unk> Dergisi / ISSN: 1309-9302 / dergipark.org.tr/tr/pub/odusobiad • National Education (Esen, 2020: 3-4). With the cooperation of UNICEF and MEB, the "Guidebook for Guidance Services for Individuals with Temporary Protection Status" was prepared and it was aimed to inform Syrian refugees in this context (<unk>ahin, 2020: 388-389). By expanding the content of the first circular, the circular "Educational Services for Syrian Citizens Under Temporary Protection" was issued on September 26, 2013. However, these two circulars were not sufficient to eliminate the problems in the implementation of the education services provided to Syrians (Esen, 2020: 3-4). Therefore, with the circular numbered 2014/21 published on September 23, 2014, titled "Educational Services for Foreigners", certain standards have been set for the education services to be offered to Syrian children (Aksoy, 2020: 37). In addition to these, "Temporary Protection Regulation" was published in the Official Gazette on October 22, 2014. With the 28th article of the circular, the education of Syrian children has been comprehensively discussed (Esen, 2020: 3-4). The circular titled "Execution of Services for Foreigners under Temporary Protection", published by AFAD on 18 December 2014, also determined issues regarding the solution of the education problem of Syrians. It has been determined that the purpose of the education and training activities to be carried out for Syrian students is to make up for the education they had to take a break and to ensure that they do not lose any years when they go to their country or any third country. Planning, coordination and control of the educational services to be carried out is left to the responsibility of the Ministry of National Education. It was decided that the need for teachers would be met by the Ministry of National Education, when necessary, by appointing people who can speak Arabic in accordance with the conditions (Ertaş & K<unk>raç, 2017: 104-105). The "2015-2019 Strategic Plan" published by the Ministry of National Education was the first document containing plans for the education of refugees (Aksoy, 2020: 37). The increase in the number of Syrian students attending school between 2011 and 2020 is the result of strategic planning as well as legislative studies. According to this; for the first time, the "Strategy Document and National Action Plan for Combating Irregular Migration" was announced to cover the period of 2015-2018, with this document, strategic targets were determined in Türkiye's fight against irregular migration and studies were carried out for its implementation. Secondly, the "Strategy Document and National Action Plan for Combating Irregular Migration" covering the years 2021-2025, including the next period, was prepared and entered into force (T.C. <unk>çişleri Bakanl<unk> Göç <unk>daresi Başkanl<unk>, 2022). The legal arrangements made in this process were carried out in parallel with the changes in needs over time. For example, Y<unk>K initially admitted undocumented Syrian students to universities with special student status, but later provided the opportunity for lateral transfer to those with a certificate. The Ministry of National Education announced that it was not possible to apply a curriculum other. --- 2462 At<unk>f / Citation: <unk>KTEN, Serkan. (2023). "Educational Policies and Practices Regarding the Adaptation of Syrians Under Temporary Protection to Türkiye" ODÜSOB<unk>AD 13 (2), 2457-2482, Doi: 10.48146/odusobiad.1232477, Doi: 10.48146/odusobiad.1232477 than the Turkish education system curriculum at the beginning, but in the circular it published later, it was stated that the education program would be prepared by the Syrian National Coalition Higher Education Commission under the control of the Ministry of National Education (Seydi, 2014: 297). As a result of all these efforts, although school enrollment has increased, it is seen that a very high number of children are still out of school. Numerous researches and reports on this issue are presented and provide data for determining the necessary policies. --- Projects In addition to the legislative arrangements, project activities for the education of refugee children are also carried out effectively. "Project for Supporting the Integration of Syrian Children into the Turkish Education System (PICTES)" organized by the Ministry of National Education (Mert ve <unk>ubuk, 2017: 77-78). Within the scope of the PICTES project, the Ministry of National Education recruited personnel who teach in Turkish, stating that children aged 4-5 should attend pre-school education in order to ensure the integration of Syrian refugee children into society and language learning (Canpolat, 2020: 116). The first phase of the PICTES project, which is one of the most important studies to support the education of Syrian children, was started in 2016 and was completed at the end of 2018. After this date, the second phase was started with the PICTES II project. A contract was signed between the MEB and the European Union Delegation to Türkiye within the framework of the "Agreement on the Financial Assistance Program for Refugees in Türkiye" (FRIT), and MEB activities began to be supported within the scope of this project (Esen, 2020: 3-4). The relevant general managers of the Ministry of National Education, lecturers and school administrators who are Syrian students in their schools attended the "MEB Administrative Personnel Training" meeting held in Antalya between January 30 and February 3, 2017, where the project was evaluated. Within the scope of the project, it was stated that teachers and school administrators gained very important personal experiences in their communication and interaction with traumatized Syrian children (Mert ve <unk>plak, 2017: 77-78). --- Education Policy Application Areas Within the scope of the education policy for Syrian children and youth in Türkiye, lessons have been started in five main platforms. These are; Temporary Education Centers, Public Schools, Syrianowned private schools, open education institutions and higher education institutions. --- Temporary Education Centers (GEM) In Türkiye's border provinces with Syria, units affiliated to the Ministry of National Education and some Non-Governmental Organizations (NGO) are working together to determine the education levels and equivalents of refugees. Through the commissions established within these NGOs, temporary education centers have been established in the cities, whose education curriculum is compatible with Syria. Refugees were provided with trainings by using the infrastructure of some schools outside of normal education hours. There are also schools established for just this purpose. ODÜ Sosyal Bilimler Araşt<unk>rmalar<unk> Dergisi / ISSN: 1309-9302 / dergipark.org.tr/tr/pub/odusobiad • First of all, training was provided through volunteer trainers, and then, with the joint initiatives of the United Nations and the UNICEF, teachers were paid a low wage (Taş & <unk>zcan, 2018: 47). NGOs and AFAD took the lead in the establishment of Temporary Education Centers under provincial/district national education directorates. There is a Syrian and a Turkish administrator in the management of GEM. Since the Syrian curriculum is given here in Arabic, the majority of the staff consists of Syrian teachers. As a result of a research that included interviews with administrators in Temporary Education Centers, Turkish and Syrian teachers, as well as Syrian students, it was seen that the correctly chosen curriculum, various practices such as parent meetings, and the appointment of Syrian teachers at the center increased the sense of belonging to the school and supported the socialization of students. On the other hand, it was determined that the lack of opportunity for Syrian students to communicate with their local peers at GEMs caused them to move away from the goal of integration (Kaya, 2019: 12). In the 2016-2017 academic year, policy changes were made and it was decided not to open new temporary education centers and to close the existing ones after a certain period of time. In addition, it is planned to enroll middle class students studying in these temporary education centers and registered in the Foreign Student Information Operating System (Y<unk>B<unk>S) to state schools in the upper class by providing effective Turkish education. In this context, increased Turkish education was started to a total of 25,278 students in a total of 23 temporary education centers in Adana (1), Ad<unk>yaman (1), Hatay (19), and <unk>stanbul (2) in 2019 (T.C. Cumhurbaşkanl<unk> <unk>letişim Başkanl<unk>, 2020: 26). --- Public Schools Public schools belonging to the Republic of Türkiye are a second alternative for the education of --- Syrians. Within the scope of the Ministry of National Education's Education and Training Services Circular for Foreigners, Syrian children have the opportunity to enroll in public schools by presenting the foreigner identification document they have received. The language of instruction in these schools is Turkish. --- Private Schools Syrians have been given the opportunity to provide education in accordance with the Syrian curriculum in buildings that are suitable for education in Türkiye, either allocated to them by the state or found by them. --- Open Education Foreigners in Türkiye have the opportunity to receive education in line with the relevant provisions of the regulations of open education institutions. In this respect, this opportunity is also valid for Syrians (<unk>akmak, 2018: 467). At<unk>f / Citation: <unk>KTEN, Serkan. (2023). "Educational Policies and Practices Regarding the Adaptation of Syrians Under Temporary Protection to Türkiye" ODÜSOB<unk>AD 13 (2), 2457-2482, Doi: 10.48146/odusobiad.1232477 --- Higher Education At first, the view that Syrians would return to their countries in a short time led to the temporary planning of higher education discourse and policies in this context. However, not knowing how long the Syrians will stay with the prolonged period and the increasing number of refugees have created the need to produce permanent and sustainable solutions. At the first stage, planning was made so that Syrian students would not receive a diploma but would not fall behind in education. In this context, in 2012, Syrian students were given the right to attend classes with special student status for one year. In 2013, this limitation was removed and lateral transfer opportunity was provided for those who have a certificate. In addition, 460 Syrian students were provided with dormitories and scholarship opportunities together with the university (Seydi, 2014: 296). In this context, many educational activities for refugee students were organized by state universities and foundation universities under the coordination of the Council of Higher Education (Y<unk>K), and support was provided to help them cope with financial and cultural difficulties (<unk>ahin, 2020: 388-389). Despite the provision of these opportunities, the problem of not knowing the official procedures or not being able to provide the necessary documents to benefit from this right to education has been the biggest obstacle for Syrian youth to continue their university education. The way to overcome this situation is to grant the right to register based on declaration. However, while the right to be a higher education student requires passing a difficult exam for Turkish citizens, it is likely to cause a reaction by the Turkish citizens because of this right will be given to Syrians by declaration (<unk>opur & Demirel, 2017: 24). The "Foreign Academic Information System" project was initiated by Y<unk>K in 2016 for the employment of Syrian academics in Türkiye at Turkish universities. 5 thousand academics from 31 different countries, most of them Syrian nationals, were registered in the system. However, it has been determined that only 345 academicians are employed in universities in Türkiye, 320 of which are in theology and Arabic departments (Tanr<unk>kulu, 2018(Tanr<unk>kulu, : 2601)). In addition, Gaziantep University has founded different faculties for Syrian students in the regions cleared of terrorism in Syria by the Turkish Armed Forces. One of them is the Faculty of Economics, Administrative and Social Sciences in Al-Bab. Here, 2 Turkish and 7 Syrian academicians teach students who are all Syrians (Gaziantep Üniversitesi El-Bab <unk>ktisadi ve <unk>dari Bilimler Fakültesi, 2022). In addition, the same university has an education faculty in Afrin, Syria (Gaziantep Üniversitesi Afrin E<unk>itim Fakültesi, 2022). However, the number of Syrian academics employed except for the theology and Arabic departments is considerably less than number of all the Syrian academics in Türkiye. This situation was also expressed in the workshop "The Problem of Arab Academics and Suggestions for Solution" held by the Mediterranean Research Center. As a result of the workshop, it was stated that there were requests for the employment of Syrian Academics, generally in Arabic language and theology faculties of universities in Türkiye. There is a great potential for academics in other disciplines to work in their own fields of expertise. For this, it is necessary to be founded departments that teach Arabic in ODÜ Sosyal Bilimler Araşt<unk>rmalar<unk> Dergisi / ISSN: 1309-9302 / dergipark.org.tr/tr/pub/odusobiad • universities or to be founded a Turkish/Arab university. However, there are bureaucratic obstacles to founding departments providing education in Arabic (Tanr<unk>kulu, 2018(Tanr<unk>kulu, : 2601)). Studies conducted in different cities and universities about Syrian higher education students starting education reveal that students experience difficulties in this process. In a study conducted on Syrian students studying at Uşak University in 2018, it was revealed that students perceive social exclusion on issues such as economic, spatial, city-based, education and health. On the other hand, it has been seen that Syrian university students have multiple disadvantages as a result of the combination of the problems arising from being refugees with the educational problems they encounter at the university (Gül & Kayl<unk>, 2020: 985). As a result of another research, it was seen that Syrian and other immigrant university students in Konya tried to overcome difficulties such as communicating with academic and administrative staff, being marginalized culturally and socially, and understanding the thought system of another culture. On the other hand, it has been determined that they experience satisfaction problems such as stress, psychological trauma, financial dependence, identity confusion, language problems, cultural shock, inability to integrate into different social norms and values, changes in the education system, loneliness, lack of social and spiritual support (Altunbaş, 2020: 72-73). The media has a great role in the problem of marginalization and segregation of Syrian university students. The media produces wrong content about Syrian students. The media uses exclusionary rhetoric towards Syrian refugees. An example of this is the publication of false information, such as Syrian students being placed in the quota of Turkish students at universities and scholarships given by the state. For this reason, in order to prevent disinformation on these issues, it is necessary to provide accurate information to the public periodically by authorized institutions (Gül & Kayl<unk>, 2020: 1004). Despite all the problems experienced, it is seen that Syrian university students under temporary protection move into the social cohesion process more easily and quickly than other Syrians. A study revealed that some of the most important factors affecting the social adaptation of Syrians to Türkiye are social and psychological support resources, individual characteristics, university education, and social activities at the university (Kaya, 2019: 157). --- Problems in Education Policy Implementation After the implementations made within the framework of the regulations, some technical and social problems also caused a complicating effect in achieving success. However, since the regulations made are of a general nature, they have also revealed different applications and results in different cities and schools within their original conditions. In this context, individual efforts and achievements towards solving problems are also observed in general. --- Technical Issues --- Difficulties in the Registration Process With the migration influx, bureaucratic difficulties experienced in the school enrollment process for At<unk>f / Citation: <unk>KTEN, Serkan. (2023). "Educational Policies and Practices Regarding the Adaptation of Syrians Under Temporary Protection to Türkiye" ODÜSOB<unk>AD 13 (2), 2457-2482, Doi: 10.48146/odusobiad.1232477, Doi: 10.48146/odusobiad.1232477 schooling. The numerous documents required for school registration and the lack of any registration system for registration caused some students not to go to school. In order to solve this problem, the Ministry of National Education published the circular numbered 2014/21 "Educational Services for Foreigners" in 2014. With this circular, children in Türkiye under temporary protection have the right to enroll in school only with a foreign identification document without a residence permit. The Information Operating System for Foreign Students (Y<unk>B<unk>S), which was prepared to facilitate the registration process of foreign students, reduced the difficulties experienced in the school enrollment process (Tunga et al., 2020: 329). However, although the Ministry of National Education enables Syrian children to enroll in any public school with the foreign identification document given to them, it is stated by the school administrators that there are some difficulties in the registration process when this document is not available (Levent & <unk>ayak, 2017: 32). --- Equivalency and Grade Level Determination Difficulty One of the problems that Syrian students face in education is the problem of equivalence. The differences between Syrian and Turkish education systems and curricula cause problems regarding the grade levels that Syrian children will attend (Levent & <unk>ayak, 2017: 33). In addition, due to the fact that they take a break from education until they lead a regular and safe life during migration, the time they spend without education is long. After this long break, it is a problem at which grade level they will be enrolled. This problem arises when Syrian students, who do not speak Turkish, are enrolled in the first grade of primary school for language learning, or they are enrolled in the upper classes appropriate for their age, although they do not speak Turkish at all (Emin, 2018: 78). --- Access to Education and Participation Problem Lack of data on school-age children, transportation and service problems of students outside the camps, intercity mobility caused by migration from the camp to the city can negatively affect the continuity of education (Balc<unk>, p.2020). For families, the fact that the schools they will send their children to are far away can create various security concerns and may cause them not to send their children to school (Akp<unk>nar, 2017: 28). The transition to distance education during the coronavirus epidemic affected Syrian refugee children more than other children. Children and their families Who don't know Turkish, television broadcasts that are in Turkish, lack of television, internet, computer at home due to poverty, crowd family population, and obligation to children for taking an active role in housework negatively affected the education of refugee children. These situations, which affect the access to Education Information Network (EBA) courses, have also caused the children's level of benefit from the courses to be low. On the other hand, the fact that EBA courses can be recorded and watched again reduced the problems experienced by children in expressing themselves in face-to-face environments (<unk>ekiç, 2019: 794-796). --- Language Issue In general, in the adjustment of students at school; Factors such as the behavior of teachers and administrators, friendship relations with their peers, the rate of feeling safe, and relations in and out ODÜ Sosyal Bilimler Araşt<unk>rmalar<unk> Dergisi / ISSN: 1309-9302 / dergipark.org.tr/tr/pub/odusobiad • of the classroom are effective (Aykut, 2019: 176). This is even more important for immigrant children from different languages and cultures. In order to achieve this, knowledge of language is the first requirement. The majority of immigrants from Syria, except Turkmen, do not speak Turkish. In addition to the absence of Arabic-speaking staff, teachers, the number and quality of teachers to teach Turkish are insufficient. Therefore, there is a communication problem between local teachers and students. The continuity of teachers' attendance is insufficient. It is a quality problem for those who do not have pedagogical training and who are not teachers to teach in Arabic (Balc<unk>, 2020). In addition, people involved in the education of refugee children should be equipped with education in emergencies (Yavuz & M<unk>zrak, 2016: 194). All these situations negatively affect the education quality of Syrian students, especially in public schools. The success of social integration has decreased as a result of the fact that Syrian students attending school mostly prefer TECs due to language problems and social adaptation problems (K<unk>z<unk>l & Dönmez, 2017: 234). There is a wage problem due to the fact that Syrian teachers working in temporary education centers cannot be paid directly by the Ministry of National Education. Syrians who work as teachers are paid a small amount by various international and local institutions and organizations such as Türkiye Presidency of Religious Affairs, Municipalities and UNICEF (Emin, 2016: 21). Teachers who have Syrian students in their classrooms suggest that these students should be taken to a preparatory class before being distributed to other classes and they should be taught Turkish with the Latin alphabet (Erdem, 2017: 39). In addition, in order to increase the motivation of students for language learning while providing this lecture, care should be taken that the number of Syrian students in the classrooms is not too high when compared to the number of general class students. Because students making groups among themselves and communicate in their own language. As a result of this, it has been determined that the level of need for Turkish has decreased and the Turkish language learning of Syrian students has therefore fallen behind (<unk>akmak, 2018: 74-75). A second factor in the language learning and adaptation process is the family attitudes of refugee for their children. However, families who do not have sufficient knowledge of Turkish may also be unable to communicate with the school (<unk>akan et al., 2018: 292). In terms of families, not knowing Turkish or knowing Turkish at a low level causes them to be unaware of their legal rights and duties and to have difficulty in handling their official affairs. This situation can affect refugees at obtaining a valid identity document, receiving services from the public, benefiting from social assistance and social services, etc. Not knowing the language causes women to participate less in society, and therefore, they cannot adapt to social life (Aslan & Güngör, 2019: 1624). --- War trauma recovery challenges and need for specially educated instructors Studies have shown that the mental health of the children of emigrant families is adversely affected due to the traumatic events they experience during and after the migration process. Problems such as forcing girls to marry at a young age, not being sent to school, being exposed to all kinds of violence, lack of nutrition, not being able to benefit from their health rights adequately, and being forced into labor are the negativities experienced in this context (Karanfilo<unk>lu, 2019: 109). Those who have the right to go to school may have less chance of being affected by these negativities and traumas. However, the content of the education and the knowledge and attitude of the educator are important here. The ability of refugee students who have survived great difficulties at an early age when basic knowledge and skills are learned and adapting to the country they live in is related to the approaches of their teachers (Sa<unk>lam & Kanbur, 2017: 312). The children of refugee families who came to Türkiye by fleeing the war in Syria need to see a different approach compared to the children of refugee families from other countries. According to the teachers and administrators who participated in a study, the reason for this was the death of Syrians in their country and the fact that they came to Türkiye to escape from the war after experiencing fear. Iranian, Afghan and Iraqi students are in Türkiye as a result of their families' desire to have better living conditions and educational opportunities in Türkiye. The families of these students have better financial opportunities and higher education levels. Therefore, they have fewer problems in adapting to Türkiye (Sar<unk>taş et al., 2016: 225). In order to provide effective education to Syrian students under temporary protection who have experienced war trauma, it is important that school counselors are educated and experienced in working with children and adolescents from different cultures. This education should include multicultural counseling, advocacy and trauma (Ka<unk>n<unk>c<unk>, 2017(Ka<unk>n<unk>c<unk>, : 1773)). In this sense, it is important that not only teachers and school counselors but also nurses have special training on the education of refugee children and families. In this context, nurses should have knowledge about the chronic and infectious diseases of refugee children, basic human rights, and growth and development processes. In addition, nurses should work on providing safe playgrounds for children, raising the awareness of the society about the effects of migration for children and families, enabling refugee children to communicate with people and eliminating the prejudices of the society (Ayd<unk>n et al., 2017: 13). ODÜ Sosyal Bilimler Araşt<unk>rmalar<unk> Dergisi / ISSN: 1309-9302 / dergipark.org.tr/tr/pub/odusobiad • <unk>t is possible to say that a large number of professionals have gained field practice in the ten years since the Syrian refugees first arrived to Türkiye. --- Curriculum and Teaching Materials Problems In terms of the quality of education, the number of students per teacher should be 40 at most (Yavuz and M<unk>zrak, 2016:195) In addition to the building deficiencies, there is a lack of tools and equipment in learning Turkish (Balc<unk>, 2020). Educators suggest the implementation of level classes that will classify students according to grammar in Turkish education and reinforcement courses for Syrian children (Göksoy, 2020: 465). Various aids are provided to Syrian students by civil society and international aid organizations. UNICEF and Turkish Red Crescent are at the forefront of these organizations. In this context, besides a certain amount of financial aid, stationery and clothing aid is also provided to the students who come to the school (Eren, 2019: 220). School needs and stationery sets are distributed to children by the Association for Solidarity with Asylum Seekers and Migrants (SGDD), and mediation is provided for helping other aid organizations and charitable citizens (Ak<unk>ll<unk> & Dirikoç, 2015: 264). --- Current Diploma Issue in Temporary Education Centers It is a problem that students studying in temporary education centers are not given a valid diploma. Because of this problem, if the refugee student goes to his own country or a third country after education, he cannot provide the equivalence of education. In the solution of this problem, especially international organizations should play a role due to the political crisis between Türkiye and Syria (Yavuz & M<unk>zrak, 2016: 196). --- Lack of Coordination, Information and Guidance First of all, Syrians and parents do not know about public school opportunities and their knowledge of their rights and responsibilities in education is insufficient (Balc<unk>, 2020). This is due to the inadequacy of informing the public about the activities carried out by official institutions in the field of education for Syrians in Türkiye (Seydi, 2014: 297). In order for the aid and training activities to be regular and effective, coordination between public institutions and non-governmental organizations should be improved (Emin. 2016: 22). --- Social Issues --- Disadvantaged Children Children who do not have a parent or family or who are disabled need special education and rehabilitation (Balc<unk>, 2020). --- Socio-Economic Disability and Child Labor Due to factors such as the low education level of Syrian families, poverty, lack of social security, and cultural differences with the local people, it makes it difficult for children to adapt to the society they live in (Ereş, 2015: 17). In fact, it can be seen that families make their children work and even beg money from the principal and teacher at school due to poverty (Balc<unk>, 2020). At<unk>f / Citation: <unk>KTEN, Serkan. (2023). "Educational Policies and Practices Regarding the Adaptation of Syrians Under Temporary Protection to Türkiye" ODÜSOB<unk>AD 13 (2), 2457-2482, Doi: 10.48146/odusobiad.1232477, Doi: 10.48146/odusobiad.1232477 The biggest victims of the war in Syria are refugee children. They are deprived of their right to education because they have to be partners in their families

In 2011, after the internal conflicts in Syria, mass migration to Türkiye started. Syrians entering the country are under temporary protection status. In the face of this situation, policies that required social cohesion were implemented and education became one of the most important cohesion policies. In this context, revealing how education is provided for immigrants in Türkiye has been important in demonstrating the success of integration policies. In this study, it is aimed to reveal what are the policies and implementation results for the inclusion of Syrians under temporary protection in the Turkish education system. In the method of the study, official statistics, legal regulations, news and academic research were handled and document analysis was used. In this context, all dimensions of the subject from primary education to higher education were categorized. When the findings were shared systematically, it was seen that the Syrians, who were initially considered temporary, continued their education in the camps without obtaining a diploma or document. With the end of the war, legal arrangements such as "Educational Services for Foreigners" and "Execution of Services for Foreigners under Temporary Protection" were made in order to include Syrians in the Turkish National Education System and Higher Education. In the field studies examined in the literature analysis, it was seen that despite the legal regulations, technical and social problems continued in practice and the participation in post-primary education was low. In addition, it has been revealed that the success of education policies for Syrians under temporary protection is related to the cultural structure of the cities where Syrians live in Türkiye, their geographical proximity to Syria, the population of Arab origin in the city, the status of schools, and the qualifications of teachers. As a result of the implications of the findings; It is recommended to provide effective language learning to all age groups, to ensure continuity in school after primary school by taking measures against situations that prevent participation in the lesson, and to encourage Syrian academicians and higher education students, who are the most susceptible group.

dvantaged Children Children who do not have a parent or family or who are disabled need special education and rehabilitation (Balc<unk>, 2020). --- Socio-Economic Disability and Child Labor Due to factors such as the low education level of Syrian families, poverty, lack of social security, and cultural differences with the local people, it makes it difficult for children to adapt to the society they live in (Ereş, 2015: 17). In fact, it can be seen that families make their children work and even beg money from the principal and teacher at school due to poverty (Balc<unk>, 2020). At<unk>f / Citation: <unk>KTEN, Serkan. (2023). "Educational Policies and Practices Regarding the Adaptation of Syrians Under Temporary Protection to Türkiye" ODÜSOB<unk>AD 13 (2), 2457-2482, Doi: 10.48146/odusobiad.1232477, Doi: 10.48146/odusobiad.1232477 The biggest victims of the war in Syria are refugee children. They are deprived of their right to education because they have to be partners in their families' struggle for survival. Children are worked in fields and gardens in rural areas, and in workshops and on the streets in cities. In border provinces such as Gaziantep, Kilis, <unk>anl<unk>urfa, and Hatay, Syrian child workers can be found working in knitwear workshops, textile factories, dried fruit factories, shoe manufacturing workshops and car repair shops, agricultural labor, and in the streets, in jobs such as paper tissue and water sales (Akp<unk>nar, 2017: 23). --- Social Acceptance and Adaptation Problem It can be seen that Syrian families do not want to enroll their children in school for fear of being bullied. In order to prevent this situation, it is necessary to carry out studies to ensure social integration before physical adaptation (Balc<unk>, 2020). --- It is seen that Syrian refugees, who have economic opportunities and have established a business in Türkiye, easily rent a house, their children continue their education and establish good relations with the local people through their workplaces. This makes them more advantageous than Syrians who do not have access to economic resources. Adaptation processes for themselves and their children are slow for those whose economic situation is not good (<unk>imşek, 2018: 383). It is important that the educated refugees contribute more to employment and the development of the country in order to solve the problem of adaptation to the society they live in. This will enable them to benefit more from the country's opportunities socially and economically and to be more accepted by the local people (Ertan ve Ertan, 2017: 23-24). From this point of view, considering that most of the Syrian refugees who have been in Türkiye for more than ten years will stay in the country in the next period, it is important to plan an education that will ensure adaptation and employment for them. Schools and educational environments are important for refugee children to recognize the new culture that they met in the first place. Therefore, it should be known that adaptation to school is the beginning of the cultural transformation process (<unk>imşek ve Kula, 2018: 18). In addition to the correct planning of the educational content, necessary measures should be taken to eliminate the obstacles to school attendance. --- Factors Affecting Success Neglects such as not knowing the language, adaptation, irregularity, uncertainty, lack of planning, absenteeism from school, lack of boundaries, health problems, irregularity, transition to higher education without examination, family indifference, differences in practice, and mental chaos in the society on this issue lead to failure in education for Syrian students (Göksoy, 2020: 463). In addition, the tension created by the new life that comes with war, migration and displacement all over the world negatively affects the inner world of children. Children's longing for their homeland, the differences between their old and new lives, the physical and social deprivation of their new places, the tension and hostility caused by cultural incompatibility, social pressures such as discrimination, exclusion, and marginalization cause severe and traumatic consequences for children (Gencer, 2017: 848). Therefore, the attitudes and behaviors they encounter at school are an important ODÜ Sosyal Bilimler Araşt<unk>rmalar<unk> Dergisi / ISSN: 1309-9302 / dergipark.org.tr/tr/pub/odusobiad • tool in overcoming this trauma. In this regard, as well as trainers, local peers and families who will motivate them to have a positive attitude are needed. --- Different City Examples Apart from these general problem determinations, different practices and results have emerged for the education of Syrians in different cities of Türkiye in line with the efforts or the level of knowledge of the trainers and the conditions of the school and city. The differences in this direction have been revealed by various studies. In a study conducted in Samsun, which is located in the Black Sea region of Türkiye, it was determined that refugee children had difficulties in communicating with their teachers and peers. As a result, it has been observed that Syrian students are largely deprived of the knowledge, skills and attitudes provided by the school. In addition, it was determined that they could not socialize as a result of difficulties in getting along with their peers and they were mostly alone. Another finding is that Syrian children's inability to meet basic care skills such as cleaning due to socio-economic impossibilities creates discomfort for other families. It was observed that when teachers had the opportunity to meet with parents, they could not establish a healthy dialogue (Uzun & Bütün, 2016: 79). As a result of the interviews held with teachers, school administrators, Syrian students and parents in Alt<unk>nda<unk> District of the capital Ankara, the issues of communication, education process, classroom climate, physical structure of the school, school attendance, enrollment, behavior, cultural differences, academic success, bullying-exclusion and marginalization were discussed. It has been revealed that there are problems on these issues (Songür & Olgun, 2020: 2969). In the research in Nevşehir, one of the cities of the Central Anatolia region, the classroom teachers stated the problems encountered in the education of Syrian students; language problem, registration, bullying, adaptation, attendance, cleanliness, separation and being unwanted by Turkish parents (<unk>zenç & Ferhat, 2019: 68-70). In a study conducted in Afyonkarahisar province in the west of Türkiye, it was revealed that teachers are in need of professional development in terms of analysis of the content of the education to be given to refugee students, teaching strategies, lesson tools and materials, development and evaluation of measurement tools. It has been stated that this situation arises from the fact that teachers receive university education for education in monocultural and monolingual environments and gain experience in this context (Erdem, 2017: 39). As can be seen, in various researches conducted in different provinces from the north to the west of Türkiye, from the inner regions to the capital, it has been revealed that the Syrian students face many problems as a result of the trainings given to Syrian students. However, there are other studies that show that the training provided has positive results or that different applications can be more successful. At<unk>f / Citation: <unk>KTEN, Serkan. (2023). "Educational Policies and Practices Regarding the Adaptation of Syrians Under Temporary Protection to Türkiye" ODÜSOB<unk>AD 13 (2), 2457-2482, Doi: 10.48146/odusobiad.1232477 A study on refugee children who came for religious education in the Beykoz district of Istanbul in 2020 showed that the content of the education provided ensured the interaction and harmony of refugee children with the society they live in. This result revealed that common religion courses are of a quality that will make Syrian students feel like a foreigner in Türkiye. However, it has been stated that the content of religious education in question aims to protect refugee children's ties with their past without isolating them from the culture of their own country (Sever, 2020: 1). In a study based on the opinions of teachers working in the 2017-2018 academic year in <unk>stanbul and Bursa, the important industrial cities of Türkiye's Marmara Region, it was seen that teachers believed that there was no cultural conflict between Syrian and Turkish students studying at primary school despite cultural and social differences. Considering that the problem in cultural cohesion stems from not knowing the language, it has been concluded that teachers benefit from group work, values education, social activities and children's games for language improvement (Jafari et al., 2018: 134). As a result of the research in question, the results in line with the prediction that the problem of adaptation will be solved with the solution of the problem of not knowing the language of the teachers were also revealed in a study conducted in <unk>stanbul. Accordingly, in the research conducted through interviews with some teachers working in primary schools in <unk>stanbul's Ataşehir district, it was concluded that the attitudes and perceptions of Syrian students' friends towards them were prejudiced and exclusionary at the beginning, but a supportive and sharing attitude emerged with the increase in communication with the efforts of the teachers and the development of the language (Karaa<unk>aç & Güvenç, 2019: 557). In a study conducted in Samsun, it was seen that teachers carried out activities to strengthen the communication skills of refugee children, to ensure their active participation and to facilitate their adaptation to the social order. For this purpose, it has been determined that they have developed methods such as using illustrated text, language teaching with games, setting concise rules, exhibiting a moderate approach, creating a need for Turkish by creating a purposeful seating arrangement. In addition, it has been determined that teachers use techniques such as technology use, repetition and concretization to make it easier for students to understand (<unk>akan et al., 2018: 290). On the other hand, there are opinions in the literature that sports activities can contribute to the positive development of the personality of refugee children. Accordingly, since participation in sports is a means of socialization, sports activities are helpful in the psychosocial development of children and in solving socio-cultural problems (Sag<unk>n & Güllü, 2020: 95). It has also been demonstrated by other research results that being able to speak a common language and share in common social activities such as sports increases harmony between local and refugee students. It has been determined that in the 2019-2020 academic year, Syrian students studying in the <unk>arşamba district of Samsun encounter a general acceptance within the school, and their adaptation problems decrease. It has been observed that as the duration of students going to school in Türkiye ODÜ Sosyal Bilimler Araşt<unk>rmalar<unk> Dergisi / ISSN: 1309-9302 / dergipark.org.tr/tr/pub/odusobiad • increases, their sense of belonging to Türkiye also increases (Aslan, 2020: 42). This situation shows the importance of taking measures to ensure the attendance of more Syrian students to school. Although it is thought that sharing on common religion and belief is effective in harmony, it is not as effective as speaking the common language. In the research conducted with the teachers of Religious Culture and Moral Knowledge, who work in various cities of Türkiye and have Syrian students in their classes, different results have emerged. The majority of the teachers participating in the research stated that Syrian children experience various problems due to language barriers and that they encounter problems such as labeling, marginalization and violence in social cohesion (Alt<unk>ntaş, 2018: 469). This example has shown that although common religious education is unifying in terms of harmony in terms of content, it still prevents being successful enough if there is no practice of speaking the same language. On the other hand, integration with local people can be more successful for Syrians living in cities in Türkiye, which have linguistic, historical and cultural, even ethnic and kinship ties with their country. A study conducted in Ankara and Hatay provinces is an important example in showing the effect of living in cities close to their own culture on adaptation. In a study comparing the situation of Syrian children living in Ankara and Hatay provinces, it was determined that children living in Hatay province, which is at border of Syria, have less difficulties and have more freedom. For example, Syrian girls living in Hatay feel more secure and can go out of the home more. It is thought that the reasons for this situation are that Hatay is culturally and geographically close to Syria, that there is a large number of people who can speak Arabic, that there are kinship ties with Syrians and that Hatay is a smaller city. It has been stated that the fact that Syrian children in Hatay experience their own language and cultural characteristics more intensely here facilitated the adaptation process of the children, but on the other hand, it made it difficult for them to learn Turkish and to comply with the country's rules (Gencer, 2019: 280-281). --- Conclusion and Recommendations After the migration to Türkiye that started in 2011, many legislations have been published regarding the education of Syrian children and young people. In addition, Syrian people have implemented various projects in cooperation with the Ministry of National Education, the European Union and non-governmental organizations to ensure their adaptation to the Turkish education system and their access to education. Within the scope of the education policy for Syrian children and youth in Türkiye, education have been started on five main platforms. These are; Temporary Education Centers, public schools, Syrian private schools, open education institutions and higher education institutions. In this context, despite the legislative changes and the opportunities provided in all kinds of training platforms, technical and social problems arose in practice. At<unk>f / Citation: <unk>KTEN, Serkan. (2023). "Educational Policies and Practices Regarding the Adaptation of Syrians Under Temporary Protection to Türkiye" ODÜSOB<unk>AD 13 (2), 2457-2482, Doi: 10.48146/odusobiad.1232477, Doi: 10.48146/odusobiad.1232477 Technical problems are difficulties experienced in the registration process, difficulty in determining equivalence and class level, problems in accessing and participating in education, language learning problems, difficulties in overcoming war trauma and the need for specially trained trainers, lack of curriculum and teaching materials, valid diploma problem in temporary education centers, the lack of coordination, information and guidance. Social problems, such as disadvantaged children without family or disabled children, early marriage and child labor as a result of socio-economic insufficiency, social acceptance and adaptation problems, uncertainty, lack of planning, absenteeism from school, health problems, irregularity, transition to higher education without examination, family indifference, etc. are other elements. In the studies showing the results of the application, it has been determined that solving the language problem in many different city examples generally ensures the social cohesion of the students and their success in education. In conclusion, The Ministry of National Education, together with many international organizations in Türkiye, carries out important studies for Syrian refugees in terms of adaptation to the national education system (K<unk>z<unk>l and Dönmez, 2017: 234). However, in order to solve the problems and problems experienced, decision makers should not only focus on the physical initiation of school for refugees, but also develop multi-dimensional policies to ensure social cohesion (Yavuz & M<unk>zrak, 2016: 194). In solving the problems here, state institutions and NGOs; The joint work of Syrian refugees and local governments will contribute (<unk>akmak, 2018: 467). There is a need for a culturally sensitive and inclusive education policy that will support the schooling of Syrian children and support their education processes. It is necessary to carry out child-oriented studies that will take place in school-family-society cooperation. In order to reduce the current problems of Syrian students, school social work practices should be carried out based on their needs, individual differences and potentials and evaluating them with all aspects in a bio-psycho-social framework (Gencer & <unk>zkan, 2017: 229). Also, compliance is not a one-sided situation. Therefore, it requires the common will and practices of the local people and refugees. Therefore, it is important that the social acceptance and adaptation process be carried out simultaneously and through policies that complement and support each other. In this context, while making policies and regulations regarding Syrians under temporary protection in Türkiye, decision mechanisms should be established where their opinions will be taken (Y<unk>ld<unk>r<unk>malp vd., 2017: 124). In conclusion, this study shows that the Syrian refugees, who were initially thought to be temporary, were planned to continue their education in the camps and without obtaining a diploma or document, and then, with the prolongation of the war, the necessary legal arrangements were made to include Syrians in the Turkish National Education System and Higher Education. Despite the arrangements made, it is seen that technical and social problems continue in practice and the desired level of participation in education after primary school cannot be reached. However, it has been revealed that the results of education policies for Syrians under temporary protection vary with factors such as the ODÜ Sosyal Bilimler Araşt<unk>rmalar<unk> Dergisi / ISSN: 1309-9302 / dergipark.org.tr/tr/pub/odusobiad • cultural structure of the cities where Syrians live in Türkiye, their geographical proximity to Syria, the population of Arab origin in the city, the status of schools, and the qualifications and experience of teachers. On the other hand, it has been understood that the most important factor in ensuring success and harmony in education, wherever it is, is grammar. It has been revealed in almost all researches on the subject that the efforts made to ensure sufficient success in this regard cannot provide the desired result to a large extent. It has been determined that it is important to ensure that an increasing number of Syrians receive education in higher education and that Syrian academics take part in the academy, even if it is not at the desired level. The increase in this group as an educated and most adaptable group to contribute to employment has been found to strengthen the bond formation between both communities. In this respect, effective language learning in all age groups, ensuring school attendance and increasing the number of academicians and students in the academy should be the priority policy issues for harmonization. Sosyal sorunlar ise ailesi olmayan dezavantajl<unk> veya engelli çocuklar, sosyo-ekonomik yetersizlik sonucu erken yaşta evlilik veya çocuk işçili<unk>i, sosyal kabul ve uyum sorunlar<unk>, belirsizlik, plans<unk>zl<unk>k, okula devams<unk>zl<unk>k, sa<unk>l<unk>k sorunlar<unk>, düzensizlik, s<unk>navs<unk>z yüksek ö<unk>renime geçiş, aile ilgisizli<unk>i vb. di<unk>er unsurlard<unk>r. --- Genişletilmiş <unk>zet E<unk>itim politika uygulama sonuçlar<unk>n<unk> gösteren çal<unk>şmalarda birçok farkl<unk> şehir örne<unk>inde dil sorununu çözmenin genel olarak ö<unk>rencilerin sosyal uyumunu ve e<unk>itimdeki başar<unk>lar<unk>n<unk> sa<unk>lad<unk> tespit edilmiştir. Bu kapsamda Milli E<unk>itim Bakanl<unk>, Türkiye'deki birçok uluslararas<unk> kuruluşla birlikte Suriyeli mülteciler için ulusal e<unk>itim sistemine uyum aç<unk>s<unk>ndan önemli çal<unk>şmalar yürütmekte oldu<unk>u söylenebilir (K<unk>z<unk>l ve Dönmez, 2017: 234). Ancak yaşanan sorunlar<unk>n ve sorunlar<unk>n çözümü için karar al<unk>c<unk>lar<unk>n sadece mülteciler için fiziksel olarak okul aç<unk>lmas<unk>na odaklanmas<unk> de<unk>il, ayn<unk> zamanda toplumsal uyumu sa<unk>lamaya yönelik çok boyutlu politikalar geliştirmesi gerekmektedir (Yavuz ve M<unk>zrak, 2016: 194). Buradaki sorunlar<unk>n çözümünde devlet kurumlar<unk> ve STK'lar; Suriyeli mülteciler ve yerel yönetimlerin ortak çal<unk>şmas<unk> katk<unk> sa<unk>layacakt<unk>r (<unk>akmak, 2018: 467). Suriyeli çocuklar<unk>n okullaşmas<unk>na ve e<unk>itim süreçlerine destek olacak, kültürel aç<unk>dan duyarl<unk> ve kapsay<unk>c<unk> bir e<unk>itim politikas<unk>na ihtiyaç vard<unk>r. Okul-aile-toplum işbirli<unk>i içinde yer alacak çocuk odakl<unk> çal<unk>şmalar<unk>n yap<unk>lmas<unk> gerekmektedir. Suriyeli ö<unk>rencilerin mevcut sorunlar<unk>n<unk>n azalt<unk>lmas<unk> için okul sosyal hizmeti uygulamalar<unk>n<unk>n ihtiyaçlar<unk>, bireysel farkl<unk>l<unk>klar<unk> ve potansiyelleri temel al<unk>narak ve tüm yönleriyle biyo-psikososyal çerçevede de<unk>erlendirilerek gerçekleştirilmesi gerekmektedir (Gencer ve <unk>zkan, 2017: 229). Ayr<unk>ca uyum tek tarafl<unk> bir durum de<unk>ildir. Dolay<unk>s<unk>yla uyum yerel halk<unk>n ve göçmenlerle ortak irade ve uygulamalar<unk>n<unk> gerektirir. Bu nedenle toplumsal kabul ve uyum sürecinin eş zamanl<unk> ve birbirini tamamlayan ve destekleyen politikalar arac<unk>l<unk>yla yürütülmesi önemlidir. Bu ba<unk>lamda Türkiye'de geçici koruma sa<unk>lanan Suriyelilere ilişkin politika ve düzenlemeler yap<unk>l<unk>rken görüşlerinin al<unk>naca<unk> karar mekanizmalar<unk> oluşturulmal<unk>d<unk>r (Y<unk>ld<unk>r<unk>malp ve Cemal, 2017: 124). --- Sonuç olarak bu çal<unk>şma, uyum kapsam<unk>nda e<unk>itim politikas<unk> belirlenirken başlang<unk>çta geçici oldu<unk>u düşünülen --- Conflict of Interest Statement There is no financial conflict of interest with any institution, organization or person related to my article titled "Educational Policies and Practices Regarding the Adaptation of Syrian Under Temporary Protection to Türkiye".

In 2011, after the internal conflicts in Syria, mass migration to Türkiye started. Syrians entering the country are under temporary protection status. In the face of this situation, policies that required social cohesion were implemented and education became one of the most important cohesion policies. In this context, revealing how education is provided for immigrants in Türkiye has been important in demonstrating the success of integration policies. In this study, it is aimed to reveal what are the policies and implementation results for the inclusion of Syrians under temporary protection in the Turkish education system. In the method of the study, official statistics, legal regulations, news and academic research were handled and document analysis was used. In this context, all dimensions of the subject from primary education to higher education were categorized. When the findings were shared systematically, it was seen that the Syrians, who were initially considered temporary, continued their education in the camps without obtaining a diploma or document. With the end of the war, legal arrangements such as "Educational Services for Foreigners" and "Execution of Services for Foreigners under Temporary Protection" were made in order to include Syrians in the Turkish National Education System and Higher Education. In the field studies examined in the literature analysis, it was seen that despite the legal regulations, technical and social problems continued in practice and the participation in post-primary education was low. In addition, it has been revealed that the success of education policies for Syrians under temporary protection is related to the cultural structure of the cities where Syrians live in Türkiye, their geographical proximity to Syria, the population of Arab origin in the city, the status of schools, and the qualifications of teachers. As a result of the implications of the findings; It is recommended to provide effective language learning to all age groups, to ensure continuity in school after primary school by taking measures against situations that prevent participation in the lesson, and to encourage Syrian academicians and higher education students, who are the most susceptible group.

Introduction With the emergence of feminism, more and more people are becoming aware of the importance of gender equality. However, gender inequality still persists in visual culture that guides people's mindsets. The visual culture of gender inequality revealed in many films is subliminally reinforcing the cultural norms of the audience. According to Dowd, cinema is a teaching machine, especially the way in which it socializes people in a subtle way and provides informative representations of how people should behave [1]. The messages of gender inequality and visual culture are implicit in many Hollywood films, which are becoming very popular with the audience. Based on the theories of feminism and visual culture, this paper briefly explores how the visual culture of gender inequality is reflected in Hollywood films, focusing on three aspects, namely the unequal representation of men and women in the films, the imbalance of male and female film workers and their incomes, and the predominance of male perspectives in Hollywood films and the gender stereotypes. Hollywood has also focused on gender equality for a long time. Several scholars in academia have shown that in the early days of film history, men and women were equal and women played a key role in the film industry. Stamp said that in the silent era of film history, women occupied key positions, like directors, writers, and producers [2]. Hill's Never Did describes the various important tasks that women performed behind the scenes of the film industry in the beginning. He argued that the persistence of feminized areas of labor in the film industry today, such as work casting, script supervision, publicity, and editing, is mostly based on women's "creative service" in the early days of the film industry [3]. Smyth told that the heyday of the film business was run by women [4]. Despite the significance of women in the history of cinema and the gender equality they showed, today's cinema still has many inequalities between men and women. This exemplifies the changing positions of women in the film industry and the growing gender inequality. This paper, therefore, examines the issue of gender inequality in visual culture, focusing on a few representative Hollywood films and using relevant statistics to demonstrate how gender inequality is reflected in the visual culture of Hollywood films. --- The Inequality between Men and Women in Hollywood Films In today's innovative, open and inclusive film industry, women still possess fewer positions and achieve less income compared to men. --- The Unequal Time Distribution In Hollywood movies, the time male actors occupied is almost twice as long as female actors. These data are obtained through the machine learning system. The machine learning system is jointly developed by the University of Southern California, Gina Davis media Gender Research Institute, Google charity, and other organizations. This system uses language recognition and faces recognition algorithms. It can distinguish the gender of actors in a film and record their appearance time. Because of a lot of training, the system can be accurate in seconds. According to the statistics, in the three years from 2014 to 2016, there were 100 live-action films with the highest box office in Hollywood, and female actors only accounted for 36% of these films [5]. In most films, male actors occupy a dominant position in terms of the number of people and the responsibilities they undertake. Male actors occupy the dominant power, so their discourse power is more, and their importance is higher than that of women, leading to the relatively low sense of the existence of female actors in films and the marginalization of female roles. In Hollywood, there are many films with the theme of women, for example, Ant-Man and the Wasp is a film that can be said that men and women have equal status [6]. As a female superhero, the hornet woman has the same ability as the male superhero. Her destiny is in her own hands, and she does not need to rely on the protection of others. This phenomenon is extremely rare in films. Ant-Man and the Wasp is also known as a film in which female characters in Marvel have occupied the longest time. However, even so, the heroine van Dane's appearance time in this film is only 35 minutes [7]. Therefore, although gender equality is constantly improving with the development of the times, it is still not the mainstream idea, and gender equality is only reserved at the theoretical level, not practical. In Hollywood movies, most of the protagonists are male. The study The Film World of Men using a sample of 2,300 characters from the top 100 domestic box office films of 2013, found that 85% of the leading characters in Hollywood films in 2013 were men [8]. This is especially true in action movies, in this genre. Movies like Spiderman, Iron Man, and Hulk all have male heroes [9][10][11]. IMDB's top 20 movies are almost exclusively male. Stacey Smith, a professor at the University of California, Santa Barbara, looked at the films of 2014 and found that only 21 of the top 100 grossing films had a female lead or co-lead [12]. In Hollywood movies, the camera setting is usually dominated by the male protagonists, and the angle of view is also guided by them. Male characters act as the gods of Hollywood movies. were women, and there was only one female director in every 23 directors. In the past 12 years, the number of female directors did not exceed 10% each year. In 2008, the number of female directors reached the highest, accounting for 8%. In 2018, female directors accounted for only 3.6% [13]. Men also account for the majority of behind-the-scenes positions in photography, editing, props group, film dubbing, and so on. This situation has not been improved compared with more than ten years ago. The long-term gender discrimination has a negative impact on the female behind-the-scenes staffs. Even most people think that in the behind-the-scenes work of the film, women will only act as assistants, clothing designers, and makeup artists. According to the 2015 Hollywood diversity report, 83% of the senior executives of major Hollywood film production companies are male, and more than 6000 members of the Academy of Film Arts and Sciences are all male. Only 22.8% of Hollywood's top-level workers are women and less than 6% of them are non-white women [14]. The ability of female behind-the-scenes staffs is equal to that of men. They can even create greater value, but the ability of women is often ignored. This kind of thinking does not mean that women and men have physiological differences, but that women are not able to undertake this job. Women are deprived of the right to obtain fair social distribution. --- The Income Gap between Male and Female Actors In Hollywood movies, the income gap between male and female actors is also incredible. American actress Natalie Portman, who won the best actress award at the 83rd Academy Awards, said publicly after she starred in the film Wireless Love, that her remuneration was only one-third of that of the actor Ashton Kutcher [15]. In the 2013 film American Hustle, the remuneration of the female stars Jennifer Lawrence and Amy Adams was far lower than that of other male supporting roles in the film [16]. During the film's release, the roles of actresses in propaganda were far greater than those of male actors, but they still received much lower salaries than male actors. In recent years, more and more actresses have come forward to speak for the issue of "equal pay for equal work" for Hollywood actors and actresses. At the same time, many films showing strong female images have emerged endlessly, but they have not really changed the current situation of gender discrimination in Hollywood. --- Gender Stereotypes in Hollywood Films Gender stereotypes can be reflected in the plots portrayed in many Hollywood films. Women are often portrayed as inherently weak characters who are dependent on their men. Moreover, they are also the representatives of sex. Dowd argues that sex and desire are central elements in the portrayal of women in the Hollywood filmmaking system [1]. Denzin thinks that classic Hollywood films have never been kind to women, and women have always been given stereotypical roles, either good women or bad women, good mothers or bad mothers, sinners, or saints [17]. Das discusses that women are often depicted as traditional roles, such as family members, companions, sex partners, and janitors, and their positions are subordinate to men [18]. Greenberg argues that women are also portrayed as thin and sexualized [19]. This shows that gender stereotypes can be deeply reflected in traditional Hollywood films. However, Hollywood films do not really eschew gender stereotypes. With the influence of feminism, Hollywood films have not directly portrayed women as completely dependent on men for survival, but gender stereotypes and gender inequalities can still be seen in the films. The portrayal of female characters is still bound by gender stereotypes in terms of dress and pursuits. Many people consider the film Legally Blonde to be clearly a feminist film and a big step forward in Hollywood's portrayal of women as strong, independent, and critical thinkers. However, according to History in Politics, the film is still full of stereotypes of women [20]. For example, women like pink, and the pursuit of love is a woman's main goal in life and investing in fashion and beauty is a woman's second goal in life. While the film does attempt to attack the "stupid blonde" stereotype, other stereotypes of women persist. In terms of the plots that shape the destinies of women, Hollywood films still cannot escape gender stereotypes. Wartenberg states that an adventurous and independent woman in the first ten minutes of a Hollywood film is likely to evolve into a more submissive and supportive assistant to the man over the course of the film [21]. Such plots are evident in classic Hollywood films such as Pretty Woman, in which the resourceful Vivian, is far more excellent in the early stages of the film than wealthy businessman Edward, but she gradually changes into a more traditional supporter of Edward after she loves Edward [22]. In the film The Maid of Manhattan, the female protagonist in the film named Marissa is a single mother working as a maid in a four-star Manhattan hotel [23]. Despite the low status of her job, Marissa is resourceful, intelligent, and friendly, and is admired by her peers. She is also a loving presence in the life of her precocious young son. Things change, however, when she meets an up-and-coming politician, Christopher Marshall. When he leaves her in the second act, her world falls apart. When her son intervenes with Christopher and asks him to forgive his mother for her mistake, the politician brings her back and Marisa becomes a successful businesswoman. The film embodies the default gender stereotype that women's success depends on the intervention of a male sponsor. Thus, gender inequality persists in Hollywood films. Women are always the objects of sight and behavior of male protagonists in movies. The female figure is shown from the male perspective. The audience's viewing angle is male because the frame of the film is always focusing on male protagonists. The male gaze phenomenon is particularly prominent in Hollywood films. The male gaze refers to the view from the angle of heterosexual men through whom females are sexually objectified to please male audiences. Research published by the Annenberg School for Communication and Journalism also found that female characters are more often used as "sex objects" in films than men. Nearly 28 percent of all female characters appeared in sexy costumes and only 8 percent of men did this. Only 9.1 percent of male characters showed nudity, while 26.4 percent of female characters showed nudity [24]. For example, in the movies Spider-Man, Hulk, and Iron Man, the heroes rescue the female characters in trouble to show their strength and bravery. The Film World of Men found that only 15 percent of Hollywood films last year featured female protagonists. The researchers said that female characters in films mostly exist to serve male characters and are rarely portrayed as leaders [8]. Overall, the gender stereotypes portrayed in Hollywood films did not disappear with the advent of feminism. On the contrary, gender equality in Hollywood still has a long way to go. --- Conclusion This paper has argued that in Hollywood films, the visual culture of inequality between men and women is mainly reflected in the distribution difference between men and women and their representation in films. By observing some Hollywood movies, a series of phenomena show that women are still treated unfairly in Hollywood movies. This shows that in Hollywood movies, there is still a long way to go to achieve real equality between men and women. Although the theoretical framework is not systematic, this study can make more people aware of the unequal treatment women face. Some phenomena reflected in Hollywood movies are closely related to mainstream ideology. Because women are in an unequal position in society, the images of women in the movies are also unequal. --- With the power of film culture, more people can pay attention to the interests and situations of women. Future studies can look into female filmmakers who show and give play to their abilities, contribute to the film industry, and let the audience in front of the screen feel the power of women.

The main content of this paper is to explain how the visual culture of gender inequality is reflected in Hollywood films. This paper refers to current research and reports about Hollywood and analyzes some specific movies, including Ant-Man and the Wasp, Spider-Man, Iron Man, and Hulk. The paper concludes that the visual culture of gender inequality is mainly reflected in the differences between men and women in the distribution of time, work, and income. Specifically, compared to men, women get fewer chances to play in a film, have lower positions in the film industry, and earn less. Furthermore, it is found that most movies apply the perspective of men and repeat gender stereotypes. Therefore, this paper hopes that more people can pay attention to the interests and situation of women.

Introduction Since the emergence of the first case of coronavirus disease (COVID-19) infection on December 31, 2019, the Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) has spread rapidly throughout China [1]. There is no strong evidence that the coronavirus originated in China, but the epidemic outbreak in China was found to exist in an epidemiological link to the wholesale seafood market in Wuhan, China [2]. By March 26, 2020, the number of confirmed cases in the city of Wuhan had reached 50,006 people, and accounted for 61.48% of all confirmed cases in the country [3]. This made Wuhan a high-risk region for the epidemic in China. Nearly all of the infected patients emerging from other regions had some sort of connection with Wuhan, either by having traveled to Wuhan, or by coming into contact with infected patients from Wuhan [4]. The Chinese government implemented appropriate and timely response measures to contain the outbreak, by quarantining Wuhan City and setting up inspection points in other regions for people with a travel history to Wuhan or its surrounding regions [2]. This method of inspection relies on the public to submit reports voluntarily, but its effectiveness does not appear to be satisfactory. Several cases of withheld information were uncovered subsequent to the onset of an individual's illness or an outbreak of COVID-19 in localized clusters. For example, a man from Jinjiang, Fujian, lied that he was returning from the Philippines when he was in fact traveling home from Wuhan. He carried out regular activities and attended gatherings, which resulted in the home quarantine of over 4000 people. Another individual from Ya'an, Sichuan, became ill after returning from Wuhan. Despite being repeatedly questioned by medical staff whether they had visited Wuhan or its surrounding regions, Hou never admitted the truth. Exposure to this individual resulted in the quarantine of over 100 people, including more than 30 medical staff. Another person in Hebei, who withheld their history died after missing the opportune time for treatment, leading to 77 close contacts being quarantined. Tens of thousands of people have been forced to undergo quarantine and to shoulder the risk of infection, resulting in a devastating outcome and consequences that still cannot be properly estimated. The act of withholding travel history to high-risk regions is posing a difficult challenge for epidemic prevention and control, and has greatly undermined epidemic response measures. Worse still, the epidemic is no longer limited to China. More than a hundred countries including the United States, Italy, and the United Kingdom have reported cases of the infection, and the number of afflicted countries and cases continues to rise. A study on the public's willingness to report on travel history to high-risk epidemic regions will provide support for countries that are facing or will possibly face the epidemic. Therefore, this study's goals were as follows: 1) to elucidate and analyze the public's willingness and beliefs associated with reporting travel history to highrisk epidemic regions; 2) to provide effective suggestions and measures for encouraging the public to report their travel history in response to the World Health Organization's (WHO) advocacy for non-pharmaceutical interventions. --- Methods --- Questionnaire design After conducting the necessary literature research and receiving expert advice, we developed a selfadministered questionnaire (Additional file 1). To confirm the effectiveness of the questionnaire, we invited seven experts from fields such as health emergency, epidemiology, and public psychology, as well as three health administration workers who are currently in charge of epidemic prevention and control, to engage in two rounds of reviews of the questionnaire. In addition, we formed a team of 30 people to carry out a two-week reliability re-test (Cronbach's <unk> = 0.75). The team was recruited voluntarily from Harbin, Heilongjiang Province, China, based on the principle of convenience. They responded to the entire questionnaire and again two weeks later. The data obtained was not used for the final analysis. This questionnaire included the following parts: (1) Socio-demographic information; (2) Willingness and beliefs associated with reporting on travel history to high-risk epidemic regions; (3) Cues that promote reporting on travel history to high-risk epidemic regions. The dependent variable in this study was the respondents' self-rated willingness to report on travel history to high-risk regions, which was evaluated by the item: 'Supposing you have a travel history to Wuhan, would you report this to the designated department, facility, or personnel?' using a five-point Likert scale ranging from 1 to 5 (definitely would not report, would not report, unsure, would report, definitely would report). During analysis, for the purpose of comparing respondents who had a positive willingness toward reporting with those who did not, we sorted those who selected 4 and 5 into the category'report', and those who selected 1, 2, and 3 into the category 'not report'. In the results of this study, the number of respondents who chose each item of the dependent variable was as follows: 1-5(0.37%), 2-13(0.97%), 3-73(5.43%), 4-107(7.96%), and 5-1146(85.27%). Socio-demographic data were collected, including the respondents' gender, age, education level, marital status, place of residence, living arrangement, and religious belief. Among them, the religious belief data was obtained by the item: 'Do you have a religion', using Yes or NO as the answer option. In addition, we used the MacArthur Scale of Subjective Social Status to evaluate the respondents' social status. This scale yields comprehensive ratings for the level of income, occupation, and education [5]. The scale contains two items; respondents rated their perceived social status within their country and community from 1 (the bottom) to 10 (the top), with the total points falling within a range of between 2 and 20. Respondents were divided into two groups according to the median: <unk> 11 was lower class; > 11 was upper class. The respondents' beliefs associated with reporting on travel history was assessed through nine prompts, to which they indicated their degree of agreement by selecting a rating between 1 (strongly disagree) and 5 (strongly agree). The respondents' overall perceptions of travel history and the COVID-19 infection were evaluated by the following prompt: 'Travel history has an impact on contracting COVID-19'. The respondents' perceptions of the benefits of reporting were assessed by two prompts: 'Reporting can confirm whether I contracted COVID-19 earlier' and 'Reporting can help discover earlier potential patients infected by coming in contact with me'. The respondents' perceptions of obstacles to reporting were assessed by five prompts, including 'Reporting makes me feel stigma'. In addition, another prompt judged respondents' perception of the consequences of withholding travel history: 'Withholding travel history to high-risk epidemic regions will result in legal liability.' During analysis, all responses to the prompts were converted into binary variables: respondents who rated 4 and 5 were sorted into the category 'agree', whereas respondents who rated 1, 2, and 3 were sorted into the category 'disagree'. In the questionnaire, we also designed an item to investigate which cues would promote respondents to report. The item allowed multiple selections, including cues such as 'persuasion by family or friends', 'community public speaking (via loudspeaker, bulletin board, etc.)', and 'people around me showing potential symptoms of COVID-19'. --- Sample and data collection The cross-sectional study was carried out in the format of an anonymous web-based questionnaire. The questionnaire survey was conducted among residents of nine provinces with varying epidemic levels, selected based on the cumulative data of confirmed COVID-19 cases publicized by the National Health Commission of China, the health commissions of various provinces, cities, and districts, and the governments of each province, city, and district, as of February 11, 2020. Provinces where the epidemic was more severe were Henan, Zhejiang, and Guangdong (cumulative confirmed cases, <unk> 1000 people); provinces where the epidemic was moderately severe were Heilongjiang, Anhui, and Yunnan (cumulative confirmed cases, 100-999 people); provinces where the epidemic was less severe were Jilin, Inner Mongolia, and Gansu (cumulative confirmed cases, <unk> 99 people). Based on the ranked cumulative numbers of confirmed cases of the cities within each province, a city where the epidemic was severe (cumulative number of confirmed cases exceeding the mean value of all cities in the province) and a city where the epidemic was relatively mild (cumulative number of confirmed cases being lower than the mean value of all cities in the province) were drawn from each province. Using the method of convenience sampling, questionnaire administrators were recruited from the sample cities, who were undergraduates or postgraduates of Harbin Medical University and postponed their return to school due to COVID-19. In addition to recruiting two administrators in the three sample cities, all the other sample cities recruited an administrator, for a total of 21 administrators. They sent a link to the questionnaire to the residents in their communities through social networking software such as WeChat (WeChat is the largest social media in China similar to Facebook and Twitter, and the number of monthly active accounts has reached to 1.165 billion [6]), alongside an explanation of the study's intention and anonymity. Those who received the link could voluntarily decide whether to respond, and we required at least 100 people in each sample city to receive the link. The data were collected between February 12 and 19, 2020. Finally, a total of 1965 people received the link. In addition, according to the questionnaire network platform, 1823 people clicked on the link and 1481 people responded to the questionnaire (potential response rate = 75.37%). The researchers carefully reviewed the questionnaires and eliminated those submitted by participants under the age of 18 years and participants who selected the same answers to every question. A total of 1344 valid questionnaires were finally collected (effective response rate = 90.75%). --- Data analysis Descriptive analysis was used to show the sociodemographic characteristics of the respondents, their willingness and beliefs associated with reporting on travel history to high-risk regions, and cues that could potentially promote reports. Chi-squared test and Fisher's exact test were used to analyze the correlation between the participants' willingness and beliefs to report. SPSS 22.0 (IBM, 2010) was used to conduct the analysis. Statistical significance was set at P <unk> 0.05. --- Results --- Characteristics of the respondents More than half of the respondents (60.19%) were women. The ages of the respondents were concentrated between 21 and 30 years old (37.80%). Among the respondents, 61.76% had an education level of undergraduate or above, and 56.03% were single. More than half of the respondents (74.48%) normally resided in cities, and a majority lived with family or friends (89.73%). Almost all of the respondents (90.10%) were nonreligious. The majority of the respondents (74.33%) perceived themselves to be lower class in social status. 6.77% of the respondents chose not to report when they had a travel history to Wuhan, and 93.23% of the respondents chose to report. Gender, age, educational level, marital status, place of residence, and living arrangement were significantly correlated to willingness to report on travel history (Table 1). 9.35% of men chose not to report, which was higher than that of women (5.07%). Respondents under the age of 20 had the highest non-reporting rate (12.79%), while those aged 41 to 50 had the lowest rate (1.92%). In terms of education level, the proportion of people who chose not to report was the highest among college graduates (9.33%), but the lowest among those with master degree and above (3.01%). Single respondents had a higher percentage of non-reporter (9.56%) than married and widowed/separated/divorced. In addition, about one-tenth of the respondents who live in countryside (10.20%) chose not to report, which was higher than that of those who live in city (5.59%). The proportion of people who chose not to report among the respondents living with others (28.57%) was higher than living alone (7.69%) and living with family or friends (6.30%). --- Beliefs associated with reporting on travel history to high-risk epidemic regions Overall, 83.04% of respondents believed that having a travel history to high-risk epidemic regions would increase chances of contracting COVID-19. Over four out of five respondents agreed that reporting on travel history can help to confirm whether they had contracted the infection sooner (83.26%), and discover potential patients who were infected by coming in contact with them earlier (87.13%). About half (57.44%) of the respondents believed that reporting would make them feel stigmatized. Other percentages of respondents believed reporting would result in high follow-up checks or treatment expenses (13.84%), did not know how to report (17.34%), and believed reporting was very inconvenient (8.26%). In addition, 5.88% of the respondents expressed fear of potential quarantine after reporting on their travel history. Overall, nearly one-fourth of the respondents (23.96%) did not believe withholding information on travel history to high-risk regions would result in legal liability. Table 2 displays the factors that are significantly correlated with a willingness to report on travel history to high-risk epidemic regions. --- Cues that promote reports of travel history to high-risk epidemic regions Figure 1 displays the cues that respondents perceived may encourage their reporting of travel history to highrisk epidemic regions. 'Hearing about cases where withholders of travel history were punished by law' was the most selected option (75.97%), followed by 'people around me showing potential symptoms of COVID-19' (73.21%), and 'advocacy on television, the internet, and other media' (69.57%). --- Discussion Our study evaluated the current willingness and beliefs associated with reporting on travel history to high-risk epidemic regions among the Chinese public, as well as the correlation between their willingness and beliefs. A proportion of the public (6.77%) was found to be more inclined to withhold their history, although the majority agreed to report. Nonetheless, considering that any withheld history case can result in unpredictable outcomes, this result already deserves great attention [7][8][9]. After the emergence of frequent incidences of withheld history, Chinese authorities implemented more severe measures, including restricting population movement in cities and communities, rewarding those who provide clues to withholders, and instigating systematic blanket searches in communities [10,11]. However, in all of these actions the public plays a relatively passive role. Therefore, research into the association between the public's willingness and beliefs to report on travel history is essential for promoting the public's voluntary reporting. This study discovered several factors associated with the willingness to report on travel history to high-risk epidemic regions. First, is the belief of benefiting from reporting. Apart from being able to test whether one is infected as soon as possible, a more valuable benefit of timely reporting on travel history, is that it helps to identify other potentially infected patients. The department responsible for epidemic prevention and control would be able to investigate the public transportation that reporters took, the recent events in which they participated, and the number of people that they had been in contact with. This response would allow potentially infected patients to be examined and prevent them from further spreading the virus [12]. The results of the study reflect that the proportion of the public that recognized these benefits demonstrated a greater willingness to report (P <unk> 0.05). However, our research cannot determine whether the reporting behavior of the public could be attributed to egoistic behavior or altruistic behavior. Emphasizing all the benefits of reporting in public advocacy may be the better option for promoting a willingness to report. Furthermore, prospect theory in behavioral economics asserts that people have a mentality of aversion toward loss; their level of sensitivity toward losses and gains are unequal, and the pain they feel in the face of loss is far greater than the joy of gaining [13]. Thus, instead of promoting 'the benefits of reporting', the government can promote 'the risks of not reporting' in public advocacy to activate the loss-averse mentality in the public and encourage action. Second, considering the perceived obstacles to reporting, the feelings of stigmatization that reporting brings and the fear of potential quarantine are factors that influence the willingness to report. The proportion of the public that believed reporting would cause stigmatization and feared the quarantine that they may face after reporting expressed a lower level of willingness to report (P <unk> 0.05). The stigmatization and discrimination of potential patients are phenomena that were present since the early stage of epidemic outbreak [14]. Reporting on travel history might cause the reporter to be labeled a potential patient, which lowers their willingness to report [15]. The exaggerated and arbitrary description of risk associated with having a travel history to Wuhan provided by China's official statements and media reports may also worsen the public's isolation of people with such an experience, despite it not being the original intention. Thus, social support and social policies should be available for potential patients, including the reporter of travel history, which would alleviate the social stigma attached to the travel history reporter and ease the discriminative atmosphere. For example, the objective, open and continuous epidemic information publicity and communication with the public based on reliable medical investigation and experience, as well as non-discriminatory treatment to the travel history reporter by epidemic prevention and control departments and staff may improve the situation. The government and media should also draw connections between regions and the infection with discretion to prevent the situation from worsening into regional discrimination. In addition, this study and past studies concur that the public fears restrictive epidemic control measures, such as quarantining [16][17][18]. On some level, this fear is an extension of the underlying anxiety surrounding an infectious disease with an unknown cause and possible fatal outcome [19]. Members of the public with a travel history would generally be subjected to the prevention and control process of quarantine after reporting, until the incubation period passes without incident. Fear of this has also become one of the factors that impact on the willingness to report. Therefore, effective health education strategies and public risk communication are necessary to allay such fear. Third, this study showed that the members of the public that agreed that withholding travel history to highrisk epidemic regions would result in legal liability demonstrated a greater willingness to report (P <unk> 0.05). To combat COVID-19, China implemented Class A (the highest level) epidemic prevention and control measures. Each province and city successively implemented firstlevel responses to the major public health emergency. During this time, all units and individuals were required to comply with epidemic-related measures enforced by disease prevention and control facilities and medical facilities, and truthfully provide relevant information. Deliberately withholding travel history to high-risk regions would result in a conviction for the crime of 'endangering public security by dangerous means', based on the severity of the outcome [20,21]. Shanghai even introduced China's first COVID-19 legislation: those who evaded quarantine by withholding their travel history would be diligently pursued for applicable legal liability based on the law [22]. Understanding that their action may result in serious consequences enforced by the legal system may suppress the public's desire to withhold information. Thus, based on the existent legal framework on withholding travel history, the government should further expand public campaigns to explain legal obligations to report on travel history and emphasize the consequences of withholding information. In investigating the cues that may encourage the public's reporting of travel history, we found various types of effective cues, with the greatest number of respondents admitting to the effectiveness of hearing about cases where withholders of travel history were punished by law. This outcome could enhance the public's acknowledgment of the illegality of withholding information, as well as prompt public consideration of law enforcement dynamics. However, this study was unable to provide a conclusion regarding the public's sensitivity toward different penalties and frequency of hearing cases of people being prosecuted, as well as the threshold that affects their ultimate actions. This aspect would require further exploration in future. Seeing others showing potential symptoms of COVID-19 also encourages individuals to report their travel history, as withholding information increases the concern of negatively impacting others. Epidemiological analysis from the Chinese Center for Disease Control and Prevention shows that there are more than 1000 cluster COVID-19 cases in the country, 83% of them are family-based, or occurring in common gathering places, such as medical institutions, schools, shopping malls, and factories [23]. Another cue was the public speaking by communities. Thus far, grassroots communities, including rural communities, have played a unique role in China's epidemic prevention and control as well as in social governance [24]. The power of grassroots leaders made it possible to manage residents of each jurisdiction systematically to monitor the epidemic's advancement, promote epidemic information, and educate residents on healthy behavior. The community is the public's direct recipient, processer, and promoter of travel history reports to high-risk epidemic regions, and as such, most members of the public report to their communities first if they have the willingness to report. China's leader Xi Jinping has commented as follows: 'The community is the frontline in epidemic prevention and control, and is the most effective line of defense in blocking external infectious sources and containing internal spread' [25]. Under the current situation, where the epidemic is spreading globally, each country can consider mobilizing their communities for prevention and control as China has done-to place the power of prevention and control in the community. Nevertheless, this study has the following limitations. First, the public's willingness and beliefs associated with reporting on travel history to high-risk epidemic regions may change with the course of the epidemic. A crosssectional study is limited in its usefulness in capturing this type of dynamic change. Future research may consider a longitudinal design. Second, this study only analyzed the effects that socio-demographic characteristics and beliefs associated with reporting travel history had on the willingness to report. There may be other factors affecting the public's willingness to report. Third, considering the cost and convenience of conducting research, this study only selectively investigated a portion of provinces and cities based on the severity of the epidemic outbreak. For more universal results, a study on a larger scale may be necessary. Fourth, this study only measures and analyzes the respondent's subjective social status. Future research may need to consider the impact of objective social economic status, which will contribute to more precise crowd intervention. --- Conclusions Our study on the Chinese public's willingness to report on travel history to high-risk epidemic regions showed an inclination to withhold travel history, albeit in a small proportion of the public. Considering that any incident of withheld history can result in unpredictable outcomes, this finding demands attention. Our study also indicated that the belief of the benefits of reporting, obstacles to reporting, and legal consequences of withholding travel history would affect willingness to report. Therefore, on the one hand, suitable public risk communication and public advocacy strategies should be carried out to reinforce the understanding that travel history reporting allows the individual and others to receive infection screening and treatment earlier, and alleviate the fear of potentially being quarantined after reporting. On the other hand, social support and social policies should be made available to potential patients, including the reporters of travel history, to eliminate the feeling of stigma that may arise from reporting travel history. The government and media should draw connections between regions and the infection with great discretion, to alleviate the phenomena of stigmatization and discrimination, to which potential patients are subjected. Finally, reinforcing the legal accountability of withholding travel history and strengthening systematic community monitoring are measures currently taken by China to promote reports on travel history to high-risk regions. These actions provide internationally relevant experiences to countries that are currently facing the spread of the epidemic and countries at risk of its potential spread in response to the WHO's advocacy for nonpharmaceutical interventions. --- Availability of data and materials The datasets generated and analyzed during the current study are not publicly available due to privacy restrictions. Respondents were informed during the consent process that the data they provide would be available only to the Harbin Medical University. --- Supplementary information Supplementary information accompanies this paper at https://doi.org/10. 1186/s12889-020-09282-4. Additional file 1. Questionnaire. Questionnaire developed for this research. Abbreviations COVID-19: Coronavirus Disease 2019; SARS-CoV-2: Severe Acute Respiratory Syndrome Coronavirus 2; WHO: World Health Organization Authors' contributions MJ, SS and QW contributed to the conception and design of the study. LW, ZS and YW contributed to the data collection. LW, ZS and GZ contributed to literature search and data quality control. HJ and SZ did the statistical analysis and drafted the original manuscript. YL revised the manuscript for important intellectual content. YW, CL and LW prepared the manuscript and supplementary material. All authors contributed to data interpretation and rewriting the paper. All authors have read and approved the manuscript. LW, ZS, YW contributed equally to this work. MJ, SS, QW are corresponding author. --- Ethics approval and consent to participate This study was reviewed and approved by the Research Ethics Committee of Harbin Medical University. Informed consent was obtained from all participants, which was in the form of online questionnaires and obtained through online responses of the participants. The informed consent form described the purpose and method of data collection and ensured that the data would be kept confidential. The Research Ethics Committee of Harbin Medical University approved the procedure for obtaining informed consent. --- Consent for publication Not Applicable. --- Competing interests None to declare. --- Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Background: The Coronavirus Disease 2019 (COVID-19) that first occurred in Wuhan, China, is currently spreading throughout China. The majority of infected patients either traveled to Wuhan or came into contact with an infected person from Wuhan. Investigating members of the public with a travel history to Wuhan became the primary focus of the Chinese government's epidemic prevention and control measures, but several instances of withheld histories were uncovered as localized clusters of infections broke out. This study investigated the public's willingness and beliefs associated with reporting travel history to high-risk epidemic regions, to provide effective suggestions and measures for encouraging travel reporting. Methods: A cross-sectional study was conducted online between February 12 and 19, 2020. Descriptive analysis, chi-squared test, and Fisher's exact test were used to identify socio-demographic factors and beliefs associated with reporting, as well as their impact on the willingness to report on travel history to high-risk epidemic regions. Results: Of the 1344 respondents, 91 (6.77%) expressed an inclination to deliberately withhold travel history. Those who understood the benefits of reporting and the legal consequences for deliberately withholding information, showed greater willingness to report their history (P < 0.05); conversely, those who believed reporting would stigmatize them and feared being quarantined after reporting showed less willingness to report (P < 0.05).

Background Globally, intimate partner sexual violence (IPSV), a form of intimate partner violence (IPV) has been declared as a public health issue considering its long-term physical, biological, psychological and neurological consequences on victims in society [1][2][3]. IPSV is defined as sexual acts committed or attempted by an intimate partner without the consent of the victim or against someone unable to give consent [4]. It involves rape, unwanted pressured penetration, intentional sexual touching, and non-contact acts of sexual nature [4]. IPSV is common among women globally although, some men could also be victims; an estimated 1 out of 10 men experience IPSV while 1 out of 4 women experience IPSV in their lifetime [4]. Also, a recent literature review reported that the prevalence of IPSV among women range from 9.4% -15.2% in several countries in the Americas [5]. Several factors have been indicated to contribute to high levels of IPSV among women. For instance, it is established that cohabiting partners report higher rate of sexual violence compared with married women [5]. Furthermore, socio-cultural beliefs of women being recognized as the property of men, and other sociodemographic characteristics such as poverty, financial insecurity, lower level of education, and smoking among others have also been documented to be associated with IPSV [1,2,[4][5][6]. Cigarette smoking is indicated to be prevalent among women who are victims of IPSV [4,7]. Several studies have established sexual violence to be one of the significant predictors of cigarette smoking among women [2,7,8]. Smoking is assumed to ease the stress women undergo in abusive relationships [1,2,4]. A longitudinal study has found that women who smoke are likely to report sexual violence compared to women who do not smoke [4]. Papua New Guinea (PNG) is known to have high prevalence of sexual violence against women [9,10]. About 41% of men admitted raping a woman while one third of women have suffered sexual violence [10]. Populationbased studies in PNG have demonstrated high prevalence of IPSV and IPV in general [9,10], however they have scarcely considered the influence of IPSV on cigarette smoking among women in union. Thus, irrespective of the high prevalence of intimate partner violence in PNG studies linking IPSV and cigarette smoking among women in union are limited. This study extends the present literature by examining the association between IPSV and current cigarette smoking among women in union in PNG. This is an important public health issue for many reasons. While IPSV has been associated with an elevated risk for physical and mental health problems, cigarette smoking increases the risk of adverse health outcomes. The aim of the present study is to examine the prevalence of IPSV and its association with cigarette smoking among women in union in PNG by controlling for demographic, social and economic factors. Thus, the study seeks to test whether IPSV is significantly associated with cigarette smoking and the role of socio-demographic and economic factors in the association. This study focused on sexual violence because it is one of the most prevalent forms of gender-based violence in PNG [10]. Furthermore, some studies have suggested that while sexual violence may include physical violence, factors associated with each violence form could differ and as a result it is imperative to separately focus on each form of IPV [1,11,12]. --- Methods --- Data source, sampling technique and sample size This cross-sectional study used data from the Papua New Guinea Demography and Health Survey (PNGDHS) conducted from October 2016 to December 2018. This is the first demographic and survey conducted in PNG. The PNGDHS aimed to generate comprehensive data on demographic, maternal and reproductive issues such as fertility, family planning awareness and practices, breastfeeding practices, health behaviors, immunizations, domestic and intimate partner violence, among others. Through the Demographic and Health Survey (DHS) programme, technical support for the execution of the survey was provided by Inner City Fund (ICF), with the financial support of Papua New Guinea Government, Australian Government Department of Foreign Affairs and Trade, the United Nations Population Fund (UNFPA) and UNICEF [13]. The sample for the 2016-18 PNGDHS was nationally representative and covered the entire population that lived in private dwelling units in the country. The survey used the list of census units (CUs) from the 2011 Papua New Guinea National Population and Housing Census as the sampling frame and adopted a probability-based sampling approach. Specifically, a two-stage stratified cluster sampling procedure was followed. The methodology and selection procedure details have been reported in the PNGDHS final report. In summary, each province in the country was stratified into urban and rural areas, yielding 43 sampling strata, except the National Capital District, which has no rural areas. The division paid particular attention to urban-rural variations. Samples of census units were selected independently in each stratum in two stages. In the first stage, sorting the sampling frame within each sampling stratum to achieve implicit stratification and proportional allocation using a probability proportionalto-size selection was done. In the second stage of sampling, a fixed number of 24 households per cluster were selected with an equal probability systematic selection from the newly created household listing, resulting in a total sample size of approximately 19,200 households. To prevent bias, no replacements and no changes of the pre-selected households were allowed in the implementing stages. In cases where a census unit had fewer than 24 households, all households were included in the sample. A total of 17,505 households were selected for the sample, of which 16,754 were occupied. Of the occupied households, 16,021 were successfully interviewed, yielding a response rate of 96%. In the interviewed households, 18,175 women age 15-49 were identified for individual interviews; interviews were completed with 15,198 women, yielding a response rate of 84%. In this present study, the sample comprised 9,943 women who were in union (either married or cohabiting) during the survey. Thus, our analysis used data only on women who were in union during the survey. --- Data availability and ethical consideration The data have been archived in the public repository of DHS. The access to the data requires registration which is granted specifically for legitimate research purposes. Consent forms were administered at household and individual levels, in accordance with the Human Subject Protection. The dataset can be accessed at https:// dhspr ogram. com/ data/ datas et/ Papua-New-Guinea_ Stand ard-DHS_ 2017. cfm? flag = 0. --- Main outcome and predictor variables Current cigarette smoking was the outcome variable in this study. This was measured as having smoked cigarette in the last 24 h before the survey. Women in union were asked the question: Smoked cigarette in the last 24 h? Women in union current smoking status were classified as "No" (0): no current smoking in the last 24 h or "Yes" (1): smoking in the last 24 h. The key explanatory variable in this study was IPSV. This variable was derived from the optional domestic violence module, where questions are based on a modified version of the conflict tactics scale [14,15]. Questions asked are in relation to physical, sexual or emotional violence experiences. In this study, the focus was on the experience of sexual violence. Three standard items including whether the partner ever physically forced the respondent into unwanted sex; whether the partner ever forced respondent into other unwanted sexual acts and; whether the respondent has been physically forced to perform sexual acts she did not want to were used to generate the experience of intimate partner sexual violence. For each of these items, the responses were 'never' 'often''sometimes' and 'yes, but not in the last 12 months. However, for our analysis purpose, we created a dichotomous variable to represent whether a respondent had experienced sexual violence in the past 12 months. This was done by recoding the following responses: 'never' and 'yes, but not in the last 12 months' as "No" (0) and 'yes', 'often' and'sometimes' as "Yes" (1). --- Confounding variables Theoretically and empirically relevant demographic and socioeconomic variables were included as confounders. In all, we included twenty socioeconomic and demographic variables to adjust for in the modelling. These variables included age, region, religion, place of residence, highest educational level, literacy, marital status, residing with a partner, number of partner's wives, partner's age, partner's education, health insurance cover, internet access, mobile phone ownership, watch television, listen to radio, read newspaper/magazine, occupation and wealth index. The selection of these variables was informed by their statistically significant associations with sexual violence and cigarette smoking in previous studies [1,2,4,16,17]. (See Table 1 for the details on the coding of the covariates). --- Statistical analysis Before the analysis, all missing data were removed. Both descriptive (frequencies, percentages, mean and standard deviation) and inferential (chi-square and modified Poisson regression) analytical frameworks embedded in STATA software version 13.0 (StataCorp LP, College Station, Texas, USA) were used. The statistical analysis followed some essential steps. We performed descriptive statistics such as frequencies and percentages to describe the sample. The Pearson's Chi-square test was done to examine the differences in smoking cigarette by socio-demographic characteristics and IPSV. A modified Poisson regression, adjusting for demographic, social and economic variables, was also performed to model the association between IPSV and cigarette smoking, to estimate the relative risk (RR) of cigarette smoking [18,19]. The study used the modified Poisson regression that incorporates the robust error variance procedure to optimize the accuracy of the estimates [18], as direct estimates of relative risk produce from modified Poisson regression modelling may be a preferred method for estimating population-level risk [19]. We fitted four regression models. Model 1 included dependent and independent --- Results --- Background characteristics of the participants The mean age of the participants was 32.68 <unk> 0.08 years. About 20% of the participants were aged between 25-29 years, 28.6% were from the Highlands Region, 75.5% resided in the rural area, 49.7% had primary level of education, and 99% were Christians. Again, 83% of the participants were married, 86.6% were currently living together with their partner, and 33.8% had between 1-2 kids and 62.1% were not working (see Table 1). --- Distribution of cigarette smoking across exposure to IPSV Table 1 shows the distribution of cigarette smoking across IPSV. The results showed significant disparities in cigarette smoking and IPSV at p <unk> 0.001. Specifically, 25.9% of women were exposed to IPSV while 26.8% of women who were exposed to IPSV smoked cigarette. --- Association between exposure to IPSV and current cigarette smoking among women in union in PNG Table 2 shows the results of the association between IPSV and current cigarette smoking among women in union in PNG. We found in all the four models that women who were exposed to IPSV were more likely to smoke cigarette. In Model 1, the study revealed that those who had experienced IPSV have a higher risk of smoking cigarette compared to those who had not experienced IPSV. In Model II, demographic variables were added to the variable in Model I, the study revealed that those who had experienced IPSV, those from the Highlands region, those residing in urban areas, those with no religion and those whose partners have 3 or more wives have a higher risk of smoking cigarette compared with their counterparts. Also, participants aged 45-49 years and those who are able to read a whole sentence have a lower risk of smoking cigarette compared to their counterparts. In Model III, when other social variables were added to all variable in Model II, the study revealed that those who had experienced IPSV, those residing in the Momase region, those living in urban areas, those with no religion and those whose partners have 3 or more wives have a higher risk of smoking cigarette compared with their counterparts. Again, participants aged 45-49 years and those who are able to read a whole sentence have a lower risk of smoking cigarette compared with their counterparts. In the final Model (Model IV), when economic variables were added to all variables in Model III, the study revealed that those who had experienced IPSV, those from the Momase region, those with no religion, those residing in urban area and those whose partners have 3 or more wives have a higher risk of smoking cigarette compared with their counterparts. Further, participants aged 45-49 years, those who rated their wealth index as 'Middle', those who were Clerical officers and those who are able to read a whole sentence have a lower risk of smoking cigarette (see Table 2). The key take home message from the result is that, after adjusting for diverse demographic, social and economic characteristics at different levels of the Model(s), IPSV remained significantly associated with cigarette smoking among the participants as the direction of the association remained. --- Discussion This study examined the association between IPSV and cigarette smoking among women in intimate unions in PNG. The present study adds to the current literature on IPV and cigarette smoking. The study found that women who had experienced IPSV had a greater odds of smoking cigarette. Research on the association between IPSV and health risk behaviors especially cigarette smoking are well documented [20] and consistent with the findings of this study. For instance, in a cross-sectional study that examined the association between intimate partner violence experience and cigarette smoking, Zhang and colleagues [20] found women experiencing intimate partner violence were more likely to smoke cigarette. In the US, a meta-analysis of 31 peer-reviewed studies to evaluate the relationship between intimate partner violence victimization and cigarette smoking revealed victims of intimate partner violence are at greater risk of smoking with a composite side effect of d 1<unk>4 0.41 [21]. Thus, across the collected and analyzed literature, victims of IPV are significantly more likely to engage in smoking behavior than non-victims. Sexual violence has been found to be more closely linked to activities such as cigarette smoking than other types of intimate partner abuses. Sexual IPV victimization exhibited the most pronounced connections with cigarette smoking, according to a study done to investigate the health status and health risk behaviors related with experiences of psychological, physical, or sexual IPV among women getting care at a medical center [22]. Victims of IPV are more likely to smoke cigarette than offenders of IPV [21]. It is important to also note that because of the circumstances surrounding intimate sexual assault, it has great impact on victims' psyches, and the psychological repercussions last longer [21]. Studies have shown that victims of IPV experience mental health problems such as depression, generalized anxiety disorder, suicide risk, and post-traumatic stress disorders [23][24][25], loneliness, sleeping problem and short sleep [26] leading to variety of drug use disorders [24]. These psychological outcomes associated with IPV are indicated to profound in women than in men [25]. Victims of IPV in most circumstances resort to health risk behaviors such as cigarette smoking as coping techniques against the stresses experienced [21]. Women who are victims of IPV also find consolation in smoking when they are unable to report the abuse to family members or law enforcement authorities [9,11]. Furthermore, victims of IPV are sometimes forced by an abusive partner to use drug substances including smoking cigarette [27]. A survey conducted by a national center on domestic violence, trauma and mental health in the US found many victims of IPV are forced or coerced by abusive partners to use substances [28]. Our findings support the assertion that women are more likely to smoke as a psychological coping mechanism when they suffer stress, anger, or despair due to IPV, and the association between stress and cigarette smoking has been reported to be stronger in women than in men [29]. The present study also provides evidence to demonstrate that certain demographic and socioeconomic positions including age, wealth index, occupation, partner's number of wives, region, place of residence, religion and literacy play a role in cigarette smoking among women in union. Evidently, when compared to their counterparts, those from the Highlands region, those who live in urban areas, those with no religious affiliation, and those whose partners have three or more wives, had a greater odds of smoking cigarette. Wilson [30] posits that lower IPV is sometimes attributed with marriage, urban residency, and increasing age. Although all women can be victims of IPV regardless of their age, marital status, level of education, income status, place of residence and country of residence [31], Bhona et al. [32] found that women who have greater educational and socioeconomic levels are less likely to be victims of partner violence. The findings of this study suggest that women between the ages 25-29 years are mostly affected by IPSV and engaged in cigarette smoking as compared with other age group. This supports the assertion that IPSV affects people of all socioeconomic backgrounds, but youth from lower socioeconomic backgrounds are more likely to be exposed and suffer [24]. Moreover, women of lower and higher socioeconomic status, have more cigarette smoking tendencies as compared to those of middle socioeconomic status [33] as suggested by the findings of this study. Additionally, the findings corroborate with a study which revealed that women with no formal education, primary-level, or secondary-level qualification have a larger chance of being smokers than women with a higher education [34]. Perhaps, women with high education are more likely to have accessed information on the negative consequences of smoking. Some strengths associated with the present study need to be remarked. The study utilized a nationally representative data to examine the association between IPSV and cigarette smoking in PNG, thereby increasing the generalizability of its findings. Methodologically this study is also associated with some strengths. The present study uses a relatively new analytical approach by applying the modified Poisson regression that incorporates the robust error variance procedure to establish the association between IPSV and cigarette smoking. The modified Poisson regression approach can be regarded as very reliable in terms of both relative bias and percentage of confidence interval coverage [18]. Also, extensive discussion in much of the literature has reached a consensus that the relative risk is preferred over the odds ratio for most prospective studies with binary outcomes as logistic regression modelling overestimates the odds ratios [18,[35][36][37][38]. In that regard, the use of Poisson regression has been a promising alternative. Of course, there are some limitations that need to be commented. Our study does not explore any causal relationship between IPSV and smoking, as PNGDHS data are cross-sectional. In addition, the present study relied on self-reported data which may be subjected to recall bias. Even accurate selfreported measures may reflect individual differences not associated with health per se. Also, the IPSV variable was collected through an optional domestic violence module and as a result such type of sensitive information could not be reported or misreported by the participants. Moreover, in this study current cigarette smoking was defined as smoked cigarette in the last 24 h before the survey. This could have the potential to exclude women with a non-daily smoking pattern. Despite these limitations, the research presented here is suggestive and represents important progress. It calls attention to an association with scarce empirical examination in part due to limited research especially in the context of PNG. --- Public health and policy implications This study offers a number of policy implications that need to be acknowledged. For the purpose of this study, implications of the study have been grouped into three key areas; 1) public health and practice implications 2) health policy implications 3) research implications. First of all, in relation to the public health and practice implications, gender-based institutions and groups in collaboration with the PNG National Department of Health should organize health education and awareness creation campaign on cigarette smoking and IPSV in PNG. The health education and awareness intervention should primarily center on health, social and economic risks associated with smoking cigarette among women who experience IPSV. Based on the findings of this study, the health education program should target more of women from the Momase region, those residing in urban area and those whose partners have three or more wives since they were having a higher log count on cigarette smoking in PNG. Also, the proposed health campaign should target men and further educate them on the need not to expose women to IPSV because of the dangers associated with it. Again, since our findings showed that cigarette smoking and IPSV are significantly associated, increasing health campaign against IPSV could scale down cigarette smoking. As part of this health campaign, gender-based institutions and groups in collaboration with PNG health institutions should have a panel discussion with both men and women on why men expose women to IPSV. To the best of our knowledge, this form of discussion on experience of IPSV among women would help provide a framework to guide public health education on IPSV and its association with cigarette smoking. Concerning the health policy implications, we propose that the development/formulation of health policy that aims to reduce cigarette smoking among women who experience IPSV should include other significant demographic, social and economic variables such as region of residence, place of residence (rural/urban), religion, number of wives of partners, age, wealth index and nature of employment. This is because aside from IPSV, the above significant factors play a major role in cigarette smoking among women in PNG. Lastly, in terms of the research implications, since this study did not look at the following important research areas due to the nature of the dataset employed in this study, future research should investigate the following areas; 1) perpetuators of IPSV and associated factors; 2) knowledge of women experiencing IPSV on the health, social and economic risks associated with cigarette smoking; 3) the moderating role of selfrated health in the association between experience of IPSV and cigarette smoking among women; 4) enablers/ facilitators of cigarette smoking among women experiencing IPSV. Such studies can provide a comprehensive understanding of IPSV, cigarette smoking and the association between these experiences capable of influencing policies and interventions. --- Conclusion The rates of IPSV and cigarette smoking among women in PNG in the current study were relatively high. IPSV is shown to be a significant predictor of cigarette smoking among women in union in PNG. Understanding this association suggests that policymakers and relevant authorities in PNG can act to address IPSV to reduce cigarette smoking. Counseling to help victims of IPSV cope with stresses, awareness creation, service provision and program design on intimate partner violence are urgently required to minimize cigarette smoking among women in union in PNG. --- Availability of data and materials The data that support the findings of this study are available from the DHS. However, restrictions apply to the availability of the data, which were used under license for the current study; thus, the data are not publicly available. However, they can be made available from the authors upon reasonable request with the permission of DHS programs. --- Supplementary Information The online version contains supplementary material available at https:// doi. org/ 10. 1186/ s12889-022-13003-4. --- Additional file 1. Authors' contributions BY-AA performed the conception, the design of the work, the acquisition and the analysis. EBA, PP and WA-D performed the design of the work and the creation of tables. DOO, DV and PP performed the design and drafted the work. All authors reviewed and edited the final version of the manuscript. The author(s) read and approved the final manuscript. --- Funding The current research received no specific grant from any funding agency, commercial or not-for-profit source. No other entity besides the authors had a role in the design, analysis or writing of the current article. --- Declarations Ethics approval and consent to participate Ethical approval was not required for this study since the data used for this study are secondary data. Necessary permissions and survey data were obtained from the DHS programs. The DHS data upheld ethical standards in the research process. Consent for publication NA. --- Competing interests Authors declare that they have no competing interests. • fast, convenient online submission • thorough peer review by experienced researchers in your field --- • rapid publication on acceptance • support for research data, including large and complex data types • gold Open Access which fosters wider collaboration and increased citations maximum visibility for your research: over 100M website views per year --- • At BMC, research is always in progress. --- Learn more biomedcentral.com/submissions Ready to submit your research Ready to submit your research? Choose BMC and benefit from:? Choose BMC and benefit from: --- Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Background: Intimate partner sexual violence (IPSV) is a prevalent public health problem affecting millions of people each year globally, particularly in developing countries like Papua New Guinea (PNG). Although over two-thirds of women in PNG are estimated to experience some form of sexual violence in their lifetime, empirical evidence is limited on the association between IPSV and cigarette smoking. Thus, the present study aims to examine the prevalence of IPSV and its association with cigarette smoking among women in union in PNG.This cross-sectional study used data from the first demographic and health survey of PNG conducted between 2016 and 2018. A total of 9,943 women aged 15-49 years in intimate unions were included in this study. We estimated the relative risk of smoking cigarette using modified Poisson regression models with a robust variance and 95% confidence intervals.The rates of IPSV and current cigarette smoking were 25.9% and 26.8%, respectively. The modified Poisson regression results showed that IPSV was significantly associated with an elevated risk for cigarette smoking. Women with IPSV history were more likely to smoke cigarette relative to their counterparts with no IPSV history (RR: 1.33, 95% CI: 1.18-1.50) in the absence of covariates. After controlling for demographic, social and economic factors, the association between IPSV and cigarette smoking remained statistically significant (RR: 1.24, 95% CI: 1.08-1.42).The rates of IPSV and cigarette smoking among women in union in PNG in the current study were relatively high. Irrespective of diverse demographic, social and economic factors, IPSV was still significantly associated with cigarette smoking among women in union in PNG. The findings presented call the attention of policy-makers and relevant authorities in PNG to an important association that needs to be addressed. Counseling, awareness creation, service provision and program design on IPSV are urgently required to minimize cigarette smoking and IPSV among women in union in PNG.

Introduction Sexual violence (SV) is defined by the World Health Organization (WHO) as any unwanted sexual act directed against a person, regardless of the aggressor's relationship with the victim and the context in which it occurs [1]. The forms of SV may include sexual assault, unwanted sexual advances, sexual coercion or sexual abuse. Thus, the term encompasses a wide variety of circumstances and configurations. For example, it includes rape within marriage and by strangers, as well as abuse of minors and of persons who are mentally or physically incapacitated. Unwanted sexual advances or sexual harassment, refusal to use protection against pregnancy or sexually transmitted diseases (STDs), as well as many others, are also included in their definition of SV. The main reasons for the study of SV are the physical and/or psychological consequences suffered by the victims. Other than the possible injuries produced by any violent experience, unwanted pregnancy and the transmission of STDs, the appearance of phobias, depressive and anxious symptomatology, impoverished self-esteem, sexual dysfunctions, social difficulties and post-traumatic stress disorder are the most prominent consequences for victims of SV [2]. For all these reasons, SV is considered a social emergency that can have serious health consequences. Historically, SV has been conceptualized as a type of violence carried out by a man against a woman, based on a conception of power that, to a large extent, characterized the second feminist wave. This perspective argues that men use violence against women as an extension of the patriarchal structure and as an expression of control and power over them [3], focusing on the intersection of class and gender systems of oppression [4]. This theory received several criticisms. One of the most notable was from black feminism, which added the intersection of the racial/ethnic axis to take into account the particularities experienced by black women in the patriarchal and capitalist system since the traditional conception neglected such experiences [5]. Gradually, axes of oppression were added, whose intersection explained the particular experiences in terms of race/ethnicity, class, gender, age, sexuality, functional diversity, religion, etc. Specifically, intersectionality, as a branch of feminist theory, is concerned with identifying how various systems of power, operating at multiple levels (e.g., structural and interpersonal), combine and interact to confer disproportionate risk on populations with marginalized social identities [6]. Although stated in a very brief and simplified way, as it is not the main focus of this paper, we can see how the interest in the intersectional study of SV, as opposed to the traditional feminist theory, allows us to explain why some men might suffer victimization experiences. Even so, only a small part of the existing research on intimate partner violence and SV to date is grounded in the theoretical framework and methods of intersectionality [7], so the collective of gay and bisexual men (GBM), the subject of this study, has been particularly neglected. In relation to this, most studies based on the feminist framework have neglected men rape, and this maintains and reinforces patriarchal power relations and hegemonic masculinities [8]. In terms of sexual orientation, we could conclude that the study of SV still has a strong heterosexist bias [9]. Additional resistances can be identified that block the study of SV in this group. First, there exist a number of male rape myths and beliefs, which are organized into three main themes: denial myths ("men cannot be raped"), blaming myths ("men are responsible for being raped") and trauma myths ("men who are raped do not really suffer discomfort") [10]. The internalization of these beliefs would result in a decrease in interest in the subject by the scientific community and, consequently, a reduction in the number of studies addressing it. Second, the reluctance on the part of researchers to address unwanted or non-consensual sex in GBM, as well as in any other LGTBIQA+ group, could be related to an attempt to avoid reproducing heterosexist ideas that define sexual minorities as deviant, criminal, predatory, pedophilic, and rapist [11]. In fact, there is an intention on the part of the scientific community to reconceptualize same-gender sexual relationships as healthy, legitimate and non-essentially violent in order to debunk such myths [12], which although necessary, inevitably obscures the violence that might occur within the collective. Third, given the HIV epidemic, there was a great interest and need to focus the study of men who have sex with men on the prevention of HIV transmission. Although this branch of study is interested in exploring SV [13], it tends to prioritize addressing the effects of HIV; an incomplete part of the comprehensive study of SV in the collective. In fact, there is an overwhelming frequency in which certain variables appear in the quantitative literature related to the GBM collective, especially putting the focus on HIV and STDs [14]. The study of these variables is indispensable, but it is also necessary to expand the study of health and well-being among the GBM population to understand many other factors that can be crucial, such as SV. Although interest in the quantitative and qualitative study of sexual violence in GBM has increased in recent decades, it has been mostly in the context of intragender violence or domestic violence in same-sex relationships [15,16], in addition to prison-associated SV [17] and child abuse [18]. In fact, to our knowledge, there is no Spanish study at present that addresses sexual violence in the GBM collective, although there are a few that do discuss intragender violence [19,20]. There are, however, more international studies on sexual violence focused on the GBM population, regardless of the bond between victim and perpetrator. Some of their results report a 35.2% prevalence of non-consensual sex among men who have sex with men [21], 54.0% prevalence of sexual assault [22], 15.5% of physical SV, sexual assault, and stalking in the past 6 months [23], frequencies for sexual assault of 10.1% for cis gay men and 9.5% for cis bisexual men in the past year [24], 39% lifetime contact sexual violence and 29.2% lifetime non-contact sexual violence for bisexual men and 17.7% and 33.3% for gay men [25], prevalence of SV reaching 31.64% in gay men and 29.44% in bisexual men [26]. Although all the studies mentioned above conclude that SV in the GBM population is drastically higher than in heterosexual cissexual men and similar to heterosexual cissexual women, and also that trans people have a higher risk of victimization, the figures differ greatly among them due to a lack of clear delimitation in the concept of sexual violence (e.g., sexual coercion, sexual aggression, unwanted relationships, intimate partner violence, etc.). It has been noted that while some studies define it very narrowly and only include some of its most obvious manifestations, others include a much wider range of situations [27]. There are a number of factors that indicate that the figures on SV may underrepresent the reality of this group. These factors are related to several barriers that affect GBM in relation to reporting experiences of SV. First, individuals who identify as male are particularly reluctant to admit psychological consequences that are inconsistent with male gender role expectations [28], resulting in poor communication of experiences. Such communication would actually be selective so they would need a higher level of distress to communicate such experiences [27]. In this sense, we should highlight the role of male socialization in understanding the differences in consequences and coping strategies between men and women when victimized [29]. Second, GBM are so stigmatized by their sexual orientation that they may encounter additional difficulties in communicating their experiences of SV for fear of further discrediting themselves [30]. It is also important to take into account the role played by how sexual relations and consent are conceptualized in this group since it is necessary for a SV situation to be labeled as such in order to communicate it in a relevant way. In fact, the most common method used by GBM to give consent is not to resist [31]. This makes it difficult to differentiate a consensual situation from a violent one, both for the perpetrator and the victim. There are some perspectives that claim that there is a "culture of silence" in the collective [32] that accepts, normalizes, and consequently obscures SV. Although it seems that some people in the GBM collective accept all these situations and do not cause them discomfort, some others are resisting such normalization [11], and they are having trouble labeling some experiences as sexual violence because they have them internalized. There are a number of factors that would result in an intentional ambiguity in defining what is and what is not SV in sexual minorities. Such factors would include stereotypes like gay men always want to have sex, gender norms regarding masculinity, a particular vision of queerness that could, in fact, foreclose the role of power dynamics, and the risk of social isolation when someone refuses sexualized social relations [33]. The aim of this study is to make SV visible for GBM individuals in Spain. To do so, the prevalence of victimization and the frequency in which certain specific violent situations occur, as well as some of their main characteristics have been studied. Emphasis has been placed on the communication of experiences; it is of interest to know to what extent victims communicate and to whom they turn in order to understand the low visibility of the problem and to propose preventive strategies and interventions. It is not the purpose of this study to pathologize the sexuality of men who have sex with men nor to suggest that violent behaviors are the primary framework from which we can generalize all GBM sexual experiences. Instead, it is intended to give voice to the discomfort experienced by some GBM individuals in order to promote a discourse that supports positive sexual experiences and the emotional well-being of GBM individuals. --- Materials and Methods --- Study Design and Data Collection This is a cross-sectional study aimed at obtaining quantitative information through a self-administered questionnaire. Participation was anonymous and voluntary and aimed at the GBM population living in Spain. The final sample consisted of 550 respondents. --- Measurement Tool The questionnaire was built using Google Forms v0.8 software. The definition of SV from the WHO was used, through which SV situations were extracted and transformed into items [1]. Thus, the pilot version of the questionnaire consisted of these items. It was administered for the first time in a small sample of GBM who had suffered some type of violence (10 participants) to ensure that there were no problems with the comprehension of the items and that the participants could give their opinions in general. Interviews were conducted with questions about the appropriateness of the items for the GBM collective in order to make them more sensitive to the reality of this particular group. Based on their recommendations, the questionnaire was reconstructed in such a way that new situations of sexual violence and more response possibilities were added to define their characteristics, as well as eliminating those questions that were labeled as uncomfortable or out of place. It was also decided to divide the experiences into two separate blocks (experiences with and without physical contact) to ensure that the mildest situations were asked first and that the most sensitive ones were presented little by little. The final version of the instrument was intended not only to collect data but also to serve as a tool for participants to share their perceptions of the experience at length if they so wished. It can be seen as an attempt to provide a tool with which to express themselves, given the invisibility and problems inherent in communicating the experiences of SV in this group. The internal consistency of the questionnaire was <unk> = 0.87 for the total scale. --- Variables 2.3.1. Socio-Demographic Information Information on gender identity, sexual orientation, race/ethnicity, level of education to date and autonomous community of usual residence was collected. Questions on gender identity and sexual orientation were part of the exclusion criteria and sent some participants to the end of the questionnaire according to their answers. The responses referring to the autonomous community of habitual residence covered all those autonomous communities and cities that are part of the Spanish state. To avoid bias, the questions that formed part of the exclusion criteria were asked before explaining the goals of the study and the target population. In this way, we tried to reduce the probability that participants who did not meet the admission criteria could falsify their identity, since once these variables were answered, participants were directed to the end of the questionnaire and could not participate again from the same email address. --- Prevalence and Frequency of SV Experiences To study the prevalence and frequency of perceived SV, participants were asked to what extent they have experienced a series of violent situations. The information about the frequency offers us not only information about what proportion of the sample has suffered each situation, but also informs us about the extent to which each participant has suffered those experiences. These frequencies were recorded on a Likert scale (1 = Never/2 = Once/3 = Sometimes/4 = Often/5 = Many times). Contrary to other studies, we avoided asking about the exact number of times each situation occurred. Firstly, it is difficult for participants to get the count of each situation right, especially when dealing with potentially traumatic experiences or almost normalized experiences. Secondly, because this study focuses on the victim's perception: a situation may have happened five times but for one person it may be labeled as "sometimes" and for another as "many times". The blocks of experiences have been differentiated as done by other studies to allow comparison into unwanted insinuation, sexual coercion, and sexual assault. The situations that were part of unwanted insinuation were (1) Unwelcome sexually explicit comments (Comments) and (2) Inappropriate exhibition of genitalia (Exhibition). Sexual coercion included (3) Excessive insistence to have sex (Insistence), (4) Verbal recrimination (e.g., insults) motivated by the refusal to have sex with the perpetrator (Verbal Recrimination), (5) Blackmail and threats in order to convince the victim to have sex (Blackmail and Threats), (6) Physical punishment motivated by the refusal to have sex with the perpetrator (Physical Punishment), (7) Victim incapacitated (physically or mentally) to give consent or to resist (Self Incapacitation) and (8) Victim intentionally drugged by the perpetrator for sexual purposes (Chemical Submission). Sexual assault included (9) Unwanted touching with sexual intent (Touching), (10) Attempted rape: victim managed to escape the situation (Attempted rape), (11) Forced physically without penetration, whether oral, anal or vaginal (Forced without penetration) and ( 12) Forced physically with penetration, whether oral, anal or vaginal (Rape). Items (13) Intentional and unwanted injuries during sex (Injuries), ( 14) Image-based sexual violence: the perpetrator shows or posts sexual images of the victim without consent (Images) and ( 15) Refusal to use protection or not using it without the victim's consent (Protection) did not correspond to any of the previous blocks. To avoid ambiguity to the greatest extent possible, the items have been written specifying that there must be a feeling of discomfort. In this way, an attempt has been made not to record those situations that were appropriate for the context. For example: "... without my consent", "... and I did not want it", "... and I think it was inappropriate". If any participant answered "Never" in all situations, they were redirected to the end of the questionnaire because the rest of the questions were not applicable. --- Situational Characteristics Some information about the violent person has been examined. In order to keep the questionnaire brief, these characteristics have been recorded, as well as all of the following, for the overall situations of SV instead of asking for the characteristics of each of the situations separately. For this reason, it was reminded in these sections that in cases of having experienced several situations of violence or being protagonized by different perpetrators, the participant should mark all the relevant answers. Regarding the age difference with the perpetrator, exact numbers were again avoided with the same justification: victims may have problems remembering all the exact ages (especially in cases where there are several abuses, several abusers or if he/she was incapacitated) and an age difference of five years, for example, may or may not be a clear age difference depending on the person and their relationship with the perpetrator. The gender identity of the perpetrator and the bond with the victim at the time of the abuse were also explored. It was asked if any of the violent situations had been group aggressions and if any of them had been carried out by the same person more than once. Information about the place where the violent situation occurred was also collected. --- Communication and Social Support Participants were asked if they have shared their experiences, and if so, to whom they have shared them. A general question was asked about the potential usefulness of support groups (if they existed) for GBM individuals who have suffered violence. It was evaluated on a Likert scale (1 = Not useful at all/5 = Very useful). --- Participants The participants who were part of the sample met the following inclusion criteria: (1) identify as male (regardless of being cissexual or transsexual), (2) identify as homosexual or bisexual, and (3) live in an autonomous community or city in Spain on a regular basis. Any other answer to the questions related to gender identity, place of habitual residence or sexual orientation were discarded. Thus, if a total of 1220 responses were recorded, the initial sample included responses from the 690 participants who met the inclusion criteria. To ensure representation of the territorial distribution of the population, the number of participants from each autonomous community that was included in the study was proportional to the population of men in each community in relation to the Spanish total based on data from the National Institute of Statistics [34]. Thus, some responses from those autonomous communities and cities that were overrepresented were randomly discarded. The final sample that would later be analyzed was composed of the responses of 550 participants. These participants were obtained from the dissemination of the questionnaire through social networks (Instagram and Twitter) with the help of various Spanish LGTBIQA+ associations and through a profile on Grindr, a dating application for GBM, which contained the link to the questionnaire. --- Statistical Analysis The statistical program Stata version 14.2 was used for data analysis. A univariate analysis has been conducted for the experiences of SV, situational characteristics and communication and social support. Bivariate analysis is meant to uncover the differences between sociodemographic groups regarding SV prevalence and frequency. The responses to each situation of SV were interpreted dichotomously (1 = No; 2, 3, 4, 5 = Yes) for the prevalence study, following the criterion of "tolerance 0". That is, any violent situation was considered to be present if the participant responded with any value other than 1. Confidence of 95% was used for the elaboration of confidence intervals (CI) for the univariate analysis. Analysis of variance (ANOVA) was carried out to determine the differences between groups regarding experiences of SV according to their sociodemographic characteristics: gender identity, sexual orientation, age, race/ethnicity and educational level for the bivariate analysis. Statistical significance was interpreted with p-value <unk> 0.05 from Student's t calculation. Bonferroni adjustment was calculated separately for the analysis of each group. It should be noted that some category values do not have a large enough sample size to generalize the results. Situational characteristics and social support were examined from a descriptive approach. --- Ethics Considerations The questionnaire construction and management considered the ethical aspects of the Code of Good Practice of the Autonomous University of Barcelona and the recommendations of the European Charter for Researchers, based on the fundamental principles of anonymity, freedom, honesty and responsibility. The participants were aware of the objectives of the study and gave their consent to participate in it. --- Results --- Sociodemographic Characteristics Table 1 shows the sociodemographic characteristics of the participants who were part of the final sample. Most of them were cis homosexual white young men with a high educational level. --- Prevalence and Frequency of SV Experiences Table 2 provides a summary of the prevalence and frequency of occurrence of each of the vs. items and blocks presented in this study. The most prevalent situations for the GBM group in Spain are unwanted insinuations (87.18-92.38%), especially consisting of explicit sexual comments that are inappropriate (83.80-89.61%). The figures on the denial of the use of protection (42.67-51.18%) are also relevant considering the great number of STDs in this collective. The prevalence of sexual assault (59.83-68.02%) is especially high, with a 54.30-62.70% prevalence of situations of unwanted touching. Figures on rape prevalence are also alarming, ranging from 16.57% to 23.40%. Results on frequency are coherent with those found in the prevalence analysis. Note. The blocks Unwanted insinuation, sexual coercion and sexual aggression have been calculated from the average of the items that comprise them. --- Situational Characteristics The results on the situation and characteristics of the perpetrator are presented below (Table 3). It is of interest to note that the perpetrator is a stranger to the victim in most cases (63.65%) and that their most usual characteristics are being a cis man (95.09%), older than the victim (68.96%) and online using gay dating apps (58.15%); 9.63% of the sample informs us to have experienced group violence and 29.86% to have experienced violent situations with the same perpetrator more than once. Most participants were victimized when they were 19-24 years old (58.94%). Note. The sum of the percentages for each variable is in some cases >100% because participants could choose more than one response option. --- Communication and Social Support Finally, the responses regarding communication and social support are reflected in Table 4. The usefulness of the support groups for the participants was rated positively with a mean of 4.26 (95% CI: 4.17-4.36). Of all the participants of the study who have suffered any kind of violence, most of them did not share all their experiences (67.58%). Most of them trusted their friends to do so (84.48%). --- Experiences of SV <unk> Sociodemographic Characteristics The bivariate analysis that has been conducted regarding SV experiences vs. Sociodemographic characteristics is shown below with the symbols * = p-value <unk> 0.05, ** = p-value <unk> 0.01, *** = p-value <unk> 0.001. In terms of prevalence, trans people report a higher prevalence than cis people in the situations of Incapacitated ** and Rape **, and in Sexual Assault *. Bisexuals also reported a higher prevalence than homosexuals in situations Forced without penetration *. The age variable had the greatest effect on the items referring to situations of SV, with the youngest respondents showing a significantly higher prevalence. This is the case for the items Comments ***, Exhibition *, Insistence ***, Blackmail and threats ***, Verbal recrimination ***, Rape *** and in the groupings Unwanted insinuation *** and Sexual coercion **. In addition, people with no studies or only compulsory secondary education have reported significant differences showing a higher prevalence of SV, in most cases compared to all other categories. This is the case for the items Chemical Submission *, Injury *, Physical Punishment *** and Forced without penetration*. Finally, with respect to the Race/Ethnicity categories, differences were only found in the comparison between Latinos and whites, with Latinos reporting a higher prevalence of SV. Such differences are found in the variables Chemical Submission * and Attempted Rape *. Note. The sum of the percentages for each variable is in some cases >100% because participants could choose more than one response option. Regarding the frequency of occurrence, the results are similar. Trans people reported higher frequencies in the item Rape *. Bisexual people in the items Forced without penetration ***. Younger participants in the items Comments ***, Exhibition ***, Insistence ***, Blackmail and Threats *, Verbal Recrimination **, Rape *, the groupings Unwanted Insinuation *** and Sexual Coercion **. People with no education or with compulsory secondary education in the items Touching *, Injuries *** and Forced without penetration **, the groupings Sexual coercion ** and Sexual assault **. Finally, Latinos reported a higher frequency of Chemical Submission *. --- Discussion and limitations First, it is necessary to review the characteristics of the sample to see how representative the results really are. Although it was intended to be representative of the Spanish territory, it is not representative of most sociodemographic issues, so some categories are underrepresented: the vast majority of participants were cissexual, homosexual, with a higher level of education, white and young. In this sense, the sample of trans men is particularly small, so the results should be interpreted with caution, although many articles agree that this group receives more violence than cis people [35]. The sample of people with special education and compulsory secondary education/no education is also insufficient, as well as those of all race/ethnicity categories except for white and Latino men. All of these difficulties were to be expected, since first of all, very large samples are required to do intersectional studies [36] and also, the GBM population is very much in the minority, so it is difficult to get participants, and consequently, to make it large and representative. Even so, the considerable sample size of this study is considered a strength for the reliability of the results for the GBM population in Spain. Although one of the objectives of the questionnaire was to ensure its brevity, it would be much closer to reality to include non-exclusively categorical measures of sexual orientation and gender. Taking into account that from a constructivist perspective gender is constructed in each social interaction to the extent that heterosexist social conventions are repeated (or not) [37], it would be interesting to understand to what extent the tendency taken by this performance affects the directionality of the perpetration-victimization, rather than focusing exclusively on gender identity. Indeed, in that sense, the terms gay and lesbian, beyond signifying only preferences in the object of desire, could be well-recognizable social identities, perhaps authentic and proper genders [38]. There is a need to advance the study of gender from a non-categorical queer perspective in order to understand the phenomena studied in all scientific disciplines in a comprehensive way; specifically, in the field of SV. Exploring beyond a binary definition of gender could help to make violence against non-heterosexual individuals more visible. Given the variability and multiplicity of definitions and interpretations of SV, we cannot assure construct validity in this study, although the items that form part of the questionnaire were elaborated on the basis of what the WHO defines as SV [1] and were modified under the evaluation of GBM individuals. The figures on the prevalence of SV in the group are extremely high, which is not surprising considering that a wide range of situations of greater or lesser severity have been examined and that this is a study on lifetime victimization. However, the numbers could have been magnified due to the method of data collection, as most participants have been recruited by a GBM dating app and may be more likely to have experienced SV situations than those who do not use them. Comparison with other studies is not possible, firstly because the instrument for exploring SV situations is different, but also because some studies have only focused on childhood or adulthood experiences, while others have only recorded situations that occurred during a period of time (usually during the 6 months or a year before participating in the study). In addition to all these differences, it is worth noting the different conceptions of the concept of SV or sexual assault [27]. The only studies with which comparisons could be made report a 54.0% [22] and 35.2% [39] lifetime prevalence of sexual assault, although the questionnaire they use is very different, the definition of sexual aggression is slightly different, and they are from different countries. Those figures are lower than the prevalence found in this study, which is around 64% for sexual assault. On the other hand, the study on frequency is quite atypical in a work of these characteristics, so there is no quantitative data with which to compare the results. However, it is consistent with the violent behaviors that have been labeled previously as normalized and that manifest themselves as unwanted sexual advances (comments, exhibition, insistence) and touching, which are precisely the variables that have shown the highest frequency averages (>2), as well as verbal recrimination [11]. Results show that bisexual men, transgender men, those with a low educational level, and Latinos are more at risk of experiencing SV. This fact could be related to the concept of homonormativity, which refers to the privileges that some LGTBIQA+ people have for conforming to heterosexual norms, so that those cis homosexual men, without disability, with normative bodies, gender expressions and affective relationships would get more social recognition, so those "deviant" people would be more vulnerable to suffer violence [40,41]. Still, the sociodemographic characteristic that most significantly affects victimization is age. In fact, it is curious that, as this is a lifetime victimization study, it is the youngest people who report the most violence in a very significant way. Although it is not possible from this study to establish causal relationships, it could be due to (1) generational differences in the prevalence of SV, probably due to the emergence of the internet and social networks or the existence of more nightlife spaces for GBM, (2) a greater ability to label violent situations correctly due to the influence of the fourth feminist wave [42] in younger people, with movements such as #MeToo or (3) a memory bias that would make it difficult for older people to remember violence in their youth, especially for mild experiences of SV. Although the variable of victimization age, to our knowledge, has not been studied in SV within the collective, it has been studied in studies of intragender violence. The age variable is found to be the strongest and most consistent characteristic in relation to the victimization of any type of violence among gay male partners, a fact that they attribute to the acquisition of internal and external resources with age that are protective against violent situations [43]. However, in our results, it is much more frequent that the perpetrator is older than the victim. Considering that the majority of the sample in this study were young people, more studies are needed in the field of SV on GBM to determine whether age could be a power resource for the perpetration of violence against younger people. In relation to spaces, the prevalence of SV in dating apps is particularly high. In reference to this, we should consider the manifestations of rape culture in such spaces, particularly through Grindr, the geolocation-based social network most used by men who have sex with men. It would manifest itself through violent comments and unsolicited nude photos, and extend beyond online interactions in the form of sexual assault, sexual coercion, and image-based sexual violence [44]. This is consistent with articles that emphasize how Grindr encourages its users to see each other as objects to be consumed and discarded at will [45], a fact that could be considered an ethical danger in that objectifying and instrumentalizing others for one's own sexual pleasure could result in avoidance and closure towards otherness [46]. The effects that this whole scenario may have on the reproduction of vs. is evident, as it produces sexual encounters in which pleasure is centered on oneself and not on the interaction between the participants, a fact that could hinder communication and the interpretation of consent signals. The data from this study invites us to rethink the design, structure and format of apps for encounters between men who have sex with men, inasmuch as it is one of the spaces in which most situations of SV occur. Also noteworthy is the prevalence of SV in nightclubs and gay bars, which could be explained by the effect that alcohol and other drugs have on the perpetration of violence by predisposed individuals [47], as well as on the vulnerability to victimization [48]. A study of the expectations of the people who frequent these spaces could be of interest in order to investigate the relationship they may have with the perpetration of violent situations. Although several participants added in the "Other" option spaces such as saunas, chills, cruising areas, dark rooms, etc., they have not been addressed in this study because it would be too hasty to draw conclusions about SV from a quantitative point of view and could run the risk of stigmatizing these sexual practices. These are situations that could be understood as violent in and of themselves, which is why it is necessary to study them in-depth, from a qualitative framework, in order to investigate how consent operates in these contexts [49]. Almost all perpetrators, as reported by the victims, are cis males older than the victim. The bond with the perpetrator is one of the most interesting variables of the study, as it shows that the vast majority of violence is perpetrated by strangers. This implies that it does not make sense to study SV on GBM only in the area of intragender violence or intimate partner violence in homosexual relationships. The figures are very similar to those found in another study [21], especially high for strangers (33.3% vs. 63.65% in this study), casual partners/dates (29.40% vs. 33.99% in this study) and acquaintances (15.70% vs. 28.68% in this study). For future studies, it would be interesting to include more types of bonds, since the LGTBIQA+ group is very rich in forms of relationships (polyamorous relationships, exclusively sexual relationships, open relationships, etc.) and in this way, a representation closer to reality would be achieved. Finally, the fact that the vast majority of victims communicate their experiences to friends rather than to other links refers to the phenomenon of peers' communication and is consistent with existing literature [16]. In that sense, victim support groups could be a useful tool to promote such communication. It is important to note that all answers regarding situational characteristics and social support have been registered for all the situations of SV that the participants may have lived at once, so this study is not capable of attributing this data to any specific situation of SV. For example, we cannot tell from these figures if most rapes in particular are carried out by strangers or by acquaintances. Those difficulties were accepted to ensure the questionnaire's brevity, and therefore, to achieve the largest possible sample. As this is a preliminary study, it shows an overview of the problem and encourages other studies to further investigate the specific topics discussed above. Although it is inevitable that the study presents a series of limitations, it is important to emphasize the importance of being a pioneer in the study of SV on GBM living in Spain, as well as its usefulness as a tool for communication and expression for the victims. It is necessary to promote the study of sexual violence in this collective, as it is for all the letters of the LGTBIQA+ group. Addressing this issue, and consequently constructing prevention and intervention strategies focused on the particularities of this collective, would have important positive health outcomes. Not only because it would improve physical wellbeing by preventing STDs and the physical harm a violent situation may involve, but because it would also have a positive impact on psychological well-being. --- Conclusions This paper reveals that sexual violence is common among gay and bisexual men. Findings show that prevalence figures depend on sociodemographic characteristics like gender identity, sexual orientation, race/ethnicity and educational level. Prevalence would also be higher in certain situations and spaces. Some of the perpetrators' and victims' most usual characteristics have also been examined. Furthermore, it has come to light that those who have been affected face challenges when it comes to sharing their experiences with others and that they mostly rely on their peers to do so. Epidemiological research and interventions should take into account the intersections between gender identity and sexual orientation to better tailor prevention and treatment in this collective. Given the invisibility and stigma associated with this issue, this study highlights the usefulness that support groups could have in facilitating victims' communication. As a preliminary study, this paper could be useful for further in-depth

Sexual violence is an understudied issue in the population of gay and bisexual men, although the existing articles to date demonstrate that it is a problem that merits public attention. This study aims to approach the problem of invisibility around the matter, as well as presenting a number of variables that have been usually overlooked in Spanish research or have not been assessed at all. Lifetime sexual victimization, sociodemographic characteristics, situational characteristics and social support were examined among 550 gay and bisexual males living in Spain using a self-administrated questionnaire. Results analysis show that 90.00% (87.18-92.38%) of participants reported at least one experience of unwanted insinuation, 87.27% (84.19-89.94%) reported at least one experience of sexual coercion, 64.00% (59.83-68.02%) reported at least one experience of sexual assault, and specifically 19.82% (16.57-23.40%) reported being raped during their lifetime. Significant differences have been found between some categories regarding gender identity, sexual orientation, age, race/ethnicity and educational level. Overall, these results showcase sexual violence as a pervasive problem in the Spanish gay and bisexual community.

constructing prevention and intervention strategies focused on the particularities of this collective, would have important positive health outcomes. Not only because it would improve physical wellbeing by preventing STDs and the physical harm a violent situation may involve, but because it would also have a positive impact on psychological well-being. --- Conclusions This paper reveals that sexual violence is common among gay and bisexual men. Findings show that prevalence figures depend on sociodemographic characteristics like gender identity, sexual orientation, race/ethnicity and educational level. Prevalence would also be higher in certain situations and spaces. Some of the perpetrators' and victims' most usual characteristics have also been examined. Furthermore, it has come to light that those who have been affected face challenges when it comes to sharing their experiences with others and that they mostly rely on their peers to do so. Epidemiological research and interventions should take into account the intersections between gender identity and sexual orientation to better tailor prevention and treatment in this collective. Given the invisibility and stigma associated with this issue, this study highlights the usefulness that support groups could have in facilitating victims' communication. As a preliminary study, this paper could be useful for further in-depth research on the topics discussed above, such as sociodemographic factors related to the risk of victimization or the role of external situational characteristics to specific situations of sexual violence. It is necessary to promote the study of sexual violence in this group to understand the power dynamics that could underlie these situations. --- Data Availability Statement: The data sets used and analyzed in the current study are available from the corresponding author on reasonable request. --- Institutional Review Board Statement: Not applicable. The protocol of the study was not evaluated by the Ethics Committee of our university because it did not include invasive procedures; it did not include collection, use, or storage of biological samples from subjects; nor did it include collection, use, or storage of genetic information from participants. Following the current Spanish legislation, approval from the Ethics Committee is mandatory only when a study protocol includes any of these procedures. Informed Consent Statement: Informed consent was obtained from all subjects involved in the study. --- Conflicts of Interest: The authors declare no conflict of interest.

Sexual violence is an understudied issue in the population of gay and bisexual men, although the existing articles to date demonstrate that it is a problem that merits public attention. This study aims to approach the problem of invisibility around the matter, as well as presenting a number of variables that have been usually overlooked in Spanish research or have not been assessed at all. Lifetime sexual victimization, sociodemographic characteristics, situational characteristics and social support were examined among 550 gay and bisexual males living in Spain using a self-administrated questionnaire. Results analysis show that 90.00% (87.18-92.38%) of participants reported at least one experience of unwanted insinuation, 87.27% (84.19-89.94%) reported at least one experience of sexual coercion, 64.00% (59.83-68.02%) reported at least one experience of sexual assault, and specifically 19.82% (16.57-23.40%) reported being raped during their lifetime. Significant differences have been found between some categories regarding gender identity, sexual orientation, age, race/ethnicity and educational level. Overall, these results showcase sexual violence as a pervasive problem in the Spanish gay and bisexual community.

Introduction More than a third of today's teenagers are overweight or obese, with the prevalence of obesity slightly higher than that of overweight (18.1% vs. 16.1%) (Ogden, Carroll, Curtin, Lamb, & Flegal, 2010). Adolescents who are obese are at greater risk for obesity in adulthood (Merten, 2010), which is associated with numerous health complications such as hypertension, type 2 diabetes mellitus, dyslipidemia, and cardiovascular diseases (Cannon & Kumar, 2009;Ludwig, 2007). Additionally, adolescents who are overweight or obese are more at risk for developing health-compromising behaviors, such as substance use and violent behavior (Farhat, Iannotti, & Simons-Morton, 2010). There are also psychological consequences of obesity. Adolescent obesity has been associated with the development of body dissatisfaction (Paxton, Eisenberg, & Neumark-Sztainer, 2006) which can put adolescents at risk for externalizing and internalizing problem behaviors and social and attention problems. Studies have also shown that body dissatisfaction may lead to eating pathology (Stice & Shaw, 2002) and is associated with poor perceived health (Meland, Haugland, & Breidablik, 2007). Body dissatisfaction is particularly worrisome for overweight and obese adolescents. Overweight and obese adolescents with body satisfaction are more likely to do something to control their weight (Cromley, Neumark-Sztainer, Story, & Boutelle, 2010;Flock, Farhat, & Haynie, 2009) whereas adolescents with body dissatisfaction are more likely to engage in unhealthy weight management behaviors and are at a risk for weight gain (Neumark-Sztainer, Paxton, Hannan, Haines, & Story, 2006). One study of adults found that body image partially mediates the relationship between obesity and psychological distress (Friedman, Reichmann, Costanzo, & Musante, 2002). These findings illustrate potentially positive effects of fostering body satisfaction in adolescents who are overweight. Social engagement may affect adolescent body image (Gerner & Wilson, 2005). For children and adolescents, engaging with friends is a "developmental advantage" and a social resource, increasing socio-emotional maturity and fostering self-esteem and well-being (Berndt, 2002;Hartup & Stevens, 1997;Witvliet, Brendgen, van Lier, Koot, & Vitaro, 2010). Friendships are also important in the development of body image and have been associated with body satisfaction among adolescents in general (Graham, Eich, Kephart, & Peterson, 2000;Stice & Whitenton, 2002) and with psychosocial health among adolescents with chronic conditions (Helgeson, Reynolds, Escobar, Siminerio, & Becker, 2007). Supportive friendships may be particularly protective for overweight adolescents. Bowker, Spencer and Salvy (2005) found that for overweight adolescents, positive friendship quality was negatively associated with emotion-focused coping (feeling upset but doing nothing) and friendship conflict was a risk factor for internal blame. Besides improving overweight adolescents' mental and social health, peers have been shown to increase their physical activity and their motivation to exercise (Salvy et al., 2009). There has been relatively little research on the positive effects of peers on overweight/obese adolescents' body image. Most previous studies looked at teasing, or social comparisons and "fat talk" (Dunkley, Wertheim, & Paxton, 2001) where peer discussion of body weight and dieting may cause concern about one's own weight and body image. Less information exists on the role of social engagement in the association of weight status with body image. Having a supportive group of friends is an essential part of healthy adolescent development, and for overweight adolescents, could be a protective factor against body dissatisfaction (Dunkley et al., 2001). The purpose of this study is to examine whether social engagement modifies the association between weight status and body image. We hypothesize that: (1) overweight/obesity is associated with body dissatisfaction; (2) Overweight/obese adolescents with more social engagement are less likely to be dissatisfied with their bodies than overweight/obese adolescents with less social engagement. Because studies have shown that the influence of peers on other adolescent behaviors, such as drinking, is different for adolescent girls and boys, we expect these relationships to vary by gender and therefore examine the relationships in separate models for boys and girls (Dick et al., 2007). --- Method Sample The 2005/6 Health Behavior in School-Aged Children (HBSC) survey is a nationally representative school-based study conducted every four years to measure adolescent health, risk behaviors, and the context in which these behaviors occur. A three-stage stratified clustered sampling, with classes as the sampling units, was used to select a sample of American students in grades 6 to 10 during the 2005/6 academic school-year. Black and Hispanic students were oversampled to provide better population estimates for these minorities. Anonymous surveys were completed in the classrooms. Ethics approval was obtained from the Eunice Kennedy Shriver National Institute of Child Health and Human Development Institutional Review Board. The survey was completed by 87% of eligible students (n=9,227). Half of the 6 th graders who completed the questionnaire were not asked the questions on body image and were therefore excluded from the analysis, as were respondents who did not provide all essential demographic information, producing the analytic sample (n=6,909). --- Measures Body image-Body image was assessed as a continuous variable created from the mean of five items (<unk> =.87) from the body image subscale part of the body investment scale (BIS) (Orbach & Mikulincer, 1998). The BIS was piloted in Hungry and Finland in 2005 and performed with good internal consistency (Cronbach's alpha =.92 and.72). The first subscale of the BIS was constructed to measure body image feelings and attitudes and measured several aspects of body image: evaluation, attitudes, and emotions. This scale has been validated and used in other studies (Carroll & Anderson, 2002;Lamis et al., 2010) to assess body image in adolescents. Questions asked respondents how much they agreed or disagreed with five statements: "I am frustrated with my physical appearance"; "I am satisfied with my appearance"; "I hate my body"; "I feel comfortable with my body"; and, "I feel anger toward my body." (The original BIS includes six items, but only five were available in this survey. Additional analyses by the authors (available upon request) suggest that the reliability of the 5-item BIS measure is essentially equivalent to that of the original 6-item measure.) The scale ranged from 1 to 5 with higher scores indicating less body dissatisfaction. Weight status-Weight status was a dichotomized measure of the student's Body Mass Index (BMI) computed from self-reported height and weight according to the CDC (Center for Disease Control) 2000 growth chart (Kuczmarski et al., 2002). Overweight/obese adolescents with BMI at the 85 th percentile or above were compared to adolescents with BMI below the 85 th percentile. Social engagement-Social engagement was a continuous scale from the mean of eight items. This scale has been used in previous research investigating social isolation (Spriggs, Iannotti, Nansel, & Haynie, 2009) and quality of peer relationships (Iannotti, Kogan, Janssen, & Boyce, 2009;Iannotti et al., 2009). Three items asked whether it was easy to talk with a best friend, friends of the same sex and friends of the opposite sex (responses ranged from very easy to very difficult). Two items asked about the number of close male and female friends (responses ranged from none to three or more). Two items asked how many days/week they usually spent time with friends immediately after school (0 to 5) and how many evenings/week they usually spent time with friends (0 to 7). The last item asked how often they talk to friend(s) on the phone or send them text messages or have contact through the internet (responses ranged from rarely or never to every day). The eight items were z-transformed, standardized (mean= 0; standard deviation = 1) and averaged to provide the overall measure of social engagement (<unk> =.67). Sociodemographic characteristics-Variables included were age (continuous), race/ ethnicity (White/Black/Hispanic/Other), and socioeconomic status (Low/Moderate/High Affluence) measured by the Family Affluence Scale (FAS). FAS was constructed from questions about family wealth and categorized into tertiles. A review (Currie et al., 2008) indicated that the scale has good content and external validity. --- Statistical Analysis All analyses were conducted using Stata 9 to adjust for the cluster-based sampling design of HBSC. Weights were applied to provide nationally representative estimates. The prevalence and means of the sample variables were calculated, as were the bivariate associations between them. Multiple linear regressions, separately by gender and controlling for age, race/ethnicity and socioeconomic status, were conducted with an interaction term including weight status and social engagement. When the interaction was not significant, we concluded that the association between weight status and body image did not vary by social engagement and the interaction was removed from the model to examine the main effect of weight status on body image. When the interaction was statistically significant, we concluded that social engagement modified the association between weight status and body image, and we then stratified the analyses by different levels of social engagement. Significance was set at the 95% confidence level. --- Results Sample characteristics are presented in Table 1. The mean age for both boys and girls was 14.4 years, with no significant gender differences. The largest demographic groups were White (43.1%) and moderate socioeconomic status (48.3%). The prevalence of overweight/ obesity was 33.9% for boys and 27.4% for girls. While most respondents reported being satisfied with their bodies, boys reported significantly more body satisfaction than girls. Girls reported more social engagement than boys. Bivariate analyses showed that adolescents' overweight/obese status was related to body dissatisfaction and less social engagement for both boys and girls. Wald tests were used to account for the design effects (sampling and weighting) of our data. Overweight/obese boys and girls had more body dissatisfaction and less social engagement than their normal/ underweight peers (Table 2). Social engagement was, however, differentially associated with body image by gender. For boys, sociodemographic variables were significantly associated with body image: Black adolescent boys had more body satisfaction compared to White adolescent boys (<unk> =.23, p <unk>.001), and boys of higher affluence reported more body satisfaction than boys of low affluence (<unk> =.19, p =.001). In subsequent multivariable linear regression models controlling for age, race/ethnicity, and socioeconomic status, the weight status by social engagement interaction was not significant (<unk> =.06, 95% CI = -.30-.43); therefore, the interaction term was removed from the model to examine main effects (Table 3). Final results show that overweight/obese boys were more likely to have body dissatisfaction (<unk> = -.47, p <unk>.001) than normal/underweight boys. Boys with more social engagement were more likely to have body satisfaction than boys with less social engagement (<unk> =.59, p <unk>. 001). Among girls, sociodemographic variables including age and race/ethnicity were significantly related to body image. Black girls with less social engagement were more likely to have body satisfaction (<unk> =.47, p =.02) than their White peers, and Black girls with more social engagement were more likely to have body satisfaction (<unk> =.76, p <unk>.001) than their White peers. Age was significant in the model for girls with less social engagement. For younger girls, less social engagement was associated with body dissatisfaction (<unk> = -.14, p =.002). In regression analyses controlling for sociodemographic variables, social engagement moderated the relationship between weight status and body image (Table 3). Because the interaction term was significant, the data was stratified to examine the association of weight status with body image for girls with high social engagement (<unk>90th percentile) and those with low social engagement (<unk> 10th percentile): Overweight/obese girls with high social engagement were more likely to have body satisfaction than overweight/obese girls with low social engagement (<unk> = -.69, p <unk>.001) (Table 3 andFigure 1). --- Discussion This study examined the relationship between weight status and body image in a nationallyrepresentative sample of US adolescents. As posited in our first hypothesis, obesity was associated with body image dissatisfaction in boys and girls, confirming previous findings (McCabe, Ricciardelli, & Holt, 2010;Mond, van den Berg, Boutelle, Hannan, & Neumark-Sztainer, 2011;Neumark-Sztainer et al., 2002;Schwartz & Brownell, 2004). Although the prevalence of overweight/obesity was higher in boys compared to girls, adolescent girls reported significantly more body image dissatisfaction than boys. Social engagement moderated the relationship between weight status and body image for girls, but not for boys, thus partially supporting our second hypothesis. However, for boys, social engagement was associated with more body satisfaction, regardless of their weight status. Given the high prevalence of obesity, more adolescents may be struggling with body dissatisfaction. Both obesity and body dissatisfaction are risk factors for low self esteem and depression, and can lead to psychosocial and physical problems (Brausch & Gutierrez, 2009;van den Berg, Mond, Eisenberg, Ackard, & Neumark-Sztainer, 2010). They may also lead to disordered eating behaviors which can increase the risk for weight gain (Neumark-Sztainer et al., 2002). The high percentage of adolescent boys (33.9%) and girls (27.4%) who were overweight/obese in this nationally representative sample, along with the consistent association between overweight/obesity and body dissatisfaction across gender, highlight the urgency to address the increasing prevalence of body dissatisfaction in adolescents. Much of the research examining peer influences on body image among adolescent boys and girls focuses on the messages peers impart rather than on a more comprehensive measure of social engagement. However, some studies have shown that low friendship quality may contribute to body dissatisfaction in adolescent girls (Schutz & Paxton, 2007) so the peer group in general might have an important influence on body image. The desire to conform to the specific norms of a peer group places pressure on adolescents to be thin. This conflict between conforming to ideal norms of thin bodies and the reality of their actual body size may result in body dissatisfaction for boys and girls (Gerner & Wilson, 2005;Presnell, Bearman, & Stice, 2004). The positive association for boys and girls between social engagement and body satisfaction supports the idea that peers and social relationships are important for adolescent development. Despite the increasing prevalence of obesity, society remains extremely disapproving towards overweight individuals as demonstrated by the high prevalence of institutional and interpersonal weight-based discrimination (Puhl, Andreyeva, & Brownell, 2008). The stigma associated with obesity increases the probability that overweight individuals will internalize negative thoughts about themselves and develop body dissatisfaction (Makara-Studzinska & Zaborska, 2009). Even more distressing, adults who had child or adolescent onset obesity have the most significant body image disturbances (Makara-Studzinska & Zaborska, 2009). Social engagement moderated the relationship between weight status and body image for girls, but not for boys, thus partially supporting our second hypothesis. For boys, social engagement was associated with body satisfaction, regardless of their weight status. To our knowledge, no studies have investigated the association between peer groups and body image among boys but these findings suggest the need for further examining mechanisms linking social engagement to body satisfaction within this group. For adolescent girls, social engagement was a moderator of the relationship between weight status and body image. Overweight/obese girls with less social engagement had more body dissatisfaction than overweight/obese girls with more social engagement, suggesting the importance of social engagement for adolescent girls, especially if they do not meet the norms for weight status. Bowker, Spencer and Salvy (2005) showed that supportive friendships were negatively associated with emotion-focused coping while friendship conflict was positively associated with internal blame among overweight but not normal/ underweight adolescents. Overweight adolescents are more likely to be socially isolated and on the periphery of social networks (Strauss & Pollack, 2003), and overweight girls in particular may be more likely to experience peer exclusion (Neumark-Sztainer et al., 2002;Pearce, Boergers, & Prinstein, 2002) which may reinforce overweight/obese girls' perceptions of body dissatisfaction (Gerner & Wilson, 2005). Based on the literature review and our conceptual framework, we hypothesized that the major differences would be between adolescents with normative/society-approved weight status (normal weight and underweight) and those whose weight deviates from the normative/ideal standards. The weight status categories were dichotomized at the 85 th percentile but there could have been differences between the overweight and obese groups. However, both overweight and obese adolescents have been found to experience more teasing than average weight adolescents (Neumark-Sztainer et al., 2002.) In addition, prior to collapsing the underweight and normal weight categories, we tested for significant differences in mean body image across these groups, for boys and girls. Our results showed that there were no significant differences, for either gender. It is possible that body dissatisfaction works differently in underweight adolescents but our sample of underweight adolescents was insufficient to analyze separately. Future studies should look at this specific population. This study has many strengths including the use of a large nationally representative sample that is socially and economically diverse, the use of a previously tested valid and reliable measure of body image, and the examination of associations that are central to adolescent development, yet were seldom the focus of previous research. One important limitation includes the cross-sectional nature of the study which does not allow directional conclusions. Our proposed framework suggests that obesity leads to body dissatisfaction, but a reverse direction might also be true, e.g., body dissatisfaction may cause adolescents to indulge in excessive eating that could result in overweight/obesity (Neumark-Sztainer et al., 2006). Both pathways are plausible and can only be disentangled in future longitudinal studies. Additionally, it is important to note that there may be a bias in who reports social engagement. Depression, for example, may cause an adolescent to report less social engagement and body dissatisfaction. Another limitation is that weight status was calculated from self-reported, rather than actual, height and weight. Studies however, have shown that the bias in self-reported height and weight is minimal and certainly acceptable for analysis of group data (Dietz & Bellizzi, 1999;Goodman, Hinden, & Khandelwal, 2000). In summary, establishing friendships is an essential developmental task for adolescents and as this study showed, social engagement can protect adolescents from having body image dissatisfaction. Our findings show that social engagement is protective against body dissatisfaction for all boys, regardless of their weight status, and especially protective for overweight/obese girls. Therefore, encouraging adolescents, particularly overweight/obese teenage girls, to develop strong, healthy relationships with peers may prevent them from having body dissatisfaction. When faced with adolescent girls or boys, regardless of their weight status, who are dealing with body dissatisfaction, physicians, psychologists, parents and teachers could inquire about their social engagement and possibly encourage them to establish more interaction with friends, for example through involvement in clubs and team sports. Given the continuous rise in overweight/obesity among adolescents, the accompanying increase in the prevalence of body dissatisfaction, and the numerous adverse health consequences that could develop from body dissatisfaction, successful strategies that could prevent the development of negative body image are needed. Encouraging social engagement among adolescents may be one such strategy. More research is needed, however, to examine potential racial differences in the proposed framework and to evaluate the relative importance of parental influence and social engagement in the association between adolescents' weight status and body image. --- RESEARCH HIGHLIGHTS • Adolescent's overweight/obese status was related to body dissatisfaction. • Social engagement moderated the relationship for girls but not for boys. • Overweight/obese girls with high social engagement had greater body satisfaction. Association between weight status and body image by social engagement in girls. Low social engagement is defined as (<unk> 10 th percentile) and high social engagement as (<unk> 90 th percentile). Age, race/ethnicity and socioeconomic status were controlled in all analyses. Note. Low social engagement is defined as (<unk> 10 th percentile) and high social engagement as (<unk> 90 th percentile) Age, race/ethnicity, and socioeconomic status were controlled in all analyses. Body Image. Author manuscript; available in PMC 2013 March 1.

This study examined whether the association between adolescent weight status and body image varies by social engagement. A nationally representative sample of 6,909 students in grades 6 to 10 completed the 2006 HBSC survey. Separate linear regressions for boys and girls, controlling for age, race/ethnicity and socioeconomic status, were conducted with an interaction term (weight status x social engagement). Adolescents' overweight/obese status was related to body dissatisfaction. Social engagement moderated the relationship between weight status and body image for girls but not for boys. Overweight/obese boys had more body dissatisfaction compared to their normal/underweight peers, regardless of their social engagement. However, overweight/ obese girls with more social engagement were more likely to have body satisfaction compared to overweight/obese girls with less social engagement. Encouraging adolescent girls to develop healthy relationships with peers may prevent them from developing body dissatisfaction.

Background The plight of orphans a and abandoned children (OAC) is an increasing global problem that is particularly pervasive in Southeast Asia and sub-Saharan Africa [1]. Improving the educational attainment of the 153 million orphans and other vulnerable children worldwide is a key goal for development policymakers and practitioners. International declarations such as the Millennium Development Goals and the Education for All Movement indicate that the educational attainment of vulnerable children has become a global priority. Most recently, the 2011 Political Declaration on HIV/AIDS targeted increases in school attendance of orphans as an important and measureable indicator of progress [2]. To understand which policies can improve educational attainment for OAC, decision makers must first understand the determinants of and barriers to these outcomes. However, there are mixed results regarding which factors, including orphanhood itself, are significantly associated with educational attainment. Previous research shows that the loss of a parent can lead to a series of developmental disadvantages resulting in poor education [3][4][5][6][7], such as lags in grade for age and school attendance relative to non-orphans [3,7,8]. In contrast, other studies find little negative impact of parental death on child education [9][10][11] and instead find that alternative factors such as wealth, age, or the child's relationship to the head of household are better predictors of education outcomes [12]. Importantly, many of these studies are restricted to single country analyses, rendering results arguably context specific. Additionally, few studies examining the educational attainment of orphans move beyond outcomes such as school enrollment, grade for age, and attendance to disentangle what factors contribute to a child's learning [13,14]. Other outcomes may provide more meaningful information. A recent study showed that tests of cognitive development can be a useful measure in understanding OAC learning and educational experience across settings. In particular, cognitive development was found to be positively associated with increases in exposure to formal education [15]. One factor that may offer some insight into the development of OAC is mental health. Previous research, though not focused on orphans, highlights the role of adverse childhood experiences in life outcomes [16]. Cumulative effects from multiple risk factors have been shown to be more predictive of compromised early cognitive development among vulnerable children than any one risk factor alone [17]. In the research on orphans, the role of psychosocial status on multiple outcomes has been studied, though not specifically in relation to learning outcomes. Research on OAC in low and middle income countries found that orphaned children are more susceptible than non-orphans to exposure to potentially traumatic events due to lack of adequate adult protection [18][19][20][21]. A recent study found that OAC anxiety and emotional difficulties increased with additional exposure to potentially traumatic events [22]. Another recent study found associations between orphan risk of psychosocial difficulties and subsequent risk of HIV infection [23]. While there is a clear need for mental health interventions among OAC, few investigators have examined the role of mental health in predicting OACs' cognitive development [24][25][26]. To our knowledge, no study has attempted to do a cross-country analysis of community based OAC to address these questions. Through these analyses, we aim to understand better the linkages between emotional difficulties and a child's ability to participate in and gain from education in a context where children are particularly vulnerable to adverse events and subsequent emotional difficulties. In this manuscript, we employ within-country and cross-country analyses to examine associations between exposure to adverse childhood events, the emotional difficulties that OAC face, and their cognitive development. --- Methods --- Study description Positive Outcome for Orphans (POFO) is an ongoing longitudinal study following a cohort of children, starting ages 6 to 12, who live in institutional or communitybased settings in 5 low income countries: Cambodia, Ethiopia, Kenya, India, and Tanzania. This analysis used 3 years of data from the community-based sample to address the relationships stated above. --- Study sample The detailed sampling strategy and general characteristics of the sample have been reported elsewhere [26,27]. The following describes the elements of the sampling strategy applicable to this analysis. The POFO study utilized a two stage random sampling methodology to identify a representative sample of 1,480 orphaned and abandoned children living in communitybased settings in six sites across five low and middle income countries. Within each site, geographic or administrative boundaries were used to define sampling areas (clusters), 50 clusters were randomly selected at each site, and up to five eligible children ages 6-12 years were selected from each cluster. Eligible children were defined as follows: orphans were those children for whom one or both parents had died [1] and an abandoned child was one whose parents had left with no expectation of return. Eligible children were randomly selected from available lists or through a house-to-house census. One child per household was selected to participate in the study. For households with multiple age-eligible children, the child whose name started with the earliest letter in the alphabet was selected to participate. Additionally, each site enrolled 50 community-based children who were not orphaned or abandoned at baseline as a comparison group. --- Data collection protocol As previously published [22,26], the following describes the procedures of data collection relevant to this analysis. Children and each of their self-identified primary caregivers were contacted and interviewed twice per year for up to 3 years. Baseline and annual follow up surveys collected data on numerous characteristics including the child's exposure to traumatic events, symptoms of emotional and behavioral difficulties, cognitive development, and educational attainment. Additionally, caregivers reported on household socioeconomic characteristics. Ethical approval was obtained from the Duke University Institutional Review Board (IRB) and from local and national IRB's in each participating country. The primary measures utilized in this study include child self-reports of emotional difficulties, tests assessing cognitive abilities, and child and caregiver reports of exposure to potentially traumatic events, all reported at baseline, and at 12-month, 24-month and 36-month follow ups. The measures used for trauma, emotional difficulties and cognitive development were previously validated for use across cultures (see below) and field-tested using pilot interviews. --- Study measures --- Measuring cognitive development: kaufman assessment battery for children and the california verbal learning test The KABC-II is an individually administered test of intelligence and achievement that was developed with the intention of "building in sensitivity to preschoolers, minorities and exceptional populations" [28]. Three nonverbal subtests from the Second Edition of the Kaufman Assessment Battery for Children (KABC-II), Hand Movements, Triangles and Pattern Reasoning, were assessed at annual child interviews in the POFO study. The KABC was chosen because it is one of the most frequently used tests of learning ability internationally. The three non-verbal subtests were chosen to be used across the five countries as they are less dependent on language differences. The California Verbal Learning Test (CVLT-C) is a test of verbal memory, used by POFO researchers as an indicator of memory, attention and motivation [15]. POFO interviewers modified the memory list, referred to here as the Market List (ML). Locally relevant lists were developed at each site that contained 15 items a child might see at their respective markets. The test required children to encode and store information in order to repeat back what was read to them. The Market List was chosen based on observed variability of children's engagement with the tests during pilot work in East Africa, suggesting that a tool that reflects motivation and attention would be an important addition to the learning tasks on the KABC-II [15]. Previous analyses by POFO researchers validated these tests as measures of learning and performance for children living in LMIC [15]. The findings of this previous analysis provided support that across the five countries, the subtests functioned as one would expect measures of learning to function, that is, raw scores increased with chronological age. These tests were also strongly associated with years in school for age. Hence, the KABC II scores used here can be seen as an effective tool for measuring learning, which also reflects experience in the learning environment [15]. This analysis used the highest of the three KABC test standard scores (called topscore) for each child at each round as the primary outcome measure for cognitive development. This measure represents the best the child was able to do across the three subtests when tested by the interviewer. Standard scores of the KABC II range between 0 and 19, and each subtest has a mean of 10 and a standard deviation of 3 [28]. In this analysis, scores were scaled to US age standards to enable comparison across children and these five settings. The average number of items recalled in the first three repetitions on the Market List was used as an ancillary measure of learning, attention, and motivation. --- Measuring psychosocial well-being and emotional difficulties: the strengths and difficulties questionnaire The Strengths and Difficulties Questionnaire (SDQ) is a behavioral screening tool, designed for children ages 4-16, that measures psychosocial well-being across five dimensions: (1) emotional symptoms, (2) conduct problems, (3) hyperactivity/inattention, (4) peer relationship problems, and (5) prosocial behavior. Each subscale has 5 items, scored on a 3-point Likert scale (0-2). The four difficulties subscales add up to a Total Difficulties Score, while the fifth subscale provides assessment of prosocial behavior. POFO researchers chose the Strengths and Difficulties Questionnaire "for its brevity, its psychometric properties, and its frequent use in other international studies" [22]. The questionnaire can be completed in two versions, either by parents, teachers or caregiver report, or, for children ages 11 and older, by self-report [29]. With scoring from 0-2 on each individual item, the Total Difficulties scale ranges from 0 -40. This analysis used the Total Difficulties score as a continuous variable, rather than using a clinical cutoff, which is not available across these sites. The validity of the selfreported Total Difficulties scale has been assessed and confirmed in multiple contexts (Cronbach's alpha ranging from 0.73-0.89), indicating that the scale itself is internally valid [1]. In the POFO sample, Cronbach's alpha was 0.73. These analyses used the Total Difficulties Score selfreported by the child as the primary measure of emotional difficulties. Limiting the SDQ self report to ages 11 and older is in line with the recommendation of the SDQ b. --- Measuring adverse childhood experiences: the life events checklist This analysis used the Life Events Checklist, first created by the National Center for Posttraumatic Stress Disorder (PTSD) to aid in the diagnosis of post traumatic symptoms [30]. This checklist, which inquires about exposure to potentially traumatic events such as natural disasters, witnessing someone being hurt or killed, experiencing physical or sexual abuse, or being forced to leave home, is one of the most commonly used research instruments to evaluate exposure to trauma across countries and cultures [31]. Caregivers and children were independently asked at each interview whether the child had ever witnessed or experienced each of 21 types of events. A child was counted as having experienced an event if either the caregiver or the child reported it. As described previously, four categories of events were excluded from this analysis [22]. c A cumulative traumatic exposure variable was generated for this analysis, which sums the total count of up to 17 different traumatic event categories reported through any given round. Additional covariates: household wealth, caregiver illiteracy, and relationship to the child An asset checklist and other elements from the Demographic and Health Surveys (DHS) d of each site were used to derive a wealth index score for each participating household [32][33][34][35][36]. Wealth index scores are continuous, standardized for comparability with wealth index scores in each country's DHS, and indicate greater affluence as the score increases. Caregiver illiteracy was assessed based on a literacy test administered at the time of each survey. Caregivers unable to read four short sentences in the local language were classified as illiterate. The child's relationship to the caregiver (parent versus nonparent) and orphan status (single or double orphan v. abandoned) were included in the analysis. --- Analyses A linear regression model was estimated to describe the relationship between emotional difficulties (SDQ Total Difficulties score) and various explanatory variables, including orphan status, exposure to potentially traumatic events, and household wealth. Additional linear regression models with the KABC topscore and Market lists as dependent variables were used to estimate the association between emotional difficulties and cognitive development. These models controlled for age, gender, orphan status, wealth and caregiver illiteracy, and the number of prior administrations of the KABC test to account for child learning over time. Models analyzed up to four time points for each child cross-sectionally; each model specification was run separately by site and jointly for all sites. Child-level fixed effects models were estimated to describe the relationship between the SDQ Total Difficulties score and cognitive outcomes while controlling for time invariant characteristics of children that may affect outcomes. Models were run jointly across all sites and controlled for age as the only other observed time varying characteristics expected to be associated with the child's cognitive development during the study period. Additional models, run for sensitivity analysis, evaluated whether the association between emotional difficulties and cognitive outcomes differed by caregiver type (parent versus non-parent), OAC status, or study site. Effect estimates for subgroups were calculated as linear combinations of SDQ main effects and interactions with the respective indicator variables for each subgroup. Two additional fixed effects model analyzed the association of cognitive outcomes with the four SDQ subscales (which comprise the SDQ Total Difficulties score) and with caregiver reports of the SDQ. All models were estimated with robust standard errors to account for error correlations within sites and between multiple observations from each child. Child-level fixed effects models accounted for clustering at the level of the child. Weights were constructed to account for differences in the number of children and their age and gender distributions across study sites and were used in all models. --- Attrition To evaluate the extent to which attrition may have biased our estimates, bivariable logistic regression models of baseline characteristics analyzed whether children who left the study differed significantly from those who stayed. --- Results --- Descriptive statistics Table 1 shows descriptive statistics at baseline, including proportion and frequency of OAC status, frequency of caregiver type (parent versus non-parent) and the mean and standard deviation of performance on the KABC II, child self-reported emotional difficulties, mean count of exposures to types of potentially traumatic events, and other characteristics. Single orphans constitute the largest group within the sample (60.6%). While the average topscore at baseline is similar across sites (ranging from 6.6 to 7.6), there is more variation by site on the average level of self-reported emotional difficulties (range 5.3 to 14.2), with Cambodia and Hyderabad reporting the highest average scores. The average self-reported Total Difficulties score was 10.2 for the entire sample at baseline. On average, children had experienced 1.7 types of potentially traumatic events, in addition to their orphaning or abandonment. Nagaland and Kenya had the lowest average levels of reported exposure to potentially traumatic events at baseline. was negatively associated with higher levels of emotional difficulties in Nagaland. In two of the six sites, Kenya and Tanzania, exposure to potentially traumatic events was significantly and positively associated with higher levels of emotional difficulties. There were mixed results for orphan status. Being a single orphan was associated with higher emotional difficulties relative to non-orphans in 2 sites, and being a double orphan was associated with higher emotional difficulties in one site. However orphaned and abandoned children in Nagaland reported lower rates of emotional difficulties relative to comparison children. While the wealth index, exposure to potentially traumatic events, gender and age were statistically significant when looking jointly across "All Sites", no orphan status held significant associations with emotional difficulties in this specification. --- Predicting emotional difficulties --- Predicting cognitive development Tables 3 and4 show ordinary least squares (OLS) and fixed effects regressions of emotional difficulties predicting cognitive development. Since exposure to potentially traumatic events was highly correlated with emotional difficulties (rho = 0.33; also see Table 2), exposure to trauma was not included in these models. There was a significant negative association between emotional difficulties and the child's topscore on the KABC II tests in five of six study sites (Table 3). There is no evidence that orphan status holds a significant relationship to the child's topscore. Wealth is positive and significant in Cambodia and Kenya. Caregiver illiteracy is associated with lower performance on the KABC II tests, with significant associations in Kenya and Tanzania. Female gender was associated with lower KABC II scores in Cambodia and Kenya. Table 4 shows the results of a fixed effects model of the relationship between emotional difficulties and cognitive development, thus controlling for all time invariant characteristics within one child. Since most covariates in previous models are considered time invariant, age is the only additional covariate included in the model. The relationship between emotional difficulties and cognitive development remains negative and significant in the fixed effects estimation (Table 4). In sensitivity analysis, the relationship also holds for three out of the four subscales of the self-reported SDQ (Emotions, Hyperactivity and Peer Relations) as well as the caregiver reported SDQ (Model 3). The negative and significant relationship also holds for most subgroups evaluated in the sensitivity analysis. The relationship between emotional difficulties and The same models as in Tables 3 and4 were estimated for the CVLT-T Market List to test whether associations held across multiple measures of learning. The correlation between the KABC II topscore and the CVLT-T Market List is 0.373, indicating that while there is some shared variance, the two indices still likely capture different aspects of the child's cognitive development. The relationship between emotional difficulties and scores on the Market List test was negative and significant in 3 of 6 sites for the OLS model. The relationship was also negative and significant across "All Sites" in the fixed effects model, with similar results for the sensitivity analysis as were found with the topscore measure. Results are presented in Tables 5 and6. --- Attrition and missing data Attrition differed significantly across study sites (p-value 0.00) and children living with parents (single or nonorphan) were less likely to drop out (p-value 0.046). Results in Tables 4 and6 suggest that the observed associations do not differ between children living with parents and non-parents, thus it is unlikely that differential attrition between children residing with biological parents versus other children substantively biased our estimates. Further, the direction of the observed associations was similar in most study sites. There were no differences in rates of attrition across other relevant factors such as exposure to potentially traumatic events, cognitive development, socioeconomic status, gender, or the level of emotional difficulties. Using the child self-report measure of the Total Difficulties scale resulted in a number of missing data points at baseline, since children did not self-report on the SDQ until they were 11 years old. To check whether missing data of this nature might change results, crosssectional analysis was restricted to the 36-month follow up (when most children (84.3%) were old enough to self-report) controlling for baseline characteristics, with little change in results. Given that a) the caregivers of nearly half of all participating children were not their biological parents, b) caregivers may significantly underreport the behavioral effects of traumatic events experienced by children [37], and c) caregivers had been taking care of these children for variable amounts of time, the child self-reported difficulties were considered more accurate than caregiver reports. --- Discussion This is the first cross-country study to examine the relationship between orphans' emotional difficulties and their learning. Moreover, conducting such an analysis on a unique sample of vulnerable children who are susceptible to multiple instances of adversity, including trauma, provides evidence of the nature of this relationship within a context of heightened adversity and potentially higher risk of mental health difficulties. This analysis showed that OAC exposure to potentially traumatic events is an important predictor of emotional difficulties, thus supporting previous literature that the number of adverse events matters for risk of mental health difficulties. The most salient finding was a negative and significant relationship between a child's emotional difficulties and his/her cognitive development, within and across five of six study sites. Rates of exposure to adverse childhood events among the orphaned and abandoned children in this study were high, and exposure to adverse events, in addition to male gender, and lower household wealth, was associated with significantly higher rates of emotional difficulties. Interestingly, orphan status was only significantly associated with emotional difficulties in some sites and was not predictive of cognitive development, although the negative and significant relationship between emotional difficulties and cognitive development held within each orphan subgroup. These findings may suggest that when a child is more vulnerable to a number of adverse events, it is the context in which a child loses a parent, rather than the loss of a parent alone, that better explains mental health outcomes. Children who are vulnerable from orphaning and have experienced other adversity are at heightened risk of increased emotional difficulties, which is associated with lags in cognitive development. These findings offer insights into the relationship between vulnerable children's mental health and their ability to participate in and benefit from education. In the context of multiple adversity, the loss of a parent, or orphan status, does not show statistical significance in the hypothesized relationship between mental health and learning. Nevertheless, the overall findings still underscore the importance of mental health intervention for children who are especially vulnerable. It is likely that the effect of parental loss is further affected by the additional adverse events to which the child is even more vulnerable. No trauma or potentially traumatic event happens in isolation. Although causality cannot be inferred from this analysis, the results offer insight into the relationship between emotional difficulties and educational outcomes. The fixed effects model controls for unobserved time-invariant differences between children that may be correlated with emotional difficulties and may influence cognitive development. Consequently, the results in Table 4 give us an unbiased estimate of the relationship between a child's emotional difficulties and his/her cognitive development, controlling for these differences. The consistency of the association between these two factors across multiple contexts and measures and within different subgroups is striking. This underscores the importance of providing trauma support and focusing on the psychosocial development of vulnerable children as a means to alleviate strains inhibiting a child's learning. Additionally, the associations found between wealth and emotional difficulties may indicate that interventions aimed at easing households' resource constraints could help alleviate emotional difficulties in some contexts. One may hypothesize that in less stable, resource poor settings, children who were exposed to trauma may be less likely to recover from emotional difficulties. Interventions on both the child and family levels may work in tandem to improve education outcomes for orphans. If policies and programs can improve economic and emotional stability for both children and families, orphans may have a greater chance of pushing past the challenges of losing a parent and the additional traumas to which orphanhood exposes them. While economic support does not treat mental health difficulties, it may provide families with the resources needed to keep children in school, which may aide their psychosocial development. Such support may improve a child's chances of overcoming the emotional challenges that are associated with lags in cognitive development. We acknowledge several limitations. Children's cognitive development is described by only two measures, an aggregated score summarizing children's performance on three nonverbal KABC II subtests plus a measure verbal memory and attention based on the CVLT. While ability is likely not to change over time, the child's performance on the KABC II tests is not only an indication of the child's ability, but also a variety of other factors associated with motivation, self-confidence, response to authority and the child's developed non-verbal skillset [1]. We may interpret the association of test scores with the child's emotional difficulties to be related to changes in these additional factors. While this measure may not singularly isolate the child's cognitive development, knowledge of the critical role of emotional difficulties in the child's learning and subsequent academic attainment is an important insight for policies on global support for vulnerable children, including but not limited to those vulnerable due to parental death. Further, the available data do not offer adequate statistical power to describe the temporal association between the variables of interest, and therefore cannot rule out possibilities of reverse causality. It is possible that the observed relationships are circular and that, even if higher emotional difficulties impede cognitive development, lags in cognitive development in turn may heighten emotional difficulties. With adversity, psychosocial manifestations, and cognitive delays spanning much of the life span of the children in this study, the three year follow-up period does not offer sufficient within-child variation to disentangle these effects. Regardless of the direction of causality, if we know that children who are exposed to more trauma also report more emotional difficulties and perform lower on tests of cognitive ability, interventions to provide emotional support for children living in adversity will help at least one or more of these difficulties. Even though we do not know whether emotional difficulties are driving lags in cognition or the other way around, we still need to address the emotional difficulties, specifically targeting potentially traumatic events. Finally, the study findings may not be generalizable, as there are many other groups of OAC, including street children, institution-based orphans, and those that live in countries with widely different contexts than those in our sample. These associations may differ in those samples. The variability in the observed associations across the six study sites suggests that associations between trauma, emotional difficulties and cognitive development, and appropriate interventions, are likely context-specific. Nevertheless, this analysis offers new insight into the relationship between psychosocial factors, exposure to adverse childhood experiences, resource constraints and the cognitive development of orphans in wide variety of contexts and diverse settings in Southeast Asia and sub-Saharan Africa. --- Conclusions This study suggests that increased reports of exposure to potentially traumatic events among orphans and abandoned children are associated with higher emotional difficulties, and increases in emotional difficulties are associated with lags in cognitive development. Hence, exposure to trauma and emotional difficulties comprise important barriers to educational attainment for all such vulnerable children, including orphans. Higher socioeconomic status and better educated caregivers may offer buffers to these difficulties, since they are associated with fewer emotional difficulties and higher performance on tests of cognitive development. Interventions targeting both the psychosocial development of the child, vis a vis their exposure to adverse childhood events, and the socioeconomic status of the caregiver may work in tandem to improve educational outcomes for vulnerable children in a more holistic sense. Further, family based interventions to stabilize socioeconomic conditions or increase caregiver education may help overcome psychosocial challenges that otherwise would present as barriers to the child's educational advancement. Most importantly, psychosocial status may be an important actor in a child's ability to profit from and stay in school, and one factor that influences this status is a child's exposure to traumatic and other adverse experiences. These findings may provide a guide to developmental strategies for those working to improve education outcomes for children in less wealthy areas. --- Ethical approval Ethical approval was provided by the Duke Medicine Institutional Review Board (IRB); the local review boards of SaveLives Ethiopia and Stand for Vulnerable Organization (Addis Ababa, Ethiopia), Sharan (Delhi, India), and Kilimanjaro Christian Medical Centre (Moshi, Tanzania); and national regulatory agencies in all participating countries including the National Ethics Committee for Health Research (Cambodia), the Ministry of Science and Technology (Ethiopia), the Indian Council of Medical Research (India), the Kenya Medical Research Institute (KEMRI), and the National Institute for Medical Research (Tanzania). --- Informed consent Children under the age of 18 were asked for assent to participate in the study only after the guardian had first given permission for the child to participate, and for the interviewer to speak with the child. The project was explained to the child in an age-appropriate manner, and the child was given the opportunity to ask questions about the project before assent was requested. Children who were not competent (who had moderate to severe learning or mental health disorders) to give assent were not included in the study. Children who previously gave assent as minors were re-consented as adults when they reached age 18. --- Endnotes a While the definition of "orphan" varies across cultures and settings, UNICEF provides a commonly accepted definition for orphan as "a child who has lost one or both parents" (UNICEF 2004). b For more information, see sdqinfo.org. c "Hearing about a family member who has died", "had a brother or sister die", "seeing a dead body in town", and "having a painful or scary medical treatment," were excluded from the trauma count variable. Almost all children had lost a family member, hence the first event did not add to the variation in the count. For the second event, it was not clear whether the child witnessed the event themselves. For the third event, in some sites, burials are open-casket so all kids had seen these events. For the latter event, event description was not specific enough. It seemed that for some children, this included blood draws or like events that were scary to the child but not comparable to the other categories. d Data was compiled from a variety of country specific DHS data sets. See references for further information on data used. e Constants in linear regression models represent the estimated mean value for the (hypothetical) reference group for whom the values of all variables in the model are zero. --- Competing interests The authors declare that they have no competing interests. Authors' contributions ME conceived of the analysis, analyzed the dataset, conducted the literature review and wrote the manuscript. KW assisted with conceptualization of the analysis models and interpretation of the results, and provided manuscript feedback. JO assisted with conceptualization and implementation of the analysis, data analysis, interpretation of the results, and provided manuscript feedback. KOD selected the measures of emotional difficulties and learning and assisted with study implementation and interpretation of the results. All authors read and approved the final manuscript.

Background: Development policymakers and child-care service providers are committed to improving the educational opportunities of the 153 million orphans worldwide. Nevertheless, the relationship between orphanhood and education outcomes is not well understood. Varying factors associated with differential educational attainment leave policymakers uncertain where to intervene. This study examines the relationship between psychosocial well-being and cognitive development in a cohort of orphans and abandoned children (OAC) relative to non-OAC in five low and middle income countries (LMICs) to understand better what factors are associated with success in learning for these children. Methods: Positive Outcomes for Orphans (POFO) is a longitudinal study, following a cohort of single and double OAC in institutional and community-based settings in five LMICs in Southeast Asia and sub-Saharan Africa: Cambodia, Ethiopia, India, Kenya, and Tanzania. Employing two-stage random sampling survey methodology to identify representative samples of OAC in six sites, the POFO study aimed to better understand factors associated with child well-being. Using cross-sectional and child-level fixed effects regression analyses on 1,480 community based OAC and a comparison sample of non-OAC, this manuscript examines associations between emotional difficulties, cognitive development, and a variety of possible co-factors, including potentially traumatic events.The most salient finding is that increases in emotional difficulties are associated with lags in cognitive development for two separate measures of learning within and across multiple study sites. Exposure to potentially traumatic events, male gender, and lower socio-economic status are associated with more reported emotional difficultiesin some sites. Being female and having an illiterate caregiver is associated with lower performance on cognitive development tests in some sites, while greater wealth is associated with higher performance. There is no significant association between orphan status per se and cognitive development, though the negative and significant association between higher emotional difficulties and lags in cognitive development hold across all orphan subgroups. Conclusions: These findings suggest that interventions targeting psychosocial support for vulnerable children, especially vis a vis traumatic experiences, may ease strains inhibiting a child's learning. Family based interventions to stabilize socioeconomic conditions may help overcome psychosocial challenges that otherwise would present as barriers to the child's learning.

Introduction Violence against adolescent girls and young women (AGYW) is a global phenomenon and has implications on their wellbeing. According to the World Health Organization (1) more than a third of women globally have experienced either physical or sexual violence or both from an intimate partner or non-partner in their lifetime. It has also been reported that about one in three ever-partnered AGYW aged 15-24 years have experienced physical and/or sexual violence by an intimate partner [1]. In sub-Saharan Africa, about 37 percent of ever-partnered women have experienced physical and/or sexual violence [1]. A report by the United Nations Children's Fund Kenya County Office indicated that in 2010 about 76 percent of AGYW aged between 18 and 24 years had experienced sexual, physical or emotional violence prior to age 18 years while about half of AGYW aged 13 to 17 years had experienced sexual or physical violence in the 12 months preceding the study [2]. Exposure to violence has deleterious effects on AGYWs' health outcomes, including their mental, sexual and reproductive as well as social and economic wellbeing. It is one of the leading causes of death in this age group globally [3][4][5][6]. Kabiru and colleagues found that adolescent girls aged 11-15 years who had experienced violence had lower expectations of achieving their aspirations compared to those who had not experienced violence [5]. A study conducted in Malawi and South Africa showed that exposure to violence has an impact on school enrolment and performance [7]. Studies looking at risk factors for violence have found that women's economic status, and norms that place women's status in society at a lower position have a significant association with violence experience [4,8]. In their review of risk factors for AGYW's experience of violence in romantic relationships, Vezina and Hebert (4) also found that adolescent mothers and those who had dropped out of school were at a greater risk of experiencing dating violence. Having witnessed violence has been found to be a risk factor for violence experience [4,9,10]. Vezina and Hebert (4) argue that having witnessed violence as a child may contribute to perceiving violence as an acceptable way of resolving conflict. This is in line with the Social Learning Theory which postulates that individuals model behavior based on what they observe from their environments/contexts [11]. Living in urban informal settlements has been found to heighten risk of experiencing violence. Kabiru,Mumah (5) found that about one in three girls aged 11-15 years in urban informal settlements of Nairobi has experienced at least one form of gender-based violence. Urban informal settlements have been associated with high levels of poverty, crime and violence, poor health outcomes-including HIV/AIDS-as well as poor access to basic services such as schools and health care facilities. Past studies have found strong association between violence experience and poverty at the household [9,12] and community levels [9,13,14], a major characteristic of the urban informal settlements. Exposure to violence has been found to be associated with HIV acquisition [15], while HIV-positive status is thought to provoke violence in some contexts [8,16,17]. In their review, Campbell et al. argue that increased risk for HIV acquisition works through increased sexual risk-taking behaviors, forced sex with an infected partner, and compromised negotiation on safe sex practices [18]. AGYW who have been exposed to violence may have low negotiation powers, thus compromised safe sex practices. In the slum context, the risk of HIV and occurrence of interpersonal violence risk seem to be heightened. While recent data are sparse, a study conducted in 2007 showed that the HIV prevalence was higher at about 12% among slum residents compared to 5% and 6% among non-slum urban and rural residents, respectively, and this mirrored the national trend where the burden was higher among females [19]. Impersonal violence including violence against women is common slums compared to the general Kenyan population [20]. The HIV epidemic in sub-Saharan Africa cannot be brought under control without reducing HIV acquisition among AGYW, the most rapidly expanding demographic group on the continent [8]. The DREAMS Partnership is an initiative aimed at reducing the incidence of HIV among AGYW in 10 sub-Saharan African countries. It supports a core package of interventions provided "at scale" targeting AGYW, their families, wider communities, including men who are the sexual partners of AGYW [21]. The core package includes interventions aimed at addressing HIV risk behaviors, HIV transmission, socio-economic vulnerabilities and gender-based violence. The interventions are aimed at empowering girls and young women to reduce their risk of HIV infection, and include SASA! (Start, Awareness, Support, Action) and school-based HIV and violence prevention programs. The core package interventions are described in detail elsewhere [21]. Nairobi, Kenya, is among the 4 sites chosen for an independent impact evaluation [22]. Using enrolment interview data from AGYW cohorts established for the impact evaluation, this analysis aims to summarize the prevalence of experience of violence, the severity of physical violence, and to identify factors associated with experience of violence among AGYW living in the Korogocho and Viwandani informal settlements of Nairobi, within the first year of DREAMS implementation. We also sought to determine whether DREAMS is reaching those who experience violence, by comparing those who were and were not invited to participate in DREAMS. --- Methods --- Study design, setting and sample The DREAMS impact evaluation design and data collection protocol have been described in detail elsewhere [22]. Briefly, the design leverages the Nairobi Urban Health and Demographic Surveillance System (NUHDSS), a longitudinal platform run by the African Population and Health Research Center (APHRC) in two informal settlements of Korogocho and Viwandani since 2002 [23]. Both informal settlements are characterized by high levels of unemployment, sub-standard housing and crowding, limited access to education and other social services, high levels of insecurity, and inadequate water and sanitation infrastructure. Whereas Korogocho is a more settled community with many long term residents, the population in Viwandani is more mobile and youthful [23]. In spite of this known fact we took measures (such as making multiple visits, altering time of visits, and re-assigning more experienced field staff) to reach as many sample participants as possible. We used different interview tools for the age groups 10-14 years and 15-22 years, as DREAMS had a different primary package for each age group, but also some of the questions in the tool for the former group could not be asked to girls in the latter group as they were inappropriate given their age. The tool for the 10-14 years was adopted from the Global Early Adolescent Study (GEAS) tool, and has been validated for this particular age group in our setting. For the 15-22 years, the World Health Organization violence against women (WHO VAW) instrument [24] was included to measure exposure to violence. In this paper, we used round one (enrolment) data collected March-July 2017 on a randomly selected sample of AGYW aged 10-22 years. For the 10-14 year-olds, a random list of 1017 girls was generated. Of these, 333 (32.7%) were no longer eligible at the time of visit. Of the remaining 684, 46 (6.7%) were absent for extended period of time, 9 (1.3%) either refused or their parents refused their participation, and 23 (3.4%) had their structures located but respondents' whereabouts were unknown, leaving 606 (88.6%) with successful interviews. At enrolment, we targeted a minimum sample of 1000 AGYW aged 15-22 years. A random list of 2599 AGYW was generated. Of these, 695 (26.7%) were no longer eligible at the time of visit. Of the remaining 1904, 6 (0.31%) had incomplete interviews, 315 (16.5%) were absent from their residence for extended period of time, 283 (14.9%) refused to participate either by self or their parents/guardians, and 219 (11.5%) had their structures located but respondents' whereabouts were unknown, leaving 1081 (56.8%) with successful interviews. Our assessment of how those who participated compared to those who did not showed that there were no major differences by several sociodemographic characteristics indicating that there was unlikely to have been selection bias. --- Measures --- Outcome variables Among girls aged 10-14 years, experience of violence was measured using the questions listed in Box 1. The questions were of two types: 1) Whether the girl has "ever experienced..."; and 2) whether the girl "experienced in the last six months...". They were classified into psychological violence (questions 1, 2, and 6), sexual violence (questions 3, 4 and 5), and physical violence (question 7). A girl was considered to have experienced violence if she gave any of the answers "sometimes" (= 1), "often" (= 2) for questions 1-4; or "yes" (= 1) for question 5; or "yes, by both boy and girl" (= 1), "yes, by boy" (= 2), or "yes, by girl" (= 3) for questions 6 and 7. As these questions were not similar as those used for AGYW aged 15-22 years, the data for this group were analyzed separately. Consequently, the results for the two age categories are not immediately comparable. Among the AGYW aged 15-22 years, experience of violence was measured using 15 questions on a binary scale of "yes" (= 1) or "no" (= 0). The questions read like: "Has any male ever done any of the following things to you in the past 12 months?" (see Box 2). The tool has shown good psychometric properties [24][25][26][27][28][29]. The WHO classified these items into three dimensions of violence [24]: psychological violence (items 1 to 3; Cronbach's alpha = 0.67), Box 1. The items used to measure experience of violence among the early adolescent girls 10-14 years old Questions 1-4 had the answer categories "Never" (= 0), "Sometimes" (= 1), "Often" (= 2), "Don't know" (= 999), "Refused" (= 996); Question 5 had the answers "No" (= 0), "Yes" (= 1), "Can't remember" (= 998), "Don't know" (= 999), "Refused to answer" (= 996); Questions 6 and 7 had answer options of "No" (= 0), "Yes, by both boys and girls" (= 1), "Yes, by boys" (= 2), "Yes, by girl or girls" (= 3), "Cannot remember" (= 998), "Don't know" (= 999), "Refused" (= 996). --- Sexual violence 12. Touched you in a sexual way (e.g. kissing, grabbing, or fondling), when you did not want them to? 13. Try to have sexual intercourse with you when you did not want to but did not succeed? 14. Physically forced you to have sexual intercourse even when you did not want to? 15. Forced you to perform sexual acts when you did not want to? The answer options were "yes" (= 1) or "no" (= 0) https://doi.org/pone.0231737.t002 physical violence (items 4 to 11; Cronbach's alpha = 0.77), and sexual violence (items 12 to 15; Cronbach's alpha = 0.79). --- Explanatory variables Explanatory variables included self-reported invitation to participate in DREAMS, slum of residence (site), age at survey, marital status, whether the girl is currently in school, educational level, religion, ethnicity (i.e., mainly Somali, Kisii, Kamba, Kikuyu, Luhya, and Luo), recent employment/engagement in income generating activity, ever had sex, ever been pregnant, ever given birth, slept hungry at night in past 4 weeks, self-assessed household economic situation, and wealth index. For ethnicity, categories with low frequencies were grouped under "other. Wealth index was constructed using principle component analysis (PCA) with input as indicator variables on ownership of household and individual assets/items (including ownership of television, electricity, fridge, radio, bicycle, motorcycle, shoes, blanket, clothes, among others), household structure (i.e., floor, roof, and wall material), and on household's water supply and sanitation [30,31]. The wealth index was split into three categories of "poor" (= 1), "medium" (= 2), and "wealthy" (= 3). The variables 'ever been pregnant' and 'ever given birth' were excluded from the present analysis because of multicollinearity (see S1 Table ). Data on the gender of the teacher were not collected for the 15-22 year-olds. Data were collected electronically using face-to-face interviews by well trained and experienced field interviewers and supervisors who were also well conversant with the study area. For the 10-14 years cohort, only selected female interviewers surveyed them to minimize potential response bias. For the 15-24 year-old cohort, enumerators were both males and females with mean age 27 years and 24 years, respectively. The mix of gender among the interviewers catered for any respondent that may have preferred to be interviewed by a particular gender. Prior to the survey, tools were translated into Kiswahili only, and were back translated to ensure the questions did not lose their meaning; the tools were piloted and issues that arose were addressed. --- Analysis To estimate the prevalence of violence, we obtained, for each domain, the proportion of AGYW who reported to have experienced at least one of the acts of violence comprising that domain. These proportions were summarized with respect to important demographic characteristics. We also compared those who were and were not invited to participate in DREAMS. To assess the severity of violence, a woman/girl was considered to have experienced "moderate" violence if she answered "yes" to one or more of questions 4-5 in Box 2 (and does not answer "yes" to questions 6-11). A woman/girl was considered to have experienced "severe" violence if she answered "yes" to one or more of questions 6 to 11 [24]. We note here that this could only be done for physical violence (and among the 15-22 year olds only) as WHO guidelines on violence classification as moderate or severe only exist for physical violence, to our best knowledge [24,25]. However, in another study [1], all acts of sexual violence were considered severe but it is not clear what items comprised sexual violence domain. We return to this in the Discussion section. To investigate factors associated with violence, we considered psychological and sexual violence outcomes as dichotomous, that is, "not abused" (= 0) or "abused" (= 1). Physical violence was considered ordinal with three levels of "none" (= 0), "moderate" (= 1), and "severe" (= 2), as described above. Ordinal regression analysis was used [32]. For each of the three violence outcomes, we evaluated three plausible link functions, that is, logit, probit, and complementary log-log (clog-log). Based on the log-likelihood values and practical considerations, the logit link was chosen (see Summary of model fit evaluation in Supporting Information, S1 Text). Thus, all inferences presented in the next section were based on models with a logit link. We adopted a three-step approach to the analyses. First, a model was run with one explanatory variable at a time ('model 1'). Next, a model was fitted for explanatory variable adjusted for Invitation to DREAMS, site, and age ('model 2'). Finally, all explanatory variables significant at p 0.10 in the second step (i.e., in model 2) were included in a multivariable model (model 3). Explanatory variables were tested in the multivariable model using likelihood ratio test (LRT) at p 0.05 significance level. Invitation to DREAMS, site, and age were retained in the multivariable model even if they were not significant, as we wished to adjust for their effect. Data management and all analyses were performed using STATA v14 (StataCorp, College Station, TX). Computational details and model fit evaluation are presented in Supporting Information, S1 Text. --- Ethical considerations Ethics approval was provided by the London School of Hygiene & Tropical Medicine (LSHTM; Ref 211 11835) and AMREF Health Africa (No ESRC P298/2016). Written informed consent was obtained from each participant. For legal minors, assent was obtained from the minor before the guardian gave consent. The participants were given copies of the consent documents and project information sheet. The interviewer took them through the information sheet and consent form and gave them an opportunity to ask questions. Once they indicated to have understood about the project and agreed to participate, they were requested to sign the certificate of consent. Incase the respondents were not able to write they had their thumb print on the consent form (round 1), or put a mark on the soft copy of the consent (rounds 2 and 3) and the field interviewer noted on the comments section that the respondent was unable to write. --- Results --- Demographics of the adolescent girls and young women Data were available on 606 (10-14years) and 1081 (15-22 years) AGYW. The mean (median) age for the 10-14 year olds was 12.1 years (12 years), and that of the 15-22 year olds was 17.9 years (17.0 years). Of 606 girls aged 10-14 years, 53% (n = 323) were from Korogocho and 47% (n = 283) were from Viwandani. Five of the girls were not enrolled in school (<unk>1%). The majority were Christians (88.1%, n = 534), 10% were Muslims. They were of different ethnic origins including Kikuyu (32.2%, n = 195), Luo (19%, n = 115), Kamba (15.8%, n = 96), Luhya (15%, n = 91), Somali (9.2%, n = 56), and Kisii (5.6%, n = 34). About 5% (n = 29) had done chores or activities for which they got paid money over the past 6 months (e.g., worked for neighbors or friends, day labor or temporary work, worked for family (such as parents or relatives), providing services, among others). Of 1081 AGYW aged 15-22 years, about half (n = 536) reported to have ever been invited to participate in DREAMS. The majority had never been married (78%; 843/1081), were in school (57.8%; 625/1081), were Christians (84.8%; 917/1081), had never engaged in an income generating activity (71.8%; 776/1081), had had sex (59.4%; 642/1081), and assessed their household economic situation as moderately poor. The majority of AGYW were from the Kikuyu (29.5%, n = 319) and Kamba (19.2%, n = 208) communities, followed by Luos (16.3%, n = 176) and Luhyas (16.3%, n = 176), Somalis (8.3%, n = 90), and Kisiis (4.7%, n = 51). About nine in ten AGYW reported to be knowing their HIV statuses. --- Prevalence among 10-14 year olds Table 1 shows, for girls aged 10-14 years, the proportions that experienced violence in the past 6 months or ever experienced violence by demographic characteristics. Overall, psychological violence was the most prevalent, both within the past 6 months (32.8%) and life time (54%), followed by physical violence (16.3%), and sexual (7.1%). Violence was generally less prevalent among those invited to DREAMS, currently enrolled in school, did not engage in chores or activities for payment during the past 6 months, had never had sex, or family had enough food. S2 Table shows that the proportions that experienced any form of violence in the past 6 months was 37.8%, while in life time was 57.4%. --- Predictors of violence among 10-14 year-olds S4 Table summarizes predictors for violence among girls of ages 10-14 years. After adjusting for other factors, the odds of experiencing any violence were greater among girls who engaged in chores or activities for payment in the past 6 months, and among those whose family did not have enough food due to lack of money. Violence was lower among girls invited to DREAMS, and among non-Christians. --- Prevalence of violence among the 15-22 year-olds Fig 1 shows the percentage of the 15-22 year-old cohorts with a "yes" answer to each of the 15 violence questions listed in Box 1. Item 3 ("insult you or make you feel bad about yourself") was the most experienced at 26.5%; followed by item 1 ("Say or do something to humiliate you in front of others", 17.3%); then items 2 ("Threaten to hurt or harm you or someone close to you"), 5 ("Push you, shake you, or throw something at you"), 12 ("Touched you in a sexual way (e.g. kissing, grabbing, or fondling), when you did not want them to"), and 13 ("Try to have sexual intercourse with you when you did not want to but did not succeed") at about 10%. Fig 2 shows that the prevalence of psychological violence among at the 15-22 year-olds was 33.1% (95%CI 30.4-36.0%; n = 358), followed by physical violence (22.9%, 95%CI 20.5-25.5%; n = 248), and sexual violence (15.8%, 95%CI 13.8-18.1%; n = 171), in the past 12 months. About 44% experienced at least one of the 15 acts of violence in the past 12 months. Further, some of the AGYW experienced more than one type of violence, with 6.9% (95%CI 5.56-8.62%, n = 75) experiencing all three types of violence. About 9.2% (n = 99) reported both physical and psychological, 3.33% (n = 36) reported both psychological and sexual, and 1.7% (n = 18) reported both physical and sexual violence. Tables 2-4 present the prevalence of the three violence domains by demographic characteristics. They show that within each violence measure the prevalence of violence varied across the demographic characteristics, with about 45% and 33% of those who, respectively, know and don't know own HIV status experiencing any form of violence. --- Severity of physical violence among 15-22 year-olds Of 1081 AGYW aged 15-22 years, 9.2% (95%CI 7.57-11.03%) experienced moderate physical violence, and 13.8% (95%CI 11.9-15.97%) experienced severe physical violence in the past year. S4 Table shows these proportions for several demographic characteristics. In general, severe physical violence was more prevalent than moderate physical violence levels. Across the levels of the demographic characteristics, moderate violence was higher among AGYW not invited to participate in DREAMS than those invited, resided in Korogocho compared to Viwandani, 18-22 year-olds than 15-17 year-olds, previously or currently married/living with a partner compared to those who had never married or lived with a partner, not in school, ever had sex, slept hungry, and were poor. Severe physical violence was higher among Korogocho residents compared to Viwandani residents, previously married/living with partner, not currently in school, with some secondary education or lower, Christians and other religious groups than Muslims, among Luos and Luhyas, ever had sex, slept hungry at night last 4 weeks, assessed their household economic situation as not poor, or were in the poor wealth quantile. --- Predictors of violence among 15-22 year-olds Tables 234show the logistic regression model results for psychological, physical, and sexual violence, respectively. From Table 2, AGYW who were previously married/lived with partner, or engaged in employment/income generating activity last month, or slept hungry at night during past 4 weeks had greater odds of experiencing psychological violence. Table 3 shows the odds of experiencing physical violence were lower among those who lived in Viwandani slum relative to those living in Korogocho, and among Muslims; and the odds were greater among AGYW who were previously married or lived with a partner, or slept hungry at night during the past 4 weeks. It can be seen from Table 4 that the odds of sexual violence were lower among AGYW aged 18-22 years (compared to the 15-17 year-olds) and among Muslims (compared to Christians). Sexual violence was higher among AGYW who reported to have ever had sex, or slept hungry at night during the past 4 weeks. In summary, sleeping hungry at night during past 4 weeks was found to be significantly associated with greater odds for all three forms of violence, other factors held constant. Being a Muslim was associated with lower levels of physical and sexual violence. --- Discussion This study provides data on the prevalence, levels of and determinants for violence against AGYW in slums settings of Nairobi. This is in line with and a response to a call by the WHO a decade ago urging researchers to incorporate violence against women components into HIV and AIDS prevention and adolescent health promotion programs [21]. Violence against AGYW in Korogocho and Viwandani slums is common, with about four in ten AGYW aged 10-14 years and 15-22 years reporting to have experienced violence in the past six and 12 months, respectively. In both age categories psychological violence was the most experienced, followed by physical violence, and sexual violence. These levels of violence are in line with previous studies that have shown that AGYW in Kenya are predisposed to physical, psychological and sexual violence [2,33]. In a study by Mathur et al (2018), about two in ten AGYW aged 15-24 reported sexual violence by an intimate partner in the 12 months preceding the study. More concerning, however, is our finding that at 10-14 years, about six in ten of the young girls had ever experienced violence. Our finding that having ever slept hungry in the last one month increased the odds of experiencing physical, psychological or sexual violence is similar to findings from other studies conducted in sub-Saharan Africa (SSA) and elsewhere. A study conducted in Botswana found that women who were food insecure were more likely to experience sexual violence [34]. Food insufficiency has been linked to high-risk sexual behavior and sexual vulnerability among women [35,36]. The high levels of food insecurity reported within the NUHDSS [23] are therefore likely to contribute to the high likelihood of AGYW experiencing physical, sexual and psychological violence. Food insecurity puts women at greater risk of violence through: 1) path of stress by causing hunger and worry about having sufficient access to food, which might act as a trigger to interpersonal violence; 2) making it difficult to walk out of an abusive relationship due to dependence for food; 3) engaging in transactional sex as a means of getting food; and so on [37]. Marital status was associated with experiences of physical and psychological violence. Whereas being in a current marital union reduced the likelihood of experiencing psychological violence, having been in a marital union previously increased the chances of experiencing both physical and psychological violence. The finding here that the odds of psychological abuse among formerly married women were more than twice that for single women may point to abuse as a reason for termination of such relationships. The 2014 Kenya Demographic and Health Survey shows that a greater proportion of women and men who were divorced, separated or widowed reported to have ever experienced physical, sexual and emotional violence compared with those who were in marital unions [38]. Other studies have also shown that the risk of violence increases considerably when women want to leave, are trying to leave, are in the process of leaving, or have left a relationship [39][40][41][42]. There seem to be some protective norms/cultures. Religion was found to be significantly associated with experiencing violence. Overall, AGYW who identified themselves as Muslims had lower likelihood of experiencing all three forms of violence. This finding is consistent with past research where being religious has been associated with lower likelihood of perpetrating violence [43]. The modalities of most religions encourage peaceful co-existence at individual, family and community levels and also are likely to provide support services for their congregants to resolve conflicts. However, religion may also increase the vulnerability of AGYW as it discourages dissolution of marriages and thus it may encourage a victim to stay in an abusive marriage/relationship. Nevertheless, it is not clear whether they are indeed protected or it is an issue of reporting bias. A previous study found tolerant attitudes of Muslim women towards violence, which is a portrayal of the religious restrictions they have to abide by [44], and which could contribute to under reporting. AGYW who were in employment in the past one month were more likely to experience psychological violence. The autonomy and independence that often comes with financial freedom has been identifiable as risk for abuse in other studies (UNICEF, 2007). This finding reflects social/cultural attitudes towards women's employment and may be closely associated with the belief that men hold power in household allocation of resources and decision-making on household expenditures [45]. This may imply that women who work may have little or no decision-making powers on how their income is utilized. Future studies focusing on this can provide valuable data. Findings from other studies in SSA [46,47] suggest that women form a higher proportion of people working in the informal sector, and that they experience exploitation, work for long hours, are underpaid and are engaged in other forms of work beyond their contractual agreement, factors that are likely to contribute to psychological violence. Korogocho slum is characterized by higher levels of poverty, low education attainment and violent crime. These directly or indirectly have implications on the risk of physical violence amongst AGYW which, in Viwandani, is estimated to be lower by almost 50%. The fact that the propensity of experiencing sexual violence among older AGYW was lower by almost 40%, may point to evidence from previous studies that has shown that while the perpetuators are often thought to be strangers, a lot of this is by close family members or friends who take advantage of young AGYW who are not empowered to resist or report such advances. Our finding that having had sexual intercourse increased the odds of experiencing sexual violence was expected and is in line with previous literature [48]. Sexual experience in this study includes sexual intercourse that AGYW were forced to participate in, including rape, and this could explain the increased odds of sexual violence among those who had had sex. However, we did not distinguish between intimate partner and non-partner sexual violence. We found no significant association between DREAMS invitation and any domain of violence using these first round data. This could be attributed to the fact that the first round of data collection took place when DREAMS intervention had just started. Given the staggered roll-out of interventions, we do not expect DREAMS to have had prevented violence at this early stage of the program. Second, we have no 'baseline' to measure change in violence over the course of that year. In summary, our study has found high prevalence of and have identified some protective and risk factors for violence among AGYW in two Nairobi slums. The high prevalence among the younger age group (10-14 years), in particular, calls for urgent intervention as childhood exposure to violence has been reported to be a risk factor for violence in adulthood. Our findings can help the formulation and implementation of both national and sub-national policies, budgets and actions to reduce, eliminate and mitigate the consequences of violence against women. As data used in this paper are part of an independent Impact Evaluation, the learnings from these results have not been used to influence DREAMS implementation. However, our finding of no difference in violence among those invited to the program and those not invited among AGYW aged 15-22 years raises a flag for us to pay attention to measurement details which, if okay, might point to the fact that since social change tends to take long to happen, the implementation period might need to be longer. We also note that violence seems to occur together with other forms of social vulnerability such as food insecurity, which is very important in situations where single interventions might fail to lead to expected outcomes. Within DREAMS [21], promising interventions such as gender norms training, school-based HIV and violence prevention programs, and lessons and tools from SASA! intervention in Uganda [49], exist that need to be scaled-up. In rural South Africa, the Intervention with Microfinance for AIDS and Gender Equity (IMAGE) combined a microfinance program with a gender and HIV education to reduce risk of gender based violence and HIV [15]. Other promising strategies to leverage also exist [50]. This study has some limitations. Whereas all due care was taken to ensure that the tools captured the right information, we cannot rule out the possibility of misreporting. It has been shown in other studies that to avoid feeling embarrassed, women tend to under-report physical and sexual violence, such as rape, as is often the case might be blamed on the victim [51][52][53]. The slum population is generally unique with different social challenges compared to their rural and urban non-slum subpopulations. Therefore, results drawn from this study may not accurately be generalizable to the entire population of AGYW in Kenya. Whenever data allow, there is need to make comparisons across sub-populations and over time to be able to fully understand the dynamics and general trend of occurrence of violence and possibly link it to other health and social outcomes such HIV acquisition, pregnancy outcomes and schooling outcomes among others. Another limitation is that there is no unified framework for classifying different violence domains into moderate or severe. In the present study, we have classified physical violence only into "moderate" or "severe" following WHO guidelines. Standard guidelines for such classifications need to be developed for the other violence domains as well to allow for studying on severity of violence in settings such as ours. Finally, whereas this study has a unique sample given the age of the cohorts and the setting, the use of two different tools could not allow us to immediately compare the two age groups. --- Conclusions Physical, psychological and sexual violence among AGYW in the two Nairobi slums is common. The violence is intimately related to some of the social as well as cultural norms but, importantly, seem to be driven by the economic circumstances under which these girls find themselves in. Given the association between violence and HIV acquisition in young women, addressing violence against women and girls is critical to curbing the HIV epidemic overall and interventions against this should be supported and promoted. --- All relevant data are within the manuscript and its Supporting Information files. --- Supporting information

We sought to estimate the prevalence, severity and identify predictors of violence among adolescent girls and young women (AGYW) in informal settlement areas of Nairobi, Kenya, selected for DREAMS (Determined Resilient Empowered AIDS-free, Mentored and Safe) investment.Data were collected from 1687 AGYW aged 10-14 years (n = 606) and 15-22 years (n = 1081), randomly selected from a general population census in Korogocho and Viwandani in 2017, as part of an impact evaluation of the "DREAMS" Partnership. For 10-14 year-olds, we measured violence experienced either in the past 6 months or ever using a different set of questions from those used for 15-22 year-olds. Among 15-22 year-olds we measured prevalence of violence, experienced in the past 12 months, using World Health Organization (WHO) definitions for violence typologies. Predictors of violence were identified using multivariable logit models.Among 606 girls aged 10-14 years, about 54% and 7% ever experienced psychological and sexual violence, respectively. About 33%, 16% and 5% experienced psychological, physical and sexual violence in the past 6 months. The 10-14 year old girls who engaged in chores or activities for payment in the past 6 months, or whose family did not have enough food due to lack of money were at a greater risk for violence. Invitation to DREAMS and being a non-

Introduction Worldwide, over a million people acquire a sexually transmitted infection (STI) daily. Every year, there is an estimated 357 million new infections with one of the four STIs globally: chlamydia (131 million), gonorrhoea (78 million), syphilis (5.6 million) or trichomoniasis (143 million). 1 The United Nations, defines youths as those persons aged between 15 and 24 years. 2 Young persons (youths) are those in the transition period, from the dependent phase of childhood to the interdependence of adulthood. 3 Sexually transmitted infections are common among young people, with approximately 20 million new cases of STIs being reported every year in the United States, of which the majority occur among people aged between 15 and 24 years. 4 Young people are at high risk of getting an STI for the following reasons: young women's bodies are biologically more prone to sexually transmitted diseases (STDs), lack of access to healthcare, confidentiality concerns and multiple sexual partners, among others. 4 The Nigerian National Demographic Health Survey (NDHS) in 2008 reported that 16% of young women and 6% of young men aged between 15 and 24 years had sexual debut before the age of 15 years. A survey in Nigeria on the prevalence and predictors of early sexual debut among adolescents reported that an average of 18.6% had sexual debut before their 15th birthday -16.6% in boys and 20.2% in http://www.phcfm.org Open Access girls. 5 This is clearly one of the reasons for the high prevalence of STIs among young people. 6 Sexually transmitted infections can lead to severe complications beyond the immediate impact of infections. Some STIs, such as herpes and syphilis, can increase the risk of human immunodeficiency virus (HIV) acquisition threefold or more. Gonorrhoea and chlamydia are major causes of pelvic inflammatory disease (PID) and infertility in women. 1 Data on the incidence and prevalence of STIs in Nigeria are limited because of the under-reporting of STIs, especially among young persons, which is attributable to inadequate diagnostic facilities and limited access to treatment facilities, asymptomatic episodes of the infections and the stigma associated with STI. 7 Young adults are also exposed to diverse sources of influence (multiple sexual partners, lack of condom use) transecting different levels of causation. 2 The knowledge of the non-HIV causes of STDs is still lacking, and the risky behaviour practiced by sexually active young adults is on the rise. Prevention of STIs includes counselling, behavioural interventions, comprehensive sexuality education, STI and HIV pre-and post-test counselling, condom promotion and interventions targeted at key populations. 1 The prevention of STIs, especially in the regions where they are endemic, propelled mainly by heterosexual transmission, includes vaccination (for vaccine-preventable diseases) and practice of the 'ABC' approach (abstinence, be faithful to one partner and use of condom). 8 Literature on the knowledge and preventive practices of STIs in Lagos State is quite scanty, if any at all, especially in Surulere local government area (LGA), which is one of the highly sociable urban areas, with numerous bars, clubs, lounges, malls and pubs, highly populated by young persons. Hence, this study was conducted in Surulere LGA in Lagos State to determine the knowledge of, attitude towards and preventive practices adopted by young unmarried persons against STIs. --- Materials and methods --- Background information to the study area --- Study population, design, sample size determination and selection of participants The study population consisted of young unmarried youths who must have spent at least 6 months in the selected communities in the LGA. The study was a community-based, descriptive, cross-sectional study. The Cochrane's formula for descriptive study, n = z 2 pq/d 2, was used, with a standard normal deviation at 95% confidence interval (1.96), a prevalence rate of 62.5% (0.625) from a similar study carried out in Enugu, South East Nigeria, 12 and an error of precision at 5% (0.05). The minimum sample size was 360. An additional 20% of this (72) was added to allow for possible data loss. Hence, 450 respondents were used for the study. A multi-stage probability sampling technique was used to select the respondents from the study population in five stages. Stage 1 comprised the selection of five wards of the nine wards in Surulere LGA by simple random sampling using the ballot method. In stage 2, 20% of the total number of streets in each of the selected wards was selected through simple random sampling by ballot. Selection of houses was done via systematic sampling, with the sampling interval (k) being the total number of houses on the street divided by the number of required houses in stage 3. In stage 4, one household was selected from each of the selected houses, while stage 5 consisted of the selection of one respondent from each household by simple random sampling using the ballot method. --- Study instrument and data collection A pretested, interviewer-administered, semi-structured questionnaire adapted from the reviewed literature 12,13,14,15,16 was used for data collection. The questionnaire consisted of four sections: Section A consisted of questions on socio-demographic characteristics of the respondents, while Section B had questions on the knowledge of the respondents about STIs. Section C had Likert-scale questions assessing attitudes of the respondents towards STIs, and Section D comprised questions were asked to determine the preventive practices of the respondents against STIs. --- Data analyses Data analysis was performed using Epi-info 7.2.2.2. Results were represented in frequencies and percentages. Chi-square test was used to test for associations with the level of significance set at p <unk> 0.05. For the level of knowledge of each respondent about STI, every correct answer attracted one point, while every incorrect answer or non-response attracted a zero. The scores were converted to percentages and graded as poor (<unk> 50%) and good (<unk> 50%). For attitude questions on Likert scale, the mean score was calculated. Scores below the mean were classified as poor, while those above the mean were classified as good attitudes. In assessing the preventive practices against STIs, every correct answer scored one point, while an incorrect answer or non-response scored zero. The total score for each respondent was converted to percentage and graded as poor preventive practices if <unk> 50%, and good preventive practices if > 50%. http://www.phcfm.org Open Access --- Ethical consideration Ethical approval for this study was obtained from the Health Research and Ethics Committee (HREC) of the Lagos University Teaching Hospital (ADM/DCST/HREC/ APP/410). Written informed consent was obtained from each respondent, with assurance of confidentiality of information and their right to withdraw from the study at any point in time. The participants were informed that participation in the study was voluntary. --- Results --- Socio-demographic characteristics of respondents A total of 425 out of 450 questionnaires administered, which were adequately answered, were analysed with a response rate of 94.4%. The age range and mean age <unk> standard deviation (s.d.) of the respondents were 15-24 and 19.9 <unk> 2.5 years, respectively. Most (75.8%) of the respondents were Christians of Yoruba ethnicity (68.0%), and 64.0% had tertiary-level education (Table 1). Majority (84.7%) of the respondents were aware of STI, and most of them (76.7%) got information about STI from the Internet; teachers and schools (76.1%); and electronic media (68.1%). --- Knowledge of respondents about sexually transmitted infections Viruses (68.3%) and bacteria (48.6%) were the most common known causes of STIs among the respondents. Majority of the respondents (85.8%) knew that HIV or acquired immunodeficiency syndrome (AIDS) is a type of STI; however, a few believed that diarrhoea (23.9%), sickle cell disease (26.7%), tuberculosis (25.4%) and malaria (18.3%) were examples of STIs. Majority (97.8%) of the respondents rightly knew that STIs did not have any gender predilection, as it can affect both men and women, and 96.7% knew that STIs were commonly transmitted by unprotected sexual intercourse. More than half (61.9%) of the respondents also recognised blood transfusion as a means of STI transmission; however, less than half (46.9%) of the respondents identified'mother to child' route as a means of STI transmission (Table 2). More than half (58.1%) of the respondents knew that it is possible to be infected with no clinical symptoms. Pain on urination (84.2%) was the commonest STI symptom, while http://www.phcfm.org Open Access infertility (74.4%) was the most known complication of STI identified by the respondents. In addition, most (64.7%) of the respondents knew that not all STIs can be cured, while majority of the respondents (92.2%) agreed that STIs are preventable. Overall, about three-quarter (66%) of the respondents had good knowledge of STIs (Table 3). --- Attitude of respondents towards sexually transmitted infection prevention Table 4 shows that almost all (95.2%) of the respondents were in favour of educating young people on STI in academic institutions, while the majority (83.3%) agreed that people with STIs should not be isolated. A higher percentage (91.6%) of the respondents agreed with the idea of using condoms during sexual intercourse to prevent STI, while 94.2% believed that screening for STI is good, and majority of the respondents (95.2%) agreed that it was proper to contact the health personnel if one noticed STI symptoms. A large number (74.8%) of respondents agreed that watching pornography can contribute to risky sexual practices that can lead to STI. Overall, majority of the respondents (99%) had a positive attitude towards the prevention of STIs (Table 5). Age, level of education, attitude and preventive practices were all found to be statistically significantly associated with the knowledge of respondents about STIs (p <unk> 0.05). --- Respondents' preventive practices against sexually transmitted infections --- Discussion Sexually transmitted infections are among the world's most common diseases, with an annual incidence exceeded only by diarrhoeal diseases, malaria and lower respiratory infections. Each day, almost one million people acquire a new STI; more than 340 million new cases of curable STIs and even more new viral (non-curable) infections occur each year worldwide. Up to 80% of curable STIs occur in developing countries of the world, with adolescents and young adults having the highest rates of these diseases. In developing countries like Nigeria, STIs are among the leading causes of disability adjusted life years (DALYs) lost for women of reproductive age, exceeded only by maternal causes and HIV. 1,2,17,18 In this study, 450 young unmarried persons were assessed for knowledge of, attitude towards and preventive practices of STI. The mean age of the respondents was 19.9 <unk> 2.5 years and most (64.0%) had tertiary education, which is in line with similar studies carried out in Shone Town Ethiopia (18.6 <unk> 1.9) and Malaysia, where majority (78.3%) were undergraduates. 13,15 This study is however contrary carried out in southwest Nigeria, which reported a lower mean age. 14 This difference is likely because of age difference in the study population. Most (66%) of the respondents had overall good knowledge of STIs contrary to a study in Ado-Ekiti, Nigeria, which reported only 6.9% of the respondents as having overall good knowledge of STIs. 14 This finding could be because of the fact that most of the respondents in this study were undergraduates, compared to the Ado-Ekiti study, which was carried out among secondary school students. Human immunodeficiency virus or AIDS, gonorrhoea and syphilis were the major types of STIs identified by respondents in this study, which is in line with the finding of similar studies conducted within and outside Nigeria. 12,15,19,20,21 Majority of the respondents in this study knew that unprotected sexual intercourse was a major means of transmission of STIs, while less than half knew about transmission of STIs via mother to child -this finding is consistent with the reports of various other studies conducted within and outside the country. 12,13,14,19 A statistically significant association was found between age and level of education with knowledge of respondents about STIs (p <unk> 0.05), which implies that as the respondents got older or attained a higher level of education, their knowledge also increases. For example, respondents in the age group 20-24 years and those in tertiary institutions had better knowledge of STIs. This finding is similar to that reported by a Malaysian study where students in the age group of 24-30 years were more likely to have good knowledge about STIs, compared to those aged 17-23 years, 15 which is slightly different from a study conducted in Kwara State, Nigeria, where age was not found to be statistically significant with knowledge, but class do. 22 With regard to attitude, almost all respondents in this study were of the view that 'young people should be educated about STI in academic institutions' -this finding is comparable to that of an Indian study. 23 Condom use during sexual intercourse to avoid STIs was supported by most of the respondents in this study, which is similar to the finding in the Malaysian study. 15 The majority of the respondents in this study agreed that multiple sexual partners played a major role in the transmission of STIs,' while a high percentage of the participants opposed watching pornography. Similarly, other studies in Malaysia, China and Kampala reported that a higher percentage of the respondents knew that having multiple sexual partners was unsafe. 15,24,25 Overall, almost all respondents in this study had a positive attitude towards prevention of STIs, as was the case with that was conducted in Shone Town, Ethiopia13 This study found a statistically significant connection between knowledge and attitude of the respondents (p <unk> 0.05). This implies that as the knowledge of the respondents on STIs increases, they show better attitudes towards its prevention. About a quarter of the respondents in this study were sexually active during the time of this study; this is in line with the finding in a study in Ogbomoso, Nigeria, and Malaysia. 5,15 More than half (59.4%) of the respondents in this study had sexual debut between the ages of 15 and 19 years. This finding is similar to that of a study conducted in Ikeji-Arakeji, Osun State, southwestern Nigeria, which reported mean age at first sexual intercourse as 16.8 years (approximately 17 years). 26 The study in Ogbomoso, Nigeria, found that mean age at sexual debut was 15.80 years in girls and 15.40 years in boys. 5 Early sexual debut has been reported to be associated with an increased risk of STIs. Researches have shown that the probability of giving birth during the teen years is three times higher for those who had their sexual debut before the age of 16, compared to those who did not. 5 This report has shown that about 20% of women in Nigeria in 2013 were sexually active by the age of 15 years, and the median age for first sex stood at 17.7 years for women and 20.6 years for men. This declining age of first sexual intercourse has been proffered as one possible explanation for the increase in the number of STIs. 27,28 Some common misconceptions found among the respondents in this study about the prevention of STIs were: about 36% and 25% felt that contraceptive use and taking drugs before sex, respectively, are known protective measures against STIs, while alcohol intake before sexual intercourse was also regarded as a prevention method by about 18% of the respondents. In a study conducted in China, some of the respondents reported that taking antibiotics before or after sex, taking a shower before or after sex or using a sex detergent wash before or after sex could prevent STIs. 24 About half of the respondents in this study had one regular partner and the majority used condoms, and most of them used it always (at every sexual intercourse). This finding is contrary to the report of a study among youths in Nairobi where less than half of the respondents use condoms and in southern Ethiopia where 29.0% use condoms and 11.9%, respectively, use it always. 29,30 These differences probably could be because of differences in the cultures and religious beliefs in the different study settings. The study in Malaysia reported that about two-third of the respondents had only one partner. 15 These findings point to a higher rate of risky sexual behaviours among young unmarried people in the different study settings. About half (51%) of the respondents in this study have been vaccinated against hepatitis B virus, while almost all (95%) were not protected against human papilloma virus. Pre-exposure vaccination has been recommended as one of the most effective methods for preventing transmission of human papillomavirus (HPV), hepatitis A virus (HAV) and hepatitis B virus (HBV). Human papillomavirus vaccination is recommended routinely for boys and girls aged 11 or 12 years beginning at 9 years of age. 31 Overall, about two-third (66.0%) of the respondents in this study had poor preventive practices against STIs. Knowledge of respondents of STIs was found to be statistically significant with the preventive practices of STIs (p <unk> 0.05). --- Conclusion Most of the respondents in this study were aware of and had good attitude towards the prevention of STI, but gaps still exist in their knowledge of STIs. Preventive practices of STIs are very poor among the respondents. Therefore, targeted education on behaviour change communication aimed at young unmarried persons in communities is highly recommended to improve their knowledge and thereby implementation of preventive practices against STIs. --- Limitations This study had some limitations. Firstly, our study participants were from only one LGA out of the 20 in Lagos State. This study was conducted in a community, and the results represent the views of the respondents in that community alone, which might be different from others in other parts of the state. The presence of recall bias in this study is also very likely. Therefore, further research with a larger population with adequate power and sample sizes is recommended. --- Data availability statement Data sharing is not applicable to this article as no new data were created or analysed in this study. --- Competing interests The authors have declared that no competing interests exist. --- Authors' contributions O.E.O. was responsible for the concept, design, drafting, data analysis and manuscript revision, while O.D.O. was responsible for data collection, data analysis and manuscript writing. D.O. participated in design and manuscript review, and O.G.E. was responsible for editing and manuscript review. All the authors approved the final version of the manuscript. --- Disclaimer The views and opinions expressed in this article are those of the authors and do not necessarily reflect the official policy or position of any affiliated agency of the authors.

Background: Sexually transmitted infections (STIs) are of public health importance as over 1 million STIs are acquired daily worldwide. One-third of the new cases of curable STIs affect younger persons aged less than 25 years. Sexually transmitted infections can lead to severe complications beyond the immediate impact of infections as such. Aim: This study assessed knowledge of, attitude towards and preventive practices of STI among young unmarried persons in Surulere local government area (LGA) of Lagos State, Nigeria.

Background The adverse experiences and profound structural changes related to the COVID-19 pandemic have resulted in severe psychosocial stress all over the globe. There is a consensus that families have been particularly strained by the policies implemented to contain the virus, e.g., due to limited access to family support services [1] and daycare facilities/ nurseries or schools [2], which resulted in additional childcare responsibilities for parents while often having to work from home. While restriction measures have been gradually reduced or completely lifted, the long course of the crisis and the accumulation of stress factors such as existential fears and worries, loss and grief, economic disadvantages, negative personal and social changes in many life areas represent persistent stressors that can have adverse effects on e.g., mental health [3]. Accordingly, studies have found an increase in depression and anxiety symptoms among parents since the beginning of the pandemic [4,5], with caregivers of young children at particular risk for these mental health issues [6]. There is also evidence for an increase of parenting stress [7][8][9], which is closely related to the quality of parent-child-interaction [10] and both parental and child mental health [10][11][12]. Child mental health soon became a matter of concern in the early stages of COVID-19. National and international studies found significant increases of psychological problems in children and adolescents [e.g. 13,14] compared to pre-pandemic data. However, these studies largely focused on school-aged children or merely included younger children without highlighting their specific situation. Hence, comparable investigations of infants' and young children's mental health as well as their caregivers' wellbeing are sparse-possibly due to this group being less directly affected by the structural restrictions (e.g., school closures) implemented to contain the virus. However, parental psychosocial stress during the different waves of the pandemic is likely to have impacted very young children: They depend almost exclusively on the physical and emotional care and protection of their parents in a phase of life characterized by rapid brain growth and development [15,16], rendering them particularly vulnerable to stressful environmental influences [17] as evident during the pandemic-with severe possible implications for a healthy development [18,19]. One of the very few investigations with a specific focus on infants and toddlers is the CoronabaBY study which surveys psychosocial stress factors such as parenting stress and parental and child mental health problems in families with 0-3 year olds in Germany [20]. First cross-sectional results from a high incidence phase in 2021 showed that infants and toddlers showed an overall similar level of mental health problems at this point compared to pre-pandemic studies. However, a substantial number of the surveyed parents struggled with affective symptoms and reported limited emotional resources for childcare due to high levels of parenting stress. Because young children's mental health is generally closely linked to the psychosocial wellbeing of their caregivers [10,11,[21][22][23], there could still have been delayed detrimental effects of parental stressors in later stages of the pandemic. Given that children's mental health is the foundation for healthy development [24], it is of particular interest to understand the extent to which potentially harmful stressors were present in this population at different stages of the pandemic in order to provide appropriate support services following the very recent crisis. We therefore investigated psychosocial stress factors, namely perceived pandemic burden, parenting stress, and parent and child mental health outcomes in German families with children aged 0-3 years in a repetitive cross-sectional study during three different phases of the COVID-19 pandemic in 2021/22 (waves I and III = phases of high incidence rates and restriction measures and wave II = low incidence rates and relaxations). We aimed to answer the following research questions: 1. How many families experienced psychosocial stress factors during phases of higher and lower COVID-19 incidence rates/ restriction measures during the CoronabaBY study? We expected a higher prevalence of perceived pandemic burden, parenting stress, parental anxiety and depression symptoms, infant crying, sleeping, and feeding problems, and toddlers' emotional and behavioral problems among families in survey wave I, due to high incidence rates and lockdown measures compared with families surveyed in wave II during the summer with low incidence rates and restrictions, followed by a rise in the number of psychosocially stressed families during the wave III survey period with its renewed steep increase in infection rates (emerging Omicron variant). --- Which sociodemographic (e.g., income, education) and pandemic-related (e.g., restricted social support, increased family conflicts) markers might contribute to the aforementioned psychosocial stress factors? --- Methods --- Study design The --- Participants All participants were recruited and surveyed via smartphone app "Mein Kinder-und Jugendarzt" ("My pediatrician") (www. monks-aerzte-im-netz. de) which is a well-established communication tool connecting parents with their pediatrician. In a two-step recruitment procedure, all pediatricians in Bavaria using "My pediatrician" as part of their practice management were invited to participate in the study (N = 300). After giving informed consent (N = 73, response rate = 24.3%), an invitation for study participation was sent out via app to all eligible patients of the participating pediatricians. All parents of children between 3 months and 3 years who used the app and who understood the German study invitation were eligible to take part. Study invitation and detailed information were presented via app. Subsequent informed consent was also given via app. 18,531 study invitations were sent out via push-message. 3449 parents were included after giving informed consent and a remaining total of 2940 parents completed the study questionnaires. --- Measures All data were collected by standardized questionnaires via app. Participants were asked questions on general sociodemographic characteristics, perceived pandemic burden, parenting stress and parent and child mental health outcomes. --- Pandemic related restrictions and perceived pandemic burden Overall, ten questions were asked about specific restrictions and perceived burden related to the pandemic (e.g., 'During the strictest pandemic measures, how restricted did you feel with regard to social contacts?'). The perceived "pandemic burden" for parents and children was derived from the 5-point-answer (from --- Parenting stress To assess parenting stress, we applied the parent domain of the German Version of the 'Parenting Stress Index (PSI)' ("Eltern-Belastungs-Inventar" EBI; [25]). High scores indicated limited parental resources for upbringing and care for the child. The parent domain includes the following subscales: 'health' (parental health impairment as a cause or a result of parenting stress), 'isolation' (lacking integration in social networks), 'role restriction' (perceived limitations as a result of being a parent), 'parental competence' (parental doubt about their own abilities to manage upbringing and care for their child), 'attachment' (emotional relation of parent to the child), 'depression' (limited emotional availability within the parent-childrelationship) and'spouse related stress' (as a result of being a parent). Answers were given on a 5-point Likert scale ranging from 1 = strongly agree to 5 = strongly disagree resulting in a possible score range of 28 to 140. The three cut-off categories for each subscale and the whole parent domain were 'not stressed' (T-value <unk> 60),'stressed' (T-value = 60-69), and'strongly stressed' (T-value <unk> 70). Internal consistency of the parent domain has been proven to be good (<unk> = 0.93), and retest reliability after one year has been shown to be r = 0.87. Correlations with stress indicators and related constructs have resulted in the assumption of test validity [25,26]. --- Parental depression and anxiety symptoms Current parental depression and anxiety symptoms were assessed with the State-Trait-Anxiety-Depression Inventory (STADI; [27]). [28]. --- Infants' crying, sleeping and feeding problems and toddlers' emotional and behavioral problems For infants (0-16 months), the two subscales 'crying/ whining/sleeping' and 'feeding' of the Questionnaire for Crying, Sleeping and Feeding (CSF; [29]) were applied. Parents answered 38 questions on behaviors in their infants. Answers were given on 4-point-scales and mean values were calculated (ranging from 1 to 4). According to validated cut-off values, the dichotomous outcome noticeable problems and no problems were calculated for the domains 'crying/whining/sleeping' (cut-off value: 1.84, sensitivity: 87%, specificity: 92%) and 'feeding' (cutoff value: 1.27, sensitivity: 57%, specificity: 77%). The CSF also comprises questions to identify excessive crying as defined by the Wessel criterion ('rule of threes') [30]. The validity of the questionnaire has been secured by the proof of high internal consistencies of the scales as well as by correlations with behavior diaries and clinical diagnoses [29]. For toddlers (from 17 months old), the Strengths and Difficulties Questionnaire (SDQ, short form of the German Version; [31]) was used to examine emotional and behavioral problems. Parents were asked to classify the individual characteristics to be not true, somewhat true or certainly true for their child in four domains ('emotional symptoms', 'conduct problems', 'hyperactivity/inattention', and 'peer relationship problems'), resulting in a score range of 0-40 points. Cut-off values indicated child behavior to be 'no problems' (0-13 points), 'borderline' (14-16 points) or 'noticeable problems' (17-40 points). Internal consistency has been shown to range between <unk> = 0.73 and <unk> = 0.86. By means of comparison with other corresponding scales (e.g., Child Behavior Checklist), the validity of the instrument can be assumed [32,33]. --- Statistical analyses The present cross-sectional analysis is based on three survey waves of data collection: wave I: 1st of February to 7th of June 2021, wave II: 8th of June to 16th of October 2021, and wave III: 17th of October 2021 to 14th of March 2022. Selection and comparison of time periods was based on COVID-19 incidences and corresponding measures to contain the pandemic: wave I was characterized by high incidences and strict measures (e.g., closing of schools and close contact services until March, limited availability of first vaccination and a first occurrence of a COVID mutation). Wave II comprised the summer months with relatively low incidences, the opportunity to get vaccinated for everyone and relaxations of measures. Wave III was again characterized by very high incidences (emerging Omicron variant) but had fewer restrictions for vaccinated or recovered people compared to earlier high incidence phases in Bavaria, Germany. Statistical differences between the sociodemographic characteristics of the three survey wave samples were detected by using Chi Squared test for categorical and ANOVA for continuous variables. To answer the first research question, we calculated Chi Squared Tests and corresponding effect sizes (Phi coefficient <unk>) to detect potential differences of the addressed psychosocial stress factors between the samples of the three individual survey waves. To adjust for children's age and sex, logistic regression models were calculated with survey wave and children's age and sex as independent variable and the respective psychosocial outcome as dependent variable. In order to obtain clinically relevant answers, the outcome variables were dichotomized as follows: Pandemic related restrictions/ changes and perceived pandemic burden were dichotomized into high/ very high (point 4 and 5 on 5-point Likert-scale) versus low perceived restrictions (points 1-3), respectively into stressful/ very stressful (point 4 and 5 on 5-point Likert-scale) compared to less stressful (points 1-3). Parenting stress (EBI) was classified into stressed/strongly stressed versus not stressed. Parental mental health problems (STADI) were dichotomized into above average/ far above average versus average/ below average/ far below average, and toddler's emotional and behavioral problems (SDQ) into borderline/ noticeable problems versus no problems. In a second step, we addressed the question, which factors might have contributed to the surveyed psychosocial stress factors, and included both sociodemographic factors and pandemic-related factors as potential predictors. To explore if these factors predicted parenting stress (EBI total score), maternal depression and anxiety symptoms (STADI total score, T-values), infants' crying/whining/sleeping problems (total score of crying/whining/sleeping subscale), and toddlers' emotional and behavioral problems (SDQ total score), four multiple linear regression models over all three survey waves were calculated: block-wise multiple linear regression models with binary sociodemographic variables as predictors in the first block (parental education status, parental financial status before the pandemic, financial burden due to pandemic, having siblings, child age, chronical illness/ disability of the child) and pandemic related variables in the second block (survey wave, restricted family support services, increased family conflicts, restricted parental social contacts and perceived pandemic burden) were conducted. The formation of the models resulted in the calculation of beta weights and their p-values for corresponding predictor variables. Requirements for calculating the multiple linear regression models were met. For the linear regression models, independent variables were dichotomized as follows: Education status was dichotomized into high (university degree and high school diploma) and low (secondary and lower secondary school diploma). Financial status was also dichotomized into high ("large expenses possible" and "bigger additional expenses possible") and low ("smaller additional expenses possible", "little scope for additional expenses", "additional expenses not possible"). Accordingly, financial burden due to the pandemic was dichotomized (yes: small, medium or huge financial burden versus no financial burden due to the pandemic). Chronic illness or disability of the child was defined as any chronic illness (also allergy, hyperactivity) and/or disability. Since submission of questionnaires was only possible when all items were completed, we had only a few missing values because of obvious misreporting of parental age. All described results were based on an alpha level of 5%. A post-hoc Bonferroni correction was applied to control for multiple testing. Analyses were performed in IBM SPSS Statistics Version 28.0 for Windows. --- Results --- Sample characteristics In total, we examined 2940 parent-child dyads, 1,004 of whom participated in wave I, 938 in wave II, and 998 in wave III (Table 1). Overall, 92.9% (n = 2731) of the surveyed parents were mothers with a mean age of 33.5 years (SD: 4.8), 6.6% fathers (mean age: 35.7 years, SD: 7.0), and 0.5% were "grandparents and others". Children were on average 16.4 months old (SD: 11.7, range: 0-43 months) and were divided into 'infants' (n = 1404) with a mean age of 5.5 months (SD: 3.6) and 'toddlers' (n = 1536) with a mean age of 26.3 months (SD: 6.6). --- Perceived pandemic burden and pandemic-related restrictions Almost two thirds of the parents perceived the pandemic as stressful or very stressful in wave I (Table 2). This proportion significantly decreased from wave I to wave II and remained on a high level in wave III. The frequencies of perceived individual pandemic-related restrictions can be seen in Table 2. --- Parenting stress and parental mental health Parenting stress was present in 38.2% of the parents in wave I. This proportion was slightly higher in wave II (46.1%, p = 0.09) and significantly higher in wave III (51.2%) compared to wave I (Table 3). Maternal anxiety and depression symptoms were similar in all three waves and ranged between 24.2% in wave I and 28.3% in wave III (Table 3). In fathers, these symptoms ranged between 14.3% and 28.4% (Table 3). --- Child mental health (crying, sleeping and feeding, emotional and behavioral problems) In wave I, 26.5% of the infants showed problems on the crying/whining/sleeping subscale of the CSF. In wave III, this proportion was significantly higher (35.5%, p = 0.009) (Table 3). In wave I, 18.4% of the toddlers showed at least borderline emotional and behavioral problems, in wave III the proportion was 23.9. Figure 1 shows the percentage of families with psychosocial stress factors in each survey wave (Fig. 1). --- Influencing factors on parenting stress, maternal symptoms of depression and anxiety, infants' crying/sleeping problems and toddlers' emotional and behavior problems The block-wise, fully adjusted linear regression model (R 2 = 0.264, F(12, 2551) = 76.23, p <unk> 0.001) showed familial conflicts to have the highest effect size (<unk> = 0.355, p <unk> 0.001) on the outcome parenting stress (EBI total score) (Table 4). For maternal symptoms of depression and anxiety (STADI total score) (R 2 = 0.272, F(12, 2370) = 73.76, p <unk> 0.001), increased familial conflicts had the highest effect size (<unk> = 0.305, p <unk> 0.001). For infants' crying/sleeping problems (crying/whining/sleeping subscore of the CSF) the model (R 2 = 0.095, F(12, 1237) = 10.79, p <unk> 0.001) yielded having siblings to be a protective predictor with an effect size of <unk> = -0.159 (p <unk> 0.001) whereas increased familial conflicts (<unk> = 0.149, p <unk> 0.001) was the risk factor with the highest effect size. For toddlers' emotional and behavioral problems (SDQ total score) the model (R 2 = 0.175, F(12, 1356) = 23.97, p <unk> 0.001) yielded increased familial conflicts (<unk> = 0.216, p <unk> 0.001) to have the highest effect size. --- Discussion In a comparison of the three survey waves of the Coron-abaBY study, including a total of 2940 parents with their children aged 0-3 years, we cross-sectionally investigated the number of families experiencing psychosocial stress factors in relation to higher and lower COVID-19 incidences and restrictions in Bavaria, Southern Germany. Our results show that psychosocial stress factors in families with infants and young children remained highly prevalent over the course of the pandemic. While parental high pandemic burden mirrored incidence rates and respective restriction measures, the number of caregivers experiencing noticeable parenting stress increased irrespective of pandemic events. Parental mental health problems were highly evident, but remained more or less stable during the course of the pandemic. Similarly, toddlers' mental health problems were highly evident in later stages of the pandemic but did not increase significantly during the course of the study. In contrast, the number of infants with mental health problems significantly increased from one high incidence phase to another. Of several sociodemographic and pandemic-related variables, we found the increase of familial conflicts during the pandemic to have the strongest negative influence on parenting stress, maternal, infants' and toddlers' mental health. Looking at our results in detail, we found a perceived high pandemic burden in up to 65% of our sample, which appears to be a slightly higher rate than in a German comparison study (59% with high perceived pandemic burden) including parents of children younger than 14 years [34]. This result corroborates findings that parents of young children are particularly vulnerable to experiencing the pandemic as stressful (e.g., [2]). Partly as expected, the number of families with a high pandemic burden roughly mirrored the restrictions during the individual survey waves with a prevalence peak during survey wave I, a significant decrease in survey wave II but only a slight renewed increase in survey wave III. These findings are in line with another study showing a peak of disease-related distress early during the virus outbreak and a decrease as time proceeded [35]. A different picture emerged for parenting stress across the three survey waves: From the beginning of the study, with 38.2% of the families being affected, parenting stress was highly pronounced compared to pre-pandemic data (see [20]). Contrary to our initial assumption, the number of families with elevated parenting stress significantly increased in wave II despite lower COVID-19 incidences and fewer restrictions and only non-significantly rose to a peak in wave III. Although the effect sizes were small, overall more than half of the surveyed participants (51.2%) were affected by parenting stress. While there are no studies that are fully comparable with regard to design, survey periods, and target group, a longitudinal study on parenting stress found an increase with a longer duration of the pandemic in the year 2020 [36]. In addition, a German longitudinal investigation of the general population found that psychological distress did not decrease between two high incidence phases in 2020 despite lesser restrictions during the second survey timepoint [37]. As parenting stress did not decrease despite relaxation of measures to contain the virus, it is likely to be a longer-term stress factor. Since it is closely related to parental [38][39][40] and child mental health problems [10][11][12] and negatively impacts the parent-child-relationship [38,41], intervention efforts should aim at reducing parenting stress in families with young children in the aftermath of the pandemic. Turning to parental mental health, we identified a peak of 28.3% of mothers and 28.4% of fathers with symptoms of depression and anxiety in survey wave III. The number of affected parents appeared to be higher compared to a pre-pandemic German study with approximately 20% of parents with children under the age of three experiencing affective symptoms [42]. Contrary to our assumptions, the number of parents with affective symptoms remained high almost irrespective of pandemic events, i.e., relaxation of measures. This result is roughly in line with an Austrian study on the general population showing a substantial persistence in mental health problems even 6 months after lifting of pandemic-related restrictions [43]. Taking a look at children's mental health, the number of toddlers with borderline or noticeable emotional and/or behavioral problems rose from 18.4% in wave I to a maximum of 23.9% in wave III, however these increases were not significant. Still, while the number of toddlers with borderline and noticeable mental health problems almost exactly corresponded to the norm in survey wave I, this appears to be no longer the case in later periods of the pandemic [32,44]. Prevalence rates for infants' crying/sleeping, and feeding problems were already in the upper range compared to pre-pandemic data in survey wave I (compare [20]). The number of infants with excessive crying or feeding problems did not change significantly in relation to different survey periods. However, the proportion of infants with crying/sleeping problems was significantly higher in survey wave III than in survey wave I. One of the vanishingly small number of studies targeting infant crying, sleeping and feeding problems during the pandemic also used the CSF and found combined crying/sleeping but not feeding problems to be significantly higher prevalent in infants born during the pandemic [45]. While the small effect size in our study as well as differences in definition and operationalization of these problems have to be considered, with as much as 35.5% of infants affected, this rate is now higher than the ones found in pre-pandemic comparison studies [46]. These findings suggest a staggered negative impact of pandemic-related factors on young children's mental health, and policy makers should be made aware that the needs of infants and toddlers are as relevant as those of e.g., school-aged children in the context of pandemic after-care. In a second step, we explored possible influencing factors on parenting stress and mothers' and child mental health problems. For all four outcomes, increased family conflicts had the highest predictive value, with a medium to large effect on parenting stress, a medium effect on maternal affective symptoms, a small to medium effect on toddlers' emotional and behavioral problems and a small effect on infant crying/sleeping problems. This result highlights that also in the pandemic with its multiple external stressors, the family microclimate is particularly relevant for parental and child wellbeing and should be one focus when counseling burdened families, e.g., by providing conflict management strategies. Individually, we found higher parental education to be predictive of higher parenting stress, which is in line with earlier findings of the CoronabaBY study and might be associated with higher educated parents worrying more about the pandemic (compare [20]). Survey wave and parental perceived pandemic burden also influenced parenting stress, however all effect sizes were small. Mothers' symptoms of depression and anxiety were more pronounced with increasing child age. In contrast, a meta-analysis on moderating factors for maternal depression and anxiety during COVID-19 did not find child age to be influential [47]. As the found effect size in our sample of mothers was small, this result might not be of high practical relevance. Higher pandemic burden was also predictive for maternal mental health, again with a small effect size. Infant crying/sleeping problems were predicted by survey wave and parental pandemic burden, whereas having a sibling was a protective factor. The latter result could be explained by parents' greater experience in handling crying/sleeping problems when having more than one child and has also been shown by another study on infant crying/ sleeping problems during the pandemic in Germany [45]. Again, corresponding effect sizes in our study were small. Toddlers' emotional and behavioral problems were slightly more pronounced for children with a chronic disease or a disability, which is line with earlier findings [48,49]. There are several possible underlying causes for the trajectories of the surveyed psychosocial stress factors: First, the decreasing number of parents perceiving a high pandemic burden over time might reflect the reduced visibility of COVID-19 with the gradual relaxation of measures, but could additionally indicate the development of a habituation effect as a coping strategy. Second, given the permanence of COVID at the time of our survey, the increase in parenting stress over the course of the pandemic as well as the constantly high number of parents with mental health issues might be explained by lack of prospects, resulting in mental exhaustion-the so called 'pandemic fatigue' [37]. Third, with a longer duration of the pandemic-and again irrespective of pandemic events-the number of young children with deteriorated mental health has increased to a worrying level, while at the beginning of the study the numbers for both infants and toddlers still corresponded to the norm. As both parenting stress [10][11][12] and parental mental health issues [21][22][23] are known to be related to child mental health problems, the results corroborate the possibility of a staggered negative influence of high parental burden in this regard. However, these considerations remain theoretical as they cannot be answered on the basis of the data available in this study and will need to be addressed by future studies. This study has strengths and limitations that have to be considered when interpreting the found results. The CoronabaBY study was the first and largest German investigation on psychosocial well-being during the pandemic specifically targeting families with children aged 0-3 years. Besides including a large number of participants, parenting stress and parental and child mental health were assessed with validated standardized psychological questionnaires without notable missing values. In addition, the individual survey periods as well as the whole study duration of 13 months were considerably longer than of many other comparison studies, enabling a broader perspective that goes beyond a momentary snapshot. Also, we investigated later stages of the pandemic while a majority of the pandemic-related research was conducted around the first outbreak of the virus. We recruited a high number of financially well-off families with good education, German background, and mostly mothers (compare [20]). These aspects have to be considered with regard to the generalizability of our study results. Another limitation might be that the CoronabaBY study started a year after the initial outbreak of the virus and thus may not capture the most severe pandemic-related stress experiences. However, a significant proportion of the surveyed sample exhibited psychosocial stress factors, even if the crisis was no longer a new phenomenon, and the number of burdened parents and infants even increased during the course of the study. As the number of fathers in our sample was relatively small, we cannot provide specific results for them. Future studies should aim to shed light on fathers' mental health during and in the aftermath of the pandemic. Children's mental health was assessed via parental report and we cannot rule out the probability that parents with higher stress levels perceive their child's behavior as more troublesome. Finally, due to the cross-sectional nature of our data, we cannot draw causal conclusions with regard to the impact of the pandemic on the surveyed psychosocial stress factors. However, this study sheds light on trajectories of psychosocial stress in families with infants and toddlers in different stages of a global crisis and adds knowledge essential to provide adequate support for this vulnerable group. --- Conclusion Taken together, our results show that psychosocial stress factors in families with infants and toddlers remained highly pronounced or even increased with a longer course of the pandemic -despite relaxation of restrictions and lower incidences. Although the pandemic might start to fade into the background in public perception, the ongoing support needs of families with young children must not be allowed to fall out of focus. Because of their particularly high vulnerability and susceptibility to parental burden, promoting infants' mental health has to be a top priority for policy makers and health professionals in the aftermath of COVID-19. Future investigations need to explore the underlying causes of the trajectories found and specifically examine the relationship between parental and child outcomes in a longitudinal design. Support measures should focus on promoting a positive microclimate in the family, e.g., through conflict counselling, by reducing parenting stress and by strengthening parental resources so that they can be fully available for the signals and needs of their young children. • support for research data, including large and complex data types • gold Open Access which fosters wider collaboration and increased citations maximum visibility for your research: over 100M website views per year --- • At BMC, research is always in progress. --- Learn more biomedcentral.com/submissions Ready to submit your research Ready to submit your research? Choose BMC and benefit from:? Choose BMC and benefit from: --- Availability of data and materials All data generated or analysed during this study are included in this published article. --- Abbreviations --- Declarations Ethics approval and consent to participate The study protocol was approved by the Ethics committee of the Technical University of Munich (vote no. 322/20 S) and pre-registered in OSF (https:// osf. io/ searc h/?q= tksh5 & page=1). Study invitation and detailed information were presented via app. Subsequent informed consent was given via app. --- Consent for publication Not applicable. --- Competing interests The authors declare that they have no competing interests. --- Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Background Families with young children are particularly vulnerable for the stressors induced by the COVID-19 pandemic. However, studies on their psychosocial situation during the course of the crisis are still sparse.In a comparison of three survey waves (wave I and III = high COVID-19 incidences), we cross-sectionally investigated the proportion of families (N total = 2940) with children aged 0-3 years experiencing pandemic burden, parenting stress, and parental and child mental health problems in relation to COVID-19 incidences and restrictions in Southern Germany via validated questionnaires. Potential influencing factors were also explored.The number of parents with a high pandemic burden decreased over the course of the pandemic with a peak of 65.3% in wave I (significant changes except wave II versus III). Participants with high parenting stress significantly increased from 38.2% in wave I to 51.2% in wave III. The number of parents with symptoms of depression and anxiety remained constantly high with a maximum of 28.4% being affected. Infants with crying/sleeping problems increased significantly from 26.4% in wave I to 35.5% in wave III. Toddlers' emotional and behavioral problems showed a peak of 23.9% in wave III (no significant changes). Increased family conflicts were the strongest predictor for parenting stress (ß = 0.355), maternal (ß = 0.305), infants' (ß = 0.149) and toddlers' (ß = 0.216) mental health problems during the pandemic. Conclusions Psychosocial stress factors in families with infants and toddlers remained highly pronounced and even partly increased irrespective of pandemic events. The findings suggest a staggered negative impact of pandemicrelated factors on young children's mental health. Promoting infants' mental health as well as strengthening parental resources by reducing parenting stress should be a top healthcare priority in the aftermath of COVID-19.The study was pre-registered in OSF (https:// osf. io/ searc h/?q= tksh5 & page=1)

Introduction Spousal death (SD) has a wide range of health effects on older adults like mental, social, behavioral, and biological issues [1]. It is closely related to the risk of death in elderly [2] and increases the mortality of various diseases [3]. In particular, the mortality of old people increases in the first few months after a spouse's death [4][5][6]. The lifestyle changes caused by SD increase the incidence of almost all types of cancer [7]. It is also a risk factor for stroke [8] and is related to the severity of cardiovascular disease [9]; moreover, widowed men have an increased risk of type II diabetes [10]. In 1980, a study demonstrated that in addition to medical technology and drugs, a decrease in the proportion of SD was a reason for the decline in disabilities among older adults in the United States [11]. Although the emotional response of older adults after their spouses' deaths may gradually decrease over time [12], their period of grief is several years long rather than temporary [1,[13][14][15], and the sadness caused by a spouse's death affects the occurrence of depression [16]. Similar to sorrow, depressive symptoms continue to occur in elderly at a relatively high level [17,18]. The psychological, social, behavioral, and physical conditions of the widowed are worse than those who are currently married or cohabiting. This includes depression, decreased social connections, unhealthy lifestyles, and poor cardiovascular health [1]. Another study also indicated that spouse loss increases the degree of depression and that it will last for several years [14]. Depression is a common mental illness among older adults [19] and a populationbased cohort study also showed that age-standardized depression prevalence in older adult was 6.8% [20]. Old people with depression have a relatively high rate of disease comorbidity as well as increased mortality and risk of disability and suicide [21][22][23]. Older adults living independently from family or friends are particularly susceptible to depression [24]; however, social isolation is more likely to cause depression in this population compared with people who live alone [25]. Although family support can reduce the difference in depressive symptoms between old people who experienced SD and other older adults [26], the absence of children does not increase loneliness or depression in widowed elderly [27]. In addition, help from children and the sharing of living arrangements with them yield reduced depression symptoms and favorable self-rated health [28]. This study explored the major effects of SD on older adults' depression and whether, in addition to SD, the changes in lifestyle habits, interpersonal relationships, or living environments after a spouse's death are influencing factors. The results may provide health care strategies for the successful aging of older adults. --- Materials and Methods --- Study Design and Sampling The TLSA (Taiwan Longitudinal Study of Aging) data source of this study is the Institute of Family Planning, which was the predecessor of the Health Promotion Bureau of the Ministry of Health and Welfare, Taiwan. Stratified three-stage random sampling was adopted to select 4412 individuals from the registered population of Taiwan who were older than 60 years at the end of 1988. Of the selected individuals, 710 completed the questionnaire before and after their spouses' deaths. A questionnaire survey was conducted by special investigators through interviews to collect baseline data. For selected participants who could not respond to the questionnaire because of conditions such as unconsciousness, severe illness, or deaf-mutism, family members or caregivers who understood their situations could answer for them. --- Questionnaire A total of six longitudinal surveys were conducted in 1989, 1993, 1996, 1999, 2003, and 2007. Collected data included age, sex, ethnic group, education level, financial status, smoking status, alcohol consumption, and self-rated heath status. The simple version of the 10-item Center of Epidemiological Studies Depression Scale (CESD) was used to measure depression, and a score of <unk>10 indicated depressive symptoms. In the mobility investigation, participants were asked if they had difficulty squatting, raising both hands, picking up objects with their fingers, lifting an 11-kg object, walking up two to three floors, or walking 200-300 m, with higher scores indicating poorer mobility. In the survey on relative and friend (RF) support, participants were asked if they believed that their RF were willing to listen to their worries, whether their RF cared about them, whether they were satisfied with the level of care expressed by their RF, and whether their RF were hypercritical to what they did. In addition, the participants were asked whether their RF would ask for their opinions when making decisions or during discussions. In terms of the quantification of support from RF, a higher score indicated lower support. Regarding the questionnaire items, reduced-rank regression was first adopted to obtain the factor loadings for SD, after which the scores for mobility and RF support were acquired. Therefore, mobility and RF support scores were used to represent mobility and RF support in the later section of this study. The dimensions measured in the questionnaire are shown in Table 1. --- Relative and friend (RF) support The RF were willing to listen to their worries The r RF cared about them Satisfaction with caring from RF The RF were hypercritical to what they did --- Statistical Methods We conducted reduced rank regression to analyze the factor loading of mobility items and social support items. Higher RRR score in mobility means worse mobility function and higher RRR score in social support means higher level of social support. The McNemar's Chi-square test was conducted to determine whether each participant presented significant differences before and after SD, and the effect of variable adjustment on SD was analyzed using multiple logistic regression. SAS 9.4 (SAS Institute Inc., Cary, NC, USA) was employed with a significance level of 0.05. --- Results --- Factor Loadings of Mobility and RF Support Table 2 presents the calculation results for the factor loadings of mobility and RF support. The factor loading for each mobility was positive and greater than 0.2, indicating a positive correlation between the degree of disability and depression. It can be found that the factor loading of lower limb activity items were greater than 0.43 (walking 200-300 m, climbing up 2-3 floors and squatting), while the upper limb activity items were between 0.35-0.42 (grasping with the fingers, carrying 11 kg weight and raising the arms up). This means that lower limb activity items have a higher impact on depression than upper limb activity items. The factor loadings for the surveyed items of RF support were negative and greater than 0.2, implying a negative correlation between support level and depression. It can be found that the absolute value of factor loading of "care from significant others" was greater than 0.5, while other items were between 0.27 to 0.4 ("talk", "ask opinion", and "complain"). This means that "care from significant others" have a higher impact on depression than other social support items. --- Participants' Characteristics Table 3 presents the participant demographic variables and the descriptive statistics for SD time point and Time since SD. Of the 710 participants, 256 were men and 454 were women. The participants were divided into six groups (on the basis of the spouses' age at death), four education levels (from primary school to university and above), and four ethnic groups. The majority of the participants were female (63.9%), educated at the primary level (41.6%), and Fukinese (71.8%). The years 1989, 1993, 1996, 1999, 2003, and 2007 were used as the points of tangency for the time of SD occurrence. As shown in Table 3, the highest percentage of SD was between 1999 and 2003 (27.6%). This situation also reflects that participants will experience spouse death from the time they started participating in the study to a certain age. Time since SD was divided into five levels, namely less than 3 months, 3-6 months, 6-12 months, 12-24 months, and more than 24 months. There were a total of 42.7% of participants where time since SD was more than 24 months. --- Pre and Post Spouse Death (SD) Table 4 presents data before and after SD according to participant sex. No significant difference was observed in income satisfaction between men and women before and after SD. A decline in the proportions of both men (ORm = 4.38, p <unk> 0.0001) and women (ORm = 7.0, p <unk> 0.05) who smoked was observed. However, neither men nor women exhibited a significant difference in alcohol consumption before and after their spouses' deaths. In terms of self-rated health, men did not exhibit a significant difference before and after SD, whereas the proportion of women who believed that they had favorable health declined after SD (ORm = 1.40, p <unk> 0.05). In Table 4, diff represents the result of subtracting the post-SD score from the pre-SD score. Men exhibited a significantly decreased degree of depression after SD (mean = 1.52 <unk> 6.09, p <unk> 0.0001), whereas the women's degree of depression increased significantly (mean = -1.31 <unk> 7.49, p <unk> 0.01). A mobility score diff of <unk>0 indicates improved mobility. Men exhibited reduced mobility after SD (mean = 0.33 <unk> 1.25, p <unk> 0.0001), whereas the mobility of women improved (mean = -0.52 <unk> 1.50, p <unk> 0.0001). Neither men nor women exhibited a significant difference RF support scores after SD, indicating no significant change in support from relatives and friends after SD. --- Effect of SD on Depression Table 5 presents the multinomial logistic regression analysis of the effect of SD on depression. Model 1 consists of the simple effect without any variable adjustment; the mobility score was adjusted in Model 2, self-rated health was adjusted in Model 3, and the RF support score was adjusted in Model 4. Under the Model 1 SD effect, the proportion of depression among the participants after SD was 1.7 times that of before SD (p <unk> 0.0001), whereas the SD effects of Models 2, 3, and 4 resulted in a proportion of depression after SD 1.5-1.7 times that of before SD, and the increase was significant. The odds ratios (ORs) of the adjustment variables in Models 1-4 were all significant, indicating that worsened mobility (OR = 1.3, p <unk> 0.01), declined self-rated health status (OR = 0.5, p <unk> 0.01), and higher RF support (OR = 1.5, p <unk> 0.01) were related to a higher proportion of depression. After variable adjustment, the SD effect on men remained significant, and the ORs of the adjustment variables were nonsignificant. After the mobility score (Model 2) and favorable self-rated health status (Model 3) were adjusted, the OR of the SD effect on women became nonsignificant, whereas the ORs of the adjustment variables became statistically significant (OR = 1.3, p <unk> 0.01; OR = 0.4, p <unk> 0.01). Mobility, self-rated health, and RF social support are associated with depression in total subjects and in female; but in male, mobility, self-rated health, and RF social support are not significantly associated with depression. In female, the SD effect on depression disappeared after adjusting the mobility variable. Table 6 indicates that the proportion of post-SD depression was 6.0 times, 0.9 times, 1.7 times, and 1.3 times that of pre-SD depression when Time since SD was less than 6 months (OR = 6.0, p <unk> 0.0001), 6-12 months (OR = 0.9, p > 0.05), 12-24 months (OR = 1.7, p <unk> 0.05), and <unk>24 months (OR = 1.3, p > 0.05), respectively. The proportion of depression decreased rapidly after 6 months. In the sex-stratified analysis, only men had a significantly higher proportion of post-SD depression compared with pre-SD depression in the 12-24 months after SD (OR = 2.38, p <unk> 0.05), whereas women exhibited a significantly higher proportion of post SD depression less than 6 months after SD (OR = 4.0, p <unk> 0.01). The remaining variables yielded no significant changes. Regardless of gender, the depression proportion is obviously high in six months after spouse death, which declines after six months and without statistical difference for proportion after 24 months. Men suffered depression more than women in every period, and in particular, SD effect on depression, in period 12 and 24 months after SD, is statistically significant in men. In the analysis of post-SD living arrangements, Table 6 indicates that the proportion of depression among widowers and widows living alone and those living with family members was 2.3 times (OR = 2.3, p <unk> 0.01) and 1.5 times that before SD (OR = 1.5, p <unk> 0.01). A further sex stratification analysis revealed that men who live alone after SD exhibited more depression (OR = 3.0, borderline significance) and men who lived with their families after SD exhibited significantly more severe depression compared with before SD (OR = 2.3, p <unk> 0.01). In the stratified analysis of SD effect on depression in different living arrangements, the results for female (OR = 2.0; OR = 1.3) are similar to the trends for male, but there is no statistically significant association under "living alone" or "living with family members". In terms of the stratified analysis of spouse's health status, Table 6 indicates that a significantly large proportion of participants whose spouses had a favorable health status exhibited post-SD depression (all participants; OR = 2.0, p <unk> 0.01, men: OR = 2.3, p <unk> 0.05, women: OR = 1.8, p <unk> 0.05). However, no significant difference was observed in the proportion of widowers and widows with post-SD depression whose spouses were in poor health. --- Discussion Our findings indicate that male sex, spouse's health, and the period of 6 months after SD are risk factors for depression in elderly. The maintenance of mobility, positive self-rated health status, and a shorter period of depression after a spouse's death result in more favorable adaptability among women. Living with family members after SD can alleviate depression in older adults. The strength of our study: (1) as a longitudinal study, this study can provide clearer causal relationship between SD and depression, (2) TLSA is a national survey that has a representative sample of old people in Taiwan, (3) this study used self-matched to explore the effect of SD on depression in older adults. Smoking among men and women declined after SD, whereas alcohol consumption among the participants exhibited no significant difference, which was inconsistent with the finding of increased tobacco and alcohol abuse after SD in previous studies [29][30][31][32]. The proportion of self-rated health decline after SD increased significantly among women. This was consistent with the findings of previous studies indicating that widowed elderly have poorer self-rated health [33,34] and that SD negatively affects health [35]. Men exhibited an obvious decline in mobility after SD, whereas the mobility of women improved, indicating a sex difference. Previous studies have demonstrated that women's mobility after SD is inferior to that of women whose spouses are still alive [36,37]; however, another study revealed that widows in Taiwan are more active in recreational activities compared with married women [38]. The disparity between the present research results and the literature may be attributable to cultural differences. Although RF support did not differ significantly after SD, depression and the degree of support from relatives and friends were positively correlated. This may be explained by the increased care that participants received from older adults due to their depression. However, RF support had little influence on SD effects after variable adjustment, indicating that such support cannot significantly alleviate depression. According to attachment theory, support from friends cannot compensate for the loss of someone to whom an individual had an attachment [39], and attachment problems are particularly prevalent among old people [40]. These studies may explain why support from relatives and friends cannot effectively alleviate depression. Men exhibited a decline in depression after SD whereas women presented increased depression. However, after adjusting for mobility, self-health rating, and RF support, the proportion of men who had depression after SD was greater than that of women, indicating that SD had a significant effect on depression among men (more than 2.3 times that before SD). In other words, adopting approaches to alleviate the effects of SD on depression is more difficult for men. The proportion of women with depression after SD increased significantly (1.5 times); however, after adjusting for mobility and self-rated health, the original effect of SD on depression was explained by the adjustment variables. In addition, the adjustment variables were significantly correlated with depression. In other words, the effects of SD on depression could be alleviated or reduced in women through more favorable mobility and a favorable perception of their own health, a phenomenon not observed in men. This result is consistent with the findings of previous studies, which demonstrated that depressive symptoms caused by SD have a stronger effect on men, whereas women have superior adaptability [41][42][43]. Moreover, recreational physical activity has a preventive effect on depression in women [44]. Increased participation in recreational activities is beneficial for widows, and changes in leisure activities after SD have a greater effect on health and disability compared with aging [45]. These findings corroborate the aforementioned research results. The proportion of depression was obvious within 6 months after SD; however, this trend gradually slowed after 6 months. For women, this increase was not significantly different after 6 months (however, a significant difference was observed among men between 12 and 24 months after SD). After 24 months, both men and women exhibited no significant difference in depression before and after SD; therefore, the effect of SD can be alleviated over time. A previous study demonstrated that older adults experience relatively high levels of depressive symptoms within 2 years after their spouses' deaths [17]. The proportion of depression among women remained high within 6 months after SD; however, this proportion decreased to pre-SD levels after 1 year [18]. These conclusions are consistent with our research results. This study indicated that SD has a shorter effect on depression in women compared with men. Past research has also demonstrated that the amount of time after a spouse's death affects morale and social engagement [46]. The relatively short duration of depression among women after SD may cause them to have a more positive self-rated health status or participate in social engagement, which may also explain why women have more favorable adaptability to depression caused by SD [41,42]. The proportion of depression increased after SD regardless of the participants' living arrangements; however, depression in old people living alone after their spouses' deaths exhibited a greater increase than those who live with family members. Although the aforementioned trend persisted after sex stratification, depression among only men who lived with their families exhibited statistical significance. Therefore, both living alone and living with family members significantly affects depression. However, previous research has demonstrated that living alone is not correlated with depression [25]. After adjustment for demographic variables, health status, social support, and financial status, the effect of living alone on depression disappeared [47]. Although the same trend was observed in this study, participants who lived alone still exhibited a relatively high proportion of depression. Older adults whose spouses were healthy exhibited significantly higher proportion of depression after SD, a trend that was clearly observed in both men and women. However, older adults whose spouses had poor health exhibited no significant change in the proportion of depression after SD. Past studies have demonstrated that elderly who have lost their spouses are more likely to be depressed if they were not the primary caregiver of their sick spouses [48], and the accidental death of a spouse may increase the risk of depression [49]. The aforementioned results can be used to illustrate the phenomena revealed in this study. By conducting a long-term investigation of widowed elderly, this study demonstrated that SD is a risk factor for depression among older adults in Taiwan, and men and women adapt differently to their spouses' deaths. Women's shorter duration of depression after SD, their ability to maintain their mobility, and their more favorable self-rated health result in an increased willingness to participate in activities. This enables them to have superior adaptability to the effects of SD. Older adults are at risk of depression within 6 months after SD. Relatives, friends, and social workers should increase their attention to and care of them during this period. In terms of living arrangements, encouraging elderly to live with family members or friends after SD can reduce the incidence of depression. Important scientific contributions and implications of this study are summarized. Firstly, six months after spouse death is a high-risk period for depression. Families, friends, relatives, or social workers, have to pay more attention to widowhood elderly in six months. Secondly, older men more vulnerable to depression than women after spouse loss. In male, life assistance, instrumental and emotional supports are important practical issues after spouse loss. Third, those who live alone after spouse death suffered more depression than those who live with their family members. The living arrangement is very important to elderly, which affects the rate of adjustment after spouse loss. Fourth, the spouse was in good health before death, which means the great psychological impact of this "unexpected" event. Frequent contact, more care and support are needed for widowhood facing this unexpected event. There are some limitations though we explore several important issues for spouse loss and depression. Firstly, marital satisfaction, marital relationship, marital conflict, and the length of time they were married, were not collected in this study. Lacking these marital factors might limit us to elucidate their important roles in the rate of grief and grief adaptation. Secondly, grief and major depressive disorder share similar symptoms, such as intense sadness, insomnia, poor appetite, weight loss, etc. This study uses the CESD-10 depression scale, without the grief scale measurement, so it is difficult to distinguish grief and depression. Further studies may compare the differences of depression scale and grief scale in widowhood, or perform single-item analysis of "intense sadness", "insomnia", "poor appetite", or "weight loss" in widowhood, to distinguish grief and depression. According to our findings, if the spouse was in good health before death, his/her death ("unexpected" event) will make the great psychological impact to the other. People around these older adults should pay more attention to them. We also found that living with families after SD can alleviate depression in elderly. We suggested that these older adults who have experienced unexpected widowhood can take turns to live with their children, or travel with good friends, and receive care from families and friends. --- Conclusions Our findings indicate that male sex, spouse's health, and the period of 6 months after SD are risk factors for depression in older adults. The maintenance of mobility, positive self-rated health status, and a shorter period of depression after a spouse's death result in more favorable adaptability among women. Social workers or family members should focus on elderly people whose spouses died unexpectedly or within the last 6 months. Living with family members after SD can alleviate depression in elderly. --- Data Availability Statement: The datasets generated during the current study are not publicly available, but data are however available from the applicants upon reasonable request and with permission of the Ministry of Health and Welfare in Taiwan. --- Informed Consent Statement: Informed consent was obtained from all subjects involved in the study. --- Conflicts of Interest: The authors declare no conflict of interest.

In addition to increasing the mortality among older adults, spousal death (SD) increases their risk of depression. This study explored the factors affecting depression among widowed older adults to provide health care strategies for successful aging. A total of 710 adults older than 60 years completed a questionnaire before and after their spouses' deaths. The survey data included age, sex, ethnic group, education level, financial station socioeconomic status, SD (including time point), smoking status, alcohol consumption, self-rated health status, Center for Epidemiologic Studies Depression Scale score, mobility, and degree of support from relatives and friends. The proportion of participants with depression after SD was 1.7 times that of before SD (p < 0.0001). Worsened mobility (odds ratio [OR] = 1.3, p < 0.01), low self-rated health status (OR = 0.5, p < 0.01), and a high degree of support from relatives and friends (OR = 1.5, p < 0.01) had a significant positive correlation with depression after SD. The proportion of depression that occurred within 6 months after SD was 6.0 times higher than that of depression before SD. Participants who lived alone after losing their spouses who were healthy before their deaths exhibited a significantly increased proportion of depression after their spouses' deaths. Male sex, spouse's health, and the period of 6 months after SD are risk factors for depression in older adults. The maintenance of mobility, positive self-rated health status, and a shorter period of depression after a spouse's death result in more favorable adaptability among women. Social workers or family members should focus on older adults whose spouses died unexpectedly or within the last 6 months. Living with family members after SD can alleviate depression in older adults.

INTRODUCTION There is an imperative to tackle avoidable differences in health (White et al, 2009). This is now recognised in research design and reporting (Welch et al, 2012). Increasingly, studies are investigating the application of an 'equity lens' when reporting the effects of interventions in randomised trials (O'Neill et al, 2014;Welch et al, 2017). This is important as it permits the assessment of generalisability of trial findings to wider or specific populations. Collecting relevant data on social deprivation status also enables evaluation to determine whether deprivation is a treatment effect modifier and importantly whether an intervention might (albeit inadvertently) contribute to health inequity (Smith et al, 2020). Furthermore, from an ethical perspective Principle 14 of the Declaration of Helsinki (World Medical Association, 2013) states: populations which are underrepresented in medical research should be provided with appropriate access to participation in research. Accordingly, there is an ethical imperative to ensure that this data is collected to ensure it is possible to evaluate how representative research is to the wider community. To promote the collection and reporting of social deprivation data, frameworks have been developed to explicitly consider equity and social stratification in research conduct and reporting. One such framework is the PROGRESS framework (place of residence, race/ethnicity/culture/language, occupation, gender/sex, religion, education, socioeconomic status, and social capital) (Welch et al, 2012). It is acknowledged that participants in clinical trials are often not representative of the populations for which the interventions are intended (Witham et al, 2020). We previously reported a bibliometric analysis of 402 trials recruiting people with MSK disorders published between January 2019 to June 2020. [5] This assessed the frequency of reporting indexes such as the Index of Multiple Deprivation(Office for National Statistics, 2015), the USA Social Deprivation Index (Butler et al, 2013) and the New Zealand Index of Deprivation (NZDep2013) (Atkinson et al, 2014), in addition to measures/markers of deprivation. We reported that trials rarely reported deprivation indexes (Smith et al, 2020). Only two trials (1%) reported deprivation indices (Darlow et al, 2019;Hewlett et al, 2019). When assessed by individual measures of deprivation, 164 trials (41%) reported one or more measures. The most reported individual measures of deprivation were employment status (18%), highest level of educational attainment (15.5%) and morbidity (20%). Race (7%), ethnicity (6%) and annual income (1%) were infrequently reported. Community-based deprivation indicators such as premature death, quality of life, crime, location to amenities, housing quality, air quality and community status on road traffic accidents were not reported in any trial. One trial (Singh et al, 2019) presented subgroup results by deprivation measures. A number of important research priority were presented. Firstly, whilst there was wide diversity in deprivation measures reported, there was no consensus on what should be measured or on how trialists selected which measures to use. It is possible that poor understanding of the importance of collecting relevant indicators of deprivation (or their proxies) influences poor reporting of these measures. This has important implications for stratification and sub-group analyses to explore potential treatment effect-modification across people from different levels of deprivation. The purpose of this study was therefore to begin to address these uncertainties by asking researchers their thoughts around the measurement and reporting of deprivation in studies recruiting people with MSK disorders. Specifically, we sought to understand: • Which indices and measures of deprivation do MSK researchers know, use and report? • What factors affect the collection and reporting of deprivation indices and measures in MSK research? • Is there an association between the knowledge, collection and reporting of deprivation indices and measures, and the characteristics of respondents i.e. country of origin, academic background and clinical speciality? --- METHODS An international online, cross-sectional survey was undertaken. Participants were recruited through two methods. Firstly, we identified corresponding authors from 402 studies which formed the basis of our previous bibliometric analysis (Smith et al, 2020). Each corresponding author was emailed an invitation to complete the survey, including a hyperlink to the survey. Secondly, we advertised the survey in weekly tweets through Twitter. The tweets included the survey hyperlink. Consent was ensured through an opening page, prior to the first survey question. The online survey was delivered using Google Forms (Google LLC, California, USA). The survey is presented in Supplementary File 1. Potential participants were asked to read through an initial information page which detailed the survey's objectives and processes and explained how data would be stored and used. Following this, they were asked to provide signed consent. For those who did so, the survey was released for completion. The survey gathered information including: • Basic demographic information: country of residence; clinical/academic position; clinical background; highest educational award; method of research most associated with (quantitative/qualitative/mixed methods); years since receiving highest education award; years since most recent empirical study was published. • Knowledge of deprivation measures and indices: which of these deprivation indices have you: (1) heard of; (2) collected in a research project; (3) reported in a research paper/report? • Barriers to reporting deprivation indices and measures: why do you consider that deprivation data are rarely reported in MSK research publications? • Facilitators to reporting deprivation indices and measures: what do you feel could improve the reporting of deprivation data in MSK research publications? The survey was developed by the research team to identify factors which were hypothesised to be important in the update/reporting of social deprivation measures by MSK researchers. The proposed social deprivation measures presented to respondents were generated from the team's previous assessment of potential social deprivation measures reported in clinical trials (Smith et al, 2020). The survey took approximately 20 minutes to complete. It was piloted with five MSK physiotherapists and researcher who had previously collaborated with the research team. Their assessment did not change the content or format of the survey but informed the anticipated duration of the survey. Their responses were not included in the reported results. The survey link was posted on 20 th October 2020 and closed on 27 th November 2020. After three weeks of opening, a reminder email was sent to all non-responding corresponding authors from the first approach strategy to provide another opportunity to complete the survey. Weekly tweets were sent throughout the five weeks the survey was open. --- Data Analysis Descriptive statistics were used to summarise respondent's characteristics and their responses. --- RESULTS --- Respondent characteristics In total, 42 people responded to the survey. Respondent characteristics are presented in Table 1. In summary, the majority (42%) were based in the UK, 12% from Australia, 7% from the Netherlands and 5% from New Zealand, Brazil and the United States of America. Nine respondents originated from other countries across Europe, Asia and North America. Over half (60%) of respondents classified themselves as academics or researcher, with 69% being educated to doctoral level. Fifty-five percent considered clinical trials as their principal area of research, with 55% from a physiotherapy background, 17% orthopaedic surgery and 12% rheumatology. --- Awareness and usage of deprivation index and measures Over half of respondents had not heard of the six deprivation measures presented in the survey (Supplementary File 1; Table 2). The Index of Multiple Deprivation was the most well-known measure (26%) whereas the Carstairs Index was the least known (10%). Seven respondents (17%) responded that they had reported data from the Index of Multiple Deprivation, two (5%) had reported data on the New Zealand Index of Deprivation whilst one respondent (2%) had reported data on the Townsend Material Deprivation Score. Respondents identified other indices which were not originally proposed. These included the Index of Community Socio-educational Advantage (ICSEA, 2013) and Socio-Economic Indexes for Areas (SEIFA, 2008) by one respondent, and items from the Financial Literacy Core Questionnaire (FLCQ, 2013) by a second respondent. Table 2 summarises the results on the awareness and usage of deprivation measurements in MSK research. As this illustrates, the most commonly reported measures of deprivation reported were employment (60%), education (60%), ethnicity (50%) and morbidity (43%). The least frequently reported measures included social factors including local amenities (0%), air quality (0%) and community status on road traffic accidents (0%). Whilst a number of measures such as car ownership, proportion of people living in poverty, community statistics to premature death, housing quality and crime were known to over 90% of respondents, they were reported in research papers by less than five percent. Respondents identified additional measures not initially listed in the survey. These included: occupation and social capital (one respondent) income source (one respondent), use of food banks (one respondent) and health literacy such as the Health Literacy Questionnaire (Osbourne et al, 2013). --- Barriers and facilitator for adoption of deprivation index and measures A summary of the barriers and facilitators for adoption of deprivation measures and indices are presented in Table 3. The most common barriers to collecting these data were ranked as: prioritising other measures of contextual factors which are perceived as more important (79%), uncertainty on what data should be collected (71%), fear of missing data due to participants not wishing to provide this information (33%), time and costs to collect this data (33%) and societal attitudes towards minority and diverse communities (33%). One participant felt that not understanding the relevance of these data was an important barrier. Twenty-six percent of respondents believed that reporting deprivation was not important. The most frequently reported facilitator to collecting deprivation measures/indices was if there was a consensus on the measures to be collected and reported (88%). Seventy-nine percent of respondents felt that a greater awareness and understanding of the relationship between deprivation and MSK health would improve the likelihood of collecting these data, whilst 67% felt that mandating the use of reporting checklists that included measures of deprivation could facilitate the collection and reporting deprivation measures. --- DISCUSSION The findings of this survey indicate that people who research MSK conditions may largely have limited awareness of collecting and reporting measures of deprivation. This may explain why these data are infrequently reported in academic papers. Whilst researchers are aware of individual measures which indicate deprivation, these are not consistently reported to enable a full understanding of the characteristics of their patient cohorts. Factors such as believing deprivation data is a low priority, uncertainties around what data should be collected and concerns that they and their participants may feel uncomfortable about providing such data which could appear intrusive, are also acknowledged barriers. Increasing agreement on what measures or indexes should be collected, increasing awareness of the importance and requirement for collection, and a better understanding of the best ways to collect these data, were considered key facilitators to addressing this short-coming. The results of this survey reflect the findings of our previous bibliometric analysis of 402 MSK clinical trials (Smith et al, 2020). The low reporting of measures of deprivation indices but higher reporting of individual measures particularly morbidity, employment and education were reported. However, the survey indicates a far higher reporting of these measures than our bibliometric analysis. For instance, the bibliometric analysis reported the frequency of morbidity, employment and educational status in 20%, 18% and 16% of studies compared to 43%, 60% and 60% in this survey. This may reflect respondents being a self-selected sample of individuals more interested in deprivation measures than non-respondents. Nonetheless the current findings suggest deprivation measures are not frequently collected and reported, providing further evidence of this deficit in research conduct. A third of respondents reported concerns that participants in MSK trials may not feel comfortable about providing data on deprivation measures. Discomfort surrounding the social etiquette of collecting data on race and ethnicity within clinical encounters has been previously reported (Hasnain-Wynia et al, 2021). While Baker et al (2005) suggested there is support to collect this data, reservations about what the data will be used for and whether if appear discriminatory for both patients and healthcare professionals (Baker et al, 2005). Providing a full explanation of why the data is being collected may alleviate some of those concerns. Making this explanation directly related to improving care for the most disadvantaged to MSK interventions could be a most effective strategy to change this perspective (Baker et al, 2005). Advocates for the collection and reporting of deprivation data argue this is critically important for two key reasons. Firstly, reporting this data enables the reader to more-clearly understand the characteristics of study cohorts. This may facilitate greater generalisability of findings to specific populations, thereby aiding targeted research translation into practice for people with particular need (Welch et al, 2012;Smith et al 2005). For instance in MSK practice, a significantly higher prevalence of chronic pain is reported by black ethnic groups in the UK compared to other ethnic groups (Public Health England, 2018). Bangladeshi, Pakistani, Chinese and Black groups in the UK are twice as likely to be living on low incomes compared to the white population which also highlights systemic inequality (Public Health England, 2020). There is a higher prevalence of chronic pain in people living in deprived areas compared with those in least deprived areas and unemployed people with disability versus employed people (Public Health England, 2020;Karran et al, 2020). These combined factors may result in multiple health disparities (Newman et al, 2017). Thus, the reporting of deprivation measures is particularly pertinent in MSK practice. Secondly, collection of deprivation measures permits the analysis of trial findings by these potentially important factors. Poverty has been associated with increased symptom catastrophisation, pain severity, pain interference and disability in people with chronic pain (Newman et al, 2017). Greater deprivation is also associated with higher prescribing of slow-release, high-dose opioids (Curtis et al, 2019), despite benefits being minimal in chronic pain (Ballantyne et al, 2016;NICE, 2020). Accordingly, deprivation may be a major confounding factor in MSK research suggesting its role as an effectmodifier to outcome. For example, individuals with chronic pain and lower levels of education, literacy and working memory may benefit more from cognitive-behavioural therapy than a pain psychoeducational approach (Van Dyke et al, 2019). By not collecting data to catagorise deprivation level, it would remain unclear whether patients from different deprivation strata respond differently (or not) to specific interventions. The consistent collection of such data would permit such important subgroup analyses, to allow a better understand of whether such effects occur. Respondents identified several factors which may facilitate the collection and reporting of measures of deprivation in MSK trials. The key activity recommended was the increase in understanding of deprivation measures in MSK symptoms, why it is important that they are collected and determining what data should be collected, and how. This aligns with findings from stakeholder workshops to promote inclusive research in-which a need for a set of information resources and training to support research teams was identified as a key priority area (Public Health England, 2020). Organisations such as the UK National Institute for Health Research (NIHR, 2020) now request that trialists collect data on measures of social equity including: geographical location, age, ethnicity, socioeconomic status and access to health or social care, but on the understanding that this will assess inclusion into trials rather than to categorise/subgroup analyses. Nonetheless, this is an important step towards promoting deprivation measures as a standard element of data collection in MSK research. To take this work forward, further study to gain greater representation through consensus group activities is required. This would include engagement with trial stakeholders such as patient groups, funders, researchers and journal editors, to gain consensus on what deprivation measures should be reported in the least intrusive way. This would also include how to collect these data with different population, how to talk about the measures with participants and the etiquette needed around proxy measures of deprivation, such as income. Whilst this survey provides important findings to indicate how researchers may be supported to collect and report deprivation measures in MSK research, a number of important limitations should be considered. Firstly, the cohort consisted of 42 respondents. Whilst they represent views from across the world, this is a relatively small number of respondents in total. Furthermore, this cohort may be consisted self-selecting, as individuals volunteered to complete the survey and therefore we may only have gained the views of people interested in deprivation reporting, and not those who may be least interested in collecting or reporting these data. Consequently, there were insufficient data to explore the potential relationship between respondent demographic characteristics such as clinician versus academic positions, country of origin, educational or professional background, and their responses to the use of deprivation indices and measures. Such an analysis may have provided granularity on potential explanatory variables to reporting. --- CONCLUSION There is low awareness and reporting of measures of deprivation in MSK research. This is partly due to researchers not viewing this as important contextual information and uncertainty regarding which specific data would best illuminate social deprivation differences. There may be concerns that participants may not wish to report these data. Better promotion of why it is important may help to avoid issues of missing data. Activities are now warranted to improve reporting. This includes work to gain consensus on what data should be collected and how this may be used to allow assessment of generalisability and effect modification from interventions to treat MSK disorders. --- FIGURE AND TABLE LEGENDS --- (2.4) A guideline for collecting information on social deprivation and reporting guideline can be useful 1 (2.4) --- Supplementary File 1: Survey The collection and reporting of social deprivation measures in musculoskeletal research: an international survey study. Version 2.0 -dated: 08 October 2020 Section 1: Respondent Characteristics --- Q1.1 Where is your principal country of residence? • FREE-TEXT --- Q1.2 What is your principal job role? • Academic/Researcher • Clinicians • Clinical Academic • Other: FREE-TEXT --- Q1.3 What is your highest educational award? • BSc (Hons) • BA (Hons) --- Q1.4 How many years has it been since you received your highest educational award? • Free-Text YEARS --- Q1.5 What is the principal areas of research you are involved in? • Clinical Trials (e.g. randomised or non-randomised) --- • <unk>FREE TEXT EMAIL ADDRESS>> We would like to thank you for completing these questionnaires. Your contribution is much appreciated. If you have any questions or queries about this form, please contact: Email: toby.smith@uea.ac.uk

The reporting of deprivation measures is typically poor in musculoskeletal (MSK) research. This survey explored MSK researcher's perspectives on the deprivation indices and measures that are, or could be, collected and reported in their studies, and potential barriers and facilitators to collecting these data. An online international survey was undertaken to determine knowledge, use and reporting of deprivation indices and measures by MSK researchers and the factors which influence this. Data were analysed using descriptive statistics. 42 respondents from 16 countries completed the survey. The Index of Multiple Deprivation was the most well-known measure (26%) although only 17% had reported data from this index. Most commonly reported markers of deprivation were: employment (60%), education (60%) and ethnicity (50%). Most common barriers to collecting these data included: uncertainty on perceived importance of deprivation measures (79%), what should be collected (71%), and concerns on missing data and sensitivities from participants reluctant to provide this information (33%). Consensus on necessary measures to be collected and reported (88%) and improved awareness of the relationship between deprivation and MSK health (79%) were considered key activities to improve deprivation recording in MSK research. To conclude, there is poor awareness of the collection and reporting of deprivation measures in MSK research. Greater understanding on the importance of these data in reducing inequalities in MSK care is needed to facilitate improvement. This would enable greater assessment of generalisability and to assess whether interventions have different effects in people from different socio-economic groups.

Introduction Resilience, especially community resilience, is a function of an ecosystem's absolute ability to recover from various stresses and disturbances [1], particularly in relation to resisting shocks, economic downturns, climate change, globalization and environmental disasters [2,3]. The resilience of the community system; therefore, is not only a reflection of the pressure resistance, adaptability and innovation of the rural system but also is one of the breakthrough points, or at least a major turning point for current research into urban and rural development in China [4,5]. The rural community results from the coupling between people and the natural environment [6]. The internal structure and function of the rural community maintains a relatively balanced state depending on the resilience of the system and is reflected in economic, ecological and social subsystems [7]. The rural system is mainly composed of three subsystems: ecological, economic, and social [8]. Among them, the ecological subsystem resilience is mainly reflected in the stock of cultivated land, the coverage of forest and the storage capabilities of hilly ponds (fish ponds formed under natural low hill terrain) [9]. Furthermore, the economic subsystem resilience is generally measured using the per capita income of farmers and the yield of economic crops [10]; whilst that of the social subsystem is based on the rural social network and is a reflection of the rural people-land relationship [11]; thus, the social subsystem is significantly affected by the mobility of people, especially from rural villages. Resilience theory, which is at the core of the research reported in this paper, was originally employed to study the anti-interference ability of ecosystems [12][13][14][15], in other words, the maximum tolerance within which an ecosystem can maintain its stable state when facing change. In the 1970s, Gunderson and Holling introduced the resilience concept into social ecological systems. Holling stated that, resilience is the ability of a system to resist interference and maintain its function and control (i.e., the interference level that the system can resist while maintaining its function or the measure of capacity that the ecosystem absorbs the change and maintains its state), and further stated that, such a resilient system can effectively deal with the uncertainty that it could face in the future through the resilience of its management system [16][17][18]. As a comprehensive reflection of social, economic and ecological subsystems etc. in time and space, the rural community system has specific characteristics such as complexity, ability to self-organize, and diversity. With an increasingly close relationship and increasingly apparent mutual influence between people and nature, studying the response and adaptability of the system (with the social ecological system as the research object) to external interference from the perspective of resilience has recently attracted the attention of sustainable development research [19][20][21][22]. The prominent international academic organization "Resilience Alliance", led by Holling, used the theory of adaptive cycle to describe and analyze the dynamic mechanism of the social ecological system and put forward that such a system will successively go through four stages, namely development, protection, release, and update. These development stages constitute an adaptive cycle based on certain rules [23][24][25][26][27]. In the development phase, the system has just formed, and its resilience is gradually improved by continuously absorbing elements and establishing the connection between elements. In the protection phase, the system gradually developed, with the potential for growth decreasing and resilience weakening. In the release phase, the system is affected by a large number of interference factors and becomes disordered. At this time, system's resilience is low. Interference factors have some ruinous capacity to destroy the system and are the sources of the update phase. The system makes use of its resilience to obtain an opportunity of reorganization by learning and adapting to changes, and then realizes reorganization. Then, the system goes into the development stage again and goes back and forth, or the system collapses due to the lack of enough resilience, so as to exit the cycle [24,25,27]. In related research, the large-scale rapid outbreak of interference factors, such as the Sichuan earthquake in 2008, can serve to arouse the research interest of researchers and, consequently, the associated literature also appears in this stage [28][29][30]. However, in recent years, for the rapid and accidental external shock events, there has emerged an active interest in the adaptation and cultivation of the concept of resilience. Resilience research has shifted to the study of chronic and complex factors, with more emphasis on active human agency, using the "bottom-up" community-led development model [14,31,32]. In terms of research content, there are an increasing number of studies on the resilience of rural communities. Some scholars investigated the resilience of rural agriculture in the "agricultural" and "non-agricultural" processes, and concluded that resilience is driven by internal and external agricultural processes [33]. Furthermore, some scholars have formulated normative requirements and practical solutions of rural resilience development on the basis of known influencing factors of resilience, and connected the theory to practice using the community resilience framework [34,35]. The rural policy conducts critical research and proposes normative requirements regarding the ability of digital technology to help rural development, the provision of rural service solutions and the challenges of implementing localism [36]. In terms of research methods, scholars have used frameworks to assess the resilience of rural communities. For example, Jurjonas and Seekamp [37] proposed the rural coastal community resilience (RCCR) framework which include maintaining rural livelihoods, creating job opportunities, and addressing highly vulnerable populations in eastern North Carolina; Wilson [18] built on a conceptual framework to assess community resilience in rural China, and argued that Chinese government policies need to be substantially realigned if the resilience of rural communities is to be improved. Research on resilience theory has moved on from discussions of concept connotation and construction of a theoretical framework to the establishment of a resilience evaluation system [38]. However, despite there being many frameworks for evaluating community resilience using qualitative analysis, currently, only a few frameworks use a quantitative approach in their research. In addition, in the context of urban-rural integration and rural revitalization proposed by China, the existing community resilience evaluation framework has a relatively small scope of application, and there is no research on the evaluation of the resilience of urban fringe villages; furthermore, the research on system resilience function evaluation is relatively deficient due to the complexity and diversity of social ecological systems [39]. The aim of the research in this paper; therefore, is the measurement of the resilience regulating ability of rural communities close to the urban areas by constructing a new index system for the evaluation of social-ecological system resilience, thus developing a quantitative research approach, which combines the visualization function of the Geographical Information System (GIS) to reveal the resilience distribution characteristics of various subsystems on a village scale in order to provide a new research perspective for the development and revitalization of rural communities. --- Materials and Methods --- Study Area Zengcheng District is situated in the eastern part of Guangzhou City, which is located in central Guangdong Province to the south of China (Figure 1). After Zengcheng was annexed to Guangzhou as a district in 2014, in order to comply with the urban division adjustment of Guangzhou, Guangzhou Municipal Government produced the eastward development planning strategy and gradually transferred superior urbanization resources (such as the input of pillar industrial resources and structures) that accumulated over the years in Zengcheng District. As a core economic development area, Licheng Subdistrict became a district with prominent ecological, economic and social contradiction problems. In this paper, the remote sensing images of Licheng Subdistrict, taken in 2014 and 2018 were processed and an analysis of the characteristics of changes in the size of the built-up area was carried out. A comparison of the built-up area (Figure 2) in 2014 and 2018 shows an obvious trend of expansion that is both typical and representative. For these main reasons, the 24 administrative villages in the Licheng Subdistrict of Zengcheng District were chosen as the research object for this paper. The rural community system, which is located in the core urban development zo is an area of centralized exchange and interchange of urban-rural elements and present distinct interference landscape variation phenomenon on economic, ecological and soc subsystems. The massive influx of new urban resources effectively broke through interface of traditional rural communities and forced rural grassroot communities to tu from closure to opening-up. The resilience of rural communities; therefore, is faced w unprecedented challenges. --- Data Source The research underpinning this paper takes all of the administrative villages (24) the Licheng Subdistrict as evaluation units; uses basic village data from the latest sta tics sourced from the town government of Licheng Subdistrict together with the 20 village data provided by the Zengcheng Urban and Rural Planning Design and Resea Institute (Table 1). However, due to the Zengcheng Gualv Lake water conservation p ject, the residents of the villages of Luogang, Mingxing, Taiping, Guangming a Xigualing were relocated which meant that these villages no longer had the attribute o The rural community system, which is located in the core urban development zone, is an area of centralized exchange and interchange of urban-rural elements and presents a distinct interference landscape variation phenomenon on economic, ecological and social subsystems. The massive influx of new urban resources effectively broke through the interface of traditional rural communities and forced rural grassroot communities to turn from closure to opening-up. The resilience of rural communities; therefore, is faced with unprecedented challenges. --- Data Source The research underpinning this paper takes all of the administrative villages (24) in the Licheng Subdistrict as evaluation units; uses basic village data from the latest statistics sourced from the town government of Licheng Subdistrict together with the 2016 village data provided by the Zengcheng Urban and Rural Planning Design and Research Institute (Table 1). However, due to the Zengcheng Gualv Lake water conservation project, the residents of the villages of Luogang, Mingxing, Taiping, Guangming and Xigualing were relocated which meant that these villages no longer had the attribute of a traditional rural area. Therefore, the basic data pertaining to these five villages were ignored, leaving the remaining 19 administrative villages as the research object. This research takes the economic subsystem resilience, ecological subsystem resilience and social subsystem resilience as primary indexes and designates secondary indexes according to the attribute features of the subsystems to construct the rural community resilience function evaluation index system. The characteristics of changes in geographical and human landscapes that the ecological, economic, and social subsystems show under the interference from urbanization are considered comprehensively for index selection and are quantified for the designation of secondary indexes (Table 2). Through the three primary indexes and eight secondary indexes, the resilience index system is constructed to analyze the resilience function of the rural community system for Licheng Subdistrict. Indexes that help strengthen the ecological environment, economic income and social stability are positive indexes. The higher in magnitude of a positive index value implies that the resilience is stronger. On the contrary, indexes that are bad for the development of the rural ecological social system are negative indexes. The higher in magnitude negative index value implies that the resilience is weaker. For the standardized processing of the original data, this paper adopts the range method as expressed in Equation (1): Y ij = X ij -X min /(X max -X min ) X max -X xj /(X max -X min )(1) In Equation ( 1), Y ij is the standardized value of the index in the year; X xj is the original value of the index, X max and X min are the maximum and minimum values of the index, respectively. The entropy weight method is used to determine the weight [40,41], so as to effectively avoid any influence of the expert's subjective judgment errors on the weight analysis and, thus, to make the evaluation result more objective. The various calculations are as follows: Equation ( 2) is used to calculate the proportion of Index j in Year i P ij = Y ij <unk> m i=1 Y ij(2) Equation ( 3) is used to calculate the entropy of Index i: e j = -k m <unk> i=1 P ij <unk> ln P ij(3) Equations ( 4) and ( 5) are used to calculate the weight of each index based on the result from Equation (3): g j = 1 -e j(4) w j = g j / m <unk> i=1 e j(5) Equation ( 6) is used to calculate the resilience function value for the subsystems of each administrative village: Z ij = w j <unk> Y ij(6) Equation ( 7) is used to calculate the total system resilience value for each administrative village: S ij = Z ij + Z ij + Z ij(7) The data for the administrative villages of Licheng Subdistrict after standardized processing are listed in Table 3. Using the entropy method as detailed in Section 2.3.2, the weights for the various indexes were determined and used in the calculation of the resilience function measurement results, which are listed in Table 4. The visual resilience function values determined using the GIS Nature Breaks method are shown in Table 4 and Figure 3. --- Strength and Spatial Distribution Characteristics of Resilience of Rural Subsystems --- Ecological Subsystems For the resilience function values of the ecological subsystems for Licheng Subdistrict, the strong value range is 0.3001-0.4000; the relatively strong value range is 0.2001-0.3000; the moderate value range is 0.1001-0.2000; and the weak value range is 0-0.1000. The resilience function values of the ecological subsystems for most villages in Licheng Subdistrict fall into the moderate and weak value ranges. There are 15 administrative village units in this range, accounting for 78.9% of the total. It can be seen in Figure 4 that the ecological subsystems for Licheng Subdistrict lack resilience and the ecological environment is generally highly fragile. The ecological environment is closely related to a good agricultural foundation. The strong resilience value range of the ecological subsystems is centralized in the villages of Qun'ai, Tangxia, Mutan, and Longjiao in the northern part of Licheng Subdistrict, which is the main production area for large scale agricultural units, enterprises, farming cooperatives and production bases of the wider Zengcheng District. Adopting completed ecological environment protection measures, Qun'ai Village is the main cultivation base --- Results --- Rural System Resilience Strength and Spatial Distribution Characteristics --- Rural System Resilience Strength Figure 3 is the rural community system overall resilience function values, from which it can be seen that Licheng Subdistrict is in the value range 0.1462-0.4647, a moderate resilience state. According to the classification standard from relevant research [42,43], resilience strength can be divided into four grades: The strong value range is 0.3751-0.5000; the relatively strong value range is 0.2501-0.3750; the moderate value range is 0.1251-0.2500; the weak value range is 0.0000-0.1250. No village in Licheng Subdistrict has yet fallen into the weak value range. However, five villages fall into the strong range and five villages fall into the moderate range, accounting for 31.58% of the total. Ten villages fall into the relatively strong range, accounting for 47.37% of the total, the highest proportion. It can be found that most villages in Licheng Subdistrict have good adapting and learning abilities in addition to a good regulation ability. --- Rural System Resilience Spatial Distribution Characteristics The spatial distribution of the resilience strength of administrative villages in Licheng Subdistrict is characterized by dispersion in whole and aggregation in part. This means that villages with different resilience levels are distributed dispersedly, but villages that fall into the strong value range are aggregated in northern central areas (such as Mutan village, Longjiao village, Qiaotou village and Qunai village) and some eastern areas (such as Chengfeng village, Xiajie village). It can be seen that villages with strong resilience are characterized by linkage and sprawl, which helps northern and central regions of Licheng Subdistrict absorb urban resources and promote rural revitalization. --- Strength and Spatial Distribution Characteristics of Resilience of Rural Subsystems --- Ecological Subsystems For the resilience function values of the ecological subsystems for Licheng Subdistrict, the strong value range is 0.3001-0.4000; the relatively strong value range is 0.2001-0.3000; the moderate value range is 0.1001-0.2000; and the weak value range is 0-0.1000. The resilience function values of the ecological subsystems for most villages in Licheng Subdistrict fall into the moderate and weak value ranges. There are 15 administrative village units in this range, accounting for 78.9% of the total. It can be seen in Figure 4 that the ecological subsystems for Licheng Subdistrict lack resilience and the ecological environment is generally highly fragile. --- Economic Subsystems The function values of the economic subsystems for Licheng Subdistrict are in the range 0.010-0.200, implying a relatively strong overall resilience function with 84.21% of the administrative villages reaching moderate or higher levels. However, the function values show a large span, and the resilience function ranges of the economic subsystems differ significantly, as can be seen in Figure 5. For instance, the strong economic resilience function value range is located in the central part of the subdistrict, mainly including Xiajie Village and Sanlian Village. Whereas the relatively strong function value range is located in the northernmost part of the subdistrict (i.e., the Tangsha Village-centered The ecological environment is closely related to a good agricultural foundation. The strong resilience value range of the ecological subsystems is centralized in the villages of Qun'ai, Tangxia, Mutan, and Longjiao in the northern part of Licheng Subdistrict, which is the main production area for large scale agricultural units, enterprises, farming cooperatives and production bases of the wider Zengcheng District. Adopting completed ecological environment protection measures, Qun'ai Village is the main cultivation base of mesona chinensis benth (one of the "Ten Treasures in Zengcheng"). The central area basically falls into the low resilience function value range and an aggregation block with Xiajie Village in the center is formed. --- Economic Subsystems The function values of the economic subsystems for Licheng Subdistrict are in the range 0.010-0.200, implying a relatively strong overall resilience function with 84.21% of the administrative villages reaching moderate or higher levels. However, the function values show a large span, and the resilience function ranges of the economic subsystems differ significantly, as can be seen in Figure 5. For instance, the strong economic resilience function value range is located in the central part of the subdistrict, mainly including Xiajie Village and Sanlian Village. Whereas the relatively strong function value range is located in the northernmost part of the subdistrict (i.e., the Tangsha Village-centered continuous rural community); the centrally distributed moderate function value range is located between the strong value range and relatively strong value range. --- Social Subsystems The resilience values of the social subsystems for Licheng Subdistrict are obviously weak in the whole and are mostly centralized in the low value range of 0.0000-0.0750. Villages with the weakest social subsystem resilience values are in a zonal distribution and in a layered layout around villages with higher economic levels (Figure 6). Secondary indexes of social subsystems are obtained through quantitative processing and analysis of the situation of rural population loss, talent introduction efforts and cultural heritage protection. Rural social resilience is based on the connection degree and social familiarity of population settlements (i.e., the acquaintance society network) [44,45]. Since the 1990s, after Guangzhou entered into the stage of high-speed urbanization, villages close to cities were brought to the urbanization system resulting in a strong economic pull factor that further attracted the labor force from less developed surrounding villages. The frequent population flows in rural communities; therefore, increased the heterogeneity of these communities [46,47]. Under such unstable conditions, the stability of existing social relations in rural communities and the knowledge of rural subjectivity are especially Comparing the villages of Xiajie, Sanlian and Liantang, which are all in the strong economic resilience function value range, it is found that the annual agricultural income of Sanlian is more than RMB 20 million, whilst Xiajie and Liantang have almost no agricultural income but the per capita income of farmers in these villages ranks top as most of their incomes come from rent and the ecological leisure tourism industry. Tourism matching industries, such as farm tourism and native products, are developed by relying on the unique ecological resources. Rural economic development no longer is based on a single agricultural production methodology but has evolved towards diversified production modes. --- Social Subsystems The resilience values of the social subsystems for Licheng Subdistrict are obviously weak in the whole and are mostly centralized in the low value range of 0.0000-0.0750. Villages with the weakest social subsystem resilience values are in a zonal distribution and in a layered layout around villages with higher economic levels (Figure 6). Secondary indexes of social subsystems are obtained through quantitative processing and analysis of the situation of rural population loss, talent introduction efforts and cultural heritage protection. Village and consciously adhere to its rules and regulations. This continuously helps to promote rural social resilience. --- Discussion Based on the calculation results above, it can be concluded that rural communities in Licheng Subdistrict have a good overall resilience, but the development of subsystems is extremely unbalanced and there are many problems that need to be solved. Villages close to the core urban development area have a relatively strong overall resilience, good adapting and learning abilities that can generally meet the basic living demands of villagers. The resilience of subsystems, however, is unbalanced. Although the resilience of economic subsystems is relatively strong, the resilience of social and ecological subsystems differ significantly and the overall situation; therefore, is not optimistic. This is especially the case concerning the resilience of the ecological subsystem, which is apparently weakened against the long-term development of rural communities. The diversified economic development modes provide an appreciation space for villages to attract the inflow of capital. However, a destructive impact has been made on the ecological environment due to the imperfect rural market mechanism, which is not conducive to the stable development of the rural ecological social system. Currently, rural communities in the subdistrict are in the phase of high-speed transformation of urban-rural elements. In the process of urban sprawl, traditional rural life, dominated by farming culture with blood-tied relationships, has been eroded by modern culture. New planning, design, and construction achievements are separated from the original cultural landscape planning of traditional villages. The conflict between traditional civilization and modern civilization is exactly a reflection of the rural community system resilience being subjected to "Strong Interference". The research carries out index quantification for such an "Interference Degree", measures it through the threshold range, explores the strength of the overall resilience function of rural communities under the interference of urbanization through Rural social resilience is based on the connection degree and social familiarity of population settlements (i.e., the acquaintance society network) [44,45]. Since the 1990s, after Guangzhou entered into the stage of high-speed urbanization, villages close to cities were brought to the urbanization system resulting in a strong economic pull factor that further attracted the labor force from less developed surrounding villages. The frequent population flows in rural communities; therefore, increased the heterogeneity of these communities [46,47]. Under such unstable conditions, the stability of existing social relations in rural communities and the knowledge of rural subjectivity are especially critical for the promotion of system resilience. For example, in Xiajie Village, which is in the strong social resilience value range, there are many cultural heritage assets such as the Xiajie Ancient Post Road and Huaiyin Hall. Strengthening cultural awareness can contribute to the cohesion of communities. In addition, Xiajie Village attaches great importance to talent introduction. Aiming to protect traditional ancient villages and restore rural vitality, the Ancient Village Friendship Association of Xiajie Village has assembled a team of professional planners who are proud of their native land and are actively participating in events to promote rural revitalization. Besides, activities periodically held in Xiajie Village provide consultation services for migrant workers including health examinations and parent-child communication activities to enable migrants who live there to develop a sense of belonging to Xiajie Village and consciously adhere to its rules and regulations. This continuously helps to promote rural social resilience. --- Discussion Based on the calculation results above, it can be concluded that rural communities in Licheng Subdistrict have a good overall resilience, but the development of subsystems is extremely unbalanced and there are many problems that need to be solved. Villages close to the core urban development area have a relatively strong overall resilience, good adapting and learning abilities that can generally meet the basic living demands of villagers. The resilience of subsystems, however, is unbalanced. Although the resilience of economic subsystems is relatively strong, the resilience of social and ecological subsystems differ significantly and the overall situation; therefore, is not optimistic. This is especially the case concerning the resilience of the ecological subsystem, which is apparently weakened against the long-term development of rural communities. The diversified economic development modes provide an appreciation space for villages to attract the inflow of capital. However, a destructive impact has been made on the ecological environment due to the imperfect rural market mechanism, which is not conducive to the stable development of the rural ecological social system. Currently, rural communities in the subdistrict are in the phase of high-speed transformation of urban-rural elements. In the process of urban sprawl, traditional rural life, dominated by farming culture with blood-tied relationships, has been eroded by modern culture. New planning, design, and construction achievements are separated from the original cultural landscape planning of traditional villages. The conflict between traditional civilization and modern civilization is exactly a reflection of the rural community system resilience being subjected to "Strong Interference". The research carries out index quantification for such an "Interference Degree", measures it through the threshold range, explores the strength of the overall resilience function of rural communities under the interference of urbanization through the combination of both qualitative and quantitative methods, and provides strategies for strengthening community resilience according to the analysis results (Figure 7). social relations. In order to pursue employment opportunities and a good living environment, most villagers flow to areas with higher economic levels and more comfortable ecological environments. The rural population flow has the characteristic of selecting the nearest optimal location, which is reflected in the layered decline of the resilience of the social subsystem, forming a typical core-edge structure. The optimization strategy for the resilience function of the social subsystem of rural areas can start from the cultivation of rural subjectivity [26]. The arrival of the so-called global competition age, served to interfere with the endogenous development of rural communities which were essentially squeezed by a more powerful foreign culture, resulting in those rural communities losing the internal driving force of development and autonomy. A prominent manifestation of this is the massive loss of villagers from several villages in Licheng Subdistrict and the inflow of a foreign population. Therefore, it follows that strengthening the subjectivity of rural communities is especially important for managing "non-acquaintance society" relations. With a long history, the villages in Licheng Subdistrict were mostly built in the Qing Dynasty. Many ancestral halls are set up and the profound ancestral idea foundation is kept in these villages. At the same time, showing significant cultural potential for the whole subdistrict, there is one historical and cultural site protected at the provincial level, two sites are protected at the municipal level, and 33 sites protected at the district level (36 in total). Therefore, in respect of the strategy of adjusting the resilience of the rural social subsystem, the rich cultural resources can be used to cultivate the cultural identity of villagers, reshape the human network of the social system, and strengthen the regulating ability of social resilience. For example, Licheng Subdistrict has responded positively to the construction of the cultural line of the ancient post road carried out in the last two years. Professionals in planning, local elites and the government have cooperated actively to carry out investigations and obtain on the spot evidence so as to arouse the villagers' recognition of their own culture imperceptibly. In addition, social activities such as the post road activity also promote the regeneration of the rural social system. --- Conclusions The aim of the study was to assess the resilience regulating ability of rural communities by constructing a new index system for the evaluation of social-ecological system The strong value range of rural system resilience is mainly distributed in the northern region and partially centralized in the central region. Due to the differences in the abilities of the economic, social, and ecological systems to cope with interference, the resilience of each subsystem has different spatial distribution characteristics. The ecological subsystem has an obvious polarization in resilience. The northern region comprises many modern agricultural bases which have better soil and water conservation capacities. Therefore, the ecological system in the northern region is less damaged. Through spot investigation, it is found that Chengfeng, Jinxing, Qingfeng, and Wuyi around Xiajie are all urban villages. Agricultural production has seceded from the main local economic structure. Working outside and leasing houses for rent are the main means of livelihood of the villagers. As the earliest core area that takes over the outflows of urban resources, the central area of Licheng Subdistrict is undergoing increasingly severe urban sprawl that is causing a significant decrease in cultivated land which further causes a reduction of the eco-environmental quality in rural areas and directly affects the resilience of the local rural ecological system. Rebuilding a stable ecological pattern is critical for the current weak rural ecological subsystem resilience function. The original ecological bases of most villages were gradually changed and original farming landscapes were devoured by industrial landscapes under the trend of urban sprawl. The blurring of urban and rural landscape boundaries and the integration of rural spatial features are serious problems in the current rural development process. Restoring the rural landscape is the basis for a good system cycle. To reshape the ecological pattern, the coordination mechanism between urbanization and ecological environment should be established. External tension focuses on strengthening population quality and legal norms, technological innovation, and industrial upgrading, while internal tension depends on residents' awareness of environmental protection. The enhancement in the resilience function values of the rural economic subsystems for Licheng Subdistrict can be attributed to an influx of new activities during the process of urbanization. These activities, such as the rental of houses, ecological leisure tourism, and industrial development, have created a favorable environment for structural and functional reorganization of the system, so as to adapt the systems to the urbanization development mode. Loreau et al. stated the importance of species richness of the system [48]. The immigration of alien species can improve the adaptability of the ecological system to changes, causing a fundamental reaction of the ecological system and in so doing, changing its performance characteristics. The accumulation of adaptability allows for improvements in system resilience [49,50]. Rural economic development no longer is based on a single agricultural production methodology but has evolved towards diversified production modes. This is consistent with the theoretical analysis on the mutual influence between biodiversity and resilience [51]. In view of the high average economic system resilience of rural communities in Licheng Subdistrict, villages need to use their own resources to find a breakthrough in the market, change the role of passive recipients, and take the initiative to use diversified urban resources to improve the creativity of the system. The capital investment frenzy caused by urbanization has become an irresistible trend. If the resilience of rural economic systems cannot actively filter and absorb the outflow of resources, it will inevitably be swallowed up by market competition and fall victim to the capital cycle. Therefore, when facing capital invasion, the capital opportunity of the market can be utilized to enhance the resistance ability of the economic system and adapt to its competitive and changeable interference. The sustainable development of rural areas cannot rely on a single industry. The current technological developments have greatly reduced the manpower requirements for agricultural production, and the redundant human resources need more employment opportunities to make a living. The diversified economic development mode can not only solve the problem of a surplus of farmers, but also reduce the centralized destruction of interference sources. It is through the change of economy and ecology that has changed the original rural social relations. In order to pursue employment opportunities and a good living environment, most villagers flow to areas with higher economic levels and more comfortable ecological environments. The rural population flow has the characteristic of selecting the nearest optimal location, which is reflected in the layered decline of the resilience of the social subsystem, forming a typical core-edge structure. The optimization strategy for the resilience function of the social subsystem of rural areas can start from the cultivation of rural subjectivity [26]. The arrival of the so-called global competition age, served to interfere with the endogenous development of rural communities which were essentially squeezed by a more powerful foreign culture, resulting in those rural communities losing the internal driving force of development and autonomy. A prominent manifestation of this is the massive loss of villagers from several villages in Licheng Subdistrict and the inflow of a foreign population. Therefore, it follows that strengthening the subjectivity of rural communities is especially important for managing "non-acquaintance society" relations. With a long history, the villages in Licheng Subdistrict were mostly built in the Qing Dynasty. Many ancestral halls are set up and the profound ancestral idea foundation is kept in these villages. At the same time, showing significant cultural potential for the whole subdistrict, there is one historical and cultural site protected at the provincial level, two sites are protected at the municipal level, and 33 sites protected at the district level (36 in total). Therefore, in respect of the strategy of adjusting the resilience of the rural social subsystem, the rich cultural resources can be used to cultivate the cultural identity of villagers, reshape the human network of the social system, and strengthen the regulating ability of social resilience. For example, Licheng Subdistrict has responded positively to the construction of the cultural line of the ancient post road carried out in the last two years. Professionals in planning, local elites and the government have cooperated actively to carry out investigations and obtain on the spot evidence so as to arouse the villagers' recognition of their own culture imperceptibly. In addition, social activities such as the post road activity also promote the regeneration of the rural social system. --- Conclusions The aim of the study was to assess the resilience regulating ability of rural communities by constructing a new index system for the evaluation of social-ecological system resilience. This approach contrasts with most other resilience evaluation frameworks that are based on qualitative analysis, for example, using NVivo (QSR International, Melbourne, Australia) to analyze data obtained through semi-structured interviews [52]. The quantitative approach using the new index essentially contributes to broadening the scope of resilience measurement research and appears to be especially beneficial in the analysis of rural areas that are close to urban areas. We also analyzed the evolution mechanism and driving mechanism of rural communities from the perspective of protection and development strategy in the new dimension of resilience, and explored a new perspective for rural rejuvenation in practice. The main contributions of this paper are as follows: (1) Rural systems close to urban areas have a stronger resilience strength, its spatial distribution shows dispersion in whole and aggregation in part. (2) The resilience of most rural economic subsystems are higher, while the ecological and social subsystems are generally weaker. (3) In addition, we proposed some policy suggestions to promote the sustainable development of rural areas and realize rural revitalization. The influences of multiple factors need to be considered for the establishment of the resilience index system, which for the research area selected in this paper may have some limitations in the determination of the rural community resilience function values due to limited data. More comprehensive

Resilience of rural communities is becoming increasingly important to contemporary society. In this study we used a quantitative method to measure the resilience regulating ability of rural communities close to urban areas-in Licheng Subdistrict, Guangzhou City, China. The main results are as follows: (1) Rural systems close to urban areas display superior adapting and learning abilities and have a stronger overall resilience strength, the spatial distribution of which is characterized by dispersion in whole and aggregation in part; (2) the resilience of most rural economic subsystems can reach moderate or higher levels with apparent spatial agglomeration, whilst the ecological subsystem resilience and social resilience are generally weaker; the spatial distribution of the former shows a greater regional difference while the latter is in a layered layout; (3) some strategies such as rebuilding a stable ecological pattern, making use of urban resources and cultivating rural subjectivity are proposed on this basis, in order to promote the sustainable development of rural areas and realize rural revitalization. This work also gives suggestion for the creation of appropriate and effective resilience standards specifically targeted for rural community-aiming to achieve the delivery of local sustainability goals.

are based on qualitative analysis, for example, using NVivo (QSR International, Melbourne, Australia) to analyze data obtained through semi-structured interviews [52]. The quantitative approach using the new index essentially contributes to broadening the scope of resilience measurement research and appears to be especially beneficial in the analysis of rural areas that are close to urban areas. We also analyzed the evolution mechanism and driving mechanism of rural communities from the perspective of protection and development strategy in the new dimension of resilience, and explored a new perspective for rural rejuvenation in practice. The main contributions of this paper are as follows: (1) Rural systems close to urban areas have a stronger resilience strength, its spatial distribution shows dispersion in whole and aggregation in part. (2) The resilience of most rural economic subsystems are higher, while the ecological and social subsystems are generally weaker. (3) In addition, we proposed some policy suggestions to promote the sustainable development of rural areas and realize rural revitalization. The influences of multiple factors need to be considered for the establishment of the resilience index system, which for the research area selected in this paper may have some limitations in the determination of the rural community resilience function values due to limited data. More comprehensive data indexes, such as the ratio of agricultural income and the ratio of secondary and tertiary industries, can be considered in subsequent research. In addition to the expansion of objective data, on-the-spot tracking investigations can be enhanced to verify the accuracy of the results in practice. The future planning should not only optimize the material space, but also use a comprehensive perspective to analyze the rules of the rural social ecological system, stimulate the driving mechanism of rural development, and encourage more villages to move towards a new phase of growth and protection. --- Data Availability Statement: Not applicable. --- Conflicts of Interest: The authors declare no conflict of interest.

Resilience of rural communities is becoming increasingly important to contemporary society. In this study we used a quantitative method to measure the resilience regulating ability of rural communities close to urban areas-in Licheng Subdistrict, Guangzhou City, China. The main results are as follows: (1) Rural systems close to urban areas display superior adapting and learning abilities and have a stronger overall resilience strength, the spatial distribution of which is characterized by dispersion in whole and aggregation in part; (2) the resilience of most rural economic subsystems can reach moderate or higher levels with apparent spatial agglomeration, whilst the ecological subsystem resilience and social resilience are generally weaker; the spatial distribution of the former shows a greater regional difference while the latter is in a layered layout; (3) some strategies such as rebuilding a stable ecological pattern, making use of urban resources and cultivating rural subjectivity are proposed on this basis, in order to promote the sustainable development of rural areas and realize rural revitalization. This work also gives suggestion for the creation of appropriate and effective resilience standards specifically targeted for rural community-aiming to achieve the delivery of local sustainability goals.

Introduction Around one billion people worldwide are disabled, i.e. experience moderate to severe functional limitations [1]. The prevalence of disability is expected to further increase due to factors including demographic change and growing importance of non-communicable diseases and injuries [1,2,3,4,5]. Although ''for the society as a whole, 'function[ing]' must be a major effect'' (p. 881), comparative effectiveness research and, surprisingly, even disparities research and social epidemiology have rarely made use of comprehensive measures of functioning [6]. Switzerland, as one of the richest countries of the world [7], provides a good example for studying the role of social inequalities in functioning. If we find social inequalities of functioning in a country with a high expenditure on health care and social security [7], we would expect them to be even more pronounced in less wealthy countries. Several previous epidemiologic studies have shown social gradients of disability [8,9,10,11,12,13,14,15,16] and chronic pain [17]. These were, however, mostly confined to the elderly population [8,9,10,11,12,13]. Comparative data on younger and middle-aged populations are still lacking. Predictors have usually been comprised of measures of socio-economic status (SES) [12,13], with indicators of social integration being used less often [8,16]. So far, studies have assessed disability mainly by dichotomous outcome variables, e.g. asking respondents whether or not they experienced ''a long lasting condition that substantially limits one or more basic physical activities [...]'' (p. 698) [12]. Even if more complex scales were used, information was often reduced to a binary outcome [1,13]. In the case of Switzerland, several reports on social inequalities and health are available. Based on data from the Swiss Health Survey 1997, a social gradient of disability was shown in persons aged 65 and older, but younger respondents were not examined [18]. Information on impairments and activity limitations was combined and the outcome was again dichotomized. A report based on the Swiss Health Survey 2007 showed increasing problems in relation to activities of daily living with older age but did not account for SES [19]. Rather than viewing disability as a static attribute of an individual or population, the WHO's International Classification of Functioning, Disability, and Health (ICF) proposes a dynamic and multi-dimensional approach [20,21,22,23]. Disability can be conceptualized as a continuum of problems in functioning, varying across life situations and environments [22]. This concept of functioning encompasses the actual state of body functions and structures, as well as activity and participation (A&P) in a given environment. In a recent study of British women aged 64-83, Dale and colleagues [24] found, for instance, that only restrictions in participation and limitations in complex activities were associated with increased risk of mortality in their fully adjusted models, while impairment was not. Drawing on the ICF framework, we propose a theoretical model (Figure 1) to guide our empirical analysis. While controlling for demographics and health behaviours (pathway 1), we hypothesize (a) that the components of functioning (i.e. impairment, pain, and A&P limitations) are determined by SES and social integration (pathway 2); (b) that A&P limitations are influenced by impairment and pain (pathway 3), and (c) that impairment, A&P limitations, low SES and poor social integration exert a cumulative effect on perceived health [25] (pathway 4). --- Methods --- Design We analyzed data from the Swiss Health Survey (SHS) 2007 [26] obtained from the Federal Statistics Office of Switzerland (Bundesamt fu <unk>r Statistik: Schweizerische Gesundheitsbefragung 2007). The SHS 2007 adopted the ICF framework by including measures of functioning for population surveys developed by the United Nations' Washington Group on Disability Statistics [27]. --- Sample & data collection The survey was based on a representative stratified random sample of all private Swiss households with fixed line telephones. Household members aged 15 and above were randomly selected. The survey was completed by 18,760 persons, corresponding to a participation rate of 66%. People who did not speak German, French or Italian, were asylum seekers, had an extremely poor health status or lived in nursing homes were excluded [26]. Data were collected with Computer Assisted Telephone Interviews. People aged 75 or above were interviewed face-toface [26]. --- Study variables Functioning. The SHS 2007 included several measures of impairment, pain and A&P limitation with heterogeneous scales of questionnaire items (nominal, ordinal and metric variables). Therefore, we dichotomized all respective information such that a value of 1 indicated the presence of problems and a value of 0 corresponded to none. Following the ICF framework, we then constructed 3 separate sum indices. The impairment index comprised problems in the following areas: vision, hearing, speaking, body weight (body mass index,16 kg/m2 or.30 kg/m2), urinary continence, defecation, energy and drive, sleep, and heart rate. The A&P limitation index was based on problems in the following activities: walking, eating, getting up from a bed or chair, dressing, toileting, bathing, preparing meals, using a telephone, doing the laundry, finances/accounting, public transport, major household tasks, and shopping. The pain index used information on anatomical locations of pain including head, chest, stomach, back, hands and joints. Graphical and statistical tests (using Stata command nbvargr) showed that these indices shared a Poisson distribution. Perceived health. Perceived health was measured by asking respondents how they would rate their health in general (ordinal rating scale with the 5 options: very good, good, fair, poor, and very poor). SES. Level of education, income, paid employment, gender and migration background were used as indicators of SES. Education was categorised according to the International Standard Classification of Education as the total number of years of formal education including school and vocational training [28]. Income was measured by net equivalent household income according to OECD criteria [29]. Participants were asked whether or not they were currently in paid employment. Information on migration background was gathered by determining whether at least one parent was of foreign origin. Social integration. We constructed 2 sum indices for social network utilization and perceived social support individually. The social network utilization index encompassed information on whether the following activities were performed at least weekly: visits of family members, phone calls with family members, visits of friends, phone calls with friends, and participation in clubs or associations. The perceived social support index comprised information on four items: having someone to turn to, having at least one supportive family member, feeling lonely (reversely coded), and missing someone to turn to (reversely coded). In addition, we considered whether respondents were married or not. Social class. In order to test for an overall social gradient in disability, we integrated information on income, education and social integration and constructed a social class index with 4 groups. We thus followed Bourdieu's concept of social class representing core dimensions of economic, cultural and social capital [30]. Persons with income, education and a combined social support and network index below average were assigned to the first group (lowest class). People with 2 out of 3 indicators below average were assigned to the second group (low to middle class). Respondents with one indicator below average were assigned to the third class (middle to high class); and those with all indicators above average to the fourth group (highest class). The 2 indices on social integration were combined to avoid overrepresentation of social capital. The resulting distribution of social classes in the Swiss general population had high face validity with most respondents being assigned to the middle classes and the lowest percentage to the highest class (Table 1). --- Data analysis All computations were made with Stata 11 (Stata Corp, College Station, USA) with the exception of the missing value imputation which was performed with R. The indices described above were checked using Stata's confirmatory factor analysis module confa [31]. For each index, the index solution was tested against the null hypothesis of a diagonal structure of the covariance matrix (i.e. independence). To test our hypotheses, we calculated separate mixed effects regressions for the 4 outcomes of impairment, pain, A&P limitation, and perceived health. The 26 Swiss Cantons (territories equivalent to counties) were nested according to language regions (German, French and Italian). Five Cantons are bilingual and in order to obtain a fully nested model were subdivided into smaller areas according to the dominant language. Random intercepts were introduced for the language region and Canton (or sub-area for the 5 bilingual Cantons) to account for unobserved heterogeneity due to clustering of respondents by language region and Canton. We used Poisson links to model pain, impairment, and activity limitation, and an ordinal probit link to model perceived health, assuming a latent metric variable underlying the ordinal scale distribution [32]. In order to assess the models' fit to the data, we calculated Pseudo R 2 values by comparing the log likelihood of the full model with that of an empty model. [33] For the Poisson type models, we used the Stata command xtmepoisson and for the ordinal probit model, the command gllamm [34]. We calculated 4 models for each outcome: an overall model and 3 separate models for persons aged 15-30 (younger group), 30-54 (midlife group), and 55 and older (older group). The age cut-offs were chosen to allow comparability with other research, particularly on elderly persons. Bonferroni correction was used to account for multiple testing, i.e. empirical p-values were multiplied by 4. To test for an overall social gradient in disability, we used the models described above to predict the average number of impairments, pain problems, and A&P limitations in the four groups of social class. To correct for multiple testing, 99% confidence intervals were used. Sampling weights provided by the Federal Statistics Office were applied to the data. The variables used for the construction of the social class index were treated as observed, and all other variables were set to the class mean. Average marginal counts were computed using Stata's postestimation command margins [35]. Across all study variables, less than 1% of values were missing, corresponding to 2,687 cases with one or more missing values. Proportions of missing values ranged from zero (gender, age) to 6% (income). In order to avoid list-wise exclusion of cases with some missing values, all missing values were imputed using the multiple imputation procedure MissForest [36]. While we report results from the imputed data set, the un-imputed data set was used in a sensitivity analysis and the results compared; no relevant differences in parameters were found. All analyses were controlled for age in years (to account for variation in age within the age groups) and health-related behaviours (alcohol consumption, smoking and low physical activity). We initially considered two economic macro-level variables, the Gini coefficient and gross domestic product per Canton. Since both variables failed to improve model fit, we decided to exclude them from the analyses. --- Results The study population is described in Table 1. More than two thirds of the Swiss population experience at least one type of impairment or pain. Around 16% experience at least one A&P limitation, with growing proportions in older age groups (Pearson's r = 0.25; p,0.0001). Results of the mixed effects regressions are presented in Tables 2 and3. Based on Pseudo R 2 values, the best overall model fit was obtained for the model explaining A&P limitation, followed by perceived health, impairment, and pain. The regression models fitted the data better in the midlife and older age group than in the younger group. --- Impairment Male gender, paid employment, higher education, higher social support and social network utilization were all associated with a decreased number of reported impairments (Table 2). The number of pain problems was positively associated with the number of reported impairments. No significant effects of income or migration background on impairment were observed. Overall, results were consistent across age groups. Apart from social network utilization, all effects of these variables on impairment were more pronounced in the midlife and older groups (considering the betas and intercepts). The effect of 'being married' was confined to the midlife group. --- Pain The number of pain problems decreased with perceived social support, but increased with the number of impairments (Table 2). Being married was associated with a higher number of pain problems. While being male reduced reported pain, education and paid employment did not strongly influence the number of reported pain problems in the overall model. Income and migration background again did not reach statistical significance. Higher education, however, reduced the number of reported pain problems in the midlife group. Being in paid employment increased the reported number of pain problems in the younger group, but decreased it in the older group. Perceived social support, in turn, exhibited a consistent, strong inverse relationship with number of pain problems across all age groups. --- A&P limitation The number of A&P limitations increased among persons who reported more problems in pain and impairment (Table 3). Being in paid employment, having higher education or higher income was associated with lower levels of A&P limitation in the overall model. Migration background did not play a role. Stronger social network utilization was related to lower levels of A&P limitation, which was consistently observed across age groups. Stronger social support was associated with more A&P limitation in the overall model which seems to be due to the older group. Pain, impairment and SES variables (with the exception of income and migration background) displayed strong effects across all age groups. In contrast to impairment and pain, male gender was associated with less A&P limitation. --- Perceived health A&P limitation, pain, and impairment were the strongest negative 'predictors' of perceived health, while being in paid employment and perceiving high levels of social support were the strongest positive 'predictors' in the overall model (Table 3). Higher education and income as well as social network utilization (Bonferroni corrected p-value = 0.1) further contributed to better perceived health. Being married and being male were associated with worse perceived health in the overall model. These effects were particularly pronounced in the elder group. Migration background did not assert significant influence when it was adjusted for the other predictors. --- Overall social gradient of disability Figure 2 shows the numbers of problems in impairment, pain and A&P limitation across 4 social class groups based on the above models (marginal counts). The overall picture suggests a social gradient of disability in the Swiss general population with higher social class being associated with less disability. --- Discussion In this cross-sectional study of a representative sample of the Swiss population, we found that measures of SES and social integration were strongly associated with several dimensions of disability and perceived health. Socially deprived and isolated people were at higher risk of disability. Conversely, the negative effects of impairments on pain, of pain on impairments, and of impairment and pain on activity and participation could be compensated by higher income, better education and social integration (Tables 2 and3). To a large extent, this finding was consistent across all age groups. Migration background did not play a significant role when it was adjusted for other measures of SES and social integration. Gender, however, was among the strongest predictors of disability and perceived health, even when we adjusted for all other SES and social integration variables. Being male positively influenced impairment and pain but negatively affected A&P limitation and perceived health, when it was adjusted for the remaining disability components. This finding suggests differential response behaviours of men and women, potentially influenced by interactions of gender with other SES and social integration variables. To our knowledge, this is the first investigation that comprehensively explores potential social determinants of functioning and disability in a representative sample of young, middle-aged and older Swiss residents of both sexes. It also pioneers the application of the ICF framework for disability research using a large national database. Most of our findings were in agreement with our a priori hypotheses, thus supporting the ICF's multidimensional approach towards disability (Figure 1). Our results are in line with studies from Greece [8] and Estonia [14] showing a social gradient of disability in the general population. They also support findings from previous studies reporting a social gradient of disability in the elderly Intercepts (in standard units) were 215.19*** for the threshold between ''very bad'' and the rest; 29.13*** for ''very bad'' and ''bad'' vs. the rest; 22.07 for ''very bad'', ''bad'', and ''fair'' vs. the rest; 10.61*** for all other categories vs. ''very good''. OProbit = Ordinal Probit. Note: all models have been adjusted for age and and health behaviours (alcohol consumption, smoking, physical activity). doi:10.1371/journal.pone.0038782.t003 [9,10,11,12,13,18] but draw on more complex outcomes. In our study, most associations have the same direction but differ in strength across the age groups. These differences highlight the need for longitudinal studies in order to elucidate the evolution of disability patterns along the course of life. The hypotheses on SES and social integration were tested simultaneously, i.e. respective parameters were adjusted for each other. Consequently, our study confirms the findings from Nilsson [16] in middle-aged Danish men showing independent influence of financial assets and social support. Nonetheless, social support might compound the influences of SES on functioning and disability. We did, however, not examine respective interactions in our study. Several limitations need to be taken into account. Firstly, our theoretical model assumes unidirectional effects. However, the cross-sectional design did not allow the direction of reported effects to be tested. Secondly, we did not elaborate on gender-specific associations. Nevertheless, we observed gender-specific effects and gender is known to interact with socioeconomic variables. A more detailed investigation would be desirable but was not the focus of this analysis. Thirdly, the variables relating functional status did not allow the severity of problems to be gauged. For instance, for pain, only the number of pain problems could be used without information on pain severity or duration. Fourthly, we used 4 separate multiple mixed effects regression models and did not perform a test of an overall model, e.g. based on structural equations [32] or conditional independence graphs [37,38]. Further clarification of the structure of associations (Figure 1), including potential moderation and mediation [39], is required from future research. For example, it may be the case that the influence of impairment on A&P limitation is mediated or moderated by social position or gender. Finally, Switzerland is a high resource setting with high standards of health care and social welfare. This makes it difficult to generalize our findings to lower resourced settings where we would expect social inequalities to be even more pronounced, though typically data of good quality are not available from these settings. These limitations are balanced by several strengths. Our results are derived from a representative sample of the Swiss population stratified into three age groups and the application of a comprehensive set of variables measuring functioning. Furthermore, we applied a theoretically-based approach to linking socioeconomic and sociological factors with newly constructed sum indices of functioning and with perceived health. Robust statistical methods included confirmatory factor analysis of the indices, mixed effects regression models to account for unobserved heterogeneity, Bonferroni correction to account for multiple testing, and tests of the models' fit to the data. This study highlights consistent social inequalities in functioning and perceived health in a representative sample of the Swiss population aged 15 or above. Higher levels of social integration and of SES were associated with better functioning in terms of impairment, pain and limitations of A&P and better perceived health. Our findings are challenging in view of the high living standard in Switzerland. Indeed, given the social gradients in disability observed in this setting, they are likely to be even more pronounced elsewhere. --- Author Contributions Conceived and designed the experiments: JR EvE. Analyzed the data: JR. Contributed reagents/materials/analysis tools: EvE CF JS. Wrote the paper: JR EvE CF JS. Error bars represent 99% confidence intervals. Data are weighted with sampling weights and adjusted for age, gender, being married, employment, migration background, smoking, alcohol consumption and leisure physical activity. In addition, impairment is adjusted for pain, pain for impairment and A&P limitation for impairment and pain. doi:10.1371/journal.pone.0038782.g002

Many people worldwide live with a disability, i.e. limitations in functioning. The prevalence is expected to increase due to demographic change and the growing importance of non-communicable disease and injury. To date, many epidemiological studies have used simple dichotomous measures of disability, even though the WHO's International Classification of Functioning, Disability, and Health (ICF) provides a multi-dimensional framework of functioning. We aimed to examine associations of socio-economic status (SES) and social integration in 3 core domains of functioning (impairment, pain, limitations in activity and participation) and perceived health. We conducted a secondary analysis of representative crosssectional data of the Swiss Health Survey 2007 including 10,336 female and 8,424 male Swiss residents aged 15 or more. Guided by a theoretical ICF-based model, 4 mixed effects Poisson regressions were fitted in order to explain functioning and perceived health by indicators of SES and social integration. Analyses were stratified by age groups (15-30, 31-54, $55 years). In all age groups, SES and social integration were significantly associated with functional and perceived health. Among the functional domains, impairment and pain were closely related, and both were associated with limitations in activity and participation. SES, social integration and functioning were related to perceived health. We found pronounced social inequalities in functioning and perceived health, supporting our theoretical model. Social factors play a significant role in the experience of health, even in a wealthy country such as Switzerland. These findings await confirmation in other, particularly lower resourced settings.

Although its meaning has changed, the belief that women and men are held to different standards of sexual conduct (i.e., the sexual double standard) remains pervasive in contemporary U.S. society (Bordini and Sperb 2013;Crawford and Popp 2003;Marks and Fraley 2006). When Reiss (1956) began researching the double standard over a half-century ago, premarital sex and the stigma attached to women who engaged in it were primary concerns. In the ensuing decades, the sexual revolution largely destigmatized female premarital sex per se (Risman and Schwartz 2002), shifting double standard research to the gendered consequences of sexual permissiveness (i.e., number of sexual partners) and casual sexual encounters (e.g., "hook ups"; Manning, Giordano, and Longmore 2006). Today, studies of gendered sexual scripts and the sexual double standard continue at a rapid pace, both because empirical evidence of a double standard has been mixed (Crawford and Popp 2003;Marks and Fraley 2005) and its implications for identity development and gender inequality remain profound (Armstrong, Hamilton, and England 2010). An area that has received relatively little investigation is the sexual double standard in the context of adolescent development. This omission is particularly striking given that most youth have reached sexual maturity by early adolescence and more than 30% begin engaging in sexual behaviors by 9 th grade (Centers for Disease Control 2012; Walvoord 2010). Although the well-studied topic of sexual permissiveness is less relevant at this age (Carver, Joyner, and Udry 2003), peer evaluations of emergent sexual behaviors and perceived sexual precociousness are likely to have large and lasting impacts on adolescents' adjustment (Carpenter 2005). Moreover, the emergence of sexual behavior in adolescence occurs in the context of increasing psychological investment in peer relations for companionship, selfconcept, and a sense of belonging (Giordano 2003). Thus, the manner in which the peer ecology reacts to sexual debut is particularly important both for the adolescents who initiate sexual activity and for the development of attitudes and norms among those who do not. In our study, we draw on sexual script theory (Simon and Gagnon 1984, 1986, 2003) and longitudinal social network data to document sexual double standards in adolescence in the United States. We take advantage of yearly longitudinal data from 921 youth to measure within-person changes in adolescents' friendships and sexual behavior from 6 th to 9 th grades. Such analyses help to elucidate the proposed mechanism underlying the double standard: that peer responses to individual sexual behaviors differ by gender. We also move beyond a unitary conception of sexual behavior by examining changes in peer acceptance following two sexual behaviors: having sex and "making out." Distinguishing these behaviors is theoretically important because "light" sexual activities-such as holding hands, kissing, and making out-may be consistent with female adolescents' traditional sexual scripts (i.e., they fit into a romance narrative and are considered relatively safe behaviors for girls, leaving virginity "intact") and associated with increased peer acceptance (Tolman 2002;Wiederman 2005). Alternatively, male adolescents engaging in similar behaviors, particularly if these behaviors do not lead to sex, may be viewed as losing their independence and agency (central facets of hegemonic or heteronormative masculinity) and see decreases in peer acceptance (Eder, Evans, and Parker 1995;Miller 2008). To our knowledge, no study has yet tested this reverse double standard hypothesis for romantic noncoital behaviors. --- Sexual Script Theory and Empirical Research Sexual scripts are socially constructed cognitive schema that define normative sexual behaviors and inform individual actions in sexual situations (Simon and Gagnon 1984, 1986, 2003). At the cultural level, traditional sexual scripts are gendered prescriptions for appropriate sexual conduct, which scholars commonly measure through qualitative and survey analyses (Bowleg, Lucas, and Tschann 2004;Masters, Casey, Wells, and Morrison 2013). Studies in this vein generally find opposing sexual scripts by gender. Men and boys are believed to act on innate and strong sex drives to initiate heterosexual contacts for the purpose of sex (rather than romance) and pursue multiple sexual partnerships. In contrast, women and girls are expected to desire romance over sex, value monogamy, and "gatekeep" male sexual advances within committed relationships. A sexual double standard then arises because women and girls who violate traditional sexual scripts and have casual and/or multiple sexual partnerships are socially derogated, whereas men and boys performing similar behaviors are rewarded for achieving masculine ideals. The existence of a sexual double standard has important implications for gendered sexual development. In particular, the sexual double standard likely socializes male and female adolescents to differing behavioral expectations and sexual risks. Empirical evidence demonstrates that young women who endorse the sexual double standard are less likely to have sex, perceive greater barriers to condom use, and are less likely to use condoms (Crawford and Popp 2003;Lefkowitz, Shearer, Gillen, and Espinosa-Hernandez 2014;Moore and Rosenthal 1992). Moreover, the sexual double standard can lead girls and women to feel shame and guilt if they do engage in sexual behaviors (Tolman 2002). In contrast, young men who more strongly endorse the sexual double standard tend to have more sexual partners (Lefkowitz et al. 2014). Thus, if peer ecology socializes male and female adolescents differently in reaction to their sexual behaviors, and if adolescents incorporate these messages into their sexual schemas, it might translate to future sexual beliefs and behaviors in adolescence and beyond. Recent research on sexual scripts and the double standard has focused on college-aged populations and "hook up" (i.e., casual physical encounters with no expectations for future commitment) culture (Allison and Risman 2013;Bradshaw, Kahn, and Saville 2010;Conley, Ziegler, and Moors 2013;Glenn and Marquardt 2001). The empirical evidence from these studies remains decidedly mixed (Crawford and Popp 2003;Marks and Fraley 2005). Qualitative studies generally find support for a "strong" double standard, such that interviewed young adults report that men are likely to receive affirmations as "players" or "studs" for casual and/or frequent sexual encounters whereas women are negatively labeled as "sluts" or "hos" for similar activities (Bogle 2008;Glenn and Marquardt 2001;Jackson and Cram 2003;Shoveller, Johnson, Langille, and Mitchell 2004). Quantitative studies, however, have provided more equivocal results. Although some quantitative research continues to find evidence for the sexual double standard (Bordini and Sperb 2013;Crawford and Popp 2003;England and Bearak 2014), several experimental vignette designs and some attitudinal studies do not document a strong sexual double standard. Instead these latter studies find that young adults are negatively evaluated for permissive sex regardless of gender (Allison and Risman 2013;Gentry 1998;Marks and Fraley 2005), that is, a "weak" double standard whereby the negative evaluations of young men's permissiveness are only somewhat less than the negative evaluations of young women's permissiveness (see Bordini and Sperb, 2013, for a review), or even a reverse double standard whereby young men are evaluated more harshly for their sexual behavior than are young women (Howell, Egan, Giuliano, and Ackley 2011;Zaikman and Marks 2014). The differences between qualitative and quantitative double standard studies may arise from the level at which sexual scripts are measured. Qualitative studies may be identifying cultural-level scripts of gender-appropriate sexual behavior, whereas quantitative studies may be measuring inter-or intra-personal scripts that are inconsistent with broader traditional beliefs. Alternatively, apparent distinctions between cultural-and individual-level sexual scripts may be due to the gap between ideal beliefs and actual behaviors in social settings. For example, experimental evidence suggests that young adult men and women are more likely to detect social cues that confirm rather than disconfirm cultural-level sexual double standards (Marks and Fraley 2006). Thus, survey and experimental research divorced from the social contexts in which sexual appraisals are actually made will underestimate traditional sexual scripts (Marks and Fraley 2007). Desire to apply a single sexual standard or hold sexually progressive views may not eliminate individuals' applications of double standards in practice, especially when sexual evaluations emerge in dynamic peer settings (Kreager and Staff 2009). Overcoming this methodological limitation requires data as close as possible to the social contexts in which evaluations of sexuality take place. Prior research on the sexual double standard suffers from two additional shortcomings. First, studies typically rely on convenience samples of undergraduates, which provide important information about campus sexual cultures, but do not illuminate gendered processes surrounding sexual debut, which occurs prior to high school graduation for the majority of individuals in the United States (Centers for Disease Control 2012). Furthermore, it is estimated that only one-fourth of adolescents have their first sexual experience outside a romantic relationship (Manning, Longmore, and Giordano 2005), and the number of sexual partners in adolescence remains relatively low. Thus, focusing on sexual permissiveness and casual sex, as is typical of most college-based and young adult studies, does not generalize to adolescence. Norms and social reactions to sexual precocity, rather than perceived promiscuity, would be of more theoretical interest during adolescence. This shift also places greater emphasis on early transitions from virgin to non-virgin than number of sexual partners or frequency of sex. Second, double standard research has predominantly focused on sexual intercourse behaviors (e.g., casual sex, sexual permissiveness). Such emphasis is warranted because these behaviors are clearly associated with traditional sexual scripts, with female permissiveness and casual sex expected to be socially sanctioned through gossip and lowered peer acceptance. In contrast, similar male behaviors should generate social rewards and increased peer acceptance. Missing from this research, however, are behaviors consistent with traditional female sexual scripts but inconsistent with traditional male scripts. Female adolescents are expected to "be desirable but not desiring" (Reid, Elliott, and Webber 2011), meaning that dating and romantic non-coital behaviors (e.g., kissing, holding hands, making out) should be socially rewarded through increased friendships and improved reputations. These behaviors simultaneously demonstrate a girl's other-sex attractiveness and gatekeeping abilities (Wiederman 2005). For male adolescents, "light" sexual and romantic behaviors that do not progress to sex could symbolize passivity and a sublimation of hegemonic masculinity to feminine control. Participation in non-coital romantic activities may then lower boys' esteem in the eyes of peers because they are inconsistent with group expectations for competitive and aggressive masculinity (Eder, Evans, and Parker 1995;Marks and Fraley 2007). This reverse double standard may be even more dramatic in adolescence, when most couples are transient and peer groups are predominantly same gender. Indeed, early romantic relationships may originate as means of garnering same-gender acceptance or to fulfill group expectations (Brown 1999). Understanding the consequences of both romantic and sexual behaviors in these peer-driven contexts should illuminate the contours of contemporary sexual double standards. Understanding a sexual double standard in adolescence is also of interest because early socialization to gendered sexual scripts likely has lasting impacts on future sexual attitudes and behavior. Peer influence and the importance of peer acceptance peak during early adolescence (Giordano 2003), sustaining the social fishbowl in which sexual behaviors are monitored and defined. Moreover, the gender segregation typical of child and early adolescent friendship networks adds to the policing and reinforcement of traditional sexual scripts (Eder et al. 1995;Mehta and Strough 2009). Not only does the gender-segregated world of teen peer culture increase the salience of gendered sexual scripts and make it difficult to maintain sexual secrets, but gossip and evaluations of sexual (mis)conduct quickly diffuse through dense peer networks (Brown 1999). As a result, female adolescents' first experiences with sex not only may be less desired and satisfying than male adolescents' first experiences (Guggino and Ponzetti 1997;Sprecher, Barbee, and Schwartz 1995), but also may result in lower peer acceptance as prior friends seek to distance themselves from the sexual transgressor. Kreager and Staff (2009) approached the adolescent sexual double standard through an innovative network lens. Using data from the National Longitudinal Study of Adolescent Health (Add Health), they tested associations between the number of self-reported sexual partners and peer-reported friendship nominations (i.e., peer acceptance). They found evidence for a double standard in that the number of self-reported sexual partners was positively associated with peer acceptance for male adolescents and negatively associated for female adolescents. Furthermore, they found that the associations varied by students' socioeconomic status (SES) and by gender of the peers who nominated the student as a friend. The double standard was more likely to occur for high SES females and low SES males. The friendship benefits associated with male permissiveness and the friendship costs associated with female permissiveness were driven by females' friendship nominations. Kreager and Staff (2009) concluded that the double standard not only depends on the gender of sexual actors, but also their sociodemographic characteristics and the gender of peer evaluators. --- Network Approaches to the Adolescent Double Standard Although suggestive, Kreager and Staff's (2009) study has several limitations. Its crosssectional design is a primary concern because the direction of the reported associations is uncertain. In addition, the associations could be spurious due to unobserved between-person heterogeneity. For example, low-esteem female adolescents may have low peer acceptance and seek approval through sexual relationships, whereas high-esteem male adolescents may have high peer acceptance and greater access to multiple sexual partners. Another study limitation lies in the aging Add Health data. The sexual double standard is subject to cultural shifts (Risman and Schwartz 2002), and thus patterns observed in Add Health's twodecades-old data may not hold in more recent youth cohorts. Finally, similar to other studies in this area, Kreager and Staff (2009) narrowly focused on sexual permissiveness, operationalized as number of self-reported sexual partnerships. Lyons, Giordano, Manning, and Longmore (2011) also approached the adolescent sexual double standard from a network perspective. They did not find significant associations between sexual partnerships and self-perceived peer popularity, interpreting these findings as "some contradictory evidence regarding the basic notion that violating the sexual double standard is associated with heavy social costs" (p. 447). However, sub-sample interviews suggested that respondents did perceive, and often contribute to, a sexual double standard. To resolve these apparently contradictory findings, Lyons et al. asserted that sexually permissive female adolescents are able to maintain close friendships with peers who share similar behaviors or sexual attitudes even though the larger peer community evaluates such behaviors negatively. The Lyons et al. study benefits from its mixed-methods design, longitudinal assessment, and more recent data collection. However, their sample was not school-based and therefore did not include peer-nominated status measures. Compared to self-reported popularity, peer friendship nominations can provide more variability, and they are less likely caused by psychological constructs like self-esteem and depression. Associating multiple sexual behaviors with a continuous and longitudinal measure of peer acceptance would advance prior tests of the adolescent double standard. Another advantage of analyzing the sexual double standard with longitudinal friendship network data is that researchers are able to distinguish changes in individuals' self-perceived friendships (i.e., self-reported outgoing friendship nominations) from changes in their received friendships (i.e., peer-reported friendship nominations toward the adolescent) (Holland and Leinhardt 1981;Rulison, Kreager, and Osgood 2014). This distinction between self-and peer-reported friendships is particularly advantageous for isolating social responses to individual behaviors or characteristics (e.g., stigmatization) over time. As Link and Phelan (2001, p. 366) emphasize when reviewing stigma research, a primary challenge to operationalizing the concept is that prior studies commonly take an individualistic focus and measure "the perceptions of individuals and the consequences of such perceptions for the micro-level interactions." Missing from these approaches are the socially constructed consequences of non-normative behavior for the stigmatized individual, which Link and Phelan argue are typically social rejection and status loss. Measuring changes in an adolescent's peer-reported friendships after sex while controlling for his or her self-reported friendships then locates the source of potential status gains or losses in peer reactions rather than in the adolescent's actions or attributes. Network data then allow us to more accurately measure social exclusion or acceptance processes and attend to Goffman's (1963, p. 3) call to view stigma as "...a language of relationships, not attributes...." --- Current Study In our study, we examine the within-person association between peer acceptance and sexual behaviors in a longitudinal sample of male and female adolescents. We derive two competing hypotheses for whether gender moderates the sexual behavior-peer acceptance association. From sexual script theory, we propose that female adolescents' peer acceptance will decrease following sex, whereas male adolescents' peer acceptance will increase (Hypothesis 1a). The alternative hypothesis, derived from several recent studies of young adults' sexual attitudes and vignette evaluations (Allison and Risman 2013;Gentry 1998;Marks and Fraley 2005;Zaikman, Vogel, Vicary, and Marks in press), asserts that peer acceptance of male and female adolescents will both decrease following sex (Hypothesis 1b). From literature describing "normative" sexual scripts, we generate another hypothesis for non-coital "light" sexual behavior, asserting a reverse sexual double standard whereby female adolescents' peer acceptance will increase and male adolescents' peer acceptance will decrease following reports of making out (Hypothesis 2). Following findings from Kreager and Staff (2009), we also test if the association among gender, peer acceptance, and sexual behaviors is moderated by student's SES and the gender of peer nominators (Hypothesis 3). Finally, we conduct sensitivity analyses to test (a) if sexual behaviors are associated with out-of-grade friendships (Hypothesis 4a) and (b) if the associations between peer acceptance and sexual behaviors are moderated by adolescents' dating status (Hypothesis 4b). The former test is important for ruling out the possibility that sexually active youth replace their in-grade friends with older or out-of-school friends, whereas the latter test is important because sex may be less stigmatizing for female adolescents involved in a romantic relationship than for female adolescents who have sex outside a romantic relationship (Tolman 2002). In sum, our study represents a comprehensive analysis of the gendered social consequences associated with adolescent sexual activity. --- Method Participants and Procedure Data came from a sample of adolescent participants in the PROmoting School-communityuniversity Partnerships to Enhance Resilience (PROSPER) longitudinal study (Spoth, Greenberg, Bierman, and Redmond 2004;Spoth et al. 2007). PROSPER is a large-scale prevention effectiveness trial aimed at reducing adolescents' substance use. It follows two successive cohorts of 6 th grade students living in 28 rural communities in Iowa and Pennsylvania. Each community was selected because it (a) had a public school district with 1,300-5,200 enrolled students and (b) had at least 15% of students eligible for reduced-cost lunches. One of the Pennsylvania schools did not agree to participate in the network portion of the study necessary for the current research, resulting in 27 school districts available for analyses. Students completed confidential pencil-and-paper questionnaires administered during school hours in the Fall of 6 th grade, and in the Spring of 6 th, 7 th, 8 th, and 9 th grades (totaling approximately 10,000 students per wave and 15,000 students completing at least one survey). The in-school survey included friendship network items used to construct our dependent measures. Across the five waves, participation rates ranged from 86% to 90% of enrolled students, and eligible students completed on average 4.18 of 5 in-school surveys. In addition to the in-school survey, a random sample of 2,267 participating families were invited to complete in-home written questionnaires for the focal adolescent, mother, and if present, father (Lippold, Greenberg, Graham, and Feinberg 2013). Adolescent in-home written questionnaires included self-reported sexual behaviors. Of the invited families, 979 (43%) completed at least one in-home assessment and 914 (93%) of these families included an adolescent who also completed an in-school survey and comprise our analytic sample. Many invited families declined in-home surveys, which may bring into question the generalizability of our results to the PROSPER population. Youth participating in the inhome interview were no different from other PROSPER youth with regard to free-lunch receipt, gender, living with two biological parents, and substance use initiation (Lippold, Greenberg, Graham, and Feinberg 2013). However, they were less likely to be delinquent, had higher grades, and had higher peer acceptance than other PROSPER respondents. Table 1 displays descriptive statistics for all variables by wave and gender. Approximately 33% of girls and 34% of boys attritted from the sample by wave 5. --- Measures Peer acceptance-During the PROSPER in-school survey, all students nominated up to seven best or close friends in the same school grade and reported how many comparably close friends they had outside their grade or school. Friendship nominations were collected using an open name generator where students wrote the first and last names of each friend on the survey form, which were then matched to student rosters. Names were matched 83% of the time. We operationalized peer acceptance at each wave with the number of incoming friendship ties received (i.e., network indegree) from same-grade peers. Our primary dependent variable is thus a measure of sociometric likeability (Cillessen and Marks 2011). Across all waves, the average female respondent received 4.1 friendship nominations and the average male respondent received 3.3 nominations, which was significantly different (see Table 1). Nominations ranged from 0 to 16 for females and 0 to 15 for males, and the distributions were right skewed. For both girls and boys, the mean number of received friendship nominations peaked in 7 th grade (wave 3). We also created measures of same-gender and cross-gender peer acceptance. Because approximately 75% of the sample did not receive any cross-gender nominations in any given wave, we dichotomized the cross-gender peer acceptance measure into "no cross-gender incoming ties" (coded 0) and "one or more crossgender incoming ties" (1). Sexual behaviors-We measured students' self-reported past-year sex with the item, "During the past 12 months, how many times have you had sex?" Numerical responses were dichotomized into "none" (coded 0) and "once or more" (1) for each wave. As indicated in Table 1, the prevalence of sex was extremely rare in the early waves, approaching zero for both 6 th grade (wave 1) girls and boys, but rose to 10% for both genders by 9 th grade (wave 5). The data therefore capture sexual onset for a minority of respondents. To better identify students who transition from no sex to first sex, we also created a First Sex variable that is coded as "0" for all waves prior to self-reported sex and "1" for all waves after sex is reported one or more times. There is a small chance that respondents who reported no sex at the first wave lost their virginity prior to the 6 th grade. The near-zero prevalence of sex in the first two waves makes this possibility unlikely, but data limitations mean that we cannot rule out the possibility. By the 9 th grade, 17% of girls and 12% of boys were classified as nonvirgins. It is also possible that, given the wording of the question, respondents interpreted "sex" as "heavy" non-coital sexual behaviors other than intercourse (e.g., oral or anal sex). Throughout the text, we therefore refer to this variable using the broader label of "had sex" rather than more specific labels of "intercourse" or "coitus." Light sexual behavior was operationalized at each wave from the item, "During the past 12 months, how many times have you made out?" Responses were dichotomized into "none" (coded 0) and "once or more" (1). Although uncommon in the early waves, a majority of the girls and over 40% of boys reported making out at least once by the 9 th grade. As one would expect, the overwhelming majority (over 95% at every wave) of adolescents who reported sex also reported making out and between 10-25% of those who reported making out also reported sex at each wave. Control variables-We include several time-varying control variables that prior literature suggests are correlated with peer acceptance and/or sexual behaviors and thus may make our hypothesized associations spurious (Boislard, Dussault, Brendgen, and Vitaro 2013;Kapungu, Holmbeck, and Paikoff 2006;Rostosky, Wilcox, Wright, and Randall 2004;Schvaneveldt, Miller, Berry, and Lee 2001). These variables were measured in the in-school survey. Outdegree is the number of friendship nominations (ranging from 0 to 7) that respondents make at each survey wave. For models predicting peer acceptance from same-gender and other-gender peers, we include measures of same-gender and cross-gender outdegree and dichotomized the cross-gender measure into "no cross-gender outgoing ties" (coded 0) and "one or more cross-gender outgoing ties" (1). Family Relations at each wave is the grand mean of five subscales (alpha=.81), standardized to wave 4 means and standard deviations to provide a common metric, assessing child-parent affective quality (12 items, e.g. "During the past month, when you and your [mom/dad] have spent time talking or doing things together, how often did [she/he] let you know [she/he] really cares about you?"), joint activities (6 items, e.g. "During the past month, how often did you work on homework or a school project with your mom or dad?"), parental monitoring (5 items, e.g. "During the day, my parents know where I am."), inductive reasoning (3 items, e.g. "My parents give me reasons for their decisions."), and family cohesion (7 items, e.g. "Family members really help and support each other."). Grades at each wave are operationalized from the question "What grades do you generally get in school?" Responses ranged from "Mostly lower than D's" (coded 1) to "Mostly A's" (5). Religious Attendance at each wave is derived from the question, "How often do you go to church or religious services?" Responses ranged from "Never" (1) to "More than once a week" (8). Delinquency at each wave is operationalized through self-reports of 12 delinquent behaviors in the past 12 months. A graded-response IRT model was used to score this measure, with item parameters coming from the middle wave of data collection. IRT differentially weights items according to their seriousness (i.e., lower frequency items are given more weight) and provides scores that are approximately normally distributed (see Osgood, McMorris, and Potenza 2002). Drinking at each wave is derived from the item, "During the past month, how many times have you had beer, wine, wine coolers, or other liquor?" Responses ranged from "Not at all" (coded 1) to "More than once a week" (5). School change indicates that a respondent changed schools between survey waves, a transition likely associated with changing peer and sexual contexts (South, Haynie, and Bose 2005). Finally, we account for grade and age trends in our outcome with measures for Wave and Wave 2. A major advantage of the longitudinal network data in PROSPER is that it allows us to examine how within-person changes in peer acceptance are associated with within-person changes in making out and having engaged in sex. In such analyses, individuals are compared to themselves across measurement occasions, thus these models statistically control for all time-stable factors that might be related to both peer acceptance and sexual behavior. For instance, at least some of the effect of sexual behavior on peer acceptance may be driven by relatively stable between-person differences in attractiveness, athleticism, race, or socioeconomic family background. An added advantage of our within-person analytical design is that it correctly specifies the temporal ordering of primary concepts. The dependent measures, friendship nominations, are measured concurrent with survey administration and therefore chronologically follow the primary independent variables measuring sexual behavior occurring in the year prior to each survey wave. This temporal specification reduces potential reverse causal ordering (i.e., that changes in friendships increase or decrease the likelihood of sexual behavior over time). In fixed effects estimation, mean values are calculated for each youth over time for each outcome (i.e., overall-, same-and cross-gender peer acceptance) and predictor variable (i.e. making out, sex, family attachment, grades, etc.). These person-specific mean values are then subtracted from observed values at each time point. Within-person deviations in outcomes are then regressed on within-person deviations in predictors, controlling for time (i.e., Wave and Wave 2 ). Formally, the model appears as follows: where i identifies each person, t indicates wave, y is one of the peer acceptance dependent variables, x is a vector of time-varying independent variables (including sex, making out, and wave), <unk> is a vector of estimated coefficients, and <unk> are person and person-wave error terms. An additional advantage of fixed-effects models is that estimated coefficients remain unbiased if sample attrition is correlated with the individual error term, <unk> i, and in unbalanced panel designs (i.e., t varies over individuals). We estimated these models with the "xtreg, fe" command in STATA. This command yields coefficients that capture within-person changes in peer acceptance as a function of changes in sexual activity and wave. We added a small constant to the peer acceptance measure and then logged it to account for the right skew. We estimate robust standard errors to correct for potentially heteroscedastic and dependent error terms, and logistic models were estimated for binary outcomes. In addition, we used chained multiple imputation (using the "mi" command in STATA) to impute missing data on our control variables. The amount of missing data was relatively modest, ranging from 0% for school change between waves 1 and 2 to 14% for religious attendance at wave 1. --- Results --- Overall Sexual Activity and Peer Acceptance We first examine if yearly changes in sex and making out are associated with changes in peer acceptance from grades 6 to 9 in the pooled sample. Table 2 presents estimates from a series of linear fixed effects models of peer acceptance. In Model 1, estimates for the total sample (i.e., pooled female and male adolescents) show that, net of time-varying control variables, year, and time-stable factors, no statistically significant associations appear between having sex or making out and peer acceptance in the total sample. That is, averaging across females and males, there is no evidence that having had sex or making out are associated with changes in incoming friendship nominations. Looking at other covariates in Model 1, we also see little evidence that our individual-level time-varying covariates predict changes in peer acceptance over the observed waves. The significant positive coefficient for Wave and significant negative coefficient for Wave 2 suggest an inverted U-shaped peer acceptance pattern over time consistent with the observed peak in peer acceptance in 7 th grade (see Table 1). --- Sexual Activity and Peer Acceptance by Gender Null associations between the two sexual behaviors and peer acceptance in the pooled gender model may mask opposing patterns for females and males. To test Hypotheses 1a and 1b, we thus split the sample by gender and estimate similarly-specified models for females and males. Consistent with a "strong" double standard (Marks and Fraley 2005) and in support of Hypothesis 1a, the coefficients for having sex are in opposite directions for female and male adolescents, where females see an approximate 45% decline [i.e., exp(-. 58)] in peer acceptance in waves where they report having sex and males see an 88% increase [i.e., exp(.63)] in peer acceptance in waves where they report having sex. Moreover, this gender difference is large in magnitude and is statistically significant. A z-test of the equality of coefficients across independent samples (Paternoster, Brame, Mazerolle, and Piquero 1998) demonstrates that the females' coefficient for having sex is significantly smaller than the males' coefficient (z = -3.31, p <unk>.001). The significant positive coefficient for having sex in Model 1 for male adolescents also contradicts Hypothesis 1b, drawn from several studies of young adult sexual attitudes, that both male and female adolescents are penalized for sexually permissive behaviors (in this case operationalized as adolescent sex). Results from Model 1 for females and males also show that, net of reported sex, females' peer acceptance increases by approximately 25% [i.e., exp(.22)] in waves in which they report making out, and males' peer acceptance declines by about 29% [i.e., exp(-.34)] in waves in which they report the same behavior. As with having engaged in sex, this gender difference is statistically significant (z = 3.09, p =.002). We thus find support for Hypothesis 2 that light sexual behavior is consistent with female sexual scripts and inconsistent with male sexual scripts. The pattern of results suggests that participating in light sexual activity creates a sexual double standard opposite of that found for having had sex. Model 2 adds an outdegree covariate to Model 1, which captures the number of outgoing friendship nominations sent by each adolescent respondent. The coefficient for this variable is positive and significant, meaning that adolescents who send more friendship nominations are also likely to receive more incoming friendship nominations from peers. This result is not surprising and is consistent with ideas of social exchange (i.e., friendship investments are likely to be reciprocal; Laursen and Hartup 2002). More important for the current study was that including outdegree did little to affect the gendered associations between sexual behaviors and peer acceptance. For example, the

A sexual double standard in adolescence has important implications for sexual development and gender inequality. The present study uses longitudinal social network data (N = 914; 11-16 years of age) to test if gender moderates associations between adolescents' sexual behaviors and peer acceptance. Consistent with a traditional sexual double standard, female adolescents who reported having sex had significant decreases in peer acceptance over time, whereas male adolescents reporting the same behavior had significant increases in peer acceptance. This pattern was observed net of respondents' own perceived friendships, further suggesting that the social responses to sex vary by gender of the sexual actor. However, findings for "making out" showed a reverse double standard, such that female adolescents reporting this behavior had increases in peer acceptance and male adolescents reporting the same behavior had decreases in peer acceptance over time. Results thus suggest that peers enforce traditional sexual scripts for both "heavy" and "light" sexual behaviors during adolescence. These findings have important implications for sexual health education, encouraging educators to develop curricula that emphasize the gendered social construction of sexuality and to combat inequitable and stigmatizing peer responses to real or perceived deviations from traditional sexual scripts.

, and males' peer acceptance declines by about 29% [i.e., exp(-.34)] in waves in which they report the same behavior. As with having engaged in sex, this gender difference is statistically significant (z = 3.09, p =.002). We thus find support for Hypothesis 2 that light sexual behavior is consistent with female sexual scripts and inconsistent with male sexual scripts. The pattern of results suggests that participating in light sexual activity creates a sexual double standard opposite of that found for having had sex. Model 2 adds an outdegree covariate to Model 1, which captures the number of outgoing friendship nominations sent by each adolescent respondent. The coefficient for this variable is positive and significant, meaning that adolescents who send more friendship nominations are also likely to receive more incoming friendship nominations from peers. This result is not surprising and is consistent with ideas of social exchange (i.e., friendship investments are likely to be reciprocal; Laursen and Hartup 2002). More important for the current study was that including outdegree did little to affect the gendered associations between sexual behaviors and peer acceptance. For example, the reduction in females' peer acceptance but increase in males' peer acceptance following sex do not appear to be accounted for by changes in the number of adolescents' outgoing friendship ties. This finding suggests that the friendship consequences of having sex are not solely due to self-selection processes (e.g., female adolescents narrowing their friendship circles to be with a new sexual partner) and have a solid basis in peer responses to the sexual behaviors themselves. In unlisted analyses (available upon request), we also examined socioeconomic disadvantage (operationalized as receiving free or reduced lunch) as a potential moderator of sexual behavior and peer acceptance. We created interactions between reduced-cost lunch receipt at 6 th grade (a stable indicator of socioeconomic disadvantage) and our sexual behavior measures in Model 2 for females and males. Because the free-lunch variable did not vary over time, it did not have a main effect in the fixed-effects model. None of the interactions reached statistical significance, suggesting that, in contrast to Hypothesis 3, the observed patterns did not vary by respondents' level of disadvantage. To help visualize the gendered patterns, Figure 1 plots predicted peer acceptance values over wave for hypothetical females and males who begin with similar sexual behaviors but then diverge from one another beginning at wave 3 (all other covariates held at their gendered means). We plotted predicted values for three hypothetical male and female adolescents: a respondent who (a) does not report having sex or making out at the observed waves (solid line), (b) reports making out at waves 3 through 5 but never reports having sex (dashed line), and (c) reports making out and having sex at waves 3 through 5 (dotted line). Note that we did not include a category for having sex without making out because this group never exceeded 5% of the sample at any wave. As can be seen in Figure 1a, female adolescents who report making out without having sex beginning in the 7 th grade are expected to have approximately 0.5 more friendships by 9 th grade than are female adolescents reporting no sexual behaviors. Because the penalty for having sex is larger than the benefit of making out, females who report both behaviors are expected to have approximately 0.5 fewer friendships by 9 th grade than are females who report neither behavior. The opposite pattern is observed for male adolescents: Males who make out without having sex beginning in 7 th grade are expected to lose a little less than 0.5 friendships by 9 th grade compared to males without reported sexual behaviors. Male adolescents reporting both behaviors beginning in 7 th grade are expected to have about 0.4 more friendships by 9 th grade than are males who report neither behavior. The sex measure reported in Table 2 is operationalized as a time-varying covariate, meaning that sex reported in waves following initiation are assumed to continue to influence the association between having sex and peer acceptance. Alternatively, we can operationalize sex as the transition from virgin to non-virgin (i.e., code the measure as "0" at all waves prior to first reported sex and "1" for the wave when sex is first reported and at all following waves). This operationalization allows us to examine if changes in peer acceptance are primarily associated with the transition to first sex or with sex that continues to occur throughout the observed period. Model 1, Table 3, replaces the time-varying sex variable with a variable capturing the transition to first sex. For both female and male adolescents, the coefficient for first sex varies little compared to the sex coefficients in Model 1, Table 2. The double standard appears to occur at both first sex and in time periods following sexual initiation. It may also be that the association between having sex and peer acceptance is conditional on making out, or vice versa. In Model 2, Table 3, we examine potential moderation between sex and making out by comparing dummy indicators for (a) person-waves where respondents report having sex (with or without making out) and (b) person-waves where respondents report making out without having sex-to the reference category of personwaves where respondents do not report either behavior. We combine waves where respondents report having sex without making out with waves where respondents report having sex and making out because of the low proportion of waves where sex was reported without making out. For female adolescents, waves in which they only make out without having sex are associated with significantly greater peer acceptance than waves where they did not make out or have sex. For male adolescents, we observe the opposite pattern, such that waves where they make out without having sex result in significantly less peer acceptance compared to waves where they report neither sexual behavior. Although the coefficients for having sex (with or without making out) are not significantly different than the reference category of not having sex or making out for both genders, when the reference category is changed to making out, these associations achieve statistical significance (p =.003 for females, p =.012 for males). Again, the opposing patterns across gender are consistent with differential sexual scripts and a sexual double standard. To this point, we have predicted peer acceptance originating from all peers, whether male or female. In Table 4, we disaggregate incoming friendship nominations by gender to examine if sexual behaviors result in greater changes in same-gender versus cross-gender friendships. Because the frequency of cross-gender peers is low at our observed ages, we dichotomize these into "no cross-gender nominations" and "one or more cross-gender nominations," which are then modeled with logistic fixed-effects regressions. Looking first at the samegender peer acceptance models (Model 1 for girls and boys), we see strong similarities in estimated coefficients to Model 2 of Table 2. This result is not surprising because the majority of incoming nominations modeled in Table 2 come from same-gender peers. More interesting are the cross-gender models (Model 2). Here, we observe that, similar to samegender nominations, female adolescents who report having sex are significantly likely to lose male friends and female adolescents who report making out are significantly likely to gain male friends. However, for male adolescents, coefficients from the models of samegender peer acceptance differ from those of other-gender peer acceptance. In particular, males who report making out are likely to see significant declines in friendship nominations from other males but the coefficient for other-gender peer acceptance is not statistically significant and close to zero. Any social penalty associated with boys making out appears to come solely from other male adolescents. It is possible that sexually-active adolescents replace same-grade friendships with older or younger peer relationships. This possibility may be particularly likely for female adolescents, who are more likely than male adolescents to date older heterosexual partners (Halpern, Kaestle, and Hallfors 2007). We were able to gain leverage on this idea using items collected in PROSPER waves 4 and 5. During those interviews (8 th and 9 th grade), students were asked how many friends they had outside their school-grade, ranging from 0 to 10. In unlisted analyses, we used within-person change analyses to predict this measure with the same model specification of Table 2, Model 2, for males and females. We observed no significant associations between sexual behaviors and changes in female or male adolescents' out-of-grade friendships over the two waves. We thus found no evidence supportive of Hypothesis 4a, that sexual activity coincided with increased out-of-grade friendship ties. It may also be that the association between sexual behaviors and peer acceptance is moderated by romantic involvement with the sexual partner. Female adolescents, in particular, may be at greater risk of stigmatization for engaging in sexual activities with a non-romantic partner (Crawford and Popp 2003;Tolman 2002). Unfortunately, PROSPER did not begin collecting dating status items until wave 4, and nominated partners are not connected to reported sexual behaviors (i.e., the sexual data are not dyadic). We were therefore unable to accurately test the romantic relationship moderation hypothesis. However in unlisted analyses (available upon request), we did include past year dating status and interactions between this variable and our sexual behaviors in a within-person change model of peer acceptance similar to Table 2, Model 2, for females and males. Dating had a significant positive association with peer acceptance (i.e., dating respondents had more friends than those who did not), but the interactions between dating and our sexual behaviors never approached significance, so there was not support for Hypothesis 4b. --- Discussion Gender scripts perpetuating a sexual double standard remain a central concern among gender and sexual behavior scholars. During adolescence, a widespread sexual double standard promotes gender inequality, suppresses female adolescents' healthy sexual desires, and motivates peers to stigmatize male and female adolescents perceived as gender nonconformists (Bordini and Sperb 2013;Crawford and Popp 2003;Tolman 2002). Although the internalization and deployment of distinct and potentially harmful gender scripts occurs in childhood (Martin and Ruble 2010;Tolman 2002), the application of differential gender expectations around sexuality begins in adolescence with the emergence of romantic and sexual strivings. In our study, we addressed a significant gap in our understanding of the sexual double standard with an adolescent grade cohort. Building on recent research in this area (Kreager and Staff 2009;Lyons et al. 2011), we operationalized peer acceptance using a network measure of received friendship nominations. We then examined within-person changes in peer acceptance following self-reported sexual behaviors (i.e., having sex and making out) to estimate the peer acceptance of the two sexual behaviors and look for potential gender differences in this association. Our approach extends prior research by (a) focusing on sexual onset occurring in adolescence, (b) specifying the temporal ordering between sexual behaviors and gendered social responses from peers, (c) controlling for adolescents' outgoing friendship nominations to better isolate the social reactions to sexual behaviors, and (d) accounting for time-stable between-person differences that potentially create spurious associations between sexual behaviors and peer relations. Our results were consistent with sexual script theory (Simon and Gagnon 1984, 1986, 2003) and a strong sexual double standard. Female adolescents who reported having sex had significant decreases in same-grade friends, whereas male adolescents who reported having sex had significant friendship increases. These patterns suggest that females and males receive very different social feedback during a critical period of sexual development. Even sexually abstinent adolescents would observe the social costs associated with sex for female youth and the social benefits of sex for male youth. For females, abstinence in adolescence may be a strategy for maintaining social acceptance, whereas for males, precocious virginity loss may be a strategy for social success. Our analyses of "light" sexual behaviors (e.g., making out) add nuance to the traditional sexual double standard. Net of having sex, female adolescents who reported making out saw significant increases in peer acceptance, whereas male adolescents who reported the same behavior saw significant declines in peer acceptance. We argue that this reverse double standard is consistent with gender scripts promoting romance for females and discouraging romance for males. Accordingly, "light" sexual behaviors may serve as markers of sexual desirability and maturity for female adolescents, but may signify dependence and submission for male adolescents. Moreover, because making out almost always precedes sex, female youth will tend to benefit more than will male youth early in romantic relationships. This pattern would be consistent with Giordano, Longmore, and Manning's (2006) mixed-methods study, which found that early adolescent boys involved in romantic relationships tend to have less confidence and perceived relationship power than their girlfriends. The authors argued that female adolescents have more experience than male adolescents with emotional and dyadic interactions resulting from their prior same-gender peer relationships, and they therefore adapt more easily to the new terrain of heterosexual romantic relationships. Our study adds to this narrative the idea that same-gender peers provide cross-pressures for sexual behaviors that affect romantic relationship dynamics. That female adolescents appear to be rewarded more than their male peers for making out, and that making out almost always precedes having sex, likely contributes to female adolescents' greater control over male adolescents early in romantic relationships. In this view, adolescent females' will continue to receive social rewards prior to having sex, but peer pressure and anticipated rewards contribute to adolescent males' attempts to move relationships toward sex. Differential peer reinforcement thus helps to sustain female sexual "gatekeeping" in adolescent romantic relationships (Baumeister and Vohs 2004). Although we found that female adolescents' reports of having sex and making out behaviors resulted in similar associations with being chosen as a friend by males and females (i.e., sex was associated with reduced friendships from both male and female peers and making out was associated with increased friendships from both male and female peers), the same was not true for males. In particular, making out was not associated with significant changes in male adolescents being chosen as a friend by female adolescents. This finding suggests that, for adolescent males, the social costs of light sexual activity are primarily associated with responses from male peers. An advantage of our analytical strategy was that we were able to predict changes in peerreported friendships net of changes in self-reported (i.e., outgoing) friendships. In this way, we come closer to measuring the gendered social reactions associated with adolescent sexual behaviors. Even though adolescents who reported more friendships were also more likely to receive more friendship nominations, the gendered differences in the associations between sexual behaviors and peer acceptance remained significant when controlling for the former. It is thus more valid to claim that sex creates a negative peer response for girls and a positive social response for male adolescents. Our distinguishing incoming and outgoing social ties therefore permits traction on the often-elusive stigmatization process. --- Limitations and Future Research Directions Although our study represents a contribution to gender and double standard research, it is not without limitations. Perhaps the most significant limitation concerns the unknown generalizability of results beyond PROSPER's rural Iowa and Pennsylvania schools. It may be that rural America holds more traditional sexual values and is more gender-segregated than non-rural settings. Supporting the former hypothesis, the prevalence of sexual initiation by 9 th grade in our study's sample is less than half that found in recent nationally representative surveys (Centers for Disease Control 2012; Walvoord 2010). This difference makes it essential that future researchers test if our results generalize to other regional and cultural contexts. Another study limitation pertains to the unknown social visibility of respondents' reported sexual behaviors. It is likely that adolescents attempt to manage if, when, and what their peers know about their sexual activities, particularly given perceptions of the social consequences of such behaviors. Moreover, sex only occurs in private settings, so knowledge of its existence beyond the sexual actors themselves should only occur via self-disclosure or gossip. PROSPER did not ask respondents about their awareness of peer sexual behaviors so it remains unclear if changes in peer acceptance following self-reported sexual behaviors are the direct results of peer evaluations of those behaviors. However, recent research suggests that adolescents communicate openly and often with friends about sexual behaviors and risks and that this communication typically exceeds that with parents (Busse, Fishbein, Bleakley, and Hennessy 2010;Ragsdale et al. 2014). The likelihood that sexual information circulates in peer networks is heightened by our findings that at least one sexual partner (i.e., male adolescents for sex and female adolescents for making out) benefit from such disclosures. Although our results are suggestive that sexual behaviors become public knowledge, future research would benefit from asking adolescents their perceptions of the sexual behaviors of nominated peers. Another fruitful avenue for future research would be examining if a sexual partner's social status conditions the association between sexual behaviors and peer acceptance. For example, a female adolescent who makes out with a low status male adolescent may drop in social status compared to a female who makes out with a high status male. PROSPER did not collect dyadic data on sexual behavior so we were unable to connect self-reported sex with the respective partner's characteristics. Longitudinal data of both friendship and sexual networks would be ideal for better understanding the sex-status association. Longitudinal friendship and sexual data for a global network (e.g., school or community) would also allow more sophisticated network analyses (e.g., stochastic actor-based models for network dynamics such as SIENA) that estimate a wider set of network structural properties and are specifically designed to distinguish peer selection from peer influence processes (Snijders, Van de Bunt, and Steglich 2010). It should also be noted that PROSPER's friendship nomination data allow us to operationalize peer acceptance (i.e., the number of received friendship nominations) but not peer-perceived popularity (i.e., the number of received "popular" nominations; see Cillessen and Marks, 2011, for a review). Our study therefore focuses on the association between sexual behaviors and peer likeability rather than the association between sexual behaviors and peer reputation. We are careful to confine our results to peer acceptance, likeability, or social preference, but similar processes should occur with a peer-perceived popularity outcome and we encourage future research to undertake such a study. The survey limits our ability to understand social processes surrounding the sexual debut of sexual minority youth. PROSPER did not ask respondents the gender of their sexual partners or their sexual orientation more broadly, making it impossible to distinguish other-sex and same-sex sexual behaviors. Given the low prevalence of same-sex intercourse during adolescence, we assume that the observed patterns apply only to heterosexual relationships. We thus urge future research, using an oversample of sexual minority adolescents, to evaluate peer responses to same-sex sexual debut and public displays of affection. Finally, future research should examine if the dynamics of sex and peer status vary from early adolescence to late adolescence and young adulthood. PROSPER friendship nominations have only been coded for grades 6-9, providing limited variation to examine age as a potential moderator of the sexual double standard. Future research that includes friendships during late adolescence will be able to gain greater leverage on this issue, which is particularly interesting because later adolescence coincides with a higher prevalence of sex and the prioritization of romantic and cross-gender relationships over same-gender friendships for both young women and men (Connolly and McIsaac 2011). Permissive and casual sex would also increase during this age range, further complicating peer applications of traditional gender scripts (Manning, Longmore, and Giordano 2005). Additionally, the association between precocious sexual behaviors and peer acceptance may vary in the shortterm versus the long-term. For example, Allen, Schad, Oudekerk, and Chango (2014) recently found that making out was associated with increased peer status during early adolescence, but that the correlation weakened over age. Moreover, they also found that precocious sexual and minor delinquent behaviors were associated with long-term romantic relationship difficulties and increased substance use and criminal involvement. Examining such long-term consequences of sexual behaviors is beyond the scope of current project, but should be the focus of a future study. --- Practice Implications Despite these limitations, our findings have important implications for educators and prevention scientists who work with adolescents in school or other settings. Sexual health educators should recognize that the sexual double standard not only continues to exist, but also shapes the structure of adolescents' peer networks and individual adolescents' friendship experiences. Sexuality education should include messages about men's and women's roles in romantic and sexual relationships, as already recommended by SIECUS's comprehensive sexuality education curriculum beginning in pre-adolescence with coverage of gender roles in general and continuing in adolescence with discussions of gendered messages around perceptions and evaluations of peer dating and sexuality (National Guidelines Task Force 2004). Such lessons should include messages about gender equality, but also should discourage boys from choosing to engage in sexual behaviors because they believe it is expected of them or would enhance their reputation. In an age of social media, sexual education should also emphasize the negative consequences of gossip and public dialogue that conform to the sexual double standard and stigmatize male and female adolescents for real or perceived deviations from traditional gender scripts. In sum, our results suggest that sexuality education should continue to consider gender dynamics within adolescent dating relationships given the differential reputation outcomes of sexual behavior and extend such education to cover gendered responses to sexuality within the broader milieu of adolescent peer society. --- Conclusion Our study provides a significant contribution to sexual double standard research. With longitudinal peer network nomination data, we found evidence of a strong double standard during the period of sexual onset, such that female adolescents tend to lose peer acceptance after reporting having engaged in sex and gain acceptance after reporting making out, whereas male adolescents tend to gain acceptance after sex and lose acceptance after making out. Far from disappearing, our results suggest that the sexual double standard and the gender scripts that underlie it remain alive and well during the developmental period when youth begin shaping their sexual identities and peer influence peaks. --- Predicted peer acceptance over waves by gender and sexual behavior --- First Sex --- Author Manuscript

A sexual double standard in adolescence has important implications for sexual development and gender inequality. The present study uses longitudinal social network data (N = 914; 11-16 years of age) to test if gender moderates associations between adolescents' sexual behaviors and peer acceptance. Consistent with a traditional sexual double standard, female adolescents who reported having sex had significant decreases in peer acceptance over time, whereas male adolescents reporting the same behavior had significant increases in peer acceptance. This pattern was observed net of respondents' own perceived friendships, further suggesting that the social responses to sex vary by gender of the sexual actor. However, findings for "making out" showed a reverse double standard, such that female adolescents reporting this behavior had increases in peer acceptance and male adolescents reporting the same behavior had decreases in peer acceptance over time. Results thus suggest that peers enforce traditional sexual scripts for both "heavy" and "light" sexual behaviors during adolescence. These findings have important implications for sexual health education, encouraging educators to develop curricula that emphasize the gendered social construction of sexuality and to combat inequitable and stigmatizing peer responses to real or perceived deviations from traditional sexual scripts.

In contemporary societies, direct engagement in political action is crucial to disadvantaged people's pursuit of social change. Taking part in mass movements such as the Arab Spring or Occupy Wall Street, signing an Amnesty International's petition, or casting a vote in electoral events such as the recent EU referendum in Britain are examples of actions that enable individuals to voice discontent, challenge the status quo, and improve their collective social and economic standing. Research in social psychology has shown that injustice appraisals and anger (Jost et al., 2012), and a sense of efficacy (Balch, 1974) are key factors in motivating individuals' engagement in these direct forms of political participation (van Zomeren, Leach, & Spears, 2012;van Zomeren, Spears, Fischer, & Leach, 2004). However, disadvantaged individuals may lack material and cultural resources to join political movements or engage collectively against relevant institutions. They may have low confidence in their own abilities to influence the political decisionmaking process or doubt the system's willingness to respond to their needs (Balch, 1974). As a result, they may avoid engaging directly in political action. Thus, the question remains if there are other channels through which disempowered individuals may express collective dissent from the status quo. In this research, I use the social banditry framework (Hobsbawm, 1959) and propose that support for actors who disrupt the functioning of the system (legally or illegally), and who operate outside conventional societal and political structures of power and resistance, is one such channel. Specifically, I argue that support for such actors enables disadvantaged individuals to voice their discontent against an unjust system in a vicarious form. A contemporary version of social banditry is the group of hackers calling itself Anonymous (Wong & Brown, 2013). Anonymous challenge the state's authority using trolling and other forms of misbehavior. Recently, the group was named Person of the Year in the Time Magazine, attempted to expose ISIS (Griffin, 2015), and intervened in the American presidential campaign (Gibbs, 2016). Despite their political and social relevance, little is known about what motivates people to support this group. The present research tests the social banditry framework to explain support for Anonymous as a form of vicarious dissent. --- Direct Participation in Political Action Political participation may encompass institutional (e.g., voting), noninstitutional (e.g., protests, signing a petition), or even violent (e.g., damage to property) actions (van Deth, 2014;Becker & Tausch, 2015). What these (otherwise different) behaviors have in common is that they refer to a situation in which individuals directly engage in political action to influence and alter their political contexts. According to van Zomeren et al.'s (2004) dual pathway model, two complementary but independent factors contribute to explaining individuals' decisions to engage directly in political action. The first factor is injustice appraisal. Research has demonstrated that individuals' subjective experience of disadvantage is a powerful predictor of actions aimed at confronting the disadvantage (Smith, Pettigrew, Pippin, & Bialosiewicz, 2012). At the societal level, perceiving the system as unfair increases the likelihood that individuals take part in protest (Jost et al., 2012;Rothmund, Becker, & Jost, 2016). Importantly, the perception of injustice elicits anger, which mediates the linkage between injustice appraisal and engagement in protest (Jost et al., 2012). The second factor is efficacy. In order to take action, individuals must feel able to achieve the desired change. At the societal level, individuals must perceive they can shape the political system, and that the political system is responsive to their needs and demands, a concept known as political efficacy (Balch, 1974;Campbell, Gurin, & Miller, 1954). Research has demonstrated that stronger levels of political efficacy are associated with increased participation in different forms of political action, such as voting, or attending a rally (Flavin & Keane, 2012;Smets & van Ham, 2013). Current models of political action assume inaction and passivity when individuals are unable to cope successfully with an unjust disadvantage (van Zomeren et al., 2012). There is also some evidence that individuals are more likely to opt for violent forms of participation when they believe that their group cannot obtain change through collective efforts (i.e., they lack group efficacy; Taush et al., 2011). But how do individuals express collective dissent when they perceive an unjust system as irresponsive to their political needs (i.e., they lack political efficacy)? Here, I contend that in such circumstances individuals' may opt to convey their dissent vicariously. Rather than engaging directly in political action, they may support actors who are otherwise outside the conventional realm of politics, are often targeted as criminals by the state, and whose behavior disrupt the system's functioning. To test this contention, in the present research I investigate support for the hacker group Anonymous. Specifically, I contend that support for this group represents an expression of inchoate anger and dissent when individuals do not feel they can express their political grievances by directly targeting institutions, the system, or the government. --- Social Banditry as Vicarious Dissent If social and economic arrangements prevent subordinate groups from expressing dissent against oppression and injustice, individuals must rely on alternative strategies for criticizing and resisting power (Scott, 1990). For instance, in traditional peasant societies, the masses were devoid of political voice and had few means for pursuing or expressing a desire for social change (Hobsbawm, 1973). In such a context, individuals' aspirations for social justice took the form of inchoate anger, a desire for vengeance against the oppressive establishment (Schneider & Schneider, 2008). Groups like bandits, rustlers, and brigands often became the embodiment of such anger. Hobsbawm (1959) used the label'social banditry' to describe those outlaws who, in peasant societies, were labelled as criminals by the authorities, but were supported and protected by local communities. Social bandits were sometimes regarded as heroes because their robberies and looting were construed as acts of defiance against the rich and powerful. They disputed the social order by showing that the powerful and the oppressors could be challenged. Traditionally, social bandits prospered in geographic areas that were hard to police such as mountains, forests, and deserts. Epitomized by figures such as Robin Hood or Jesse James, they were celebrated in local folklore as noble individuals who robbed the rich, gave to the poor and used violence only for self-defense and for righting wrongs. Their deeds enabled disadvantaged individuals, who felt they had no access to other means, to take part in a sort of vicarious 'protest' against the oppressors. Sociological and historical research has long debated whether certain outlaws in specific contexts were in fact 'vicarious executor(s) of the unarticulated rage of most of the rural poor' (Lewin, 1979, p. 82), or merely self-interested criminals who preyed on and exploited peasants as well as landlords (Blok, 1972;Hobsbawm, 1972;Joseph, 1990;Slatta, 2004;Schneider & Schneider, 2008). Nonetheless, regardless of its phenomenology, social banditry has an important psychological function. Support for banditry may reflect disadvantaged individuals' desire for more justice, a vicarious expression of dissent. Such a psychological dimension has yet to be empirically investigated. In this research, I test the social banditry framework and investigate support for Anonymous vis-à-vis intentions to participate in direct forms of institutional (voting) and non-institutional (protest) political action. --- Anonymous as Social Banditry Anonymous is a network of hackers that emerged from the online board '4chan' in 2006 (Coleman, 2014;Jarvis, 2014;Wong & Brown, 2013). Originally, Anonymous was known for its outrageous pranks aimed at upsetting people and creating public amusement. This behavior, known as trolling, includes spamming online forums, disrupting access to web pages (DDoS), revealing people's personal information and spreading disturbing content on the internet. Anonymous started to attract popular interest when it used these tactics to oppose Scientology's attempts to limit freedom of expression on the internet (Coleman, 2014). Since then, Anonymous has become an important global actor, waging 'war' against the US government, ISIS and other actors, and engaging in a series of 'operations' against financial companies. While the political meaning of such actions is more explicit compared to earlier hoaxes, trolling has remained an important component of Anonymous' behavior. Anonymous shares important characteristics with the traditional concept of social banditry (Söderberg, 2008;Wong & Brown, 2013). Anonymous uses both legal and illegal tactics to attack its targets and challenge authorities, although they do not have a coherent and unified political program. As traditional bandits, they operate in a space which is difficult to police and oversee, the internet (Schneider & Schneider, 2008). Importantly, Anonymous is surrounded by the same ambiguity that characterized the concept of social banditry. They may be perceived as Robin Hoodtype figures, who 'take from the powerful to empower the disempowered' (Wong & Brown, 2013, p. 1016) or merely as trouble makers, or even criminals (Tomblin & Jenion, 2016). Their operations are conducted for personal enjoyment and amusement, as well as 'punishing' institutions and corporations. --- Social Banditry, Individualism and Anonymous Social bandits were not revolutionaries and did not explicitly promote ideas about justice, or collective change. Instead, their actions sanctioned individuals' desire for vengeance against the oppressors, their ambitions to 'take the law into their own hands' and the righting of individual wrongs (Hobsbawm, 2000). Here, I propose that, compared to direct engagement in political action, vicarious dissent and support for social banditry reflect an affirmation of individualistic values. Individualism and collectivism are distinctive cultural values which place relatively stronger importance either on the individual or the harmony of the group, respectively (Hofstede, 1980;Oyserman, Coon, & Kemmelmeier, 2002;Triandis, 1995). Individualism emphasizes personal autonomy, independence and self-interest. In contrast, collectivism emphasizes interdependence, groups' common fate, and harmonious relationships. Individualism and collectivism can be further divided along the vertical and horizontal dimensions, depending on the importance of values of hierarchy or equality, respectively (Sivadas, Bruvold, & Nelson, 2008). The vertical dimension captures individuals' acceptance of hierarchical relationships and differences in status among individuals, whereas the horizontal dimension refers to the acceptance of egalitarianism. The resulting four-fold typology encompasses horizontal individualism (HI) and collectivism (HC), vertical individualism (VI) and collectivism (VC). Previous research has demonstrated such values may influence political decisions. For instance, using data from the 1990-1992 National Election Studies, Funk (1998) showed that stronger individualism (and lower collectivism) was associated with lower propensity to engage in collective actions aimed at benefitting the community. This is because people endorsing individualistic values may be less attuned to the societal interest, and more attuned to self-interest (cf. van Prooijen, 2013). Research has yet to investigate the role of individualistic and collectivistic values in the context of social banditry and Anonymous. According to the Intracultural Appropriation Theory (ICAT), social actors challenging the established order seek legitimacy through cultural beliefs and values shared among the population to gain support, justify their actions, and achieve their goals (Travaglino, Abrams, & Russo, 2017). For instance, in the Southern Italian context, Travaglino, Abrams and Randsley de Moura (2014) showed that endorsement of masculine honor beliefs among youth was associated with a more positive view of mafia-style groups. This is due to mafia's appropriation and strategic use of values of masculinity, self-reliance and violence. But which values are used by Anonymous to obtain legitimacy? The actions of Anonymous are characterized by a strong tension between individualistic and collectivistic values (Coleman, 2014;Goode, 2015). Prima facie, Anonymous seems to emphasizes collectivism. For instance, Anonymous' membership is subsumed under the umbrella of an overarching collective identity. Members all use the same sobriquet ('anonymous') when they communicate with each other, and they all wear the same Guy Fawkes mask when they interact with the public. Such measures protect members' identity in case of illegal actions. They also work as a check against potential temptations for self-aggrandizement and fame. Whereas Anonymous (as a collective identity) is well known and influential, its anonymous members are generally not. However, Anonymous' ethos places an even stronger emphasis on individualistic values. It promotes personal privacy, individual autonomy, and unrestricted freedom of speech. Consistent with the broader hacker community' ethos, Anonymous distrusts centralized authority and celebrates the 'tremendous power of the individual' (Coleman & Golub, 2008, p. 256). Golumbia (2013) characterizes Anonymous as (cyber)libertarians, a strongly individualistic form of political ideology (Iyer, Koleva, Graham, Ditto, & Haidt, 2012). Importantly, in line with the notion of social banditry, Anonymous' operations resemble more closely acts of revenge against governments and corporations, rather than a program aimed at improving the collective interest. This research examines the role of individualistic and collectivistic values in predicting support for Anonymous vis-à-vis direct political engagement. --- Overview of the Studies and Hypotheses In this research, I present two studies investigating attitudes toward Anonymous vis-à-vis engagement in direct forms of political action. According to the social banditry framework, people in an unfair context, but with no political voice, will feel anger against the system. This should in turn boost support for actors who disrupt and challenge the system, and thus who provide a vicarious voice for such anger (Hobsbawm, 2000). This framework implies that stronger injustice appraisals and lower levels of political efficacy should predict stronger anger against the system. Anger should, in turn, predict more positive attitudes toward Anonymous. These studies compare attitudes toward Anonymous to intentions to engage in non-institutional (protest) and institutional (voting) forms of direct political action. According to the dual pathways model, injustice-fueled anger and efficacy are two distinct paths that explain participation in political action. Therefore, individuals should be more likely to express intentions to protest when they report higher levels of anger and stronger political efficacy. Finally, stronger political efficacy (but not anger) should be a predictor of voting intentions because of the institutional, and system-supporting character of voting (Tausch et al., 2011). In addition, Study 2 tests the relationship between individualistic and collectivistic values and support for Anonymous vis-à-vis other forms of direct political engagement. Support for social banditry is a manifestation of individuals' desire for personal vengeance against the system, rather than a program for collective social change. Moreover, Anonymous promotes values of personal autonomy and individualism. Thus, in line with the social banditry framework and ICAT, I predict that endorsement of individualistic values will be linked to more positive attitudes toward Anonymous. In contrast, participation in direct political action isby definitiona form of collective action aimed at promoting societal interest. Thus, collectivistic values should predict intentions to engage in direct forms of political action such as protesting or voting (Kelly & Breinlinger, 1995;Funk, 1998). Because both support for Anonymous and political action represent a desire for social change and justice, both expressions of dissent should be predicted by horizontal (rather than vertical) values. Across studies, I included measures of participants' political orientation and subjective social status to control for their effects. Individuals' political orientation may influence how they perceive inequality (Rothmund et al., 2016). This, in turn, may affect individuals' intentions to engage in direct forms of political action or their attitudes toward Anonymous. Subjective social status refers to individuals' perception of their position in the social hierarchy (Jackman & Jackman, 1973), and may affect their confidence in political institutions (Cook & Gronke, 2005). Thus, it is important to control for this construct. Finally, as previous research has shown that younger people are more likely to use the internet to engage in political action (Bakker & Vreese, 2011), and because hacking is associated with a male stereotype (Tanczer, 2016), analyses control for gender and age. --- Study 1 --- Method Participants and procedure. Three hundred and four British participants (151 males, 151 females, 1 unreported) took part in the study. The mean age was 33.39 (SD = 11.17) and the majority were English (85.82%). The remaining participants were from Scotland (6.8%), Wales (5.6%) and Northern Ireland (1.98%). Participants were recruited using Qualtrics via the online platform Prolific Academic (www.prolific.ac) to participate in a survey "about social issues" (see the Appendix for further information about the sample). After completing the measures, participants were debriefed in writing, thanked and compensated for their time. --- Materials. Responses were measured on a 7-point scale (1 = strongly disagree, 7 = strongly agree) unless differently noted in the succeeding text. For each construct, the order of presentation of the items was randomized. A mean score for each construct was computed by averaging the relevant items. Political efficacy. Political efficacy was measured using seven items from various scales (Campbell, Gurin, & Miller, 1954;Muller, Jukam, & Seligson, 1982;Niemi, Craig, & Mattei, 1991;Watts, 1973;cf. Taush et al., 2011): 'The way people vote is the main thing that decides how things are run in this country', 'It is only wishful thinking to believe that one can really influence what happens in society at large' (reverse-coded), 'Government officials don't care much about what people like me think' (reverse-coded), 'It seems that whoever people vote for, things go on pretty much the same' (reverse coded), 'I feel I am quite well represented in our political system', 'How much influence do you think someone like you can have over national government decisions?' and 'To what extent do you feel that the basic rights of citizens are well protected by our political system?' (1 = none at all, 7 = a great deal). Items formed a reliable scale (<unk> =.77). Perceived justice of the system. Perception of living in a just system was measured using three items drawn from Kay and Jost's (2003) system justification scale. The items measured the perception that the UK's economic and social arrangements are just and fair. Items used were, 'In general, I find British society to be fair', 'Most of the British policies serve the greater good', and 'Everyone has a fair shot at wealth and happiness in the UK' (<unk> =.86). Anger. Participants' perceived anger toward the political system was measured with three items (1 = not at all, 7 = extremely). Participants read 'In general when I think about our political system, I feel...' and then indicated the extent they felt angry, frustrated, and outraged (<unk> =.88). --- Intentions to participate in non-institutional political action. Intentions to engage in political action were measured with three items (1 = very unlikely, 7 = very likely). Participants first read 'below are listed a series of activities people may take part in to express their voice and/or dissent in society. Please indicate how likely you would be to take part in each of these activities in the future if the opportunity arises'. Participants were then asked to indicate the likelihood to'sign a petition', 'attend a protest event' and 'participate in a public demonstration' (<unk> =.70). --- Voting intentions. Participants were asked how likely they were to vote in the next general elections (1 = very unlikely, 7 = very likely). --- Attitudes toward --- Results and Discussion Intercorrelations among variables, means, and standard deviations are presented in Table 1. Consistent with the social banditry framework, an inspection of the bivariate correlations revealed a significant relationship between attitudes toward Anonymous and political efficacy (r = -.18), the perception of the fairness of the system (r = -.23) and anger toward the system (r =.30). Path Analysis. I tested the social banditry framework using a path model, where the perceived fairness of the system and political efficacy predicted attitudes toward Anonymous through anger against the political system. In line with the dual pathways model, paths were modelled from both political efficacy and anger to intentions to engage in non-institutional and institutional political action. --- Residuals of voting intentions, collective action intentions, and attitudes toward Anonymous were allowed to covary, to capture systematic variation between the variables not accounted by the predictors (Preacher & Hayes, 2008). Gender, age, political orientation and SSS were covariates in the model. Parameters were estimated using full information maximum-likelihood method, given that there were few missing observations (<unk> 3%). Analyses were run in R using the Lavaan package emulating Mplus (Rosseel, 2012). The model had a very good fit. Chi-square was non-significant, <unk> 2 = (4, N = 304) = 3.48, p =.48. The other indices also indicated a well-fitting model, AGFI =.998, CFI = 1.00, SRMR = 0.01, RMSEA <unk>.001 (90% CI = 0.00 to 0.08, p =.77). Figure 1 summarizes the model. Perceived justice, <unk> = -.30, SE =.06, p <unk>.001, and political efficacy, <unk> = -.38, SE =.08, p <unk>.001, were negatively related to anger against the political system. This is consistent with previous findings from Tausch et al. (2011, Study 3) and suggests that perceiving the political system as unfair and irresponsive boosts anger against it. Anger in turn predicted attitudes toward Anonymous, <unk> =.22, SE =.05, p =.001. There were no direct relationships between perceived justice and political efficacy, and attitudes toward Anonymous <unk> <unk> <unk>.08, SE <unk>.09, p >.24. In line with the dual pathway model, intention to participate in noninstitutional political action was predicted by anger, <unk> =.42, SE =.06, p <unk>.001, and political efficacy, <unk> =.29, SE =.09, p <unk>.001, whereas the direct effect of perceived justice was non-significant, <unk> = -.07, SE =.06, p =.30. Finally, voting intentions were significantly predicted by political efficacy, <unk> =.16, SE =.11, p =.027, and nonsignificantly by anger or perceived justice, <unk> <unk> <unk>.012, SE <unk>.07, p >.86, suggesting that decisions to vote are not seen as an expression of protest in the British context. The indirect effects of political efficacy and perceived justice via anger were tested using 5,000 bootstraps. As expected, and consistent with the social bandit framework, both the indirect effects of perceived justice, <unk> = -.07, SE =.02, 95% CI [-.09, -.02], and political efficacy, <unk> = -.08, SE =.03, 95% CI [-.17, -.04] on attitudes toward Anonymous were significant. In line with the dual pathway model, there was a significant indirect effect of perceived justice on participation in non-institutional political action, <unk> = -.12, SE =.03, 95% CI [-.16, -.06] (see also Jost et al., 2012). Interestingly, there was also a negative indirect effect of political efficacy on participation in non-institutional political action via anger, <unk> = -.16, SE =.04, 95% CI [-.29, -.12]. 1 --- Study 2 --- Method Participants and procedure. Four hundred and ten participants (188 males, 212 females, 10 unreported) from the USA were recruited using Qualtrics and Mturk. The mean age was 34.99 (SD = 10.93). Participants were from different States, including California (9.5%), Florida (9.2%), Texas (6.3%), Pennsylvania (5.7%), New York (5.5%) and Illinois (5.2%). The other States represented in the sample each accounted for <unk> 5% of the sample (see the Appendix for further information about the sample). Data for this study was collected in December 2015, before the official start of the American presidential primaries. Materials. The measures were the same as in Study 1, except that a measure of horizontal/vertical individualism/collectivism (Sivadas, et al., 2008) was included. Responses were measured on a 7-point scale (1 = strongly disagree, 7 = strongly agree). For each construct, items' sequences were randomized and a mean score was computed by averaging the relevant items. --- Political efficacy. Seven items as in Study 1 were used to measure participants' perceived ability to exert political power (<unk> =.82). Perceived justice of the system. Three items from Kay and Jost's (2003) scale were used to measure individuals' perceived justice of the system, as in Study 1. The items were adapted to the American context and formed a reliable scale (<unk> =.87). Individualism/collectivism. Participants' orientation toward the individualism/collectivism and horizontal/vertical dimensions was measured using a slightly adapted version of Sivadas, et al.'s (2008) 14-item scale. The scale includes three items tapping HI (e.g., I am a unique individual'), three items tapping VI (e.g., 'competition is the law of nature'), four items tapping HC (e.g., 'I feel good when I cooperate with others'), and four items tapping VC (e.g., 'I would sacrifice an activity that I enjoy very much If my family did not approve of it'). Anger. Perceived anger toward the political system was measured using the same three items used in Study 1 (<unk> =.89). --- Intentions to participate in non-institutional political action. Intentions to engage in political action were measured with the same three items (1 = very unlikely, 7 = very likely) as in Study 1 (<unk> =.80). --- Voting intentions. Participants were asked how likely they were to vote on the next election day (1 = very unlikely, 7 = very likely). --- Attitudes toward Anonymous. Participants rated their perception of Anonymous using the same nine items used in Study 1 (<unk> =.94). Before completing the items, participants were also shown an identical description and image. --- Political orientation. One item measured participants' political orientation, 'Would you consider yourself a Liberal or a Conservative?' (1 = extremely liberal, 7 = extremely conservative) --- Subjective social status (SSS). Participants' perception of their own social status was measured using the same graphic item used in Study 1. --- Results and Discussion Bivariate correlations showed a significant relationship between attitudes toward Anonymous and political efficacy (r = -.16), perceived justice (r = -.39), and anger (r =.39). Intercorrelations among variables, means, and standard deviations are shown in Table 2. To check whether the scale used was able to distinguish the different cultural dimensions, a factor analysis was performed on the 14 items measuring HI, VI, HC, and VC, using principal components as the method of extraction and varimax rotation. A four-factor solution emerged explaining 66% of the variance. In line with Sivadas et al. (2008), items assessing each of the four constructs loaded on the expected factor, except the item 'My happiness depends very much on the happiness of those around me' which cross-loaded on both HC and VC (rotated factor loadings.49 and.39, respectively). In addition, the item 'Children should feel honored if their parents receive a distinguished award', whose expected factor was VC (.23), loaded more strongly on HC (.53). Thus, these two items were dropped. A factor analysis on the remaining 12 items using principal components as the method of extraction and varimax rotation yielded a four-factor solution explaining 73% of the variance. All the items loaded strongly on the expected factor (>.80) but not on the others (<unk>.19). The reliability of each of the three-item subscale was satisfactory, <unk> >.79. Therefore, the 12-item version of the scale is used in the analyses below. Path Analysis. In line with the social banditry framework, I tested a path model in which perceived justice and political efficacy predicted attitudes toward Anonymous via anger against the political system. Following the dual pathway model, distinct paths were modelled from both political efficacy and anger to intentions to engage in institutional and non-institutional political action. The horizontal/vertical individualism/collectivism subscales were added as additional predictors of political participation, voting intentions and attitudes toward anonymous. As in Study 1, residuals of political participation, attitudes toward Anonymous and voting intentions were allowed to covary. Gender, age, political orientation and SSS were covariates in the model. Because there were few missing observations (<unk> 2.19%), FIML was used to estimate parameters in the model. Analyses were run using R and the Lavaan package emulating Mplus. The model had excellent fit and is summarized in Figure 2. Chi-square was non-significant, <unk> 2 = (5, N = 405) = 7.02, p =.22. The other indices indicated a well fitting model, AGFI =.997, CFI = 1.00, SRMR =.011, RMSEA =.03 (90% CI = 0.00, 0.08, p =.67). Anger was predicted by political efficacy, <unk> = -.30, SE =.07, p <unk>.001, and perceived justice of the system, <unk> = -.29, SE =.06, p <unk>.001. In turn, anger positively predicted attitudes toward Anonymous, <unk> =.16, SE =.05, p =.002. Perceived justice significantly predicted attitudes toward Anonymous, <unk> = -.28, SE =.06, p <unk>.001. The path from political efficacy to attitudes toward Anonymous was non-significant, <unk> =.06, SE =.07, p =.31. Consistent with the dual pathways model, intentions to engage in noninstitutional political action were predicted by anger, <unk> =.36, SE =.05, p <unk>.001 and political efficacy, <unk> =.29, SE =.07, p <unk>.001. The path from perceived justice was not significant, <unk> = -.04, SE =.06, p =.55. Voting intentions was predicted by political efficacy, <unk> =.13, SE =.08, p =.017 and, unexpectedly, anger, <unk> =.18, SE =.05, p =.001. There was no effect of perceived justice, <unk> =.02, SE =.06, p =.73. --- As expected, HI was uniquely and positively associated with attitudes toward Anonymous, <unk> =.14, SE =.07, p =.005. VI, HC, VC did not significantly predict attitudes toward Anonymous, <unk>s <unk>.07, ps >.14. This is consistent with ICAT and the hypothesis that support for Anonymous reflect an expression of individualistic values. In line with the idea that positive attitudes toward Anonymous represent individuals' aspiration toward more equality, the horizontalbut not the verticaldimension predicted support for the group. In contrast, HC predicted intentions to participate in non-institutional political action, <unk> =.16, SE =.07, p =.001, and voting intentions, <unk> =.21, SE =.08, p <unk>.001. Other dimensions of collectivism and HI and VI were not significantly related to either variable, <unk>s <unk>.08, ps >.09, consonant with the fact that political participation and voting are activities driven by collective ideals. --- General Discussion This research investigated attitudes toward Anonymous, a group of hackers using different tactics to challenge and retaliate against governments and corporations. I drew on the social banditry framework (Hobsbawm, 1959), and proposed that support for Anonymous represents a vicarious expression of dissent. I hypothesized that weaker perceived political efficacy and stronger sense of injustice would be associated with more positive attitudes toward Anonymous through anger against the political system. Across two samples (N = 304 and N = 410) and two different geographical contexts (the UK and the US) results supported this hypothesis. Study 1 used a path model to test predictors of attitudes toward Anonymous vis-à-vis intentions to engage directly in institutional (voting) and non-institutional (e.g., protesting) political action. Consistent with prior research (e.g., van Zomeren, et al., 2004), stronger injustice-fueled anger and stronger levels of political efficacy significantly predicted non-institutional political action. Political efficacy (but not anger) was a significant predictor of voting intentions, consonant with the idea that, at least in the British context, voting is a non-confrontational form of political engagement. In line with my theorising and the social banditry framework, people who perceived weaker political efficacy and expressed lower justice appraisals expressed stronger support for Anonymous through anger against the political system. This pattern of results is consistent with the idea that social actors like Anonymous may become a vehicle for individuals' anger against an unjust system, and convey dissent vicariously. Study 2 largely replicated these results in a different geographical context, the US. Notably, in Study 2 anger was a significant predictor of individuals' voting intention. This is perhaps an indication of current party polarization, and the resulting confrontational nature of voting, in the American context (Kimball & Gross, 2007). In addition, Study 2 extended Study 1 by

The author thanks Clare Longhorn for helping preparing the measure for this study and discussing the manuscript with me. My thanks also go to the participants at the Culture and Psychology Conference (Sept, 2016, University of Kent), and to Libby Drury for her comments and feedback on this work.

prior research (e.g., van Zomeren, et al., 2004), stronger injustice-fueled anger and stronger levels of political efficacy significantly predicted non-institutional political action. Political efficacy (but not anger) was a significant predictor of voting intentions, consonant with the idea that, at least in the British context, voting is a non-confrontational form of political engagement. In line with my theorising and the social banditry framework, people who perceived weaker political efficacy and expressed lower justice appraisals expressed stronger support for Anonymous through anger against the political system. This pattern of results is consistent with the idea that social actors like Anonymous may become a vehicle for individuals' anger against an unjust system, and convey dissent vicariously. Study 2 largely replicated these results in a different geographical context, the US. Notably, in Study 2 anger was a significant predictor of individuals' voting intention. This is perhaps an indication of current party polarization, and the resulting confrontational nature of voting, in the American context (Kimball & Gross, 2007). In addition, Study 2 extended Study 1 by examining the role of individualism and collectivism in the context of vicarious and direct political engagement. Based on the notion of social banditry, and on evidence that Anonymous emphasizes values of individualism and libertarianism (Golumbia, 2013), I proposed that endorsing an individualistic orientation would be associated with more positive attitudes toward Anonymous. More specifically, because support for Anonymous conveys individuals' aspirations for social equality, I proposed that the horizontal dimension of individualism should be associated with support for the group. Results supported these hypotheses. Individuals who more strongly endorsed HI expressed more favorable attitudes toward Anonymous. This effect was independent of other constructs and cultural dimensions. This indicates HI's distinct role in explaining attitudes toward Anonymous. In contrast, those who more strongly endorsed HC had stronger intentions to engage in political action. These findings are consonant with the social banditry framework, and in particular Hobsbawm's (2000) observation that bandits are admired because their existence conveys a message of personal vengeance against the authority, rather than a collective program for social change. Findings also provide support for ICAT's central proposition that cultural values may bestow groups with legitimacy. Groups can then use this legitimacy to gain stronger social consensus, accomplish their goals and objectives, and obtain power and support. Future research should test the role of individualism in different political contexts, and use different groups and forms of vicarious dissent. --- Efficacy, Anger and Support for Social Banditry According to the dual pathway model, injustice-fueled anger and efficacy are independent pathways to political action (van Zomeren, et al., 2012). Evidence from this research is consistent with that assumption, but only regarding direct participation in political of action. Across studies, anger fully mediated the relationship between political efficacy and attitudes toward Anonymous. This result can be attributed to the fact that measures used in these studies tapped participants' perception of the social and political system as a whole, rather than participants' sense of efficacy regarding specific issues or groups (Jost et al., 2012). It is reasonable to expect that participants who perceive lower levels of political efficacy, and are thus dissatisfied with the system's responsiveness to their political needs, also express more anger toward the political system. This is consistent with the idea that support for social banditry is a form of vicarious dissent, whereby political grievances that cannot be otherwise voiced trigger anger against the political system, which in turn promote support for bandits. Another interesting feature of this evidence is the relationship between political efficacy and non-institutional political action. Political efficacy predicted intentions to engage in non-institutional political action both directly (and positively) and indirectly (but negatively). This suggests that anger may suppress the association between political efficacy and engagement in non-institutional political action. In turn, this may explain why some authors have failed to find a relationship between political efficacy and participation in protest (e.g., Rudig & Karyotis, 2014). Future research should further investigate the articulation between political efficacy, anger and different forms of political engagement. Moreover, this research demonstrates the key role of anger in predicting vicarious expressions of dissent. Recently, Tausch et al. (2011;Becker & Tausch, 2015) theorized that anger is a better predictor of direct engagement in, rather than of support for other agents', political actions. For instance, Tausch et al. (2011, study 2) showed that anger did not predict Muslim students' support for government policies aimed at supporting Muslim communities in India. Somewhat inconsistent with this evidence, results from these studies showed a positive relationship between anger and support for Anonymous (also when controlling for contempt, see footnote 1). This may be due to the fact that groups such as Anonymous represent a better vehicle for people's anger, compared to government policies. This finding also supports the idea that different forms of anger have different implications for political action (Taush, et al., 2011;cf. Russel & Fehr, 1994). Support for Anonymous may be driven by feelings more akin to resentment due to the fact individuals perceive the political system as unjust and uncontrollable (Feather & Sherman, 2002;Smith, Parrott, Ozer, & Moniz, 1994). Future research should better differentiate among different forms of anger vis-à-vis support for Anonymous. --- Limitations and Future Directions This paper has presented two studies investigating support for Anonymous in the British and American contexts. Results support the social banditry framework and the dual pathway model and provide important new insights about the idea of vicarious dissent. Nonetheless, this research is affected by some limitations. First, the measure of political efficacy used in this study (Taush et al., 2011) does not allow to distinguish between internal and external political efficacy. Political efficacy is a complex concept with multiple dimensions (Morrell, 2003). Future studies should include measures able to distinguish between internal and external, as well as collective, dimensions of political efficacy (Lee, 2005), and investigate the specific role of each dimension in predicting vicarious dissent vis-à-vis direct engagement in different forms of political action. In similar a vein, future research should better elucidate the role of grievances and injustice appraisals in predicting support for Anonymous. This research used items drawn from Kay and Jost's (2003) system justification scale, measuring the perception of the fairness of the system. Future research should test the effect of feelings of relative disadvantage and deprivation concerning more specific economic and social areas (e.g., issues of privacy and information). A key area for future research is the transition from vicarious to direct political expressions of political dissent. There are circumstances in which Anonymous' supporters have taken the streets (against Scientology) or joined forces with other, more traditional, social movements and protest groups (the Occupy movement). It could be that support for groups such as Anonymous ultimately provides an arena where new politicized social identities may emerge, which may, in turn, promote individuals' engagement in protest. Moreover, some anti-establishment politicians such as Nigel Farage in the UK and Donald Trump in the US, or parties such as the Five Star Movement in Italy may harness vicarious dissent to gain consensus. Indeed, such actors often use the rhetoric of'rebalancing power from large corporations and big government institutions and putting it back into the hands of the people' (UKIP Manifesto). These are important avenues for future research. Finally, it is important to highlight that, although cross-sectional data allow testing of theoretically specified relationships among variables, causal conclusions cannot be drawn from the data. Future research should use longitudinal and experimental designs to investigate support for Anonymous or other forms of modern social banditry. For example, future research could manipulate perceived political efficacy and investigate its effect on anger and support for Anonymous. --- Implications and Conclusions This research is the first to investigate vicarious dissent and support for Anonymous. Social psychological work has predominantly focused on engagement in direct forms of political action (Becker & Taush, 2015). However, disadvantaged and voiceless individuals can use different means for contesting their disadvantage (Leach & Livingstone, 2015;Scott, 1990). Support for social bandits and groups like Anonymous might be one such means. These groups may provide individuals with the opportunity to express dissent in a vicarious form. This research introduces the novel social banditry framework (Hobsbawm, 2000) to the social psychological study of political action. Theories of political action such as System Justification (Jost et al., 2012) or Belief in a Just World (Lerner, 1980) suggest that individuals may accept and justify existing (sometimes oppressive) social arrangements. Social Identity Theory contends that when individuals perceive intergroup boundaries as stable and legitimate they may avoid direct challenges to the status quo (Tajfel & Turner, 1979). There is indeed plenty of evidence that in some circumstances individuals do justify the system (Jost et al., 2010). Moreover, research indicates that perceiving the world as a just place, or believing that intergroup boundaries as unchangeable, dampen engagement in political action (Ellemers, 1993;Jost & Hunyady, 2005). However, it does not follow that when individuals are not directly engaged in political action, they are supinely accepting their disadvantage. Even the most severe conditions of powerlessness can be accompanied by silent expressions of resistance. To build a better understanding of power relations in society, social psychology must make it a priority to investigate those expressions. 1 Study 1 also included a measure of contempt (1 item, 'When I think about our political system, I feel contempt', Tausch, et al., 2011). Consistent with the idea that support for Anonymous is an expression of anger against the system, rather than contempt, contempt did not significantly predict attitudes toward Anonymous, <unk> =.08, SE =.05, p <unk>.22. Moreover, contempt did not predict intentions to take part in non-institutional political action, <unk> =.06, SE =.05, p <unk>.41. However, in line with previous work (Tausch, et al., 2011), it was negatively and significantly associated to psychology, 87, 649-664. doi: 10.1037psychology, 87, 649-664. doi: 10. /0022-3514.87.5.649 Watts, M. W. (1973)). Efficacy, trust, and commitment to the political process. Social Science Quarterly, 54, 623-631. Wong, W. H., & Brown, P. A. (2013). E-bandits in global activism: WikiLeaks, anonymous, and the politics of no one. Perspectives on Politics,11, 1015-1033. Doi: 10.1017/S1537592713002806 Think of the ladder above as representing where people stand in society. At the top of the ladder are the people who are best off -those who have the most money, most education and the best jobs. At the bottom are the people who are worst offwho have the least money, least education and the worst jobs or no job. The higher up you are on this ladder, the closer you are to people at the very top and the lower you are, the closer you are to the bottom. Where would you put yourself on the ladder? Please select the number below which corresponds to the rung where you think you stand. --- Supplementary File Sample Information

The author thanks Clare Longhorn for helping preparing the measure for this study and discussing the manuscript with me. My thanks also go to the participants at the Culture and Psychology Conference (Sept, 2016, University of Kent), and to Libby Drury for her comments and feedback on this work.

INTRODUCTION A history-of-ever catastrophic flooding occurred in Zhengzhou city, the capital of Henan province and its around areas, China on July 20, 2021, affecting 150 counties, which led to 302 deaths and 50 missing [1]. In addition to the loss of lives and properties, and physical injuries, the natural disasters brought in a huge impact on the mental health of individuals who survived through the disaster. When facing an unexpected disaster, individuals are susceptible to a series of adverse physical, emotional, cognitive behavioral reactions. These adverse responses are manifested by such as headache, insomnia, anxiety, nervousness, fear, sadness, depression, anger, irritability, inattention, memory loss and even life changes in beliefs and personality [2]. Post-traumatic stress disorders (PTSD) is a common mental health disorder that are associated with traumatic events. For example, approximately one third individuals had PTSD when they had motor cycle crashes or road traffic crash, and this disorder could persist over time [3,4]. A major earthquake disaster occurring in Sichuan several years ago led to a high percentage of orphan survivors with PTSD [5][6][7]. Preventing chronic psychological injury in the early post-trauma period is needed to alleviate or prevent the occurrence of PTSD. Accumulating evidence suggest that social support plays an important role in the prevention of post-traumatic stress disorders (PTSD). Social support may help alleviate job-related burnout in professional health workers [8,9]. A negative correlation was observed between the severity of PTSD in individuals living in the disaster area and the overall perception of social support [10]. Similarly, social support was reported to significantly reduce the severity of trauma symptom in patients suffering from rheumatoid arthritis and low back pain [11]. Several previous studies have shown that personality is associated with the severity of PTSD [12,13]. Individuals with low neuroticism had relatively weaker PTSD than those with high neuroticism when they suffered the same trauma event [12]. Marked impulsivity, aggression and a susceptibility to anti-sociality and substance abuse was observed in persons who had high negative emotionality and low constraint/inhibition when they experienced traumatic events [13]. The positive association between personality of neuroticism and PTSD is also validated, whereas a negative correlation between extroversion and PTSD exists, in the meta-analysis studies [14,15]. A higher incidence of negative emotions such as anxiety and depression frequently occur during the emergency of major public health issues. The higher the degree of anxiety or depression is, the greater the perception of stress. Natural disasters may add additional stress on persons particularly who are in the suffer, increasing their anxiety and depression. Flooding has been shown to significantly increase depression (adjust odds ratio (OR) = 7.77, 95% confidence interval (CI): 1.51-40.13) and anxiety (OR = 4.16, 95% CI: 1.18-14.70) in a previous study conducted in England [16]. In China, an intervention study showed that a systemic planned intervention, including planned settlements, intensive health service and environmental interventions, could significantly reduce depression, anxiety and flooding-associated PTSD [17]. The flooding disaster recently occurring in Henan Province has led to the huge loss of lives and properties in the local area. Attention is worthy of being paid to the mental health of the residents living in the disaster-affected area. Keeping calm emotion and offering social support to the population are important in helping the individuals to walk out of the disaster-induced psychological injuries. Yao et al. [8] has reported that professional health workers with different personality characteristics had significant performance and different outcomes when they faced job-related stress. However, it is yet to be determined how personality affects the mental health in the residents who experienced catastrophic flooding in Henan, China. The purpose of this study was to analyze the associations between residents' traumatic stress disorder, social support, mental health status, and personality type after experiencing the natural disasters. This study hypothesized that individuals with different personality had different psychological responses when they faced the natural flooding disaster, which occurred in Henan, China. --- MATERIAL AND METHODS --- Participants This study was approved by the Ethics Committee of Zhengzhou Normal University, Henan Province, China. During the period of July 30-August 4, 2021, a cross-section study was conducted in which a total of 572 residents living in Henan Province and experiencing the "July 20" devastating flood disaster were finally included, who completed the questionnaires. All the questionnaires were administered through an online survey platform named as "questionnaire star, " and informed consents were obtained from all participants online when they took part in the study. The information of the study with the link of the online questionnaire was distributed through social media (e.g., wechat). In the introduction part of the questionnaires, the demographic information include age, gender, marital status, flooding involvement, home address (e.g., zip code), properties and personal loss and so on. Of the total of 572 residents who completed the questionnaires (also known as valid questionnaires), 89 (15.6%) were men and 483 (84.4%) were women. The age of the participants ranged 16-62 years old, with an average age of 25.8 years. --- Methods --- Impact of Event Scale-Revised (IES-R) questionnaire To assess the psychological effect of the flooding disaster on the residents in Henan Province, an Impact of Event Scale-Revised (IES-R) questionnaire was used to obtain the information on residents' response to the traumatic stress. This questionnaire in Chinese version has a good reliability and validity in evaluating psychological stress in PTSD studies conducted in China [18]. The questionnaire is consisted of 3 dimensions of intrusion, avoidance and hyperarousal symptoms with a total of 22 items. Each item is scored using Likert 0-4 grade method, from 0 (never) to 4 (always). The total score for each individual is summed up of the score of each item, which ranges 0-88. A higher score indicates that the stress is more severe. According to the total score, the stress disorder was classified into 0-23 (normal), 24-32 (mild), 33-36 (moderate), and >37 (severe). The Cronbach's <unk> coefficient of the scale in this study was 0.962. --- Social Support questionnaire The Perceived Social Support Scale (PSSS) questionnaire was used to measure the social support that the participants could obtain. The PSSS is a scale measuring social support, which is assessed based on the individual's self-understanding and self-feeling. It measures the degree of social support from such as family, friends, and others [19]. There are 12 items for self-assessment in the questionnaire, and each item uses Likert 7-level scoring method. The sum score of each item was calculated for each individual. A higher sum score reflects that an individual feels more supports from the society. The Cronbach's <unk> coefficient of the scale in this study was 0.954. --- Mental health status In this study, a concise version of the Depression Anxiety Stress Scales (DASS-21) questionnaire was used to evaluate the mental health status of the participants. The questionnaire was compiled by Lovibond and Lovibond [20], which is composed of 21 items, measuring the 3 negative emotional dimensions of depression, anxiety, and stress. Among them, depression factors are related to pathological dysthymia, low self-esteem, and low-level positive emotions. Anxiety factors are related to the physical and subjective experience of anxiety arousal, and stress factors are related to negative emotions such as stress, worry, and conflict. The scale uses Likert 0-3 grade scoring method (0 -completely disagree, 1 -partially agree, 2 -mostly agree, 3 -completely agree). Scores were determined by summing the individual items and multiplying them by 2. A higher score indicates a stronger negative emotion experience. The Cronbach's <unk> coefficient of the scale in this study was 0.975. --- Eysenck Personality Questionnaire-Revised (EPQ-R) The personality type was determined using the version of the Eysenck Personality Questionnaire-Revised (EPQ-R) in Chinese, which is translated and revised by Qian Mingyi, a professor of psychology at Peking University, which has been shown with good reliability and validity [21]. Two dimensions, Introversion-extroversion (E) and Neuroticism-emotional stability (N), were used to assess the personality type. The Neuroticism/stability scale (EPQ-N) was for emotional stability when facing negative affect such as depression and anxiety, and the Introversion and extroversion vector scale (EPQ-E) for the demand of external stimulation. There are a total of 24 items each with a score of either 0 or 1. Based on the score cut offs of 43.3 and 56.7 for the E and N dimensions, respectively, 4 personality types were defined: stable introvert, stable extravert, unstable introvert, and unstable extravert. The Cronbach's <unk> coefficient of the EPQ-E and EPQ-N in this study was 0.818, 0.874. --- Statistical analysis All statistical analyses were performed using SPSS 18.0 software in this study. The numerical variables showed an approximate normal distribution after normality test (the absolute values of skew and kurtosis are both <unk>1), and M<unk>SD were used to represent its average and standard deviation. Pearson correlation analysis was performed for the correlation between traumatic stress disorder, social support, depression-anxiety-stress and personality characteristics. Machine learning of generalized linear algorithm (GLM) was performed using H2O.ai to analyze the importance of the factors in depression-anxiety-stress, respectively, in which the subjects were randomly (ratio = 0.70) divided into training and validation datasets. The GLM was fitted to estimate the set of parameters by maximizing the loglikelihood of the data for the best model. The ternary interaction of traumatic stress disorder, social support, and emotional stability on depression-anxiety-stress were also examined. The difference was considered statistically significant when a p <unk> 0.05 at the 2-sided <unk> level. --- RESULTS --- Common-method variation test The common-method variance of variables was first analyzed using Harman's single factor method. There were 14 factors with eigenvalues >1, and 28.19% of the variance was explained by the first factor. This result indicates that no severe common-method variance exists in this study based on the critical threshold of 40%. --- Distribution of traumatic stress disorder, social support and depression-anxiety-stress in the participants Table 1 shows the associations of the traumatic stress disorder, social support, and depression-anxiety-stress with several analyzed variables in the residents of Henan province. The individuals who live in the flooding areas had a significantly higher score of traumatic stress disorder than those who do not (p <unk> 0.05). Similarly, significantly higher scores of traumatic stress disorder and depression-anxiety-stress were pronounced in male than female participants (p <unk> 0.01). There was a significant association between marital status and the scores of traumatic stress disorder measured by IES-R and stress measured by DASS-21, with a higher level in married vs. single (p <unk> 0.01). The differences were significant in the scores of traumatic stress disorder, social support and depression-anxiety-stress among individuals with different personality types (p <unk> 0.001). Further pairwise comparisons showed that the score of traumatic stress disorder measured by IES-R was the highest for the individuals with residents with unstable extravert personality, followed by unstable introvert, stable introvert and stable extravert personality. Those with stable extravert personality had the highest scores of social support, followed by stable introvert, unstable extrovert and unstable introvert. The individuals with stable extravert personality had the lowest scores of depression-anxiety-stress, followed by stable introvert, unstable introvert and unstable extrovert personality. To further assess the importance of these factors in depression-anxiety-stress, the generalized linear model algorithm (GLM) was performed using machine learning in H2O (H2OFlow of H2O.ai), and the results are shown in the Figure 1. Among the risk factors of depression-anxiety-stress measured by DASS-21, the traumatic stress response measured by IES-R ranked the first, followed by EPQ-N, EPQ-E, gender, age, marital status, social support and flooding counties. The traumatic stress response, EPQ-N and age, were positively associated with depression-anxiety-stress, while EPQ-E, gender, marital status, social support and flooding counties were negatively associated with depression-anxiety-stress (Figure 1). --- Correlation between traumatic stress response, social support, neuroticism and depression-anxiety-stress Neuroticism refers that an individual has low emotional stability and strong reaction when facing stress. The Emotional stability (N) score was used in the Neuroticism/stability scale (EPQ-N) as the degree of neuroticism hereafter in the analyses. Pearson correlation analysis showed that traumatic stress response was significantly negatively correlated with social support (correlation coefficient -0.176, p <unk> 0.01), while it was significantly positively correlated with depression-anxiety-stress (correlation coefficient 0.694) and neuroticism (correlation coefficient 0.379) (p <unk> 0.01), respectively (Table 2). Social support was negatively correlated with depression-anxiety-stress (correlation coefficient -0.343). Neuroticism was significantly positively correlated with depression-anxiety-stress (correlation coefficient 0.522). Extroversion was significantly positively associated with social support (correlation coefficient 0.303), whereas it was significantly correlated with depression-anxiety-stress (correlation coefficient -0.110) and neuroticism (correlation coefficient -0.107). --- Interaction of traumatic stress disorder, social support and neuroticism in depression-anxiety-stress Table 3 shows the results of the main effect of 3 variables and their interactions. In model 1, gender, age, traumatic stress response, social support, neuroticism and extroversion were included. It showed that 56.8% of the total variance of depression-anxiety-stress scores was explained by 6 variables. Traumatic stress response and neuroticism had a significantly positive association with depression-anxiety-stress, while extroversion and women showed a significantly negative association. No significant associations were found for social support and age with depression-anxiety-stress. In the model 2, a 2-way interaction term of traumatic stress disorder, social support, extroversion and neuroticism was added beyond the model 1. The main effects of gender, traumatic stress disorder, neuroticism and extroversion variables remained significant. A significant interaction was observed between traumatic stress disorder with either social support (p <unk> 0.001) or neuroticism (p <unk> 0.001), or extroversion (p <unk> 0.001). However, there was no interaction between social support and either neuroticism (p > 0.05), or extro version (p > 0.05), and between neuroticism and extroversion (p > 0.05). In the model 3, a 3-variable interaction term was added to the model 2. Similarly, the main effects of all variables except age and the interaction between traumatic stress disorder and either neuroticism or extroversion remained significant. The 3-variable interaction term of traumatic stress disorder, social support and neuroticism but not traumatic stress disorder, social support and extroversion was also significant (p <unk> 0.001), explaining 1.4% variance of depression-anxiety-stress score. Traumatic stress disorder showed the largest positive main effect on depression-anxiety-stress (<unk> = 0.578). To more directly visualize the interaction of traumatic stress disorder, social support and neuroticism (also emotional stability) in depression, a 3-variable interaction diagram of depression was constructed based on the method described by Dawson and Richter. Figure 2 shows that as the traumatic stress response increases, the depression-anxiety-stress of the affected residents gets worse. Residents with the same level of traumatic stress disorder, high social support and low neuroticism had a lower depression-anxiety-stress score compared to those with low social support and high neuroticism. Compared with social support, neuroticism has a more obvious effect on depression-anxiety-stress when experiencing high traumatic stress; individuals with low neuroticism had a lower depression-anxiety-stress score than those with high neuroticism. --- DISCUSSION This study demonstrated the effect of history-of-ever catastrophic flooding in Henan province on depression-anxiety-stress in residents with different personality. The residents who suffered flooding attack had a significant post-traumatic stress disorder (PTSD) compared to those who did not. They had significantly higher depression-anxiety-stress than those who did not suffer the disaster. The findings in this study are in consistent with previous report that natural disasters can lead to a severe impact on people's psychology, mainly manifested as depression, stress, anxiety and fear [5]. A high prevalence of PTSD exists in many orphan survivors after a major earthquake disaster several years ago in Sichuan [5][6][7]. The risk of depression and anxiety significantly increase in residents who experienced flooding in UK [16]. Intensive health support could significantly reduce depression, anxiety and flooding-associated PTSD [17]. These findings suggest that psychological aids are needed to this population who suffered the disaster on time to resolve the PTSD. The reason is that if these short-term Model 2 -a 2-way interaction term of traumatic stress disorder, social support, extroversion and neuroticism was added beyond the model 1. Model 3 -a 3-variable interaction term was added to the model 2. psychological problems cannot be effectively resolved via all kinds of intervention programs on time, longterm mental illnesses will be caused [5]. There is a significant gender disparity in the traumatic stress disorder with a higher score in men than in women. This finding seems inconsistent with some other studies, which reported that women had higher lifetime prevalence of PTSD than men [22]. One possibility may be due to the traditional belief that men should take more responsibilities in the family and society, and they usually have greater psychological pressure. A previous study performed in China also demonstrated that men were more susceptible to psychological problems such as depression-anxiety-stress after a natural disaster [5]. A higher score of traumatic stress disorder and stress was also observed in married individuals compared to those singles in this study. Married individuals usually have more family responsibility, there may have both kids, parents and siblings to take care, more economic burden on shoulders [23], and will always worry about the safety and situation of family members during emergencies. In addition, there is a positive correlation between the traumatic stress disorder and depression-anxiety-stress. This finding is in agreement with the previous observation that the traumatic stress burden resulted in uncontrollable worries in individuals, and stress could directly trigger negative emotions such as anxiety and depression [24]. It has been reported that when facing emergency and stress, individuals with different personalities take different actions and have different responses. Those with stable emotion could calmly challenge the stress and emergency situations, displaying less fluctuations in their physiological response under stress, and showing strong self-adjustment ability [25]. Emotional unstable individuals are more likely to be impulsive and anxious than those with stable emotion [26], while extrovert individuals usually have more social networks, often participate in group activities, and get more social supports from friends and relatives. The individuals who are extroverted and stable emotion usually can keep calm in stress or emergency, and can deal with various difficulties more rationally. In contrast, those with instable emotion may have anxiety and stressful feeling under these situations. Individuals with neuroticism or unstable emotion have higher PTSD compared to those with extrovert and stable emotion [15]. It can be found that the interactions existing between traumatic stress disorder, social support and neuroticism (emotional stability) in depression-anxiety-stress when they faced the natural disaster. When the individuals faced the flooding-caused damage and loss, those with high social support and stable emotion had the best mental health status. Regardless of social support, those with stable emotions had better mental health than those with unstable emotion particularly at high PTSD. These findings suggest that emotion management and psychological aids may be an important measurement in prevention of PTSD and disaster-related depression-anxiety-stress or mental health problems. To a certain extent, the amount of social support depends on social networks an individual owns. Personality usually can affect the plenty of social networks; those with extrovert and stable emotion may have more social networks and may actively seek help from friends and relatives, whereas those with introvert and unstable emotion may be shameful to seek help from others or may have difficulties in making friends. Thus, when they faced the difficulties or the disaster-induced stress, individuals with different personalities had different outcomes in mental health. The findings in the interaction between traumatic stress disorder, social support and personality in mental health are in agreement with the previous observations that extrovert individuals with stable emotion often adopt active coping styles to solve problem and seek help when they encounter difficulties and emergencies, and they receive higher social support. Social support and friendship help play an important role in maintaining the mental health [27]. Similarly, the prevalence of depression (suicide thoughts and behaviors) increased in the building-collapse survivors when they felt desperate with less social supports [28]. Another nightclub fire disaster study also showed a positive association between negative emotion personality and PTSD [29]. Thus, social support to those who suffer natural disasters on time is an effective measurement to prevent PTSD, and improve their mental health, making them feel warm from society, relatives and friends, and fill their heart with full hope of live, they are not alone. Interestingly, in this study, there is a significant interaction existing between traumatic stress disease (PTSD), social support and neuroticism but not between PTSD, social support and extroversion in depression-anxiety-stress. One possibility is the most likely when people face such a large scale natural disaster, it is a challenge for everybody no matter how extrovert they are. However, the disaster-causing stress may deteriorate the mental health problem for those neuroticism. Thus, neuroticism individuals should be granted with high priority more social support to prevent mental problem. In this study, 2 scales of IES-R and DASS-21 were applied to detect psychological disorder for PTSD and depression-anxiety-stress, respectively. Both IES-R and DASS-21 have been shown suitable for use in clinical and compensation setting to detect PTSD and major depression disorder in a motor vehicle crashes study, demonstrating both scales with acceptable sensitivity and specificity in an injured individuals engaged in compensation [3]. However, 2 scales were applied in this study based on their applications initially designed, IES-R specifically for PTSD and DASS-21 for depression-anxiety-stress, respectively. The IES-R has 3 subscale measurements: Intrusion (i.e., intrusive thoughts, nightmares, intrusive feelings and imagery), Avoidance (e.g., avoidance of feelings, situations and ideas, responsiveness numbing) and Hyperarousal (e.g., anger, irritability, hypervigilance) [30]. This scale was used for traumatic stress disorder measurement after the flooding. The DASS-21 assesses 3 dimensions: Depression (e.g., hopelessness, lack of interest or involvement, inertia), Anxiety (e.g., autonomic arousal, anxious), and Stress (e.g., irritable, upset, over-reaction, nervous), and this scale was used to measure the comprehensive depression-anxiety-stress (not only from the flooding, but also due to other factors). Interestingly, there is a significant interaction existing between neuroticism, social support and traumatic stress disorder, suggesting that personality could modify the effect of PTSD and social support on depression-anxiety-stress in the residents living in the flooding area in Henan, China. Some limitations exist in this study. First, the crosssection study design was applied for the survey, and it is a time point snapshot for all variables. Thus, it could not be infered a causal relationship except the association, which may be accompanied only. A longitudinal or intervention study may help to address the causal inference. Secondly, the participants voluntarily signed in to fill the online questionnaires, the selection of subjects may have a bias, e.g, women are far more than men in the number (or proportion), and the samples may not well represent the population. Moreover, although 2 different psychological stress questionnaires (IES-R vs. DASS-21) were used, and could not be clearly dissected each other. However, one of the strengths of this study is that the sample size is relatively large. In addition, this study was conducted immediately after the flooding, and it may limit some recall bias and different intervention measurement effects in the survey. --- CONCLUSIONS In summary, this study demonstrated the effect of traumatic stress disorder, social support and personality on their interactions in the mental health of the residents in Henan province, where recently suffered a history-of-ever flooding. It showed that personality modified the effect of traumatic stress disorder and social support on mental health. Individuals who have stable emotion had a better mental health compared to those who have unstable emotion even when they are at the same level of traumatic stress disorder. These findings suggest that personality consideration may be helpful in improvement of mental health after the disaster when psychological aids and social support reach out. However, further follow-up studies are warranted to investigate the effect of different strategies or measurements on the prevention of post-traumatic stress disorder.

To analyze the impact of the flood disasters, social support and personality on the mental health of residents in Henan Province, China, providing fundamental knowledges for making measuring strategies to improve the psychological protection and anti-stress ability of the residents after the disaster. Material and Methods: A cross-section study was conducted via an online survey platform "questionnaire star, " which included 572 residents in Henan Province, which underwent the history of ever flood disaster on July 20. The questionnaires of Impact of Event Scale-Revised Edition (IES-R), Perceived Social Support Scale (PSSS), the Depression Anxiety Stress Scales (DASS-21) and the scales of Eysenck Personality Questionnaire-Revised (EPQ-R) version in Chinese were also administered to each participant. Generalized linear regression model was performed. Results: The residents who live in the flooding areas, are male and married had a significantly higher post-traumatic stress disorder (PTSD) score than their counterparts. The scores of depression-anxiety-stress in the residents with stable emotion were significantly lower than those with unstable emotion (p < 0.001). Machine learning showed that PTSD ranked the top risk factor, followed by neuroticism for Depression-Anxiety-Stress after disaster. The PTSD was negatively correlated with social support (p < 0.01), while it was positively correlated with depression-anxiety-stress and emotional stability (p < 0.01). There was a statistically significant interaction between PTSD, social support and neuroticism on depression-anxiety-stress (p < 0.001), with an independent effect of 1.4% on depression-anxiety-stress. Emotional stability showed the largest association with depression-anxiety-stress. Conclusions: Residents living in the catastrophic flooding areas had significant post-traumatic mental health issues, and the severity of mental problems was differently affected by post-traumatic stress disorder and social support in individuals with different personalities. Introvert and PTSD were the major risk factors for depression-anxiety-stress after the disaster.

Introduction After initiating a couple of decades ago, the percentage of transnational marriages reached the highest percentage at 30.7% in 2005. Although there has been a decrease, it is still at about 10% of the total number of marriages in South Korea, where the current intervention study was conducted [1]. Recently, international marriage had increased, in particular, Korean males who have difficulties in finding their spouses domestically tend to marry foreign females mostly from Southeast Asia, including Vietnam, Cambodia, and Uzbekistan as developing countries [2]. Korea traditionally had homogeneous culture and race [3]. Compared to other countries, such as USA, Canada, and Australia, ethnic diversity in South Korea has not existed long enough to accommodate other cultures easily [4]. Thus, there might be difficulties in accepting other culture among Korean multicultural families, whose family members have different nationalities. A study of Korean mothers-in-law showed that only 11% of them accepted their foreign daughters-in-law's culture [5]. Taking into account trends of international marriage in Korea [2], multicultural families in the current study refers to as families with Korean husbands and foreign spouses, and mothers-in-law in multicultural families meant Korean mothers on the husbands' side. Different from USA, Australia, and European countries, the married couple, a Korean husband and a foreign wife, are more likely to live in together with the Korean male's parents. Multicultural families with Korean males and foreign females are five times more likely to live with mothers-in-law at 75% than domestic families at 14.1% [6,7]. Similarly, in the South Asian cultural context, if a mother becomes a widow, she is likely to cohabit with their married son [8]. As Korean culture is different from Southeast Asian culture [9], mothers-in-law of foreign female partners in multicultural families tend to take the role of introducing Korean culture and courtesies including table manners [7]. Specifically, although both Korea and Vietnam, one of the most common countries foreign daughters-in-law are from, have a Confucian background, there are differences in language, cuisine, and culture, including uncertainty avoidance and long-term orientation [10]. For example, Koreans seem to be rigid and intolerant of unwonted or unexpected behaviours or opinions. On the other hand, Vietnamese tend to be more laid-back and flexible towards rules [11]. As seen in the previous literature [12], the elderly mothers-in-law experienced stress due to a burden of supporting their son's family and the differences in culture from Vietnamese daughters-in-law. Meanwhile, mothers-in-law might have psychological stress due to communication difficulties at more than 30% and cultural differences at slightly under 15% [13]. Moreover, they might be concerned that their foreign daughters-in-law could be running away from their sons and the couple's grandchildren be left behind, and afraid of blame from their son or others for conflicts with their daughtersin-law [14]. Thus, mothers-in-law of multicultural families may be perceived as three times more stressful than others (F = 18.84, p <unk>.001) [7]. Exposure to negative stress and failure of coping strategies can lead to depression, loss of meaning of life, demotivation, and impairment of quality of life (QoL) [15]. Compared to 20% of older adults who generally experience depression [16], mothers-in-law reported depression up to at 84%, including 75% with a high level of depression [5,6]. Accordingly, they perceive their QoL impaired [14]. According to Lazarus [15], stressed out individuals tend to perform cognitive appraisal consciously or unconsciously in order to minimise physical and/or mental harm by stress. If an individual perceives stress as a threat or realises their absence of stress coping strategies, the level of stress can be more negatively appraised than the actual stress level. A negative appraisal can lead to physiological changes, such as an increase of blood pressure and heart rate acceleration, and can interfere with adaptation to stress. Repeated negative self-assessment can even cause physical illness, anxiety, depression, impaired QoL, and social dysfunction. On the other hand, a resilient response to stress can lead to flexible coping strategies in various situations [17,18]. Resilient cognitive appraisal can improve understanding of actions and emotions under stressful situations, acceptability of others, acceptance of diversity, amicable relationships with others, belief and acceptance of life, and optimism for the future [19][20][21]. Furthermore, it can lead to decreases of cholesterol, anxiety, stress, depression, and increases of belief and acceptance of life and satisfaction [22][23][24][25]. Resilience programs help in building psychological flexibility and relaxation [22] and in enhancing the individual's inner strength so as to adapt to stressors [26]. In most of the previous literature, resilience programs aimed at promoting resilience to negative stress for children [27], college students [20,24], employees [22], and patients [25,26]. A resilience program for oncology-haematology nurses in the previous study found to be ineffective for improving their stress management and QoL in the short-term and long-term points of view [28]. However, a resilience program for farmers showed significant effects on their QoL [29]. Although resilience is manifested throughout the individual's whole life, including older adult life, and is capable of being learned [30], there is scant literature about subjects in old age, particularly in South Korea. In addition, there is a need for developing interventions for cultural adaptation stress, depression, and QoL in Korean mothers-in-law of multicultural families, which have been rarely developed and applied. While foreign daughters-in-law could be well-supported by programs or activities from multicultural family centres, improving resilience of mothers-inlaw, who have been accustomed to rigid homogenous culture, would be helpful for accepting the foreign culture and different lifestyle of their daughters-in-law in the globalised era. Therefore, considering the need from the previous literature, this study aimed to develop and implement the cultural adaptation promotion resilience program (CAPRP) for mothers-in-law in multicultural families, and to verify its effects on resilience, acculturation stress, depression and QoL among mothers-in-law in multicultural families, comparing with the control group. --- Methods --- Study design A non-equivalent control group pre-posttest design was adopted for evaluating effects of the cultural adaptation promotion resilience program (CAPRP) for mothers-in-law in multicultural families. Both groups completed pre-and post-test questionnaires, but the intervention group participated in the CAPRP. After completion of the study, the control group was provided an alternative program, only one three-hour session as needed. The study design is presented in Fig 1. --- Study subjects Participants were recruited from multicultural family support centres in the metropolitan city A. A list of nine multicultural centres in the metropolitan city A was found and two centres accepted to participate in the study after the first author either phoned or visited the centres to explain the study's process. The intervention group was randomly chosen from either C or D centres by picking out one piece of paper from an opaque box. Twenty-four participants from the C centre who consented and were enrolled to the study considered to be in the intervention group, while 25 from the D centre to be in the control group. Inclusion criteria were participants who were 1) Korean mothers-in-law, living with her foreign daughter-in-law more than six months; 2) providing consent to participate in this study; 3) without mental or cognitive impairment; 4) able to understand and answer questions in Korean; and 5) without any hearing impairment. The sample size of the study was calculated in consultation with a health statistician based on the previous studies [31][32][33]. The G Power 3.1.2 program was used for calculation at the two independent group t-test at a power of 0.80, a two-tailed significant level of.05, and the effect size of 0.80 based on the previous research in resilience programs for families of patients with chronic schizophrenia [33], for family caregivers of the elderly with dementia [31], and for trauma-exposed refugees [34]. A sample size of 21 participants in each group was required. Taking into account a dropout rate of 10%, a total of 50 eligible participants were invited in the intervention group and the control group. Among 25 invitees in each group, 21 participants in the intervention group enrolled for the program after three dropped out due to no show at the first day of the program. Of the 25 invitees, 22 participants in the control group agreed to take part in this study. At the follow-up, 21 participants from each group completed the entire study after one in the control group dropped out. --- Instruments The questionnaires of the study consisted of socio-demographic characteristics [5], resilience [35], acculturation stress [36], depression [37], and QoL [38]. Appropriate permissions from developers of each instrument were received. Socio-demographic characteristics. The socio-demographic characteristics of participants in this study included age, marital status, education level, and religion. Their perceived economic level, health state, and satisfaction of family support was answered on a five-point Likert scale. Information of whether they have any disease was also asked as well as disability due to disease on a three-point Likert scale. Questions about their daughters-in-law included the country which their daughters-in-law were from, duration of living with them, perceived daughters-in-law's Korean level on a five-point Likert scale, satisfaction with the relationships with daughters-in-law on a five-point Likert scale, and ethnocentrism, asking if their daughters-in-law should fully adapt Korean culture regardless of their original culture, on a threepoint Likert scale. Resilience. The Resilience Scale-14 (RS-14), developed by Wagnild & Young [35], was used for assessing resilience of mothers-in-law in multicultural families. This instrument consisted of 14 items, which can be divided into two dimensions-personal competence (10 items) and acceptance of self and life (4 items). Each item is scored on a seven-point Likert scale, ranging from 1 (strongly disagree) to 7 (strongly agree). A range of the total score is between 14 and 98, higher scores indicating higher resilience. Subcategories can be referred to as very low (14-56), low (57-64), moderately low (65-73), moderately high (74-81), high (82-90), very high (91-98). The Cronbach's <unk> was.93 when developed, and.91 for the current study. Acculturation stress. Acculturation stress of mothers-in-law in multicultural families was measured with a scale developed by Chung& Park [36]. This instrument includes 11 items, which can be divided into three domains-'perception of prejudice and discrimination', 'cultural conflicts', 'discomfort with daily life'. Each item is scored on a five-point Likert scale, ranging from'strongly disagree' (1 point) to'strongly agree' (5 points). A range of the total score is between 5 and 55, higher scores indicating higher acculturation stress. The Cronbach's <unk> was.86 when developed, and.94 for the current study. Depression. Depression of mothers-in-law in multicultural families was measured by the 30-item Geriatric Depression Scale Short Form-Korean Version (GDSSF-K), which Kee [37] standardised and modified for Korean older adults. This instrument consists of 10 positive items and five negative items, which are answered by yes, scored 0, or no, scored 1. A range of the total score is between 0 and 15, higher scores indicating higher depression. The Cronbach's <unk> was.88 when developed, and.79 for the current study. Quality of life. The Korean Version of WHO Quality of Life Scale Abbreviated Version (K-WHOQOL-BREF) by Min et al. [38] was used to measure QoL of mothers-in-law in multicultural families. This instrument contains 24 items, which can be grouped into physical health (7items), psychological (6 items), social relationship (3 items), and environmental (8 items) domains, as well as two additional items about general QoL and general health perception. Each item is answered by a five-point Likert scale from 1 (strongly disagree) to 5 (strongly agree). The range of the total score is between 26 and 130, higher scores indicating higher QoL. Min et al. [38] reported Cronbach's <unk> as.89, and the value of Cronbach's <unk> was.94 for the current study. --- The cultural adaptation promotion resilience program (CAPRP) Preparation for conducting the program. The CAPRP program was developed by the first author for mothers-in-law in multicultural families in this study and provided by the first author and an assistant researcher after appropriate training. For preparing for implementation of the program, the first author had completed several programs of art counselling psychology, positive psychology, resilience coaching, narratives, motivation counselling, and built experiences in developing and conducting a strength-focused resilience promotion program for older women. In addition, the first author tried to expand her expertise through attending various conferences and workshops related to psychology. Two assistant researchers were prepared for data collection by the first author for consistency. Development process and implementation of the program. Goals, components, contents, activities, and the number of sessions of the program were formulated after relevant literature review [30,31,39,40]. Professionals majored in geriatric nursing, psychiatric nursing, and psychiatry as well as doctors of positive psychology and resilience in older adults amended strategies, then approved validity. Previous literature revealed that major stressors of mothers-in-law in multicultural families were differences in culture and lifestyle from their foreign daughters-in-law, and that they have somatic symptoms, severe depression, impaired QoL, enhanced by passive and negative coping strategies. The final goals of the program thus are 'improvement in cultural adaptation resilience', 'a decrease of cultural adaptation stress', 'a decrease of depression', and 'an increase of QoL'. Based on the seven components of resilience, a goal for each session was'self-opening', 'causal analysis', 'impulse control', 'emotion regulation', 'empathy','sense of optimism', and 'utilisation of strengths'. The contents and other details of the program are presented on Table 1. The CAPRP in this study was implemented twice per week for four weeks, 60 minutes per session for a total of seven sessions. The program was provided to 21 participants in the intervention group by the first author of the present study with the help of a pre-trained assistant. An education room in the multicultural centres was selected, considering the participants might be vulnerable, so that the participants was able to feel comfortable and free to share their thoughts. The venue was equipped with internet and a beam projector, used for presenting various examples of multiculturalism. --- Data collection process Data were collected from January 04, 2016 to March 04, 2016 by two trained research assistants to avoid researcher bias. On the first day of the program, three of the enrolled participants in the intervention group were no-shows, and 21 participants completed the baseline questionnaires and took part in the program. In the control group, 22 participants attended to answer the baseline questionnaires. One participant in the control group dropped out at the followup. A total of 42 participants were included in and completed the study. A week after completion of the program, both participants in the intervention group and the control group were asked to answer follow-up questionnaires. --- Ethical considerations An ethics approval from the Institutional Review Board of the related university was obtained (2015_88_HR). Two research assistants explained the purposes of this study, its procedure, voluntary participation, confidentiality and privacy, then the written informed consent was obtained from all participants. Incentives for participation were provided with the equivalent of transport fares. --- Data analysis The collected data were analysed using the SPSS/WIN 23.0 program. The homogeneity of the two groups was analysed with descriptive statistics, chi-square test, Fisher's exact test, and ttest. The effects of the program were examined by t-test after a normality test of the dependent variable and the effect size of the CAPRP was calculated and presented with Cohen's d (95% --- Results --- Participant characteristics and homogeneity Table 2 presents the homogeneity of general characteristics for two groups, which were not significantly different. Age range of the participants in this study was from 62 to 79 years old, with the mean age of 68.48<unk>4.09 in the intervention group and 70.10<unk>5.06 in the control group. Among the participants, 47.6% of the intervention group and 61.9% of the control group were satisfied with the relationship with their daughter-in-law. A majority of participants in both groups-90.5% of the intervention group and 80.9% of the control groupresponded that their daughters-in-law should follow the Korean culture. Looking at Table 3, there were no significant differences in scores of baseline questionnaires about resilience, acculturation stress, depression and QoL between the two groups. --- Effects of the CAPRP The intervention group showed a significant decrease in acculturation stress (t = -3.63, p =.002), and depression (t = -3.05, p =.006), while resilience (t = 4.04, p <unk>.001) and QoL (t = 4.95, p <unk>.001) significantly increased compared to the control group. Table 4 presents differences in resilience, stress, and QoL of the intervention group and the control group. --- Discussion The current study assessed the effects of the seven-session CAPRP on mothers-in-law in multicultural families. The main findings of the current study showed that the CAPRP was effective on improving their resilience and QoL, whereas acculturation stress and depression was diminished. It was difficult to compare the results of this study to the previous research because of the scarce research regarding resilience programs for older females, including mothers-in-law. Conversion scores were used in the discussion, due to differences of instruments in the previous literature. The participants in the current study (mean age: 70 years old) scored resilience (converted to seven-point scale) as 4.4. This was similar to the study conducted by Kim & Na [41]. The subjects in this study were community dwellers with schizophrenia who have judgement and perception of reality, and reported with the mean score 4.3. On the other hand, patients with neurofibromatosis (mean age: 57 years old) [25], patients with breast cancer (mean age: 60 years old) [26], and Indonesian elders who experienced tsunami (mean age: 66 years old) [19], were more likely to score higher at 4.9, 5.2, and 6.1, respectively, than participants in the current study. The subjects in the studies by Vranceanu et al. [25] and Loprinzi et al. [26] were capable of utilising information about their illness, whereas the subjects in the study by Kim & Na [41] showed a difficulty in perception under a stressful situation. Seemingly, the participants in this study had lower resilience than those in previous studies because they lacked perception of and information about multiculturalism in a single culture society. In addition, not only had they difficulties in daily communication, but also insufficient support and coping resources. The 5-point converted score of acculturation stress was 2.9 among mothers-in-law in multicultural families in this study. This score was similar to the score 2.8 of Chinese international students in South Korea [42], while it was slightly lower than the score 3.1 of female marriage migrants [43]. The experiences of the participants in the current study were close to those in the previous studies. The Chinese international students expressed such difficulties as 'feeling alone', 'being sick', and 'not knowing what to do' [42]. The marriage migrants showed emotional restraint due to a different cultural point of view [43]. It can be assumed that they might be more stressed out than the international students or the mothers-in-law, requiring multiple roles as a wife, a daughter-in-law, and a mother, as well as acculturation. As a result of the CAPRP, the 5-point converted score of acculturation stress decreased by 2.3 among the mothers-in-law in multicultural families who used to restrain their emotions in a patriarchal society. Participating in the program, they learned how to express their emotions in a positive way throughout the activities such as drawing a person that everybody knows, drawing a mandala, and tearing a piece of paper. It was considered that the time when they appreciated each other's drawings might help them understand various points of view. Park & Park [42] applied cognitive art therapy, including drawing and making a story board, to the Chinese international students who had difficulties in verbal communication, resulting in a decrease by 2.4 of acculturation stress. Yang & Lee [43] applied collage group art therapy to the migrant women to help them in recognising their emotions and finding their strengths, resulting in a decrease by 2.0 of acculturation stress. The programs in these previous studies were similar to that in the current study, where participants who had difficulties in verbal expression learned to convey their emotions by drawing activities and spontaneously understood others without blame. The 15-point converted score of depression was 5.3 (mild depression), showing a decrease to 3.0 after the CAPRP among mothers-in-law in multicultural families in this study. This score was higher than older adults living at home, who scored 1.8 [44]. On the contrary, the score in the current study was lower than patients with neurofibromatosis, scoring 8.4 and 7.6 after the program [25], and with older adults from older citizen's public centres, scoring 9.7 and 9.0 after the related program [45]. Compared to previous studies, the change in depression in the current study was substantial. It might be due to the fact that the CAPRP enhanced selfperception with causal analysis and acceptability of different cultures, provided enough time for sharing their experiences with the others who were placed in a similar situation. Vranceanu et al. [25] included a problem-based approach for understanding the illness in part of their program; however, the program mainly contained emotional coping strategies, such as meditation and imagination. Lee & Park [45] included in their program notions, causes, and 11 coping strategies for stress, depression, and somatisation. The program ran for five weeks for an hour each week, which might be insufficient for getting acquainted with the information. After the CAPRP, the 5-point converted score of QoL increased from 2.9 to 3.7 among the mothers-in-law in multicultural families. The results of the current study were similar to those of a study comparing a positive psychotherapy program with a cognitive behaviour therapy program among depressed older adults [46]. The participants in this study showed a 3.0 level of QoL at baseline in both groups. The character strengths-based positive psychotherapy improved QoL to 3.6, while the cognitive behaviour therapy only increased QoL to 3.1 [46]. Similar to the previous studies [23,46], the program in this study applied coping strategies, such as developing optimism, building positive relationships, and recognising strengths, after identifying negative emotions under stressful circumstances. In addition to this feature, the program also attempted to change circumstantial factors by exposing the participants to positive experiences or positive emotions right after negative experiences, prior to dealing with the culture of their daughter-in-law. As a result of these efforts, the CAPRP was effective in improving participants' resilience and QoL despite the relatively short research period and the participants' lower level of demographic characteristics than that of previous studies. There are several limitations to the present study as follows: Firstly, the mothers-in-law in multicultural families, who participated in the current study, were unable to take part in this study for a longer period because they had to spend much time with house chores and rearing their grandchildren. It is suggested to examine the long-term effectiveness of the program in the future studies. Secondly, comparison with previous studies was limited because there were no previous resilience programs for mothers-in-law in multicultural families. Further studies, applying the CAPRP program with larger sample sizes would be recommended to examine effects of the intervention. Thirdly, the control group did not receive any treatment, so it can be recommended to compare the CAPRP program with other treatments or regular programs for the seniors. In addition, the effect of a program may decrease over time [47], thus a regular cognitive behaviour training program would be suggested for enhancing or maintaining their resilience. Fourthly, we only included mothers-in-law in the program and did not include much information about daughters-in-law's characteristics, such as age, children in the current marriage, and rural/urban residence before immigration to Korea. It is recommended to develop an integrated program for promoting cultural adaptation of both mothers and daughters-in-law in multicultural families together, taking into account daughters-in-law's individual characteristics. Lastly, the nationalities of the daughters-in-law of the intervention group were Vietnam at 90%, whereas those of them of the control group were at 69% due to random allocation. Considering the results of the present study, a cautious interpretation is needed. Future studies on effects of the program for different mothers-in-law, who have daughters-inlaw from various countries, would be recommended. In spite of these limitations, the current study has a few strengths. To our knowledge, this CAPRP is the first program for cultural adaptation of older females, that is, mothers-in-law. Previous programs for the elderly were mostly focusing on their depressive symptoms [32]. Recently, a leisure program [48] or a mourning program using art therapy was applied [49]. On the other hand, the CAPRP could enhance participants' psychological strength comprehensively based on the Lazarus, compared to other programs dealing with only limited aspects. Considering that the communication problem was one of the most important issues for them, the program contained sessions for improving communication skills. It also helped to build on their empathy with their daughters-in-law from different culture by watching videos about multicultural families, particularly about foreign female migrants, and writing cards to their daughters-in-law. Based on a thorough literature review of previous domestic and international studies, the components of resilience among mothers-in-law in multicultural families were identified as self-opening, causal analysis, impulse control, emotion regulation, empathy, sense of optimism, and utilising of strengths. This finding can be applied when developing resilience-related education programs for those with similar characteristics. Additionally, the CAPRP was effective for decreasing stress and depression as well as increasing QoL, thus it can be utilised for intervening in situations of stress, depression, and QoL of the older members of the community, given the fact that the numbers of the older population and migrants are expected to continuously increase until 2040 [50]. The program mostly adopted visual aids as educational materials in order to minimise the limitation of literacy, compared to the previous resilience promotion programs for people without cognitive impairment and illiteracy. Further, the CAPRP could be applied to older females in other countries, whose culture might be as rigid as Korea, to help them become accustomed to different heritage, values, or lifestyle. The program may also be applied to where there is a huge generation gap in any society. --- All relevant data are within the manuscript. --- Author Contributions Conceptualization: Sang-Hwa Lee, Dong-Hee Kim. Data curation: Sang-Hwa Lee, Dong-Hee Kim, Kyoungrim Kang. Formal analysis: Sang-Hwa Lee, Dong-Hee Kim, Kyoungrim Kang. Investigation: Sang-Hwa Lee, Dong-Hee Kim. Methodology: Sang-Hwa Lee, Dong-Hee Kim. Project administration: Sang-Hwa Lee, Dong-Hee Kim. Resources: Sang-Hwa Lee, Dong-Hee Kim. Software: Sang-Hwa Lee, Dong-Hee Kim, Kyoungrim Kang. Supervision: Dong-Hee Kim. Validation: Sang-Hwa Lee, Dong-Hee Kim. Visualization: Sang-Hwa Lee, Dong-Hee Kim, Kyoungrim Kang. Writing -original draft: Sang-Hwa Lee, Dong-Hee Kim, Kyoungrim Kang. --- Writing -review & editing: Sang-Hwa Lee, Dong-Hee Kim, Kyoungrim Kang.

Mothers-in-law in multicultural families tend to experience psychological burden. This study aimed to verify the effects of the cultural adaptation promotion resilience program (CAPRP) on resilience, acculturation stress, depression, and quality of life among mothers-in-law in multi-cultural families. Forty-two participants from multicultural family support centres in the metropolitan city A were assigned to either the intervention group or the control group. The CAPRP was performed for 60 minutes, twice a week for four weeks. The intervention group showed a significant decrease in the acculturation stress (p = .002), and depression (p = .006), while resilience (p < .001) and quality of life (p < .001) significantly increased compared to the control group. The intervention group reported significant improvements in resilience, acculturation stress, depression, and quality of life in comparison with the control group. The results indicated that the CAPRP, developed based on positive cognitive appraisal, was an efficient nursing intervention for mothers-in-law in multicultural families.

Background A clinical study is an investigation that uses human subjects to contribute to knowledge that can be applied to benefit society. Clinical studies typically evaluate an intervention that is applied to study subjects. The intervention might be a drug, vaccine or therapeutic or surgical procedure [1]. Clinical studies are indispensable for the progress of medicine, especially for the discovery of new pharmaceuticals [2]. However, the use of human subjects in these studies as a unit of analysis introduces certain challenges [3]. In the past, study participants have suffered from the experiments that they were subjected to. One example is the experiments that were performed on prisoners during World War II [4], which generated a series of tragedies that are still remembered today. Those experiments violated the ethical rules that are currently applied to clinical studies [5,6]. Past violations have contributed to the development of mechanisms to protect human subjects [7]. The scientific community was made aware of these violations, and various regulations have been elaborated to protect human rights and the integrity and dignity of individuals participating in biomedical research [8]. However, the unethical studies of the past have left a strong impression and are still referenced in developing countries. In Brazil, the general perception is that the risks to study subjects are greater than the benefits, and the term "human guinea pig" is commonly used to describe clinical study participants [9]. The basic principles of bioethics (autonomy, beneficence, non-maleficence and justice) [10] may be violated in clinical studies on human subjects in such developing countries as Brazil and may be a source of bias within the study [11]. The characteristics of study populations from these countries, such as poverty, illiteracy, lack of education and lack of availability/access to health care, may create inherent bias. In addition, several authors warn that populations in developing countries require special protection, due to these factors [12]. These study subjects may not understand the study process or may participate in clinical studies because they do not have access to medical care. Other authors have noted that these study participants are more susceptible to coercion and may more easily become volunteers in clinical studies [13]. Fortunately, it is clear that there has been increased emphasis on bioethics in Brazil in relationship to clinical studies of human subjects. The Clinical Study Regulation was created under Resolution 196/96 of the National Health Council (Conselho Nacional de Sa<unk>de-CNS) of the Health Ministry. As a result, the ethical evaluation process that a clinical study must pass is sufficiently rigorous and is in accordance with national laws [14]. In addition to the bioethics evaluation performed by the Ethics in Research Committee (Comitê de Ética em Pesquisa -CEP) of the study institution, international studies must also receive consent from the National Commission of Ethics in Research of the National Health Council (Comiss<unk>o Nacional de Ética em Pesquisa do Conselho Nacional de Sa<unk>de -CONEP) and the National Sanitary Monitoring Agency (Agência Nacional de Vigilância Sanitária -ANVISA) [5]. All subjects receive an Informed Consent Statement (ICS). This document explains in accessible language the research details to ensure that that the subject understands the procedures, risks, discomforts, benefits and rights involved and makes an autonomous decision [15,16]. Despite the rigor of these interventions, the idea persists that study subjects in Brazil participate in clinical studies for the wrong reasons [9]. The objective of the present study is to identify the specific motivators that inspire Brazilian volunteers to participate in clinical studies and to describe the demographic profile of these study subjects. In addition, we will evaluate whether the motivations are in accordance with established ethical and legal principles. --- Methods --- Rationale for the choice of methodology A multi-methods study [17] with qualitative and quantitative phases was used. Qualitative methods were used to identify the motivations for study participation, focusing on interviewees perception and behaviors [18]. Quantitative analyzes were used to compare the motivators for research participation between groups and the variables that are associated with this motivation. --- Recruitment of the participants A convenience sample was used. Three clinical research centers that conduct Phase I (bioequivalence) and Phase III (therapeutic) studies provided the records of subject participants. One center was in S<unk>o Paulo, another was in Campinas and the third was in Goiania. In Brazil, there are few Phase I studies. Therefore, we considered bioequivalence studies, which use healthy volunteers to compare the bioavailability of a generic medicine to a reference medicine, to be Phase I studies. The participants were selected from 10 lists with a total of 500 volunteers each and were contacted by telephone for a personal interview in the Federal University of S<unk>o Paulo (Universidade Federal de S<unk>o Paulo, UNIFESP). These lists were composed of volunteers who participated in previous studies when they left permission to be contacted for future studies. They included the names of participants of studies conducted by the centers for the two years previously. The participation rate was 89%. In total, 90 participants were contacted (30 from each center), and 80 agreed to participate. The 10 refusals to participate included 4 men and 6 women, from the 3 centers. --- Procedure for translating the data The interviews were conducted in Portuguese being recorded and later on transcribed and analyzed. The final result was translated into English by AJE (American Journal Experts) a site of experts from the scientific community with expertise in the two languages (English and Portuguese). The quotes were back translated in order to ensure the fidelity of the interviewees' statements. --- Sample size The sample was composed of 80 volunteers who had participated in at least one Phase I or Phase III clinical study. The sample size was limited by the qualitative indepth interview. Furthermore, this pilot study was not intended to be representative of the entire population of study participants. Rather, this study was designed to identify the main motivations that lead to clinical study participation and whether these motivators differ by research phase. to share knowledge about the study. The KI introduced the theme of the research, noting the peculiarities of the participants [19]. A total of ten KIs were interviewed. Three were principle investigators from the study centers, two were clinical study center coordinators, one was a representative of the Brazilian Society of Clinical Study Professionals (Sociedade Brasileira de Profissionais de Pesquisa Clinica-SBPPC), one was a representative of the Brazilian Association of Clinical Study Representative Organizations (Associaç<unk>o Brasileira de Organizaç<unk>es Representativas de Pesquisa Cl<unk>nica-CRO), one was a regional director of clinical operations of the pharmaceutical industry and two were doctors responsible for conducting the studies in public and private reference hospitals. The interviews with the KIs were unstructured, conversational interviews [20,21]. They were recorded, transcribed, analyzed and provided information that could be used to develop interview questions for the clinical study subjects and to aid in understanding the discussions of the interviewed individuals [19]. These interviews allowed for comparative analysis of the motivations for participation in research studies according to subjects and KIs. --- Data collection instrument The categories that emerged from the discussion with the KIs were used to describe the study subject volunteer profile and topics in the survey script. This survey script included sociodemographic data (age, gender and education), type of clinical study (Phase I or III), the involved pathology and location, the number of times that the volunteer participated in clinical studies, the motivation for participation, information concerning the ICS (whether it was introduced correctly and whether the volunteer could explain the study topic and remember the ICS), compensation received for participation and recruiting vehicles (how the volunteer found out about the study). Socioeconomic class was measured using the ABEP scale (Brazilian Association of Research Companies -Associaç<unk>o Brasileira de Empresas de Pesquisa), which takes into consideration the education level of the head of the household and ownership of assets [22]. This scale was used to sort participants into standardized subgroups labeled from A to E where A was the highest economic strata. Mean family income (MFI) at the E-level of the ABEP index is very low (below a 'living wage' level considered acceptable for families in the US), whereas a D-level family enjoys a MFI value roughly 1.5 times the E-level MFI. The Clevel MFI is roughly 2-3 times the E-level value, A-level and B-level is approximately 27 times and 9 times the Elevel MFI value respectively. The anonymous interviews were semi-structured, indepth conversations that were recorded with the consent of the respondent. The interviews were approximately 40 minutes in duration, and the participants were compensated R$20.00 (twenty reais) upon completion of the interview. --- Analysis of the results --- Qualitative analysis of the interviews After transcription, the interviews were submitted for content analysis, as proposed by Taylor and Bogdan [23] and Bardin [24] in accordance with the following steps: --- Initial reading The initial reading allowed for the formation of general impressions. --- Preparation of the material The material was prepared by separating and categorizing the responses according to topic. This material gave rise to independent archives for each script item: age, gender, education level, motivation for participation, study type, number of times the individual participated in clinical studies, method of recruitment, compensation for participation, understanding of the ICS and socioeconomic class. Each item included 80 responses that correspond to each sample constituent. With this information, the categories were constructed. --- Treatment of the results The frequencies were calculated, allowing for interpretation. In the present study the triangulation technique was used [19]. Each interview was coded by more than one researcher to ensure the consistency of interpretations and increase the reliability of the categories. Excerpts from the interviews with KIs and study subjects appear in italics in the results. --- Statistical analysis Hypothesis tests were conducted for each of the script variables to compare the characteristics and motivations of Phase I and Phase III study subjects. After codifying the answers from the qualitative phase, each theme was analyzed to generate up to 5 response categories. These variables were tested for the homogeneity of the distribution of responses across categories. The Student's ttest was used to evaluate average age, study phase and research motivation. The Pearson's chi-square test or the Fisher's exact test was used for the categorical variables. A significance level of 5% was adopted, and Stata version 11 software was used to perform the analyses. --- Ethics The protocol was reviewed and approved by the Universidade Federal de S<unk>o Paulo (UNIFESP) Research Ethics Committee (Protocol #0870/10), with provisions for participants to participate anonymously, to decline to participate, to leave questions unanswered, and that they could interrupt their participation at any time, according to the Declaration of Helsinki. No one of the researchers occupy dual role (they were not clinicians or clinical researcher's managers). --- Results There was a predominance of women among the respondents (58%). Table 1 shows that the majority completed high school or college (approximately 90%), were less than 50 years of age (72%), were in either socioeconomic class B or C (84%) and had participated in only one clinical study (63%). --- Analysis of the interview Motivation to participate Content analysis of the interviews identified motivating factors for participation in clinical studies. Figure 1 lists the main motivation for participation categories that emerged from the discussions with the KIs and study subjects. Both groups identified the therapeutic option and financial compensation as key motivators, suggesting that the KIs understood the motivations of the volunteers. --- Therapeutic option Key informants --- Opinion of the KIs versus the study subjects According to the KIs, the therapeutic option to have access to a new treatment and the hope that this new drug can produce an improvement is a motivator to engage in clinical studies for potential experimental treatments. The discussions with the KIs in the following transcripts reveal the opinions of these professionals. The group of patients, because they already have the illness, is always searching for something better, a new treatment. They submit themselves to the collateral effects of the medication to try a new medication, i.e., for cancer. Others opt to participate in the study because they do not have another therapeutic option. These cases are more common in oncology and rare illnesses with few conventional treatment options. --- Study subjects Motivation for participating in a clinical study according to study participants: According to study subjects, the therapeutic option is a strong motivator to enroll in a clinical study. The potential benefit of a new medication was the most cited motivator. However, it is worth noting that this motivator is restricted to those volunteers with some type of pathology (in this case, Phase III study subjects). The following excerpt is from an interview with a study participant: Motivation? It was for the benefit itself because it was a problem that I have. I thought: let's do it! --- Financial compensation Key informants According to the KIs, financial compensation influences a volunteer's decision to participate in a clinical study, especially among healthy volunteers. Although the Brazilian legislation only allows for compensation for time spent in the study, this remuneration is often attractive to a portion of the participants: I believe that the healthy volunteers, in Phase I studies and in some other situations are motivated by a financial compensation, although in accordance with resolution 196 we cannot pay a Study Subject in Brazil, but in some way this compensation is the greatest motivation to participate in the study. There is a camouflaged payment [compensation for the work-loss day], but many times it is much higher that the workday compensation. --- Study subjects Healthy volunteers identified compensation as the principal motivation for participation in a study. The majority of volunteers did not want to reveal the amount that they received for participating in these studies. It is not the first time that I have participated. They pay well, and I have confidence in the place and it is good because they do several examinations, and I a find out about how my health is. --- Access to healthcare Key informants According to KIs, many study subjects enter clinical studies because they believe that they will have improved access to health care and better quality of care. The subjects believe that they will be better able to check up on their health and will be able to avoid the time that they would spend in the public health sector. Moreover, the ability to call a study physician at any time is attractive to these patients. In general the people, who participate of clinical studies, participate in research projects because they receive excellent attention and a healthcare option that is frequently better than the healthcare medical plans. The patient feels truly taken care of, this is very difficult for me to assume, but it is what happens, the easiness in the attendance, the patients undergo examinations, they are taken care of with a special attention by the doctor... The KIs mentioned access to healthcare as one of the main reasons for engaging in a clinical trial. However, the clinical trial participants did not evaluate it the same way. They acknowledge access to healthcare as a benefit but secondarily to other reasons that they consider more important for participating in a clinical trial, such as: financial advantage or therapeutic option. --- Altruism --- Key informants The KIs believed that a small portion of the study subjects altruistically engage in these studies to improve the lives of others. I believe that there is an emotional component to this condition that emerges. For example, a father that volunteers for an AIDS vaccine because he lost a child to the disease. --- Study subjects Some of the respondents mentioned the possibility of helping other people as a motivator for participating but always as secondary motivation. It was to improve my knee and also to help in the study. First, because I have arthritis, and my healthcare plan was very bad.... and second, because I am helping the researcher, and who knows, maybe humanity. --- Other motivations Key informants The KIs also commented that the possibility for volunteers to make friends with people of a similar age and health status might lead them to participate. experiences, they talked and they started to meet outside of the clinic. The possibility of receiving the test medication for the entire illness period, even after completing the study, was another motivation that was mentioned by the KIs. --- Study subjects Few study subjects cited motivations other than a new therapeutic option and financial compensation. However, some added additional motivators, such as the recommendation of their personal physician to participate in the study. I wanted to try another treatment. The ones that I had tried had no effect. That was when my doctor said that there might be a good opportunity for me to participate in a study for a new arthritis medication. --- Informed Consent Statement (ICS) All the interviewed subjects declared that they had read and understood the ICS. They also reported that when they were unsure that there was a professional available to help them. However, when asked if they could remember the principal points of the ICS, a number of participants did not remember any of the content. Among those that remembered some parts of the ICS, none mentioned the potential risks of the study in which they participated. That I could quit whenever I wanted to, and that I could speak with a doctor at any time. I could stop when I wanted to, that I would take laboratory medicine with a name that I can't remember...I had to do the exercises correctly. I gone through the entire interview, I did all the examinations, but I do not remember what was written, I enrolled with the desire to resolve this [health] problem. --- Results of the statistical analysis Table 2 shows that participants in different types of studies have different motivations to participate (p <unk> 0.0001). For example, 94.7% of respondents that were motivated by financial compensation were Phase I study subjects. Conversely, 100% of the respondents who were searching for a therapeutic option were Phase III study subjects. Other motivations such as altruism and access to the healthcare were mentioned but were not the main motivators. Only 4% of participants alleged that these were the principal motivators. The average age of participants (Table 3) was significantly different depending on the phase of the clinical study (p <unk>0.0001). Those that participated in Phase I studies were younger (average age: 32 years; 95% CI: 28.7-34.8) than participants in Phase III studies (average age: 53 years: 95% CI: 49.1 -56.0%). The average age varied by study purpose (p <unk> 0.0001). The bioequivalence and medication studies had the youngest average ages (average age: 31.5; SD= 9.3). In this case, all were phase I study subjects (Table 4). Most of the volunteers from both phases belonged to class C. However, the proportion of class D and E subjects varied according to study phase. Only 2.5% of participants from Phase I clinical studies belonged to the classes D and E. None of the respondents belonged to class A (Table 5). No statistically significant differences in the proportions of men to women when stratifying for study purpose or research phase. In addition, there were no differences in social class or education level when stratified by motivation for entering the study (data not shown). --- Discussion Developing nations are participating in multi-center clinical studies at increasing rates. This participation is due to the reduced operational costs, ease of recruiting study subjects, ability to conduct research and regulatory capacity of these countries [26]. According to several authors, this shift is primarily due to economic considerations [20,27], suggesting that volunteers for clinical studies in developing countries are "guinea pigs" [9]. Therefore, there should be heightened concern for the ethical requirements to conduct a clinical study in these areas [21]. Further innovation in the pharmaceutical field requires these countries to be able to conduct clinical studies ethically. It is essential for international studies to be conducted in Brazil to advance the body of knowledge [2]. Despite advances in legislation and in professional development in this field, Brazil is far from being considered a major research center. The idea that Brazilian research participants are guinea pigs is not justified. Brazil conducts only 1.8% of total global clinical studies (n =1417). In contrast, the United States performs the majority of clinical studies; 54.3% of global clinical studies are conducted in the US (n= 41917) [25]. These numbers demonstrate that Brazil is still not an important clinical research center, despite having characteristics that would be conducive for performing studies. Therefore, it is inappropriate to accuse Brazil of being the supplier of guinea pigs for clinical studies [9]. However, this finding should not decrease the focus on ethical concerns for clinical studies. These advances must be made rationally and should avoid unfounded critiques and prejudices of emerging countries. Altruism should be the main reason for a subject's decision to participate in a research study 28). Ideally, the volunteer is capable of making decisions based on the information provided about the proposed study, and understands the purpose, risks, benefits, alternatives and requirements of the study. After receiving this information, the volunteer is able to decide to participate, free from coercion or improper influences [28]. However, there is concern that the benefits may interfere with the study subject's evaluation of the risks [21]. The results of the present study, although preliminary, show that health and financial benefits are the primary motivators for respondents to become study subjects. Individuals from higher socioeconomic classes were more likely to cite altruism as a secondary motivator after economic advantage (Phase I) or therapeutic option (Phase III). The same motivations were also identified by the KIs. Although several authors insist that these motivators are more common in developing countries [21], studies demonstrate that they are found in study subjects throughout the world [29,30]. In a review by Stunkel and Grady [29], 12 of 13 studies showed that financial motivation was the main reason to participate in studies. One volunteer from these studies comments, "Nobody is Robin Hood to make the good for the society." In the studies where the participants affirmed that compensation was the main motivator, no participants refused payment [31]. In a study of the motivations for healthy subjects to participate in research Vrhovac et al. [32] found that almost 80% reported compensation as the primary motivator. However, in that study, 20.6% of the volunteers denied that the money was the principal reason and identified humanitarian motivations and the desire to contribute to society. The key informants for the present study also mentioned altruism; however, it was only mentioned in exceptional situations. Alternatively, there are study subjects that believe that compensation should be commensurate with the risks of the study and that compensation should be increased for invasive procedures, such as drawing blood [33]. Almeida et al. [34] found that volunteers with low education and low socioeconomic status were more likely to cite financial motivations to participate in a clinical study. In this study, we did not find this association; there was no difference in education level between Phase I and Phase III participants. However, participants in a higher socioeconomic class were more likely to participate in Phase I studies (Table 5) (thus, according to Table 1, they were more likely to be healthy volunteers participating for financial reasons). These findings are in agreement with those of Kass et al. [30], who found that Caucasians with a higher education level were more likely to be motivated by financial benefits. These findings are in contrast to the beliefs of some authors who say that compensation attracts low-income volunteers, resulting in a disproportionate number of poor people participating in clinical studies [35]. Gelderen [36] found that the youngest volunteers were the most likely to mention money as the primary motivator to participate, which is a finding that was reproduced in our study (average age of 32 years in Phase I participants). This result is not surprising, as younger people might not be as financially established and may be looking for ways to earn money. Moreover, Phase I participants are required to be healthy, which is a criterion that is more common among younger adults. Included studies of menopause and temporomandibular pain; ANOVA. Among Phase III participants, the principal motivation for participation in clinical studies was to search for a new treatment, a motivator that was also mentioned by the KIs. Grecco and Diniz [37] have noted that clinical study volunteers in developing countries see participation in clinical studies as an opportunity to receive better healthcare, easier access to more expensive laboratory examinations and novel drugs. According to Grecco and Diniz, these motivations constitute a conflict of interests. It is worth noting that Brazilians have universal access to the Unified Healthcare System (Sistema <unk>nico de Sa<unk>de -SUS), which is funded by the federal government [38]. However, access to this system does not limit the interest in clinical studies, perhaps due to the quality of the services or the degree of individualized attention. Cabral et al. [1] have noted that in the research environment, the physician-patient relationship is inverted. Classically, the physician satisfies the needs and interests of the patient. However, in a clinical study, the patient satisfies the interests of the study and thus the researcher. Lackey [27] adds that the researcher, for the advancement of the study, must treat these subjects as scientific objects submitted to the laws of cause and effect. The study subjects may not understand this distinction and may continue to participate in the study expecting a cure when, in reality, the medicine may not have any effect. Canvin and Jacob [39] observed a group of volunteers for a study of epilepsy and noted that these volunteers were happy to help others only when they could also help themselves. The authors called this behavior "weak altruism," and McCann et al. [40] have called this behavior "conditional altruism." According to these authors, the desire to help others or to contribute to the body of knowledge does not lead a volunteer to participate in a study unless that volunteer perceives that his participation in the study may benefit him personally. The Informed Consent Statement (ICS), which is required under Resolution 196/96 (CNS/196), is provided for the volunteer to inform him about the study, including the associated risks and the responsibilities of those involved. The ICS is indispensable to the ethical conduct of the research. According to a number of authors, volunteers do not always pay enough attention to the ICS. This oversight may occur because the volunteer trusts the doctor (or at least the role of the doctor) and does not take into consideration the role of the doctor as a researcher, that is, the doctor is not necessarily focused on meeting the patient's global health needs [27]. Additionally, the patient may overestimate the benefit for himself and lose the ability to weigh the disadvantages of the study [41]. Finally, the volunteer may not understand the ICS because of a low educational level [42]. In the present study, most volunteers had a good education, and all declared that they understood the ICS. However, when they were questioned about the study details, they only remembered the possibility of speaking with a doctor and the possibility of leaving the study. None of the subjects mentioned the disadvantages of the study, such as the risks or potential adverse reactions to the medicine. These findings may indicate that the volunteers did not adequately understand the ICS. It is worth noting that a portion of the respondents did not remember that some items of ICS may be limited to the time that they participated in the study. The more "comfortable" situations associated with the study were remembered, while the situations associated with risks were not mentioned. However, the date of last participation in a clinical trial was not questioned, which may constitute a limitation in analysis of the data. --- Conclusion In the present study, study volunteers were motivated by some type of personal benefit from participating in a study (especially financial or therapeutic benefits), which is a finding has been observed in clinical study participants in other countries. According to several authors, this finding is accentuated in emerging countries, such as Brazil, due to the limitations of the health care systems, even despite the universal access provided by the SUS. Altruism was not a common motivator, and when it was mentioned, it was never the primary reason. Some have called type of motivation this "conditional altruism". The ICS was understood by all participants; however, the degree of understanding may have been limited. None of the subjects remembered the text referring to the potential harms of the procedure or medicine. The authors hypothesize that this behavior results because the individuals participate out of self-interest, rather than altruism. --- Study limitations This study was a preliminary study with a qualitativequantitative approach. The study sampled 80 participants and was not representative of the total population of clinical study participants. Another limitation is the fact that those who refused to participate in the research were not surveyed. --- Study strengths This study demonstrates that personal benefit is the primary motivation for volunteers to participate in clinical research. However, we argue that this motivation seems to be universal and is not limited to developing countries. Moreover, the theory that financial compensation disproportionately attracts poor volunteers has not been supported. Furthermore, the fact that study participants are primarily motivated by personal benefit does not suggest that they do not evaluate the risks of participation. For further consideration, if we consider altruism to be the only ethically acceptable motivation for study participation, must we evaluate whether the principle of autonomy has been disrespected? --- Competing interests The authors declare that they have no competing interests. Authors' contributions SAN conceived the study and participated in its design and coordination and wrote the final version of the manuscript. GBI collected data and ZMS performed the statistical analysis and interpreted the data obtained. All authors read and approved the final manuscript.

Background: In the past, clinical study participants have suffered from the experiments that they were subjected to. Study subjects may not understand the study process or may participate in clinical studies because they do not have access to medical care. The objectives of the present study were 1. to analyze the motives that might cause a volunteer to participate as a study subject; 2. to identify the social-demographic profile of this study subjects; and 3. to determine whether the motives to volunteer as a study subject are in accordance with the established legal and ethical principles for research in Brazil. Methods: Mixed-methods research was used (a qualitative-quantitative approach). A sample of 80 volunteers underwent a semi-structured interview, which was based on a survey script that was elaborated from discussions with key informants. The sample was randomly selected from a database of clinical study volunteers that was provided by Brazilian clinical study centers. The interviews were recorded and transcribed. Descriptive statistics were used for content analysis, including contingency tables with hypothesis testing.The motivations for clinical study participation were linked to types of benefit. The most frequently encountered motivations were financial gain and therapeutic alternative. Altruism was not a common motivator, and when altruism was present, it was observed as a secondary motivator. All participants reported that they understood the Informed Consent Statement (ICS). However, only two parts of the form were remembered by all of the volunteers: the section on being able to leave the study at any point and the section that stated that there would be some responsible professional at their disposal for the entirety of the study.The present study shows that study participants are primarily motivated by personal benefit when volunteering to participate in clinical studies. Whether these study participants had an integral understanding of the ICS is not clear.

This research also presents evidence that the complexity of local adaptation arises from the political, economic, social, cultural and institutional factors and processes that interplay within the households and communities. Sometimes, these same factors and processes work against effective local adaptation at both household and community level. However, Engle (2011) claims that adaptive capacity of a nation or community that is affected by climate variability is influenced by the institutions, management and governance. The findings presented in this thesis illustrate that at household level, women are not given a chance to contribute to household resilience to climate variability because of cultural values that suggest that women are subordinates and that therefore only men have the authority to actively participate in such developmental activities. The motivation of this research is based on the extent and intensity of floods in flood prone areas, and the growing scholarly and personal interest to understand the context and dynamics of vulnerability to floods and local adaptation. The main objective was to understand the impact of floods on sustainable development specifically how communities living in flood prone areas perceive their vulnerability to floods and how they are coping with the floods in situ. This chapter presents the research questions that this thesis sought to answer. It also presents the methodology that was used to collect data and finally, the thesis layout with brief description of what is presented in the different chapters. --- LIST OF TABLES --- Background of the study In January, 2015, Malawi experienced fatal floods which affected 15 Districts out of the 28 Districts and a state of emergency was declared by the President of the Republic of Malawi on January, 13 th, 2015 (DoDMA, 2015). During that time, 1,102,364 people were affected by the floods, 104 people were reported dead, 645 people injured, 172 people were reported missing and 336,053 people were displaced with 225 displacement sites (Ibid). In Nsanje, where the research was conducted, out of a population of 238,103, 74,250 people were displaced in 22 displacement sites, representing 28% of the total population in the district, of which 31 people died, and 153 people were reported missing (DoDMA, 2015a). A section of Nsanje District has been experiencing floods for over a decade and was declared a flood prone area the same time (GoM, 2006). Therefore, firstly, this research seeks to understand how the communities that live in the flood prone areas perceive their vulnerability to floods and how some of them have been coping with the floods since 1952, just over 6 decades. This study is important because it contributes to knowledge on understanding the changes in the extent and intensity of the floods in flood prone areas and the corresponding adaptive measures that have been implemented over the years. In 2012, communities living in flood prone areas of Nsanje District experienced devastating floods that resulted in loss of property and damage to infrastructure (DoDMA, 2012). Furthermore, following a flood in 2009, a declaration was made that TA Nyachikadza, one of my research sites, was no longer habitable because it is surrounded by a marsh and due to the geographical position of the area it was prone to extreme flooding (DoDMA, 2010). In the same year, some villages in Traditional Authority Mlolo, under Group Village Kadyamba were also declared uninhabitable due to the severe flash floods that were experienced in the areas. The Government of Malawi through officials from the Department of Disaster Management Affairs (DoDMA) advised the communities living in these areas who are mainly communities to move to upland areas, which are not prone to flooding. The first meeting was held in 2010, then in 2012 and the last one in 2015 where DoDMA officials strongly advised communities living in flood prone areas to relocate upland so that they should reduce the impact of floods on their lives. However, the communities have refused to relocate and have insisted on living in the flood prone areas. Even after the 2015 floods, the majority of the communities moved back to the flood prone areas (Nsanje District Council, 2015b). Currently, there is no data that explicitly explains the factors that motivate communities to live in flood prone areas. However, there have been claims from various individuals through the media, that suggest that the communities continue living in flood prone areas because they are used to receiving handouts and humanitarian relief support and services during floods (Nyasatimes, 2012). This research therefore secondly seeks to understand the factors that motivate communities to continue living in the flood prone areas despite the previous experience and exposure to devastating floods. For over a decade, there have been climate change adaptation programs in Nsanje District according to Nsanje District Development Plan targeting communities that live in flood prone areas to help them minimise the risk and vulnerability to floods (Nsanje District Council, 2010b). Over the years as evident by the 2009, 2012 and 2015 floods, there has been a noteworthy increase in the number of households that cannot cope with the floods. Recently, there has been an increase in the number of communities that have become more vulnerable to floods over years and struggling to cope with the floods without external support (DoDMA, 2015b). However, there has been no research that sought to understand and explain why these communities are failing to be resilient to the floods despite all the investments made through climate change adaptation programs. This research furthermore explores some of the challenges that these people face which have exacerbated their vulnerability to floods. --- Vulnerability and Adaptation to Climate Change Variability Until 2006, only 6 districts were affected by climate variability, including floods in Malawi, with no reported deaths (GoM, 2006). After close to a decade, Malawi had become more vulnerable and exposed to floods that affected a larger population (DoDMA, 2015a). According to McSweeney et al. (2008) Malawi experienced a 0.9 degrees Celsius increase in the mean annual temperature between 1960 and 2006, which reflects an average rate of 0.21 Degrees Celsius increase per decade. Furthermore, McSweeney et al. (2008) predict an increase of rainfall up to 19% by 2090 using their rainfall model. However, the authors did not explicitly predict the extent of the impact on communities living in flood prone areas, the extent of their vulnerability and how complicated local adaptation would become. It is clear that some groups of people and some regions would be more vulnerable than others although there have been various uncertainties in climate variation, there is need to come up with adaptation strategies that go beyond the activities that these people are currently implementing to cope with climate variation (Adger et al., 2004). Scholars reveal the need to prepare and plan for the unknown extent of risks to reduce vulnerability and be able to cope with the extreme weather events. However, with the recent global disasters as a result of extreme weather events, it is clear that there is need to understand the realities within the vulnerable communities in terms of their perceptions, power dynamics and local adaptation to be able to have adequate data to facilitate informed debates in climate variability, vulnerability and adaptation discourses. The Intergovernmental Panel on Climate Change, IPCC (2013) has stressed that global temperatures are increasing hence very likely that there would be huge negative impacts on the climate, environment and the various ecosystems. In addition, the United Nations Framework Convention on Climate Change UNFCCC (2013) indicate that developing countries will suffer more because they are more vulnerable to climate variability due to inadequate capacity in terms of finances and expertise to deal with the negative impacts of climate variability. Within the affected communities, people are affected differently depending on various factors including social class, gender, age, geographical position, culture and group (Paavola, 2008, Kakota et at., 2011). The studies furthermore emphasise that even though an area would be affected by the same effects of climate variability, the poorest of the poor would be affected more because of lack of alternative sources of livelihood and cash; orphans, women, people living with disability and people living with Human Immunodeficiency Virus/ Acquired Immunodeficiency Syndrome (HIV/AIDS) would be heavily affected (ibid). Due to various differences in factors contributing to vulnerability, adaptation measures are also different and they depend partly on culture, geographical positions, local and institutional changes, and power (Bougsty-Marshall, 2016). Goldman and Riosmena, 2013;O'Brien and Wolf, 2010). The differences in various factors that contribute to vulnerability and hence influence local adaptive capacity have demanded more research in various parts of the world to better understand the context of vulnerability and the dynamics of local adaptation to climate variability (Wisner et al., 2004). Trawöger (2014) emphasises that there is a need to deeply understand the local perspectives of climate variability in order to deliver climate change adaptation projects and programs that would enhance community resilience effectively and efficiently (Trawöger, 2014). The global perspective of climate change and the associated risk should not be generalised but rather should be case specific in order to design appropriate approaches to adaptation to climate variability (Ibid). There are gaps in the vulnerability literature in terms of understanding climate change impacts on natural systems and social economic trajectories, including adaptation, which necessitates the need for more case studies to understand these relationships and interactions (Tucker et al., 2015). In addition, Harrison and Chiroro (2016) stress the need to understand the contexts that shape vulnerability and how communities differ within themselves to better develop adaptation and resilience projects that foster sustainable development. This thesis therefore seeks to fill the identified gaps and contribute to literature In Sub Saharan Africa, ODI and CDKN (2015) reveal that many countries in the region, including Malawi do not invest in long term climate change adaptation strategies hence exacerbating the vulnerability to floods in the long term. In addition, lack of appropriate climate information in most developing countries to enable the flood victims to plan accordingly has been argued to be one of the factors that has exacerbated vulnerability to floods. Investing in long term climate change adaptation plans requires more financial and technical resources which currently most developing countries do not have. The UNFCCC strongly recommends that developed countries should commit themselves to helping developing countries to cope with the extreme weather events even though the reality suggests that little is done towards implementing that (UNFCCC, 2013). In other contexts, Stasavage and Moyo (2000) cite an example of how the International Monetary Fund (IMF) imposed a cash budget system in Uganda and Zambia as a means to decrease the deficit on the budgets of the countries. These two countries were directed to spend only on what they have as cash to reduce overspending by the government which results in recurring government deficit. Contrary to the direction, there was evidence that the politicians and national employees managing the funds did not agree with the system since they had their own way of implementing the national activities. These examples of how donors impose how governments in the global south should manage the donor funds intrinsically reveal power dynamics in bilateral aid and donor conditions where the developing countries receive aid but have to implement the programs the way the donors want. Bryan et al. (2009) stress that climate variability will mainly affect communities living in Africa because agriculture is their main source of income. A better understanding of community's perception of climate change, adaptation strategies and decision-making processes are critical to inform policies that focus on promoting successful adaptation strategies (ibid). This thesis emphasises that missing the underlying causes and differences in the extent of vulnerability has led to the generalisation of adaptation strategies that are not applicable in some areas. For example, adaptation projects in Malawi focus on sustainable and climate smart agriculture (DoDMA, 2013) even in flood prone areas which is purely a short-term strategy and not effective to prevent damage and harm from floods. Hunger, famine and permanent migration are factors that usually threaten poor households whilst economic losses affect mainly those who have valuable assets (Ribot, 1995). At local level, it is argued that different people are affected differently by climate variability, even though adaptation strategies are designed to address vulnerability equally amongst the affected communities. Ribot (2014) stresses that there is a need hence to study different cases separately to deeply understand the underlying causes of vulnerability to floods at local level and how the communities that are affected by the floods cope with the floods in order to inform policies and guide development practitioners on what is feasible and to whom, with evidence. This thesis explores therefore is important because it provides evidence of how vulnerability to floods is created, differentiated and exacerbated in Nsanje Districts and how the affected communities are responding to the floods. Dilling et al. (2015) reveal that vulnerability is dynamic and complex such that there is no definite longterm strategy to reduce vulnerability due to unstable ecosystems, culture, social interactions and landscape with time. Furthermore, the authors suggest that short term adaptation is possible although they doubt the likelihood of long-term resilience using fixed climate change adaptation policies and strategies (ibid). Long term resilience can only be determined by the way the people who are affected by climate variability frame their vulnerability to the floods over time and therefore take appropriate actions accordingly to address their vulnerability to floods. This thesis provides evidence that vulnerability to floods is dynamic and differentiated therefore communities respond to the floods based on the way they have been affected by the floods. In other words, the extent and intensity of floods determines how the communities respond to floods therefore suggesting that adaptation strategies are also dynamic. Fixed long term strategies that are targeted towards addressing climate variations must therefore be considered with caution because there are cases where they may not be applicable. Fixed climate related decisions that are made at different stages may affect vulnerability either positively or negatively in the long term. This thesis explores the factors and processes that affect the smallholder farmers that live in flood prone areas as they cope with the floods. These factors and processes include power dynamics within the sociocultural environment and the interaction between political and economic factors within flood prone areas. --- Problem statement Currently, there is no data that explicitly explains the factors that motivate people to live in flood prone areas. However, there have been claims from various individuals through the media, that suggest that people continue living in flood prone areas because they are used to receiving handouts and humanitarian relief support and services during floods (Nyasatimes, 2012). Floods have negatively affected the sustainability of the overall development of the district. Despite huge investment to deal with effects of floods in Nsanje district, the continued trend of floods continues posing a threat to the overall development of the district A lot of research has been conducted in the area of Floods, however in Malawi Floods and sustainable development while considering Nsanje district has not been studied conclusively. It is against this backdrop that this study would like to establish the impact of floods on sustainable development. --- Objectives of the study --- Main Objective The overall aim of the research was to understand the impact of floods on sustainable development and how communities living in flood prone areas perceive their vulnerability to floods and climate variability, to investigate the power dynamics at household and community level and to explore the complexities associated with local adaptation programs in the flood prone areas. --- Specific objectives • To find out how do communities living in flood prone areas perceive their vulnerability to floods? • To find out what motivates communities to stay in flood prone areas? • To investigate factors that have caused the communities living in flood prone areas to continue being vulnerable to floods? • To find out if flood management strategies mainstream sustainable development approaches in its implementation --- Research Questions • How do communities living in flood prone areas perceive their vulnerability to floods? • Why are communities motivated to stay in flood prone areas? • What are the factors that have caused the communities living in flood prone areas to continue being vulnerable to floods? • How does flood management strategies mainstream sustainable development approaches in its implementation? --- Significance of the Study The research findings would give guidelines to the Malawi Government in the formulation of appropriate laws and policies that would seal the loopholes that create hindrances on sustainable development To the academicians this study may contribute to the literature of floods and sustainable development within the country which already exists. In addition, the study may also stimulate further research on the area of floods and sustainable development in particular in Malawi and other countries --- Structure of the paper The study has been organized into five chapters. Chapter 1: Introduction. Chapter one gives a general introduction and background to the topic of study. It also presents the problem of the study, research objectives and research questions. Finally, it presents significance of the study and chapter summary. Chapter 2: Literature Review. Chapter two deals with the review of literature pertaining to the research study. It presents theories and concepts that are related to the study. It also presents previous work related to the study and conceptual framework of the study. Chapter 3: Research Methodology. Chapter three gives description of the research process and the methods adopted for collecting and analyzing data. Chapter 4: Research Findings and Discussion. This chapter presents results from the research questionnaires which are analyzed using SPSS and presented in tables and figures. Mainly it contains demographic results of the participants and the main results on the effects of employee participation on the organization performance Chapter 5: Conclusions and Recommendations. Chapter five summarizes the findings, conclusions and relevant recommendations are presented. --- Chapter Summary This chapter has presented the general overview of this study. It has also presented the background of the study, problem statement, research objectives and the significance of the study. In addition, the chapter has presented the overall and specific questions, the significance of the study and finally, it has presented the structure of the study. --- CHAPTER TWO --- LITERATURE REVIEW --- Introduction This chapter presents a review of literature on the impact of floods on sustainable development, particularly paying attention to floods and unpacking the concept of 'vulnerability' (UNFCCC, 1992;Ribot, 1995;Adger, 1996a). Furthermore, literature is reviewed in order to understand the different factors that contribute towards vulnerability to climate variability in general, followed by various case studies that describe vulnerability in context, local adaptation and the barriers and limits to local adaptation. Finally, the main debates in the literature are summarized and ways in which this research fits into the wider academic discourse on impact of floods on sustainable development vulnerability and local adaptation. This thesis recognizes that vulnerability in flood prone areas is created and exacerbated by several factors, including social, environmental, cultural, technological, economic, political and historical factors. In addition, the thesis acknowledges that a transformational change in the way adaptation initiatives are designed is important in order to incorporate all these factors which would help to reduce vulnerability and enhance the local adaptive capacity. --- Theoretical literature Malawi, like many Southern African countries, is experiencing increasing climate change and variability, which results into poor crop yields and/or even total crop failure due to drought and floods. Being an agro-based economy with more than 85% of the population relying on rain-fed agriculture for their livelihood and survival, the level of vulnerability has increased with the increasing extremes of drought, flooding and erratic rainfall patterns (Government of Malawi, 2006). For example, in 2012, it was indicated that floods had caused an annual loss of up to 12% of maize in the southern part of Malawi and drought destroyed 46% of the maize crop (Shumba et al., 2012). A study by Msowoya et al. (2016) predicts that there will be a 14% decrease in maize production by mid-century and 33% at the end of the century in Lilongwe District which is the largest maize growing district. Maize is Malawi's staple food that is grown during the rainy season. Rain fed agriculture will decrease food production and hence there is a need to invest in various farming technologies that are climate compatible, for example irrigation farming, growing new varieties of maize, crop diversification and natural moisture conservation strategies (Msowoya et al., 2016). Malawi being one of the poorest countries with limited resources to implement the proposed technologies, it is very likely that smallholder farmers will struggle to be resilient to climate variability. Malawi and many countries in Southern Africa have adopted conservation agriculture to adapt to droughts and erratic rainfall even though other studies reveal that there is need to adjust conservation agriculture depending on agro-ecological system (Thierfelder et al., 2016). Floods have led to displacement of communities, loss of life and assets, and have contributed to reduction of community resilience in Southern Africa, including Malawi, (DoDMA, (2012(DoDMA, ( ), 2015;;ODI and CDKN, (2015); UNDP, ( 2012)). According to Magis, (2010) community resilience refers to 'the existence, development and engagement of community resources by community members to thrive in an environment characterised by change, uncertainty, shocks and unpredictability'. Community resilience and community adaptation to various climatic shocks differ according to the communities' priorities (O'Brien and Wolf, 2010). This suggests that farming communities would have different adaptation strategies to fishing communities. Furthermore, their work suggests that communities with different socioeconomic status and perhaps culture, values and community myths, will have different coping and adaptation strategies. This thesis also examines how communities who live in the flood prone areas cope and adapt to the floods by exploring existing opportunities within the flood prone areas that enhance their livelihoods. This thesis also addresses the existing gap in literature on how communities are coping and adapting to the rapid, unprecedented and complex environmental changes (O'Brien and Selboe, 2015c) In flooding areas of Urban Lagos, Nigeria, it was revealed that women in low income households are more vulnerable than men and women in the middle and high social classes (Ajibade et al., 2013). In Nepal, women were more affected by the earthquakes than men (Shakya, 2016). Among farming households in flood prone areas, women are also more vulnerable than men (Enete et al., 2016). Bee (2016) indicate that understanding the gender dimensions of vulnerability helps researchers and practitioners to understand further the context of vulnerability and thus lead to more appropriate adaptation initiatives. Cutter (2017) advocates for more and explicit research and publications on how women and children are affected by the social and environmental injustice and particularly gender in vulnerability and climate adaptation within and among the vulnerable communities. In addition, understanding gender roles would enhance the effective implementation of adaptation programs that would like to empower the most vulnerable groups. In Iran, a study on the capacities of women in disaster management systems reveals that women play an important role in managing household chores and family livelihood, including taking care of children and cooking during and after disasters (Sohrabizadeh, 2016b). Participation of women during implementation of adaptation programs is important because it empowers them within their social-cultural and economic context to avoid misunderstandings that distract effective adaptation (Ibid). Gender relations, social power relations and knowledge processes influence women's perception of environmental changes and determines how they respond to environmental and social risks (Bee, 2016). This suggests that women in particular should be given attention in adaptation programs because of their important role at household level how they are heavily affected by disasters. There is a need to understand the roles of women in context to find a means of empowering them and equipping them to be able to deal with disasters. Reyes and Lu (2016) indicate that women provide food, keep up hope for the family, take care of the sick or injured members of the family therefore women's roles before, during and after disasters exceeds those of men. Furthermore, women from low income groups and single parents are more heavily burdened than other women (Ibid). During floods and other crises, male farmers usually migrate in search of better livelihood leaving women with children at home, where the women become responsible for household and agriculture activities (Khapung, 2016). Female farmers have limited access to agriculture extension and training programs whilst the male farmers are well exposed and actively participate in the agriculture extension programs as household heads (Mehar et al., 2016). This presents an example of how inequalities in exposure, access to resources, vulnerability, opportunities and capabilities are usually gendered and political in nature hence different groups of people are affected differently and adaptation measures should thus reflect this social differentiation, as it shapes adaptive capacity (Afriyie et al., 2017 andMadhuri, 2016). In flood prone areas, the capability to evacuate during a disaster is highly gendered as it is highly determined by the cultural constraints in female mobility and lack of physical skills and strength, in addition to various gender roles (Chung, 2017). These findings indicate that masculine strength is often greater than that of women, meaning that they can lack the capacity to escape floods on time. Furthermore, Neumayer and Plümper, (2007) indicate that when floods occur the number of women who die is more than the number of men and that floods reduce the life expectancy of women in flood prone areas. Although there is great variation from situation to situation, past research has indicated that women within flood prone areas have less capacity to cope with the disasters due to a combination of demographic, emotional and social factors that characterise them (Chandra and Gaganis, 2016). These studies suggest that gender disparities should not be overlooked in vulnerability and adaptation studies. Furthermore, they suggest that understanding vulnerability using a gender lens is critical to understanding what factors exacerbate vulnerability amongst women living in the flood prone areas and also how adaption programs can be designed and implemented to purposefully address the gendered vulnerability. Violence against women has been shown to increase in disaster-stricken communities, even though in most cases, these incidences are not recorded, investigated or shared, exacerbating vulnerability of women in these areas (Sohrabizadeh, 2016a). This suggests that in some cultures women are often disadvantaged in terms of exposure to new skills and are oppressed in terms of gaining appropriate knowledge and skills that might empower them. In a study involving fish farmers in Northern Thailand, Lebel and Lebel (2016) reveal that financial stability plays a significant role in the interactions between attitudes, analysis and emotions in making climate change decisions that are risky. This suggests that financial stability ensures right and effective adaptation measures regardless of gender, race and class (Godfrey and Torres, 2016). In Pakistan, high education status in women reduces health related sicknesses and underutilisation of health facilities in flood prone areas, emphasising the role of women in promoting good health in flood prone areas hence reducing their vulnerability to the effects of flooding (Sadia et al., 2016). The extent and context within which the women are disadvantaged need to be unpacked to help guide policy makers and development practitioners who are working on adaptation programs, in order that these programs have a meaningful impact on the lives of the most vulnerable. The distinctive role of women in flood prone areas and the gendered dimension of vulnerability to floods, including emotions and attitude towards vulnerability to floods, illustrates the complexity of vulnerability and therefore how important it is to facilitate effective adaptation in flood prone areas. This thesis therefore explores the ways in which gender affects vulnerability and local adaptation in flood prone areas to illustrate the need for effective transformation in understanding vulnerability and addressing adaptation issues which are differentiated based on local power dynamics and politics around gender, culture, class, attitude, perceptions and partly emotions. --- Adaptation to Climate Variability, Including Floods Globally, it has been recognised that developing nations are more vulnerable to the effects of climate variability because they do not have adequate expertise and finances amongst other things to deal with for example, floods, droughts, dry spells, strong winds and earthquakes (IPCC, 2013). As a result, a plea was made in 1992 to developed nations to assist developing nations through financial and technical support, including humanitarian aid in times of disasters to ease the burden of managing disasters (UNFCCC, 1992). Through the global conventions on climate change, such as the UNFCCC, The World Conference on Disaster Risk Reduction, amongst others, global and national policy on climate change and disaster risk reduction have been formulated. These climate change and disaster related policies were formulated to guide implementation of adaptation programs at international, regional and local level and to guide policy formulation at those same levels. The National Adaptations Programs of Action (NAPAs), and national climate change policies for participating countries were formulated after the Kyoto protocol in 1997. During the Kyoto protocol, some developed countries committed to reduce the production of greenhouse gases in order to reduce global warming and the associated effects of global warming. The countries that ratified the Kyoto Protocol in 1998 were encouraged to develop the National Adaptations Programs of Action (NAPA) highlighting the major climate change effects that the particular countries experience. In addition, the NAPAs were to indicate how best the countries were going to deal with the climate change effects in targeted areas following a thorough risk assessment at national level (UNFCCC, Conference of Parties 3). However, due to climate variability, the intensity of extreme weather events became so severe such that other disaster related institutions were established in order to help to manage such disasters. In 2005, parties that signed the disaster framework formed national disaster management policies with guidance from the Hyogo framework for Action (HFA) 2005 -2015; in order to reduce vulnerability to disasters and make certain that vulnerable populations are protected by their states. During the convention, it was evident that developing countries were struggling to cope with the disasters due to inadequate financial and technical capacity (UNISDR, 2005). As a result, developing countries were advised to conduct disaster risk assessments at national level to specifically identify disaster prone areas to relocate vulnerable communities to safer areas (Ibid). In 2015 the Sendai Framework for Disaster Risk Reduction: 2015 to 2030 was formulated and adopted by interested countries (UNISDR, 2015). The framework highlights areas that needed to be adjusted following a thorough assessment and evaluation on the Hyogo Framework of Vogel (2006) notes the significance of the programme in bringing interdisciplinary experts to tackle vulnerability to global environmental changes and suggest that it is a good way of enhancing effective adaptation to environmental changes. The interdisciplinary aspect of vulnerability also suggests that there are several ways of coping and adapting to climate variability, including floods. The human dimension in dealing with climate change effects has enhanced the development of climate change adaptation policies that mitigate climate change effects in a way that enhances livelihoods of vulnerable populations. The interdisciplinary approach also has resulted in development of disaster related policies that integrate the management of disasters with disaster risk reduction strategies and mitigation measures that focus on the most vulnerable populations at both international, national and local level. In addition, there have been risk assessments to find out means of reducing vulnerability of populations at risk and provide support to enable them to adapt to the environmental changes to become resilient in the long term (United Nations Office for Disaster Risk Reduction, 2001). Adaptation to climate variability, including floods has been romanticised]. O'Brien and Selboe (2015c) argue that adaptation has been presented as a problem that can be solved by technical expertise and yet the realities indicate that adaptation is a complex process that is linked to identities of the affected people, politics, power, values, beliefs and world perceptions on adaptation. The authors furthermore illustrate that the realities suggest that technical expertise alone can never solve adaptation challenges, but rather a combination of personal, political and social change depending on the spatial locations and how all these factors interplay (Ibid). As presented earlier on, several studies indicate that local adaptation is necessary to reduce vulnerability to floods. Shackleton et al., (2015) emphasise that research on barriers to adaptation needs to start asking why these barriers emerge, how they work together to shape adaptation processes, who they affect most, and what is needed to overcome them. This research therefore unpacks how various factors interplay in the flood prone areas and also explores the barriers and limitations to effective adaptation that emerge in the flood prone areas. At national level, after the development of the NAPAs, various programs have been designed and are currently funded by international organisations, both government and non-governmental organisations, companies and other civil society organisations. There are several structures through formal and informal institutions through which development aid, humanitarian aid and other forms of support are directed to vulnerable communities. However, debates continue on how effective these national and local institutions are to develop and implement the climate change adaptation programs that will reduce vulnerability of the communities and empower them to be able to effectively cope with and adapt to the effects of extreme weather events (Baudoin and Ziervogel, 2017). A study conducted in South Africa (part of Southern Africa as Malawi is) reveals that the local institutions, despite having such an important role in climate change adaptation, do not have adequate skills in resource and project management; they also lack network skills that would help them to partner with other organisations in order to work as a team (Ibid). This study emphasises the need for additional skills that go beyond managing the flood itself to partnerships in order to effectively adapt to climate variability. This thesis, explores other factors that help communities to adapt to floods and make climate related decisions beyond the flood management knowledge. Manuel-Navarrete and Pelling (2015) emphasises that the politics of transformational change in different spatial locations needs to be explored in order to understand the power dynamics and decision-making processes in various places. Several case studies conducted in vulnerable communities suggest that community adaptation programs prioritise short term challenges whilst indirectly ignoring the sustainability of such interventions (Ensor and Benger, 2009;Ludi, et al., 2014). Community-based adaptation has been defined by Reid et al. (2009) as a community-led process which is based on the priorities of the communities, their knowledge, needs and capacity. Communities are not homogenous, hence there is a need to examine how projects impact various groups in these communities, for example women, youth, elderly, the poor, orphans, people living with disabilities and many others. Blaikie (2006), amongst other scholars, illustrates how community based natural resource management as a concept and theory in forest management yields different results in different countries (Malawi and Botswana) and how the implementation and success is differentiated based on particular political, power dynamics and socio-economic relationships within the different settings. In addition, it is evident that success of the program is based on the agendas of the organisations who initiate and implement it (Ibid). There is a clear generalisation on the definition of resilience by international donors rather than the actual definition of what resilience means to the differentiated affected communities. This implies that some interventions are imposed on the affected communities based on the priorities of those who see the needs and problems from outside the communities. There is also a need to understand the perspectives of those whom these adaptation programs are designed to benefit and whether or not the targeted social group benefit from such projects. It is important therefore to unpack critical elements of local and community adaptation based on actual community problems and proposed interventions using political ecology theory to understand how community-based adaptation can reduce vulnerability to climate variability in a sustainable manner that will enhance local resilience to climate variability. Households in Nigeria for example, communities cope with floods by selling their assets, borrowing loans to diversify their livelihood, migrating to other places where they can work to get money to sustain their livelihoods, utilising support from social networks and following new farming methods and technologies (Enete et al., 2016). Critically assessing the coping mechanisms, selling assets to cope with disasters seems a very good strategy in the short term, whilst it creates vulnerability to the households in the long term. In addition, relocating to other places during flooding also does not offer long term solution. This thesis therefore explores the most reliable adaptation strategies that the smallholder farmers have being relying on over the years in order to understand how adaptation practices have changed over time and how the smallholder farmers themselves have transformed over the years due to climate variability. Dodman and Mitlin (2011) reveal the need for development practitioners to contextually understand and incorporate political structures, culture, power and gender dynamics in their development plans to adequately and effectively address issues that are affecting climate vulnerable communities. Understanding vulnerable communities is vital because meaningful adaptation has to be designed in context in order to help the communities that are affected by climate variability. This suggests that there will also be sustainability of the projects after the project implementation period is over. Furthermore, it is important to understand the extent of losses, level of vulnerability and flood coping strategies in developing countries as it will shape the future of specific, strategic and effective adaptation strategies (Enete et al., 2016). The lack of this understanding will result in unsustainable interventions that will be costly and not accepted within the communities that such interventions are meant to assist. This thesis therefore unpacks the context of vulnerability, how it is differentiated and various ways in which the affected communities monitor the floods and cope with floods. --- Flood Monitoring and Politics in Flood Prone Areas Flood management consists of various assessments including hazard, vulnerability, exposure, risk, early warning system, damage and risk mitigation planning (Rahman and Di, 2017). Rahman and Di (2017) suggest that there is a relationship between several factors that determine vulnerability and adaptive capacity of vulnerable communities. However, the study does not explicitly indicate how these factors interact to create vulnerability. Instead, it focuses on major factors that exacerbate vulnerability to floods and not adaptive capacity issues, including flood monitoring. Firstly, authors suggest that tropical countries

A section of Nsanje District has been experiencing floods for over a decade and was declared a flood prone area the same time (GoM, 2006). Therefore, firstly, this research seeks to understand how the communities that live in the flood prone areas perceive their vulnerability to floods and how some of them have been coping with the floods since 1952, just over 6 decades. The overall aim of the research was to understand the impact of floods on sustainable development and how communities living in flood prone areas perceive their vulnerability to floods and climate variability, to investigate the power dynamics at household and community level and to explore the complexities associated with local adaptation programs in the flood prone areas. This research contributes towards framing vulnerability based on the perception of the different groups of people that are vulnerable to climate variability and have experienced environmental changes throughout their everyday lives. The communities living in the flood prone areas perceive floods as part of their livelihood. The communities have experienced floods for over 5 decades and have been implementing adaptive strategies to help them cope with the floods during this time. Temporary migration is one of the effective adaptive strategies that the communities living in the flood prone areas implement during floods.

suggests that there will also be sustainability of the projects after the project implementation period is over. Furthermore, it is important to understand the extent of losses, level of vulnerability and flood coping strategies in developing countries as it will shape the future of specific, strategic and effective adaptation strategies (Enete et al., 2016). The lack of this understanding will result in unsustainable interventions that will be costly and not accepted within the communities that such interventions are meant to assist. This thesis therefore unpacks the context of vulnerability, how it is differentiated and various ways in which the affected communities monitor the floods and cope with floods. --- Flood Monitoring and Politics in Flood Prone Areas Flood management consists of various assessments including hazard, vulnerability, exposure, risk, early warning system, damage and risk mitigation planning (Rahman and Di, 2017). Rahman and Di (2017) suggest that there is a relationship between several factors that determine vulnerability and adaptive capacity of vulnerable communities. However, the study does not explicitly indicate how these factors interact to create vulnerability. Instead, it focuses on major factors that exacerbate vulnerability to floods and not adaptive capacity issues, including flood monitoring. Firstly, authors suggest that tropical countries have high flood risk and low financial and institutional capacity to afford ground-based monitoring of rainfall and river discharge, hence are more vulnerable to floods (Rahman and Di, 2017). Most developing countries use remote sensing which is sometimes not adequate enough to provide timely early warning information that led to timely evacuation and planning in most cases. Remote sensing provides information in changes in the soil moisture content over time and is usually used to determine and predict flash floods. However, remote sensing technology is not affordable to all countries. Some countries; in particular developing countries, have inadequate capacity to obtain updated remote sensing technology and space technology therefore flood management in developing countries will remain a challenge (Rahman and Di, 2017). Raju et al., (2016) indicate that remote sensing technology has advanced and is providing all the relevant information that would enhance effective planning for flood management. The soil moisture levels that are depicted over time predict the changes in soil moisture thereby predicting flash floods and floods in flood prone areas (Ibid). It is clear that flood management requires a huge investment in order to obtain effective early warning information that will reduce the effect of disasters, including floods. In flood prone areas however, where this research took place, communities use their local knowledge to monitor floods. In addition, the meteorology department does not have adequate finances to procure advance weather monitoring equipment. This thesis therefore informs flood monitoring research on how floods are monitored locally and how the current flooding trends require a change even in ways in which floods are monitored at local level in order to cope with the floods effectively. Remote sensing is an option for most developing countries; however, it depends on the willingness of the ruling governments to prioritise spending on the flood management in the national agenda. Affordability of the technology in this case is subject to various factors, including, but not limited to, national earnings and taxes, which suggests that flood management is also political. In Ghana for example, the government authorities do not actively stop people from settling informally in flood prone areas even though the authorities do not invest in flood mitigation measures in those areas (Amoako, 2016). The government's priority is to win votes from people who live in the flood prone areas (Ibid). This is the political nature of ruling governments where the focus is to remain in power, regardless of their obligation to make certain that people are safe during the floods. This also suggests that the ruling government benefits more from allowing people to live in flood prone areas because it gets more votes from the people who live there without being responsible for disaster risk infrastructure. This is a political strategy which is socially unjust to people who live in the flood prone areas. Although political and economic dimensions of governmental decision-making may differ drastically from context to context, these findings indicate how governments in some developing countries prioritise their agenda to win more votes during elections. These findings also suggest that there is a lack of proper discussions between the government and communities that are affected by floods. The people who are affected by the floods are not given an opportunity to negotiate for government led services to reduce their vulnerability to floods. As indicated earlier on, vulnerability to disasters is also socially constructed and influenced by political relations and other socioeconomic factors (Methmann and Oels, 2014). The lack of dialogue and poor government priorities in developing countries exacerbate vulnerability of those people who live in flood prone areas. This thesis reveals how political decisions and public disaster policies interplay together with other social and economic factors in the flood prone areas in order to understand the complexity of monitoring the floods and local adaptation in those areas. This thesis acknowledges the differences and complexity in governance institutions at different levels, government, community and household level. Some adaptation initiatives are delivered through government departments and established non-governmental and civil society organisations whilst other local adaptation practices are implemented by communities themselves through their particular norms, beliefs and culture. Therefore, further studies, including this research in different socio-economic, political, cultural and geographic settings will facilitate substantiated debates on vulnerability and local adaptation under various institutional arrangements in context. --- Barriers and Limitations to Effective Adaptation to Climate Variability 'The adverse effects of natural hazards on people do not only lie within the hazard itself, but derive from the position of these individuals within social, economic and political relations' (Methmann and Oels, 2014, p278). Methmann and Oels (2014) indicate the main factors that put vulnerability and local adaptation in context. This thesis unpacks these social, economic and political factors in context in order to understand the dynamics of vulnerability and how the various factors affect local adaptation. Islam et al., 2014b;Jones and Boyd, 2011;Urwin and Jordan, 2008 stress that unpacking how and why each of those underlying factors affect vulnerable communities would inform climate related policies, reduce vulnerability to climate variability and promote successful implementation of adaptation programs at local level. In addition, various scholars emphasise that understanding the complexities of vulnerability and local adaptation in context would promote transformational adaptation which is context specific and would effectively address adaptation challenges thereby reducing vulnerability to the negative effects of climate variability (Manuel-Navarrete and Pelling, 2015;O' Brien et al., 2012;O'Brien and Selboe, 2015a, 2015b, 2015c;Pelling, 2010;Pelling et al., 2015). In some developing countries, climate change and other national policies are poorly mainstreamed during implementation of projects and programs (Stringer et al., 2009). Lack of comprehensive policies at national level results in many policies that become difficult to be mainstreamed (Ibid). Some of the most important issues that need to be addressed by such policies are neglected such as the flow of money and other livelihood benefits, who is affected more, why they are more vulnerable than others, who is responsible to benefit from the implementation of such policies, under what agreed criteria and justification for purposeful exclusion (Shackleton et al., 2015). Malawi has many policies that conflict with each other. For example, the agriculture policy promotes farming close to water sources for irrigation purposes whilst the water policy indicates that farmers should not farm close to water sources to avoid water pollution and environmental degradation around the water bodies. (Government of Malawi, 2010). The inconsistency in the policies has brought confusion amongst the public and those that facilitate the implementation of development projects. The lack of clarity on such policies also affects the implementation of adaptation projects and in some cases also exacerbated exposure to floods due to soil erosion and degradation near the river banks. According to Pasquini et al. (2013), there are various factors that affect the mainstreaming of climate change issues some of which include lack of knowledge and understanding of climate change issues and lack of political will by those in authority at both constituency and local government level in Southern Africa. Globally, the focus and priorities set by ruling parties determine the major support from the national budget and other resource allocation. The national budgetary allocation towards a particular theme is where more funds at national level are allocated to the theme of interest at a particular point in time (Nagoda & Nightingale, 2017). This mainly depends on the priorities of the current presidents and/or ministers although some governments have more decentralized power structures than others. This also suggests that in some cases environmental issues are not given a priority by most politicians because most politicians focus on issues that will give impact or results in the short term so as to win people's favour and votes. Many environmental projects will usually yield benefits in the long term. For example, planting a tree will take a number of years before benefits are realised. Due to this, they are not given a priority amongst ruling governments as well as the majority of people who have crucial needs that need immediate interventions (Saito, 2013). Inter-ministerial coordination supported by the highest levels of government is more relevant to the success of mainstreaming and integrating climate change issues into the national development priorities Ibid). This also suggests that if inter-ministerial coordination is supported, district and local coordination will also be improved since these local institutions generate from the line Ministries. Therefore, this argument indicates that proper coordination is relevant to enhance effective implementation of adaptation projects at local level and programs at national level. Kosamu (2013) states that there are inadequate human and financial resources to support climate change adaptation programs in most developing countries. Not many people have the adequate skills, qualifications and experience in climate change adaptation programs hence lack effective delivery of significant contributions and outputs in climate change programs. This is in relation to cases where practically, modern science and technology methods and mechanisms are being applied as opposed to the local/ indigenous coping strategies. In Malawi, there is also little or no interaction and collaboration between climate change stakeholders such as NGOs, Civil Society Organisations and the government as evident during disaster needs assessment and response which results in lack of collective commitment in implementing climate change adaptation programs (DoDMA, 2015b). Furthermore, the post disaster assessment reveals that collaboration in disaster related issues will enhance adaptive capacity in the flood prone areas by ensuring that vulnerable communities are prepared for disasters (Ibid). However, the ability of the community or household to be able to cope with the floods or other disasters is complex and requires a multi-sectoral approach that empowers communities to be resilient to the floods in the long term. Understanding the political and social barriers that have worked against successful adaptation to climate variability is vital to assist the government and other development actors to determine distribution of costs and benefits in society (Ensor et al., 2015). For example, global climate change effects that were addressed intrinsically exacerbated social injustice through the distribution of benefits and burdens between those who create the environmental problems and those who are burdened with the consequences of the environmental changes (Page, 1999). In addition, the social policies made create intergenerational injustice that will continuously disadvantage those who suffer more and who are not capable of dealing with the environmental changes (Ibid). The intergenerational injustice, unfair and injustice distribution of climate variability burdens and the misrepresentation of vulnerability to climate variability has been noted in the case of developing countries as they continue to suffer consequences of extreme weather events that they did not create, whilst the rich continuously pollute the environment and claim that carbon trading is a solution to the problem (Bachram, 2004). In such international policies lie the realities of social equity, role of power, responsibility and rights which reveal that poor communities will continuously become more vulnerable to climate change effects whilst the richer communities will continuously contribute to global environmental changes (Bohle & O'Brien, 2006). However, Maguire and Lewis (2012) indicate that there are attempts made to address such injustices even though it is still not clear how they could be explicitly tackled. This section therefore illustrate how dynamic and complex vulnerability issues are and how dealing with them is challenging at all levels. Solutions to climate related problems therefore should not be very technical because there can never be a single way of addressing the challenges, but rather a combination of technical, social, economic, political and cultural strategies. The unclear ways of addressing such global injustice suggest that these inequalities will continuously be reflected in many adaptation programs hence creating a challenge in meeting the global, regional, national and local adaptation plans whilst exacerbating vulnerability to climate variability (Page, 1999). The author indicates that global justice does not exist and the social injustice that has been passed on from generations. The burdens of climate change will have to be shared with future generations inevitably and hence a need to plan for future adaptations to reduce vulnerability. This argument presents a case that there is need to understand the vulnerability context now to plan better for future adaptation. In some cases, vulnerability is transgenerational and communities who are vulnerable now will continuously be vulnerable to climate variability, hence empowering them to adapt to climate variability is vital. These arguments both demand the need to understand and invest in climate change adaptation immediately to lessen the impact of climate variability in the future. Environmental injustice has partly exacerbated vulnerability in communities that have limited capacity and influence to deal with the negative effects of environmental changes (Kirshen et al., 2012). In the case of East Boston Massachussets, a flood prone area, it was revealed that the communities experiencing the floods had no influence and knowledge on adaptation plans, but rather, all the decisions were made by institutions and people in authority (Ibid). This reflects some of the reasons that have caused failure in climate change adaptation programs in various communities and countries and also reflects on how vulnerable communities will remain vulnerable to climate change effects (Pelling, 1999). However, Sen, (2009) advocates that the vulnerable should be given a platform to be heard, the realities suggest that the governments and those that have the power, focus on political interests; therefore dialogue over social injustice does not realistically solve the problem (Tully, 2013). Power dynamics is therefore indeed one of the issues that needs to be addressed strategically at all levels to reduce issues of vulnerability to climate variability. Blaikie and Brookfield (1987) stress that environmental problems are complex and hence require a deep understanding of the social and political setting in which the particular environment is surrounded. This led them to develop political ecology theories that seek to understand the dynamics and complexity of environmental problems in their specific social and political contexts. Neumann (2009) acknowledges that the phenomena under investigation can be very complex and interdisciplinary, hence political ecology is important because it involves a multiscale analysis that includes political economic analysis, historical analysis, ethnography, discourse analysis and ecological field studies. In a recent study, Middleton et al. (2018) use a political ecology lens to understand the relationship between migration, vulnerability, resilience and social justice that is associated with flooding in an environmental, social and policy context in Southeast Asia. The authors focus on ways in which floods play a role in the livelihoods of the migrants in Southeast Asia in contrast to flood related policies that refer to floods as a sole driver of migration (Ibid). This thesis focuses on environmental migration as an alternative adaptation strategy against floods by smallholder farmers that live in flood prone areas. The discrepancy on perceived vulnerability to floods between the public and people affected by the floods, and temporary migration during the flooding season has been investigated in the thesis. --- Conceptual Framework --- political ecology theories This thesis presents an understanding of how vulnerability is framed based on self-perceptions that are differentiated based on locations, personal experience, age and gender, amongst other factors. This thesis also reveals the need for transformation in the way adaptation strategies and policies are perceived, designed and implemented to incorporate important aspects of culture, gender and power dynamics at household and community level that form the basis of effective local adaptation to floods. This thesis also makes reference to other scholars that have done similar work adopting political ecology theories and ideas (Adger et al., 2009;Dallman et al., 2013) focusing on vulnerability to climate variability and the dynamics of local adaptation. In addition, the thesis builds on research conducted by scholars who have focused specifically on social transformation as an adaptive strategy that has potential to reduce vulnerability to climate variability (Manuel-Navarrete and Pelling, 2015;O' Brien et al., 2012;O'Brien and Selboe, 2015a, 2015b, 2015c;Pelling, 2010;Pelling et al., 2015). These authors stress the need to understand the politics and how decisions are made within the vulnerable spatial locations such as flood prone areas and areas that are prone to war and other socio-ecological disturbances. The authors also emphasise the need to reduce vulnerability to climate variability through supporting appropriate sustainable livelihoods options that also enhance the adaptive capacity of the vulnerable communities. This thesis thus builds on the understanding that vulnerability to climate change is created by the interaction of social and natural systems, and that adaptation initiatives should incorporate cultural, political and social factors to reduce people's vulnerability to climate variability (Pelling, 2010). This research addresses the gap in knowledge and understanding of how the historical, cultural, social, economic, environmental changes, power dynamics, and political factors have exacerbated vulnerability of smallholder farmers living in flood prone areas using Nsanje District as a case study. Pelling (1999:250) suggests that 'vulnerability for individuals and social groups has three components; exposure, resilience and resistance. These components are simultaneously the products of political and socioeconomic structures and the capacity of individual actors and social institutions to adapt to hazard stress.' The focus of this thesis is to understand the extent to which smallholder farmers are exposed to floods and the opportunities that lie within the flood prone areas that have shaped their livelihood options and survival during and after the floods. This study is important because it illustrates how climate related decisions are made by community members in the flood prone areas and how the discrepancy between public perceptions and self-perceived vulnerability has resulted into inappropriate policy options that do not reflect the extent, context and intensity of vulnerability to floods. This thesis further seeks to understand why the smallholder farmers that live in flood prone areas are not able to cope with the floods in a way that their livelihood is not affected, despite huge investments over US$50 million in climate change adaptation initiatives by the government of Malawi and development partners including the World Bank, Global Environment Facility, World Food Program, Food and Agriculture Organisation, European Union and various United Nations departments (DoDMA, 2015a). This thesis explored how these smallholder farmers understand local adaptation projects, what it means for them and what is involved. In addition, to understand and learn from them the factors that would assist the smallholder farmers to be able to cope with the floods, recover after the floods and become resilient to the floods in the long term. Pelling (2010) stresses that understanding the economic, environmental, political, technical, institutional, social and cultural transformations is vital to reduce complex vulnerabilities to climate variability. This thesis therefore provides a case study of how these multiple factors interplay in the flood prone areas to create, exacerbate and reduce vulnerability to floods. --- The Local Adaptive Capacity Framework (ACCRA, 2010) Figure 1 The Local Adaptive Capacity Framework (ACCRA, 2010) Accessed at: http://community.accraconsortium.org/.59d669a8/research.html The Local Adaptive Capacity Framework is similar to the adaptive capacity wheel, (figure 2) in that it acknowledges the relevance of resources in forms of assets. Assets are very important to enable the households and communities to cope with floods In addition, it acknowledges that information and knowledge facilitates learning and room for autonomous change which facilitates innovation in which the households and communities deal with the dynamic and complex effects of floods.All the frameworks therefore indicates that communities need to be empowered economically, socially and politically in order to effectively adapt to climate variability. The frameworks also illustrate that communities and households need a multidisciplinary approach in order to deal with the effects of climate change. Policy makers therefore need to understand the dynamic and complex nature of communities in order to effectively design and implement adaption programs that will reduce vulnerability to climate variability whilst enhancing the adaptive capacity of the households and communities. The Local Adaptive Capacity framework, just like other adaptive capacity frameworks that measure and assess communities' vulnerability and adaptation, is important because it also helps researchers to understand what influences communities to adapt to floods (Adger et al., 2004). In Ethiopia, Uganda and Mozambique, the framework was introduced to measure how development interventions impact on adaptive capacity at the local level and it is mostly used by International Non-Governmental Organisations and other partners in development (Ludi et al., 2014). Usually, these assessments are done on formal institutions that govern communities and households. Formal institutions enable access to various resources within the communities including land and water, and are partly held responsible for failure in adaptation. However, it is important to understand the systems in which the context or arguments can be made, based on the social, political and ecological factors. For example, in Nsanje District in Malawi, an irrigation scheme rehabilitated after participatory consultation has benefited one community and has exacerbated flooding problem in another community (Harrison and Chiroro, 2016). This presents a situation that needs to be considered when coming up with some of the adaptation and resilience projects as they might be harmful to other communities, exacerbating their vulnerability whilst economically empowering other communities. Perhaps this also suggests that there can never be a perfect adaptation option between the government and people who are affected with the floods, and among different communities in the flood prone areas. In some cases social injustice is inevitable when designing climate change adaptation programs because vulnerability is complex and linked to other areas geographically. Addressing vulnerability issues in one area might therefore unintentionally affect other people in other spatial locations. This thesis explores the dilemma that exists in the flood prone areas, the associated political and economic issues, how decisions are made, who is consulted, who benefits and the basic understanding of environmental and social management of risks and plans that are involved. --- Chapter Summary This chapter has presented an exploration on the impact of floods on sustainable development, who is affected, why they are affected and how adaptation can be designed in order to reduce vulnerability to floods. Based on the review, it is clear that vulnerability is multifaceted and is generated by multiple factors including biophysical, socio-economic, political, cultural beliefs and norms, and other factors that are specific to a particular setting. --- CHAPTER THREE --- RESEARCH METHODOLOGY --- Introduction This chapter describes the research methodology under which this research was undertaken 'Importantly', this chapter describes the data that was collected. A mixed method approach was used to collect data relevant for this research. This is where both quantitative and qualitative data is collected to understand various research topics in context (Teddlie and Tashakkori, 2009). I used a household survey, key informant interviews, focus group discussions and participant observations to collect data for this research. The chapter provides details of the type of data that was collected, how it was collected, why it was collected that way and how it was analysed in response to the research questions. study limitations are also highlighted to indicate the jurisdictions and boundaries within which this research was conducted, and within which its findings apply. --- Research Design In this research, I used a case study design to explain, describe and explore selfperceived vulnerability to floods and local adaptation in Nsanje District in Malawi, under an Interpretivist epistemology at both household and community level (Crowe. et al., 2011 andYin, 2009). Interpretivist epistemology is an understanding that there are several realities depending on the context through which the reality is constructed and that the reality can never be objectively determined and perceived (Berger and Luckman, 1967;Carson et al., 2001;Hirschman, 1985;Hudson and Ozanne, 1988;Lincoln and Guba, 1985;Neumann, 2000). Hudson and Ozanne (1988) define ontology as the nature of reality that is characterised and influenced by several factors within a particular setting and context. Epistemology on the other hand has been defined as the way in which the reality is captured by a researcher within the particular settings that forms a reality, it is a relationship that exists between the researcher and the situation on the ground that form a reality (Carson et al., 2001). The interpretivist epistemology approach helped me as a researcher to understand how the communities living in flood prone areas perceive their vulnerability to floods. The social realities obtained enabled me to also understand how the communities have been able to cope with the floods for several decades and why they have refused to relocate to safer areas that are not prone to flooding. In my research, it was important to understand the motivations, significances, explanations and other subjective experiences that are applicable and make sense to the smallholder farmers that live in the flood prone areas. This research therefore needed to adopt the interpretivist approach it did and acknowledge that these findings are true to the communities living in the flood prone areas without any preconceived beliefs and ideas that could have shaped the findings. --- Population of the study The research areas for the study were sampled out of the whole community that live in the flood prone area through a community mapping exercise. The Community mapping exercise enabled me to identify areas that experience severe flooding to determine the study sites to focus on. Furthermore, data collection involved a series of key informant interviews, focus group discussions and a household survey where a unique sampling process for each method was undertaken, as will be discussed later on in the chapter. Firstly, I will provide details of the community mapping exercise. --- Community Mapping I conducted a community mapping exercise at the District Council Office, with the help of the District Civil Protection Committee (DCPC) members, including the District Disaster Risk Management Officer, The Environmental District Officer, The District Environmental Health Officer, The District Social Welfare Officer and Area Civil Protection Committee representatives. The Area Civil Protection Committee representatives are based in the flood prone areas and more knowledgeable on the maps and environmental changes that have occurred due to the floods. The aim was to identify, with reference to the district maps, areas of high vulnerability in relation to exposure to floods and droughts, from their experience of flood response. This exercise was done during the first planning meeting to enable me to allocate resources accordingly and also to inquire how the areas where accessible. On the maps, 46 group village headmen, under traditional authorities, Mlolo, Mbenje, Ndamera and Nyachikadza were identified by United Nations Disaster Assessment Coordination (UNDAC) team as being at risk of floods because of their location However, even though all the areas are affected by the floods, the magnitude of the impact is not the same, according to DoDMA (2012), and it was also indicated by the DCPC Members. A combination of community representatives and district officials working in disaster prone areas was essential because there was a discussion on which areas where worse off and better off, also adaptation activities that are being implemented in the areas to reduce vulnerability to floods. During this mapping exercise, the places where the people who are affected by the floods migrate temporarily during the floods were indicated. This exercise helped me to strategically identify participants for key informant interviews to understand the interdependence between the smallholder farmers living in the flood prone areas and those living in upland areas. Based on the community mapping, the DCPC also indicated non-governmental organisations that are working in the areas to reduce the vulnerability to floods. From the community mapping, it was shown that there were several climate change adaptation initiatives in TA Mlolo that included irrigation farming; whilst there was none in TA Nyachikadza. According to the DDP (2010), the areas that are under TA Nyachikadza (Part of the Blue Box on Map 3) were declared uninhabitable due to the severity in exposure to floods. The area is surrounded by a marsh hence with the frequent flood experiences, smallholder farmers were advised to relocate to upland areas through DoDMA Officials, to which they have refused. The DCPC members were able to indicate some areas that have been complying by adopting new conservation agriculture and climate change related activities more than other areas. This, according to the committee, enables them to identify new areas that need interventions in terms of coping with the adverse weather effects. In addition, it enables them to monitor progress and identify gaps in their approaches. This exercise was beneficial to me because it enabled me as a researcher to have an idea of the places I was going to conduct my research and also clarified and deepened my understanding on some of the community behaviours that I noted whilst collecting data. The community mapping exercise was also done to explore and relate what communities in flood prone areas do to reduce their vulnerability, how they become innovative and what decisions they make in relation to reducing their vulnerability. This was done to be able to compare what is documented in district documents such as the socio-economic profiles and development plans with what I would find in the field. Since 46 Group Village Headmen and their people were affected by the floods, I had to determine the actual research sites which also determined who would be eligible to participate in the research. At District level, the areas are divided into Traditional Areas, which are then sub divided into Group Villages and then Villages. In other words, a Group Village comprises of several Villages, a Traditional Area comprises of several Group Villages and a District comprises of several Traditional Areas. --- Sampling Procedure A purposive sampling was used to determine which individuals and households were to participate in the research. According to Palinkas et al, (2013);Palys, (2008) and Suri, (2011) purposive sampling is a sampling technique whereby the sample is chosen based on a certain criteria determined by the researcher's interest, in order to obtain applicable data in relation to significance, meaningful understanding and depth of the researched matter. Out of the 46 Group Village Headmen, 10 Group Village Headmen were identified as being severely hit by the floods based on the flood assessments and extent of exposure to floods. I then narrowed my research to focus on communities living in these 10 group village headmen, based on the guidance and prior knowledge obtained from the DCPC members. The same applied for Nsanje District as described in the previous section. However, not every community living in the 10 Group Village Headmen participated in the survey. Neither did all the communities along the Shire River in Traditional Authority Kunthembwe participate in the research. I had to determine the actual sample size for the research, according to the different methods that were undertaken. --- Sample size In terms of sample size, (Table 3) this research involved 227 households during the household survey and 12 focus group discussions (FGDs) with communities in addition, 57 key informant interviews (KII) were conducted. I acknowledge that the sample size varies with the type of research and size of population so there can never be a specific and fixed size. The household survey was conducted before the 2015 floods whilst the focus group discussions and key informant interviews were conducted after the floods. The floods came earlier than expected, but only after the sampling processes identified those to take part in key informant interviews and focus group discussions. I reviewed literature for sample sizes for various research as per academic researchers' experience and expertise. There has been debate on the sample size that should be considered adequate for qualitative research depending on the type of study. Whilst Charmaz (2006) indicated that 25 is the right size, Green and Thorogood (2004) suggested that a sample size of 20 should be adequate. Ritchie et al., (2003) suggested that a sample size close to 50 should be adequate because with such a bigger sample, one is certain that all the information that was expected to be captured will have been captured. In addition to this, they claim that it is very likely that no new concept will have been missed from all the 50 research participants (Ibid). Sandelowski (1995) suggested that if the sample size for research is relatively big and complex, the research presents a new and richly textured understanding of experience. Based on this understanding, I am confident that the data for this research that has been collected using mixed methods; through household survey, key informant interviews, focus group discussions and participant observation, has adequately answered my research questions. --- Data Collection Baxter and Jack ( 2008) indicated that a case study is an approach that helps a researcher to explore a phenomenon within its context, using a variety of data sources. In order to understand the cases thoroughly, data was collected through both quantitative methods, namely a household survey and qualitative methods, namely key informant interviews, focus group discussions and participant observations (Crowe et al.,. 2011). I used multiple sources of data collection so that I should also increase credibility of my research, also referred to as methodological triangulation in social science research. Methodological triangulation is a way in which multiple data collection methods are combined when collecting the same type of data, at different places and at different times in order to increase the strength of the information that is being collected in social science research (Barbour, 2001;Mason, 2002;and Stake, 1995). In this research, people living in three communities participated, two communities in Nsanje, (The Northern, top part of Nsanje district, TA Mlolo and Mbenje and the Southern, bottom part of Nsanje district, TA Nyachikadza and Ndamera). The comparative study was very important because it helped me to deepen my understanding of the different cases and be able to relate, compare and contrast critically (Miles and Huberman, 1994). Understanding how communities perceived their vulnerability to floods and how they cope with the environmental changes in different contexts also deepened my understanding and application of political ecology theories. However, in each community, based on the communities' definition of wealth status, the households were grouped according to their socio-economic status. There were three groups based on whether the households were'very poor', had 'average income', or were'relatively rich' using participatory wealth ranking which is described in the next section, under household sampling. --- Quantitative Data Quantitative data for this research was collected through a household survey using a semi-structured questionnaire i.e. household survey (Appendix 1). However, the questionnaire also contained a section where qualitative data was collected. This method was used to determine how communities perceive their vulnerability to floods, how they rank their risks in the flood prone areas and what they are doing to cope with the floods. This method was also used to explore livelihood options and community behaviours that needed to be deeply understood through participant observation, focus group discussion and key informant interviews. --- Household Survey A total of 227 households were involved in the household survey, with the help of research assistants, whose positionality is described under the research assistant section. Research assistants spent 45 minutes to 1 hour with each household. A total of 52 farming households participated in each of the 4 traditional authorities in Nsanje, District; traditional Authority Mlolo, Mbenje, Nyachikadza and Ndamera. In Traditional Authority Kunthembwe, only 19 households participated in the survey. The sample size was smaller than the other traditional authorities in Nsanje because firstly, the affected area was relatively smaller than the one in Nsanje District. Secondly, the survey was conducted to understand how the communities living in this different catchment of the Shire River Basin, Middle Shire perceive their vulnerability to floods and how they adapt. This area is also prone to floods, the communities have never experienced severe floods before January, 2015. Communities in Traditional Authority Kunthembwe experience droughts and had started practicing irrigation agriculture in 2013. Usually, there has been excessive run off from this area that contributed to flash floods in Nsanje (UNDP, 2009). However, in January, 2019, communities lost all their farm land due to the rainfall, which according to the data I collected in Nsanje had contributed to the change in the soil structure. Therefore, I decided to also conduct a household survey to understand how communities perceive their vulnerability to floods and learn from them of initiatives they are taking to mitigate the effects and manage the flood risk for sustainability. In addition, I also wanted to find out how they perceive environmental degradation, flooding and agriculture. Specifically, the household survey was used to collect information on how households understand the incidents of floods and droughts, what causes them and what the associated effects are. The survey was also used to obtain information on how households identify risks and the associated security issues in terms of their perception. This enabled me to understand the motivation and willingness to participate in activities that reduce their vulnerability. I only referred to activities that are specifically designed by the government and partners in development to increase household resilience to floods and droughts. Households were also asked to weigh the risks which they identified in relation to the issues they face, and rank each risk on a scale that ranges from very mild to extremely severe. Furthermore, the individuals

A section of Nsanje District has been experiencing floods for over a decade and was declared a flood prone area the same time (GoM, 2006). Therefore, firstly, this research seeks to understand how the communities that live in the flood prone areas perceive their vulnerability to floods and how some of them have been coping with the floods since 1952, just over 6 decades. The overall aim of the research was to understand the impact of floods on sustainable development and how communities living in flood prone areas perceive their vulnerability to floods and climate variability, to investigate the power dynamics at household and community level and to explore the complexities associated with local adaptation programs in the flood prone areas. This research contributes towards framing vulnerability based on the perception of the different groups of people that are vulnerable to climate variability and have experienced environmental changes throughout their everyday lives. The communities living in the flood prone areas perceive floods as part of their livelihood. The communities have experienced floods for over 5 decades and have been implementing adaptive strategies to help them cope with the floods during this time. Temporary migration is one of the effective adaptive strategies that the communities living in the flood prone areas implement during floods.

sanje (UNDP, 2009). However, in January, 2019, communities lost all their farm land due to the rainfall, which according to the data I collected in Nsanje had contributed to the change in the soil structure. Therefore, I decided to also conduct a household survey to understand how communities perceive their vulnerability to floods and learn from them of initiatives they are taking to mitigate the effects and manage the flood risk for sustainability. In addition, I also wanted to find out how they perceive environmental degradation, flooding and agriculture. Specifically, the household survey was used to collect information on how households understand the incidents of floods and droughts, what causes them and what the associated effects are. The survey was also used to obtain information on how households identify risks and the associated security issues in terms of their perception. This enabled me to understand the motivation and willingness to participate in activities that reduce their vulnerability. I only referred to activities that are specifically designed by the government and partners in development to increase household resilience to floods and droughts. Households were also asked to weigh the risks which they identified in relation to the issues they face, and rank each risk on a scale that ranges from very mild to extremely severe. Furthermore, the individuals were asked how as a household, they come up with decisions that reduce their vulnerability to floods and droughts, the assumptions they make, and who they involve and consult. There was a sampling process that was used to determine who was to take part in the survey as described in the next section. --- Household Sampling process The list of households periodically affected during the floods in the targeted areas was collected at the Office of Disaster Management Affairs under the Nsanje District Commissioner's Office in Nsanje. I also followed the same procedures in Blantyre District. The list was then verified at respective traditional authorities through the Area Civil Protection Committees (ACPC). The list was verified to make sure that it was up to date since there are some families/households that relocate to other areas for various reasons, including, but not limited to marriage and job opportunities. In addition, I also wanted to be certain that the people that were indicated on the list indeed live in the various villages as sometimes, the lists are altered for political reasons for distribution of campaign and other materials. The other times, the lists are combined strategically to facilitate implementation of donor projects and other requirements during relief distribution. The list is based on number of households that form a particular village in order to facilitate the distribution of developing projects and humanitarian aid. A household comprises of individuals that feed from the same pot and consider themselves to be a family. I used stratified random sampling to identify research participants in all the three groups. The stratification was important so that I should collect comprehensive data from all the social classes to understand how the households perceive their vulnerability and local adaptation. The variation in perception and local adaptation strategies was also important to understand how the social inequalities and social injustice play a role in perceived vulnerability and local adaptive capacity to the floods, droughts and extreme weather events. In addition, the inclusion of all social groups ensured inclusiveness of research participants and eliminated social biases that naturally exist in rural communities. Within each group, 17 households were selected at random. In order to do this, all households in each of the three groups was allocated a number which was written on a small piece of paper. The pieces of paper with numbers on them were rolled and placed in a small basket. Some members of the community, including the chief, and area and village committee members then picked the small pieces of paper. Only those households whose numbers were picked were eligible to be interviewed and were interviewed after obtaining their consent. Out of the selected 17 households per group, survey participants alternated between males and females to balance the gender of the research participants to get an equal number of males and females. However, there were 109 Males and 118 females who contributed to the households. There were more female research participants because there were more female headed households in the poorest of the poor groups than male headed households. Out of the 227 households, 70 households were poor based on the participatory wealth ranking exercise; from which 40 were female headed. --- Data entry and analysis I used the Statistical Package for Social Sciences (SPSS) package for data entry and simple analyses, mainly frequencies, mean, mode and simple correlations. I designed the data entry template and together with a data entry clerk, entered all the quantitative data that was collected during the household survey. Continuous data was summarized using measures of central tendency such as means, mode, median and range. Categorical data was summarized using frequency, counts and proportions. Inferential statistics was also used sparingly especially when applying correlation analyses. The findings that were drawn from the sampled research participants were used to interpret certain actions and describe relationships that exist between different factors and processes in the flood prone areas. Using NVIVO, I grouped the responses from open ended questions to identify themes that were critical during the research to enhance my understanding of how smallholder farmers perceive their vulnerability to floods and how they are coping. I also used NVIVO to identify themes from the qualitative data that was collected through key informant interviews and focus group discussions. The full details on qualitative data and analysis is given in the next section. --- Qualitative Data Qualitative case study methodology enables researchers to study complicated phenomenon within their context using descriptive data (Sandelowski, 2000(Sandelowski,, 2010)). Guiding questions (appendix 2) were used to guide the FGD and KII. The aim was to collect similar information but at different levels to enhance triangulation of the data. The focus in this case was on collective community perception to floods, droughts and extreme weather events, and local adaptation. In addition, I wanted to explore community responses to floods and droughts, and the social networks that help communities to be resilient to floods and droughts as part of adaptation strategies. The aim was to be able to differentiate the responses between households and communities, and also to triangulate and deeply understand some of the information that was collected during household interviews and participant observation. For example, to deeply understand why temporary migration as opposed to permanent migration is relevant in the flood prone areas. In some cases, some of the information collected during the focus group discussion was even more deeply understood during interviews with key informants. More especially, information on historical contexts of the floods, environmental refugees, inter-dependence between communities living in flood prone areas and those living uplands, social networks, and transfer of knowledge and information between communities in relation to adaptation to floods. --- Key informant interviews I personally conducted 57 KII with selected small participant and key stakeholders at the District level, using a question guide (Appendix 3). KII took approximately 1 hour and 30 minutes. At the community level, I interviewed, 1 traditional authority, 2 village headmen, 2 members of the civil protection committee and 5 smallholder farmers living in nearby communities, not affected by the floods themselves but have the interdependence relationship with the flood victims. For the key stakeholders, I interviewed 1 District Commissioner, 1 Disaster Management Affairs Officer, 1 Environmental Officer, 1 member of the District Civil Protect Committee, 1 District Agriculture Development Officer and 2 officers working in relief agencies, International Non-Governmental Organisation with a focus on climate change. I also interviewed a total of 23 participant who had been experiencing the floods for over a decade and have lived in the flood prone areas since birth. All these participants were identified strategically through the sampling process. In terms of gender, there were more males than females and partly it was because the males held the strategic positions that I targeted. In addition, even at community level influential people that were identified and those holding positions in civil protection meetings were men. --- Key Informant Sampling A snowballing process was used to come up with key informants that participated in the interviews based on professional expertise and community knowledge on vulnerability to floods and local adaption (Denscombe, 2003). At District level, the list of key stakeholders was strategically identified based on the positions that the individuals held. The positions had an implication on the significant role they played in disaster related issues, including the floods. I also used my personal professional knowledge through my previous role as a civil servant in the area to identify some of the key people that were interviewed at professional level. At the research sites, the first people to be contacted were the chiefs and civil protection committee members who gave me names of participants who could provide the information that I was looking for. The selection of the communities was made based on the communities experiences and understanding of the floods and local adaptation in the community. Since the wealth and social classes played a major role in the research, these people directed me accordingly, so I had a balance of views from different social classes in my research sites. Through the process, communities living in neighboring villages to the research sites were also identified as playing a major role in local adaptation to floods. It was indicated that the communities in the neighboring villages have extensive social networks with the flood victims and provide shelter to the flood victims during the floods. Information collected at the district council also indicated that there is an interdependent relationship between the affected households and those from neighbouring villages. Therefore, I included key informants from the neighbouring villages to the research sites to have a deeper understanding of how these different groups of participants relate and how they depend on each other. In addition, I wanted to understand if over the years they also have made deliberate changes in their livelihood and adaptation initiatives due to noticeable environmental changes. I also wanted to investigate if there are any long term strategies that are in place in terms of the support they offer to the flood victims, in addition to the temporal, short term strategies. Additional information to deepen my understanding was collected through focus group discussions with selected communities. --- Focus Group Discussions A series of 12 focused group discussions were conducted with smallholder farmers living in the flood prone areas. Three focus group discussions were conducted in each of the four study areas using guiding questions (Appendix 2). The discussions took approximately 1 hour. The aim was to determine how the communities collectively perceived their vulnerability to floods and how they were adapting to floods locally at community level. The first group comprised of men only, the second group comprised of women only and the last group comprised of youth (a combination of both young men and women, with deliberate efforts to balance gender representation). The gender groups, of male and females, were separated in order to give women room to discuss the vulnerability issues freely. Due to the patrilineal tradition, in most cases, women do not raise their concerns in the presence of men and are not comfortable to disagree with men's point of view. The youth groups however comprised of young men and women because due to the nature of their age, less than 35 years old, in most cases young women are able to talk freely even in the presence of young men unlike the older generation. The focus group discussions were conducted with 9 participants in 1 women-only focus group discussion in Nsanje and 11 participants in the remaining 11 focus group discussions. Only participants that were randomly selected through a list that was collected at the district council and verified at the TA participated in the focus group discussion. --- Sampling Process For the FGDs, a mixture of participnts in different social classes were identified through purposive sampling. The FGDs enabled me as a researcher to understand some community approaches and concepts that might have been missed out during the household survey and key informant interviews but were relevant to understanding participants self-perceived vulnerability to floods and local adaptations using the political ecology theoretical framework. In addition, I also observed some of the behaviours in the flood prone areas to have an understanding of the study areas and context in which they were applicable. --- Participant Observation Participant observation refers to studying households within a community, whilst living with them, with the aim of understanding the meaning behind their actions (Easterby-Smith et al., 2008). The participant observations lasted two weeks in each community in order to deepen my understanding in terms of their livelihoods, their perception towards vulnerability to floods and droughts and how they make livelihood decisions. Furthermore, it enabled me to understand how participants see their world and how they construct their reality, which then enabled me to understand the findings and discuss in context (Thorne, 2009). --- Data Entry and Analysis For all the qualitative data, I firstly gathered the data based on the different methodology. Using mostly NVivo 11, I entered all the data that I collected based on the themes that I had identified i.e. self-perceived vulnerability, vulnerability and risk, crop production and post-harvest handling, and local adaptation. For the responses under the themes, I provided different codes and colours to distinguish the different responses under sub-themes and also to enable me to isolate the responses for analysis. Based on the various responses in different codes, I then grouped similar responses to create other emerging themes from the findings which were the novel contributions of my research. --- Data Analysis Strategy Data analysis for this thesis was led by the research questions that formed the basis of the research. In addition, literature review on differentiated vulnerability and local adaptation in various spatial locations in developing countries partly influenced the way the results and discussion chapters were shaped. Firstly, it was based on how the communities who live in flood prone areas frame their vulnerability to floods. The data under this theme was further grouped and interpreted based on various factors that influenced the differentiation in the responses on how communities perceived their vulnerability to floods, thereby also influencing how they frame their vulnerability to floods, as it is presented in Chapter 5. This analysis adds more knowledge on differentiated vulnerability, mainly illustrating that self-perceived vulnerability is also differentiated therefore self-perceptions at all levels should be incorporated in climate related decisions, policies and programs in order to design programs that would adequately reduce vulnerability to climate variability. Secondly, the data was analysed to explore the various factors that motivated the communities to continue living in the flood prone areas despite the evidence that the areas are exposed to floods hence their livelihood is at risk. This analysis, that is presented in Chapter 6, further illustrates that in some cases, communities also benefit from living in flood prone areas therefore disaster related policy makers should be aware of such benefits and formulate disaster related policies that support the positive attributes rather than only focusing on the negative aspects. UNISDR (2005,2015) focus on the safety of the communities that live in disaster prone areas and suggests that relocation to safer areas is the best solution. However, in some spatial locations, promoting best livelihood options within the flood prone areas reduces further vulnerability to climate variability and is the preferred solution by the communities that are affected by the disasters. --- Limitations of the study Although there was much effort to make certain that all study limitations were taken into considerations, there are a few limitations that need to be highlighted. Firstly, the research being cross-sectional in nature, there are chances that the information collected and explained represent the perceptions of the communities during the particular research period, and may not be true for the other times. This research therefore does not depict changes in community's perception over a period of time and cannot guarantee that there can never be any changes in due course. This study therefore is limited in terms of continuity and following up of participants to deepen the understanding of the continuous effect of climate variability in perceived vulnerability, and behaviour change in addressing the floods, droughts and dry spells. Secondly, there was a gap during the data collecting period, due to floods in some research sites. As a result, some areas became impassable hence we could not visit some of the farms and areas that are extremely exposed to flooding to have an idea of what it means to be farming and living in areas that are extremely exposed to flooding. However, the data that I collected provides evidence that vulnerability to floods is subjective and that self-perceived vulnerability is influenced by various factor and processes that interplay within the flood prone areas. In addition, my study illustrates that local adaptation is complex and that communities cope with floods differently. My study provides an overview of how smallholder perceive their vulnerability to flood and how they are coping with the floods. However, I acknowledge that the extent and context of vulnerability and also the dynamics of local adapting in my study might be different in other areas. Thirdly, I acknowledge that this thesis is based on my understanding of political ecology and 'that therefore the study findings might reflect different arguments from studies that would be conducted using different approaches. In addition, the context within which the issues of gender, culture, vulnerability and local adaptation are discussed in this thesis are subjective to the context of vulnerability to floods and local adaptation in flood prone areas in Malawi; therefore I acknowledge that there are other cases that would reflect these issues differently. The findings from this research however adequately reflects that vulnerability in flood prone areas is exacerbated by floods and other factors that interplay within the areas. It is important therefore to integrate such factors during design and implementation of adaptation programs that aim at reducing vulnerability in flood prone areas. Fourthly, I would like to acknowledge that this research has limitations in the way male and female research participants' emotions are being discussed because I did not focus much on the emotions, but rather the actual responses between the two groups of people during and after flooding which are related to emotions. In addition, the issues of false consciousness in relation to some behaviours that are gender related, myths and beliefs in the case studies can be debatable. --- Chapter Summary This methodology chapter has provided detailed information on how the data for the research was collected and the analytical processes that were undertaken. It has described how the quantitative and partly qualitative data was collected through household surveys and how most of the qualitative data was collected through key informant interviews, focus group discussions and participant observations. Furthermore, the chapter details how SPSS and NVIVO were used to aid analysis of the data that was collected. Finally, the chapter presented the study limitations which provide an indication of the context within which the study was conducted and therefore the context within which findings are valid. --- CHAPTER FOUR: RESEARCH FINDINGS AND DISCUSSION --- Introduction This chapter presents wider public's perceptions of the vulnerability of the smallholder farmers who live in flood prone areas. It also presents findings and discussion on how communities living in flood prone areas in Nsanje frame their vulnerability to floods based on their own perceptions. This self-perceived vulnerability to floods is differentiated based on geographical location and position, personal experience, frequency of the floods, age, gender and emotions. The chapter also presents the factors that have contributed to some communities being more vulnerable to the floods than others, including low socioeconomic status, lack of social networks and inability to cope with the floods. Combest-Friedman et al., (2012) emphasise that households that live in flood prone areas are aware of the increase of rainfall and changes in the intensity of floods hence these people are aware of the associated risks. The important debate however is whether households in the flood prone areas are able to cope with the floods and will be resilient to the floods in the long term. Demski et al. (2017) explored the extent to which personal exposure to floods influences how the affected people perceive their vulnerability to floods and therefore engage in climate change adaptation and mitigation initiatives. Their findings indicate that statistically, those that experience floods have high risk perceptions and are likely to be involved in adaptive actions (Ibid). In some cases, however, Demski et al. (2017) found that there was no difference in perceived vulnerability to floods between those that are affected by the floods and the public. It is important to unpack the extent to which perceived vulnerability to floods is similar between communities that are affected by the floods and those that are not affected within a particular context in order to critically explore the similarities and differences that exist. The fact that in some cases personal exposure does not influence perceived vulnerability to floods suggests that in those instances, communities can continue living their normal lives without taking appropriate actions to adapt to the environmental changes they experience. This chapter therefore seeks to illustrate and understand how communities perceive their vulnerability to floods and to critically examine the factors and processes that influence the perceived vulnerability to floods in order to understand the link between self-perceived vulnerability and local adaption to floods in context. Communities in Nsanje emphasised that floods are part of their lives and livelihoods. The smallholder farmers have been experiencing floods for over six decades, during which they have on occasions had to relocate in the flooding season. It is important to note however that the discrepancy in the number of years that communities have been experiencing the floods is due to lack of specific secondary data in the district council and national documents. The period in the documents is estimated to be just over two decades when the communities in research sites indicate that it is over six decades. This in itself highlights the importance of conducting the research in the selected research sites and to some extent contributes to the discrepancy in perceived vulnerability to floods. This chapter illustrates through qualitative data and quantitative data that there is a need to understand and contextualise self-perceived vulnerability to floods and sustainability The responses from the communities elucidate the realities of vulnerability to floods and point out to the factors that can assist the government of Malawi in the design and implementation of climate change adaptation strategies that reduce vulnerability to floods for people that are affected in various locations. In addition, the insights shared in the district-level interviews reveals that most of the time, the media does not report on the most important topics that form the livelihood of the communities living on the flood prone areas. From the data, it is evident that it is vital to understand the complexity of the lives of the people living in the flood prone areas to successfully address climate change vulnerability and adaptation challenges at policy, program and project level. --- Public Discourse towards communities that Live in Flood Prone Areas Interviews with key stakeholders, including government officials, journalists, officials working in related organisations and some communities living in upland communities indicated that the communities living in flood prone areas are stubborn and enjoy receiving humanitarian aid and support during the floods. 'Communities that live in flood prone areas are stubborn. They deliberately refuse to relocate to upland areas because they know that the government will always support them during floods.' (KII #1, January, 2023). In addition, the government officials, who work at the District Council, indicated that 'We have told them to move upland several times, but they do not move. We do not know how they want us to tell them. Information was officially sent in 2020 and we remind them every now again, they are very stubborn' (KII # 53, Nsanje, June 2022. However, another key informant indicated that 'to be honest, no one understands these communities, maybe they need to be approached to really find out why they do not want to move. The government is concerned with their safety but the communities do not want to move, it is as if they are all mentally ill, how can someone not understand that they are supposed to move for their own good' (KII # 27, May, 2015). It is evident therefore that little is known about the processes and factors that influence the communities to continue living in the flood prone areas which influences public perception towards the vulnerability of communities that live in flood prone areas. This often results in misconception and inappropriate adaptation strategies which do not incorporate the realities of vulnerability in the flood prone areas. Some officials working for non-governmental organisations and civil society organisations working in the area, and others who visited Nsanje to support the people affected with the floods, indicated that they have been supporting the affected people for over 5 years but that despite this support, that the people are becoming more vulnerable to the floods, rather than better able to cope with them. 'I remember the first time I came to support the people affected with the floods here; there were only a few people who were affected and many areas that have been affected now were not affected then. That time, some of the people who were affected then have been affected again now. They were told to relocate that time but they never relocated. I personally cannot understand why they choose to suffer during floods when they could be in safer places' (KII #26, April, 2022). In addition, another official indicated that 'our organisation has been on the government's side, advising communities to relocate to safer areas. They have refused to relocate. I believe it is because they know that during the floods, the government, humanitarian organisations and other well-wishers would always send relief items to support them, they are so dependent on the humanitarian aid. I wish they could love their lives more and relocate' (KII #18, March, 2022). Another key informant also indicated that 'to some extent, I think organisations and the people themselves have contributed towards their continuous stay in the flood prone areas, as a humanitarian worker, I get busy when such incidents happen. These people in their situation have created work opportunities for us, without them, we could be jobless. The other part of us feels they should indeed move, whilst the other part of us is glad that we have work to do. It is a complicated situation' (KII #46). It is evident that communities living in the flood prone areas are not given an opportunity to express themselves in order to be understood how they perceive their own vulnerability to floods and even why they communities are motivated to live in the flood prone areas. The lack of knowledge on the communities' perspectives has created a general misconception of the livelihood of these communities. One participant indicated that 'it is unfortunate that some people out there think we love and enjoy receiving handouts in form of humanitarian aid and support. I wish we could be given an opportunity to tell them how humiliating it is to be in camps, where there is no personal freedom and the support we get is not enough for our families' (FGD #4, March, 2022). These words accentuate the point that there is a meaningful gap in knowledge of perceptions towards vulnerability between the communities living in flood prone areas and the policy makers and also key stakeholders that are implementing disaster risk reduction and climate change adaptation programs. The results from the household survey indicate that all the smallholder farmers that participated in the research would like to be supported in a way that they become independent, rather than being dependent on humanitarian aid. The proposed strategies included being supported with farming inputs and tents to put up temporary structures whilst they work towards settling down after the floods. Whilst policy makers are determined to move the communities from the flood prone areas, communities themselves indicate that they are comfortable in the flood prone areas based on reasons that are valid to them. The misunderstanding between these parties creates questions on who these policies are designed for if the ones that are affected by the policies have a different perception. The next section presents how communities living in flood prone areas perceive their own vulnerability to floods. In addition, it highlights how their perception to floods is differentiated and what factors influence the different perceptions. communities frame vulnerability to floods as part of their livelihood which is contrary to how the public presents it, as a fatal risk that requires permanent relocation to upland areas that are safe. --- Self-Perceived Vulnerability to Floods According to communities who live in flood prone areas, their perspective of the floods is influenced by the location where they live, their personal interaction with the floods, age, the timing and severity of the floods, loss of life, emotions and gender. The next sections expand on these factors to present an understanding of how they influence self-perceived vulnerability to floods in the flood prone areas. --- Self-Perceived Vulnerability to Floods by Location Some communities derived their self-perceived vulnerability based on the location where they live, which they referred to at different scales, from the district level to village level. communities in Nsanje are more vulnerable to floods. 'usually, the floods just sweep off the top soils, our homes are intact and not all our crops are destroyed, unlike the smallholder farmers in Nsanje, we hear they lose their crops and even homes are destroyed' (FGD #11, TA Kunthembwe). According to the household survey, all the respondents in Nsanje were aware that communities in Nsanje are more vulnerable to floods than them for various reasons. During focus group discussions, some of the indicators of vulnerability that were mentioned included the extent of damage caused by the floods in Nsanje as well as the way they are exposed due to the location, Lower Shire Valley. A traditional chief in Kunthembwe explained that 'We are lucky that we are upland, our friends from the Lower Shire Valley are in big trouble, all the water from upland ends there. I hear they benefit from the fertile soils after the floods but they lose everything. We are less vulnerable than the communities in Nsanje' (KII #51, TA Kunthembwe). In Nsanje, during household survey, key informant interviews and focus group discussions, communities said that they are aware of their vulnerability to floods due to their geographical position. A lady from TA Mlolo, who is also a Civil Protection Committee member explained that 'we are more vulnerable to floods because our friends in the upper Shire Catchments have cut down trees hence contributing to soil erosion and which makes water not to sink into the ground but run over the top soils during the rainy season. As a result, we are always the victims during the rainy seasons' (FGD 9, TA Mlolo). Apart from the geographical position in Nsanje, these communities attribute their vulnerability to floods due to environmental degradation that has been caused and exacerbated by the smallholder farmers living in the upper catchment of the Shire River. In addition, a male participant who is 75 years old and living in TA Nyachikadza in Nsanje District, indicated that they have become more vulnerable because of the changes in the rainfall patterns and amount. 'We have always lived close to the marshy areas all our lives and it is only recently due to changes in the way the rain is falling nowadays and the times in which the rainfall starts and end, also the amounts of the rainfall that we have been experiencing the disastrous floods frequently', (KII #14, TA Nyachikadza). A combination of geographical position, rainfall patterns and environmental degradation in the upper shire river catchment has been indicated as causing communities in Nsanje to be more vulnerable to floods than in other disticts Settling in flood prone areas, changes in land use and deforestation are some of the factors that have caused global vulnerability to floods (Kundzewicz and Kaczmare, 2000). These authors indicate that extreme hydrological events are also a natural phenomenon that will continue to occur regardless of all the underlying factors and suggesting that it is necessary to understand the extreme hydrological events and live with them (ibid). Over a decade, the data that I collected, in addition to other national documents that I reviewed indicate that although hydrological extremes occur naturally, changes in land use and deforestation in upper areas have exacerbated vulnerability to communities living in flood prone areas. Clearly, this indicates that vulnerability is more complicated with several factors that exacerbate it. Different locations are affected differently by floods based on various factors. In 2000, it was indicated that the fight against floods and droughts by all the stakeholders in the Mediterranean region had not been successful (Kundzewicz and Kaczmare, 2000). There are various reasons that have resulted to the communities not being able to cope with the floods and droughts in various places. The need to further understand the context in which communities are failing to cope with the floods and droughts helps to scrutinise the challenges and therefore often offers appropriate strategies that are strategically targeted to deal with the underlying challenges. Kundzewicz (2003) stress that higher and more intense precipitation is being observed around the globe and that some places are more exposed to such precipitation than others hence create a difference in the levels of vulnerability. This is still the case in many parts of the world where vulnerability varies depending on the extent of exposure to the rainfall and the effects. Adaptation strategies that aim at reducing the effects of floods should not adopt a one-size-fits-all approach. The approach should instead be on specific and appropriate measures that are applicable to the region, country, district, area and village in question. The more that communities are exposed to floods, the more they struggle to cope with them (Paul and Routray, 2010). In Bangladesh, people who live in areas that are not heavily affected by floods and live in better socio-economic conditions are likely to cope with the floods better than people who are more vulnerable to the floods and are worse off in terms of their socio-economic conditions (ibid). In the Philippines, households are aware that there have been an increase in rainfall, rainfall variability, an increase in the intensity and frequency of storm events (Combest-Friedman et al., 2012). However in terms of perceived risk to floods, their results indicate that spatial location and resource dependency are the factors that determine their perceived risk to floods, whilst the social economic conditions do not have any influence (ibid). Nevertheless, social economic conditions are factors that characterise households and communities and therefore very critical in adaptive capacity (ACCRA, 2010). However, the extent to which self-perceived vulnerability and social economic conditions influences perceived risk to floods needs to be unpacked in context in order to provide contextual evidence of the claims. To some extent, individuals make decisions on their vulnerability and how to cope with their vulnerability partly based on the available resources. For example, in my study area, the poorest of the poor indicated that they are more vulnerable to floods because they do not have an alternative source of livelihood, neither do they have valuable assets that they can sell during or after a disaster to help themselves. The rich however, people whose economic status is high based on their local indicators, indicated that they are exposed to the floods but are not very vulnerable. They explained that they have valuable assets which they rent out and sell during and after a disaster for their survival hence their livelihood is not hugely affected. However, Combest-Friedman et al. ( 2012) concluded that spatial location is the most significant factor that determines how households perceived their flood risk to household assets, but not perceived risk to their households. The spatial location determines how these households will be affected in terms of losses and damages to their household assets such as blankets, cups, pots and furniture. This scenario suggests that the people that are affected by the floods are not concerned about their lives being in danger and their livelihood but rather the assets they own which could be swept away or damaged by the floods. This could be another factor that motivates the households to continue living in the flood prone areas. My results on the contrary suggest that spatial location only partly influences how smallholder farmers perceive their risk and vulnerability to both household assets and livelihood therefore indicating the dynamics and complexity of self-perceived vulnerability as will be discussed in the following sections. Another man who before the January 2019 floods was categorised as a middle class farmer indicated that 'Floods have been part of us for a very long time, however, with the recent experience of floods in January, 2015, we feel helpless. Floods are now posing a real threat; all our assets have been destroyed, including our farm land. Previously, we were rejoicing when floods came because we could get fertile soils that enhanced our agricultural production. During the January, 2019 floods, our soils have changed and we do not know what to grow, this is a big blow to us' (KII #20, TA Mlolo). The results indicate that with the increase in the intensity and severity of floods over time, the smallholder farmers have realised that the floods put both their household assets and livelihood at risk and that they are more vulnerable to floods. The results further suggests that perception can change over time depending on the extent of risk and severity of the floods that determine the extent to which the smallholder farmers become vulnerable to floods. In this case, the findings indicate that floods damaged the farm lands in 2019 unlike the previous years where only household assets were damaged. These findings also suggest that the extent and type of damage varies depending on the geographical positions where communities that are in low lying areas suffer more and differently than those in upper areas. The variation in how vulnerability is framed suggests that vulnerability to climate variability is heterogeneous and dynamic and therefore that adaptation plans and strategies must acknowledge and respond to that if they are to enhance sustainable local adaptation in a dynamic environment. The comparison in the case studies clearly indicates another aspect of vulnerability whereby communities are exposed to floods due to heavy rainfall, environmental degradation and excessive runoff in other parts of the country. Contributing to wider literature, the study context illustrates how complex vulnerability issues are and that there are various levels of exposure that need to be explored in various locations to understand vulnerability levels. This understanding is needed further in order to design effective local adaptation strategies that are more relevant and suitable to the affected areas and affected communities. In addition to this, personal experience, based on age and exposure to extreme floods was observed as another aspect that influences vulnerability to floods, as

A section of Nsanje District has been experiencing floods for over a decade and was declared a flood prone area the same time (GoM, 2006). Therefore, firstly, this research seeks to understand how the communities that live in the flood prone areas perceive their vulnerability to floods and how some of them have been coping with the floods since 1952, just over 6 decades. The overall aim of the research was to understand the impact of floods on sustainable development and how communities living in flood prone areas perceive their vulnerability to floods and climate variability, to investigate the power dynamics at household and community level and to explore the complexities associated with local adaptation programs in the flood prone areas. This research contributes towards framing vulnerability based on the perception of the different groups of people that are vulnerable to climate variability and have experienced environmental changes throughout their everyday lives. The communities living in the flood prone areas perceive floods as part of their livelihood. The communities have experienced floods for over 5 decades and have been implementing adaptive strategies to help them cope with the floods during this time. Temporary migration is one of the effective adaptive strategies that the communities living in the flood prone areas implement during floods.

the smallholder farmers become vulnerable to floods. In this case, the findings indicate that floods damaged the farm lands in 2019 unlike the previous years where only household assets were damaged. These findings also suggest that the extent and type of damage varies depending on the geographical positions where communities that are in low lying areas suffer more and differently than those in upper areas. The variation in how vulnerability is framed suggests that vulnerability to climate variability is heterogeneous and dynamic and therefore that adaptation plans and strategies must acknowledge and respond to that if they are to enhance sustainable local adaptation in a dynamic environment. The comparison in the case studies clearly indicates another aspect of vulnerability whereby communities are exposed to floods due to heavy rainfall, environmental degradation and excessive runoff in other parts of the country. Contributing to wider literature, the study context illustrates how complex vulnerability issues are and that there are various levels of exposure that need to be explored in various locations to understand vulnerability levels. This understanding is needed further in order to design effective local adaptation strategies that are more relevant and suitable to the affected areas and affected communities. In addition to this, personal experience, based on age and exposure to extreme floods was observed as another aspect that influences vulnerability to floods, as discussed next. --- Self-perceived Vulnerability Based on Personal Experience of Interactions with Floods The findings illustrate the way that communities have personally experienced floods, in terms of the actual suffering and number of times they have experienced severe floods, have also contributed to the way communities perceive their vulnerability. The previous section described vulnerability at a wider community level based on spatial location, whilst this section describes individual experiences and those at the household level. 'We have been experiencing floods for over two decades, we have survived the worst floods, we are exposed and more vulnerable to floods. Floods have become part of us, we prepare for the floods, experience the floods and over the years, we have come up with strategies that will reduce the negative impacts of floods' (KII #9, TA Nyachikadza). Key informant #9, who has lived in traditional Authority all his life and is over 75 years old indicated that floods are not a new phenomenon and that they have been experiencing disastrous floods several times in his life time. In another key informant interview, an elderly man said 'Floods are indeed an issue of concern, however, it is important to note that we do not experience floods every year. Over the years, the January 2019 floods have been the worst. There are some years when we experience normal rains and we do not move. Floods are just one of those occurrences that happen once in a while so we are not too scared of them. We live a normal life.' (KII #2, TA Ndamera). The participant emphasised that the personal experience through interaction with floods over the years has created an opportunity for them to get used to the floods and learn to live with them. The communities stressed that floods are not an extra-ordinary problem but rather, a problem that is location specific and has been incorporated in their social environment. This statement suggests that personally, some smallholder farmers have accepted the floods as part of their lives and though admitting that they are affected sometimes, it has ceased to be a major concern that they would worry about every year. This aspect of how communities frame their vulnerability based on their personal experience to floods is important because it illustrates the type of attitude the people have towards the floods therefore suggesting preferred form of intervention that should be advocated for. Personal experience of floods influences the way the affected people perceive their personal vulnerability and risk to floods, in addition to their perceptions to climate change and the way they respond to floods (Demski et al., 2017). However, Demski et al. ( 2017) also highlight that there are some cases in the United Kingdom whereby some individuals who have never experienced a disaster before can perceive the vulnerability to disasters similarly to those that have experienced disasters (Ibid). These findings therefore suggests that there is need to understand ways in which personal experience affect people's perception towards floods and that such claims should not be generalised, but rather should be contextualised. My research suggests that personal experience with the floods is significant in determining how the communities perceive their vulnerability to floods and that it is different from the public that does not experience floods. During household interviews, focus group discussions and key informant interviews, the communities were able to distinguish their self-perceived vulnerability to floods based on the different times and circumstances when they had experienced the floods, the extent of damage that was caused by the floods and the intensity of their vulnerability in all those times. In addition, this thesis overall suggests that apart from the actual flooding experience, there are several other endogenous and exogenous factors that have influenced the communities self-perceived vulnerability to floods as evidenced by another study. In Chile, Lara et al. (2017) emphasise that spatial locations, timing of the floods, gender and environmental degradation in the upland areas influence how smallholder farmers perceive their vulnerability to floods. In context, whilst some factors and processes are relevant in some areas, these may be different in other areas hence the different factors that are being mentioned by different scholars to be influencing perceived vulnerability. The Nsanje case presents evidence of where location and personal experience influence self-perceived vulnerability to floods as will be discussed next. In TAs Nyachikadza and Ndamera, the communities have been experiencing flood for over 2 decades, similar to TA Mlolo. However to them, as indicated in the quotes, floods have always been a part of their lives and they were used to them until the 2015 floods, which were different due to their timing and extreme nature. The disastrous floods came during the middle of the night. An old man indicated that 'We have experienced floods for a very long time and in all those times, we were able to relocate to neighbouring villages with our belongings in good time. The only difference is that this time, the floods came unexpectedly and at night. The whole area was flooded in less than 10 minutes hence we could not think, we struggled to save our lives and lost all that we had, our household items, farm produce and boats. It is as if we have lost our livelihood. We have no idea of where to start from now' (FGD 1, TA Ndamera). The inability to prepare for the floods and relocate prior to the floods has changed the way in which some communities perceived their vulnerability to floods. This emphasises that self-perceived vulnerability to a risk is not constant, it changes over time based on the extent of personal exposure, the ability to prepare for the floods and relocate before the areas are flooded and associated damage to personal assets. Whilst some smallholder farmers indicated that floods are part of their normal lives, some of the key informant interviewees indicated that they are more vulnerable now and helpless after the January, 2015 floods. Similar findings are reported in a study in Lagos city, Nigeria by Adelekan and Asiyanbi (2016). Their findings illustrate that each city is affected by climate variability differently even though they are all in the same country, implying that context of vulnerability and therefore adaptation needs and resources cannot be generalised and should never be uniform, but rather each case should be treated differently based on adaptation needs and personal interaction with the floods. This thesis emphasises the extent of severity of floods in different locations and therefore that vulnerability to floods differs based on geographical settings. It also emphasises that the ability to cope with the floods and climate variability differs amongst different individuals based on their personal circumstances. The factors that cause vulnerability in the different settings can never be the same. The geographical boundary can go beyond cities/districts to as small as village level, as found in this thesis. It is evident that in Nsanje District regardless of location and type of hazard, individual experience with the flooding situation is crucial in vulnerability and local adaptation studies as it builds self-perception to vulnerability and over time helps those that are affected with the floods to find ways of coping. However, it was also evident that people are used to the situation where they are able to monitor the floods and have enough time to relocate. In the event where they were not able to adequately prepare for the floods, they lost all their assets and some people lost their lives. This situation suggests that communities are used to certain ways in which flooding events occur and how they prepare for the floods. In the event where there is a slight change in occurrence, mainly the middle class communities and poor communities are disturbed and become more vulnerable unlike the perceived rich communities within the same locations. Over time however, this research would suggest that the affected people would similarly find a way of coping with the new form of threat. On the other hand, people who are not able to cope with the extreme floods, mainly the poorest of the poor and the elderly would be deterred from continuous exposure to a risk and choose to relocate as will be discussed in detail later on in the chapter. During a focus group discussion with the chiefs and other influential communities that live in flood prone areas, it was clear that the some communities who are mainly old would prefer to relocate to safer areas due to lack of strength and capacity to escape during the floods. 'For some of us, we started relocating about 10 years ago. Previously, the floods did not come with much force hence we could still be in our homes during the flooding season. With time, we started relocating during floods with all our belongings, now the flooding'style' has changed. Floods are coming with no notice at all, this is dangerous for some of us who are growing old now and we do not have enough energy to run or change strategies, relocation is better for us' (FGD # 2, TA Ndamera). In addition, older communities indicated the same whilst the younger generation who were still productive and have experienced floods in various dimensions were still willing to continue taking the risk, indicating that the risk is more valuable that relocating to other places where their livelihood would be neglected. In Malawi, Kakota et al. (2011) emphasise that the groups of people who are more vulnerable to climate change include the elderly because they are usually so poor with limited alternative source of livelihood and cash to sustain themselves during and after a disaster. In relation to my study therefore, the more vulnerable a particular group is to the floods in terms of age and inadequate capacity to cope with the floods, the more they are willing to relocate to upland areas permanently due to the hassles the vulnerable groups experienced during the extreme weather events. The self-perceived vulnerability goes beyond the individual experiences with the floods in relation to how they suffer personally and how as individuals, they are able to cope with the floods, over time and with age, to the frequency of the floods in the area. --- Self-perceived vulnerability based on frequency and severity of floods and causing damage and loss of assets During all the focus group discussions in Nsanje, it was indicated that there are differences in the timings of the disastrous floods that led to relocation in Nsanje District. During focus group discussions, communities that were relatively rich indicated that they do not perceive floods as a great risk to their lives and livelihood because of the variation in their occurrence. Whilst also indicating that there are other factors that influence their self-perceived vulnerability, smallholder farmers indicated that floods that cause displacement and damage to household assets and houses do not occur annually. 'Yes, we are vulnerable to floods. However, the vulnerability to floods is not something we think about every year. There are some years that go by without experiencing floods. That is why we remember that it is a problem during the years that we are displaced' (FGD #3 TA Ndamera). In addition, another participant indicated that 'Floods are sometimes a problem. However, we do not consider it a major threat because it does not happen every year. If we were displaced every year, I think the way we think about our vulnerability to floods would have been different' (KII #17, TA Mbenje). In Traditional Mlolo however, one of the chiefs said that 'to be honest it is during the recent years that we have now started considering floods as a threat to our livelihood. We experienced disastrously floods that led to displacement in 2012, barely 3 years later, we have also been displaced. It is now becoming an inconvenience to us and hence we need to seriously think of a permanent solution to avoid being displaced and losing all our assets'. (KII #48, TA Mlolo). The responses from the different areas in Nsanje suggest that apart from spatial locations, the frequency of floods also influence selfperceived vulnerability to floods. This is very important because it also influences how farmers respond to the issues of floods and the attitude towards the urgency of some of the adaptation strategies that are advocated by the government and other stakeholders. During the January, 2015 floods, all the smallholder farmers in Traditional Authority Mlolo indicated that they had become more vulnerable to floods than in the previous years because the soil structure had changed due to huge soil deposits. In addition, some communities whose soil structure had not changed indicated that they are more vulnerable to floods, but because their houses, farm products, households' items and livestock were washed away. 'This year, 2015, we have lost everything, we have nowhere to stay, our houses, household items and livestock were washed away, previously, we had time to rescue our household items, including farm products and livestock. Our houses would contain the water to a certain level during the floods, but would dry off after a couple of days. This year, there is nothing left after the floods, we have become more affected with the floods than in the previous years' (KII #38, TA Mlolo). The communities living in Traditional authority Mlolo and Nyachikadza are used to the floods as they have been extremely exposed to the disastrous floods over a long time. During flash floods, Traditional Authority Mlolo is the first to be affected, then Traditional Authority Nyachikadza due to the geographical position of these places. However, their perception to floods changes when floods bring unexpected results including changes in soil structure (In TA Mlolo in Nsanje) and washing away of farmland which leaves communities with no land suitable for farming. The results hereby suggest that some communities do not perceive their vulnerability to floods as a serious threat to them if their agricultural land is still suitable for farming. This, therefore indicates that self-perceived vulnerability also depends on the significance and attached value to what has been affected. In this case, the farm land provides an area for farming which forms the major part of the community's livelihood. In the areas that are heavily affected by floods, mostly in Traditional Authority Mlolo and Traditional Authority Nyachikadza, houses are strategically built with mud and poles to avoid huge losses during floods. The strategic adaptation brings another aspect of self-perceived vulnerability to floods that should be accounted for in adaptation strategies. In the situation where these smallholder farmers are requested to relocate to reduce vulnerability to floods, they have temporary houses built with mud and poles that are easily replaced. The damage of houses in the case therefore is not a significant loss to them, where as it is significant to the government of Malawi, that is responsible for infrastructure including roads and bridges. These findings raise questions about who would actually benefit from the proposed relocation, the people who are affected by the floods or the government itself in terms of reduced pressure to offer support to the communities that are affected with the floods during and after the floods. During the key informant interviews, all the government officials that were interviewed lamented about the pressure that the government has during floods and after the floods due to inadequate funds. It is evident that with limited funding, the best option for the government would be for the communities that live in flood prone areas to relocate. However, this is contrary to what was evident during the research because many people would prefer to migrate only temporarily during the floods, rather than migrating permanently. In addition to this, the results illustrate that the adverse effects of floods on the people goes beyond the hazard itself to the interaction of social and economic factors within the system. Methmann and Oels, (2014) illustrate that that there are several factors and processes within a system that exacerbate vulnerability to climate variability, including social, economic and political factors. In my case studies, it was revealed that specifically, issues of gender influence how communities perceive their vulnerability to floods as discussed next. --- Self-Perceived Vulnerability to Floods according to Loss of Life and Gender Apart from losing agricultural land, loss of life during the floods also determines how smallholder farmers perceive their vulnerability to floods in flood prone areas. A participant in Traditional Authority Mlolo during the focus group discussion, a woman indicated that 'I have never experienced floods to this extent (January, 2015). My 3 children were carried away with the floods, I was helpless, for the first time I have realised that floods are dangerous and it is indeed dangerous to live in flood prone areas' (FGD #8, TA Mlolo, April, 2022). Communities realise that living in the flood prone areas is dangerous after losing their loved ones. This statement suggests that previously, the danger of living in the flood prone areas was not considered seriously until the smallholder farmers witnessed loss of lives. In an in-depth interview with a man from TA Mlolo who had lost a wife and 2 children, he indicated that 'I was deeply affected with the loss of lives of my family members, my wife and 2 children however, life must go on so I will continue farming there and live there because it is God's will that I have now lost my family members' (KII #44, TA Mlolo). This statement further suggests that for some communities, it is not just the vulnerability to floods that is associated with the death, but rather powers that they have no control over, in this case, he mentioned God. The statement illustrates that perception to floods is influenced by not only what is lost, but also the extent to which as human beings they have control over a situation and the extent to which they are able to cope with the emotions of losing household assets, family members and their farmland. The acknowledgement of the situation where communities realise that they have no control over nature poses an interesting debate in the prevention of disasters since in this case, they have a choice to either cope with the situation or do nothing about it and face the realities of environmental changes. Some women were afraid to go back to the flood prone areas to avoid the trauma. In TA Ndamera during a women focus group discussion, all the women agreed to what another woman said that 'I have now realised that floods are dangerous and that living in the flood prone area is not ideal. With the way we suffered to get to a safer place this time, I strongly agree now to the idea of relocating. Floods are dangerous, they kill and taking chances is as if we are mentally disturbed' (FGD #2, TA Ndemera). This suggests that emotions also play an important role in determining how smallholder farmers perceive their vulnerability to floods. In the case of the communities living in Nsanje, if it were not for the disastrous floods that caused death, all the communities would have still considered floods as part of their livelihood and not as a threat. During interviews and discussions, women showed and illustrated that they were heavily affected when they lost their family members during the floods, whilst men showed that they were dealing with their emotions in a way that was presented as another trauma as part of life. This might be taken to indicate that women can be more emotional in times of bereavement hence need more psychosocial and moral support where bereavement and trauma is involved in the flood prone areas. After the floods, during focus group discussions and participant observation, some women cried when they were explaining what they went through during the floods. Some women indicated that they were failing to sleep because of the memory of the flooding event, whilst none of the men I interacted with indicated that they had reached that extent. Nsanje is a patrilineal society were the men are culturally perceived as being responsible to take care of the wife and children. The women relocate and stay at their husband's land and are culturally expected to be submissive to their husbands. The discrepancy in the findings on dealing with emotions will no doubt be influenced by culture and the responsibilities either part has to play in the society, and also the fact that some men cannot openly admit that they are heavily affected emotionally. In this setting, all the women indicated that they were struggling emotionally with the loss of their beloved ones more than the men since they had to continue managing the homes whilst the men were out most of the time socialising with their friends and drinking beer sometimes. A study conducted by Lebel and Lebel, (2016) suggest that emotions are important in climate related decision making therefore they should be explored in order to understand certain behaviour within a particular environment. In Northern Thailand, women were perceived to be emotional and therefore struggled to make informed decisions when they were in an emotional state unlike men (Ibid). However, Lebel and Lebel, (2016) revealed that regardless of gender, emotions played a role in different decision making scenarios. This thesis illustrates that emotions are subjective and therefore should not be generalised based on gender, but rather to an extent within which an individual has been affected by an incidence and the realities that surround them. In flood prone areas and evacuation centres, despite being emotional, women play a major role in making certain that children and men are fed and all their household demands are met. 'During floods, all the women have the responsibility to cook for their household members, fetch water for their household and do all their household chores. We force ourselves to be strong because there is no one who can help us. Some men usually go to see friends in nearby villages whilst for us, we have to be with the children and do all the chores' (KII # 17, April, 2022). This reinforces the point that all women are not irrational but rather, many are responsible and know when to do what is expected of them, even during a crisis. For example, even after experiencing loss of family members during the floods and some struggling with their emotions, they were able to comfort their children who were also traumatised. During participant observation in the field, the women who suffered loss would cry when their children are out to play but remained emotionally strong in the presence of other family members. After the floods, the affected households evacuated to designated evacuation sites where tents were assembled for shelter. During that time, many men could not be found in tents during the day, I was told that they were going out to chat and drink with friends during the day, every day. Whilst socialising with others was perceived as a way to deal with their emotions, women never had that opportunity due to their household responsibilities. Women had to stay around the camps to cook for their families, including their husbands, and to take care of their children amongst other chores. Women therefore experience tough situations and had to fulfil all the husbands and society's expectations in the home, regardless of their experiences. Bee (2016) indicates that understanding the gendered dimension of vulnerability is relevant to design appropriate local adaptation programs that meet the needs of the communities that experience the disasters. In this case, understanding gender roles and responsibility before, during and after the floods is critical to providing specific, targeted and adequate psycho-social and emotional support to the vulnerable communities as well as guiding policy makers on the type of support that is needed and to whom, rather than generalising support to the affected people. Reyes and Lu (2016) highlights that women keep up hope for the family, care for the sick and provide home in the family. Afriyie et al. (2017b), Madhuri (2016) illustrate that different groups of people and genders are affected differently during floods. Women are emotionally stressed during and after the floods hence there is need to provide more psycho-social support to women who are bound to the evacuation sites with limited opportunity to visit friends and family members as per this thesis. The sections on various factors that influence self-perceived vulnerability to floods partly explain why many communities have been living in the flood prone areas despite the associated risks. According to the household survey and key informant interviews and focus group discussions, communities indicated that they grow various crops and are involved in some small-scale businesses. These activities are part of their livelihood in the flood prone areas that is more valuable to them and worth the risk of living in the flood prone areas. In addition, the frequency of the floods is such that the flooding problem does not occur annually hence giving the communities confidence to continue living in the flood prone areas. Subsequent chapters in the thesis discuss how communities realised that there is need for more sophisticated early warning system to prepare them during flooding and are willing to cooperate with scientists, hydrogeologists and other professionals that can assist them in the design and implementation of an effective early warning system. Officials working in government meanwhile need to understand how the affected communities are coping with the floods, identify gaps and challenges and be able to fill the gaps to achieve sustainable and effective adaptation to floods and other extreme weather events. Instead of insisting that the communities should relocate to safer areas, this research suggests that there is need to understand the complexities of vulnerability, adaptation and resilience. Inclusive and comprehensive vulnerability assessments and adaptation planning, design and implementation is necessary to effectively adapt to the floods and other extreme weather events. In TA Nyachikadza and some parts of TA Mlolo and TA Ndamera, the government stopped providing services to the communities because they were designated disaster prone areas and therefore not suitable for human settlement. However, during floods, the government helps in evacuating the communities that are affected by the floods and offer humanitarian aid and support such as food, cooking oil, legumes, buckets, plates, blankets, clothes and other essential items. The government in this case is still responsible and concerned about the safety of the people from the disasters. This case study presents a need for governments to consider other factors that either shape or reduce the vulnerability to floods in the flood prone areas. However, during elections, these areas are allocated voter registration sites and polling stations within the flood prone areas. More details on this will be presented in the next chapter as I will be discussing on the absence and presence of the government in the flood prone areas. --- MOTIVATIONS TO LIVE IN FLOOD PRONE AREAS This section focuses on factors that motivate communities to continue living in flood prone areas. The factors include community's livelihood through farming, their social networks within their original homes and their attachment to the flood prone areas. Political ecology studies suggest that emotions are tied to a particular landscape, ancestral values, culture memory and dignity, hence many people prefer living in their original homes, places of birth and where their families are rooted (Dallman et al., 2013). From work in Ghana, Addo and Danso (2017) indicate that communities that live in flood prone areas are not willing to relocate because of a variety of reasons, including low incomes and thus inability to pay rent in the new areas, not willing to lose land and social networks they have built over the years, and fear of losing their source of livelihood. My findings, as presented in this chapter, indicate that these and several other factors motivate people to live in their original homes, and thus to live with floods. Using the case study of communities living in flood prone areas, the results exemplify that these factors are variously political, socio-economic and personal in nature. In the previous chapter, it was important to understand self-perceived vulnerability because it enabled me to understand the factors that motivate communities to live in the flood prone areas regardless of the flood risk and of the Malawian government's request that they relocate to safer areas. Currently, the government officials, including the media and other stakeholders do not understand why the communities insist on living in flood prone areas despite all the associated risks. This chapter presents that the factors and processes that motivate communities to live in the flood prone areas are complex in nature. Floods in some cases are not perceived as a main problem in the areas due to their seasonality. communities also have local knowledge on the flood management hence are confident to face the floods during the flooding season. Other factors are political, social, cultural and economic in nature, which are experienced personally but based in the wider community, hence indicating that relocating to safer place is not easy for all communities living in the flood prone areas. Nevertheless, there is widespread misunderstanding of motivations to live in the flood prone areas, with the wider public perceiving the reluctance to move to safer areas as strategic so that the communities benefit more from relief items and other forms of humanitarian aid. Addo and Danso (2017) highlight that communities in Ghana prefer to live in the flood prone areas because they are not willing to lose their land, they have low income to pay rent and they do not lose their livelihood. Furthermore, Dallman et al., (2013) found that In California, people prefer to live in their original homes because of emotional ties with their ancestral values, culture, dignity and landscape. In a study of political ecology of emotions, Dallman et al., (2013) stress that even changing the physical environment within the original home is likely to destroy some cultural and sacred places where people's emotions are attached. It is clear therefore that emotions play a role in the decision to live in the flood prone areas. However, the emotions can be attached to different things, for example, sacred places, ancestral bonds, culture and tradition. --- Floods Not Perceived as the Main Problem in Flood Prone Areas The interaction between the changing environment, socio-economic, culture and livelihood amongst other factors and processes in the flood prone areas has shaped the way communities weigh the problems they experience in their daily lives. According to the household survey, 83.3% of the communities living in flood prone areas indicated that they perceived hunger as their main household problem when responding to an open ended question, asking communities what they perceive as their main problem.Apart from the areas being exposed to flooding, it is often also affected by droughts, which affect crop production. This clearly indicates that even though these communities are exposed and vulnerable to floods, the floods are not the major concern in their lives. During FGDs and KIIs, communities indicated that they perceive hunger as the main problem. 'Hunger has been a major problem in our lives for over a decade, I remember those days we used to have enough food for the whole year, nowadays our harvest is not adequate to last the whole year. For most of us that live in the areas that experience floods, we are better off because we grow fruits, vegetables and even maize in the marshy areas hence making us to have enough food throughout the year and for some households, most part of the year (KII #2, TA Mlolo, June, 2022). In another interview, another participant said that 'I would never relocate upland because I know for sure that I would not be able to have enough food for my whole family. Most of the people in the upland areas rent part of our fields, we help them to be food secure. If they come to us for help, it means our land is good for agriculture, hence we are better off here'. (KII #12, TA Nyachikadza, March, 2022). The communities that live in flood prone areas are afraid of hunger more than the floods, therefore living in flood prone areas is an advantage for them. Other studies also reveal that a source of livelihood is one of the reasons that communities that live in flood prone areas would not be willing to move and relocate upland (Addo and Danso, 2017). However, fewer studies have been conducted in rural areas where communities are living in flood prone area. This chapter exemplifies that in rural areas too, communities value their livelihood through farming and fishing, and through their sources of income from leasing their farmlands. This reveals an important aspect of livelihood decision-making; that is decisions are made based on what is actually lost during the floods and what is not lost. This chapter therefore illustrates that relocating the communities would not be an appropriate solution, as many communities would not be able to sustain their livelihood upland and would not have the capacity to resettle in the new areas. A man who lost everything during the floods explained that 'for most of us, we would struggle to settle in new areas because we have lost everything It will be difficult to obtain a farm and acquire new assets in the new areas. That will require a huge sum of money that we not have. In addition, we would not be able to buy food and bags of fertilizer that will be needed in our farms. Currently, we do not apply fertilizer in our fields and we never buy maize' (FGD# 7 TA Mlolo). This quote suggests that only communities that have other assets in the form of capital and money can afford to relocate whilst those who have nothing would suffer more if they were to relocate. This also suggest that vulnerability would increase in the new areas hence it is appropriate for communities who do not have any form of asset to remain in the flood prone areas. These findings are similar to a study that was conducted in Thailand by Tahira and Kawasaki (2017) which suggested that regardless of the damage that the floods might pose, the poor communities have inadequate capacity to recover hence relocating them would worsen their situation. Tahira and Kawasaki (2017) argue that the poor communities would struggle to establish a new livelihood in new places that they would be requested to relocate to due to inadequate adaptive capacity. Relocating to new areas requires capital, land and other assets that would help the communities to establish themselves in the new areas. In the absence of these, these communities would suffer (Ibid). Therefore, they suggest that the focus should be on improving the early warning system, dykes and offering some form of compensation for the lost earnings. My study states that communities are motivated to live in the flood prone areas because they are able to address food security issues which is the main perceived challenge to them. In addition, the communities clearly illustrate that they have devised ways of coping with the floods which also forms part of their motivation. --- Local Knowledge and Water Level Monitoring communities indicated that they have experienced floods for a very long time and that over time, they have devised mechanism of monitoring water levels in rivers that cause flooding in the area. communities explained that over the years, they had come up with ways of monitoring the water levels in the rivers that flood as part of an early warning system. 'Apart from the 2015 floods, the chiefs, the elderly, even some of the youth always observe the water levels in rivers that flood during the rainy season. We place some poles, other people place sticks a few metres from the river bank as a monitoring tool. When the water in the river is too much, the river will swell and usually water will reach the poles placed on the river banks. We move the poles whenever the water reaches them up to a certain point which then we pack our things and move upland temporarily. Using that system, we know when it is going to flood. And even before then, we have time to move our assets upland as we prepare for the floods. That is why we have never suffered any death during floods' (FGD # 7, TA Mlolo,). In a key informant interview with an old man in the village, he indicated that 'we are able to know when we are going to experience a calamity through our visit to Mbona. We have a Sacred place in TA Ngabu [One of the areas in Nsanje, refer to case study description in chapter 2] where we go and offer sacrifices to Mbona and consult him on several matters happening in our communities. For us older people, depending on the visit based on experience, we are able to know that we are going to experience floods. This is for old people like me. Many young ones do not take time to observe these things' (KII #6, March, 2022). communities have designed their own ways of understanding the extreme weather events, ranging from beliefs and myths of the Mbona to innovative ways of flood monitoring using sticks. Based on these experiences, communities have been able to evacuate from the flood prone areas in good times, before the whole area is flooded. In addition, they have been having time to save their belongings from being swept away. However, as it has been explained in the earlier

A section of Nsanje District has been experiencing floods for over a decade and was declared a flood prone area the same time (GoM, 2006). Therefore, firstly, this research seeks to understand how the communities that live in the flood prone areas perceive their vulnerability to floods and how some of them have been coping with the floods since 1952, just over 6 decades. The overall aim of the research was to understand the impact of floods on sustainable development and how communities living in flood prone areas perceive their vulnerability to floods and climate variability, to investigate the power dynamics at household and community level and to explore the complexities associated with local adaptation programs in the flood prone areas. This research contributes towards framing vulnerability based on the perception of the different groups of people that are vulnerable to climate variability and have experienced environmental changes throughout their everyday lives. The communities living in the flood prone areas perceive floods as part of their livelihood. The communities have experienced floods for over 5 decades and have been implementing adaptive strategies to help them cope with the floods during this time. Temporary migration is one of the effective adaptive strategies that the communities living in the flood prone areas implement during floods.

,). In a key informant interview with an old man in the village, he indicated that 'we are able to know when we are going to experience a calamity through our visit to Mbona. We have a Sacred place in TA Ngabu [One of the areas in Nsanje, refer to case study description in chapter 2] where we go and offer sacrifices to Mbona and consult him on several matters happening in our communities. For us older people, depending on the visit based on experience, we are able to know that we are going to experience floods. This is for old people like me. Many young ones do not take time to observe these things' (KII #6, March, 2022). communities have designed their own ways of understanding the extreme weather events, ranging from beliefs and myths of the Mbona to innovative ways of flood monitoring using sticks. Based on these experiences, communities have been able to evacuate from the flood prone areas in good times, before the whole area is flooded. In addition, they have been having time to save their belongings from being swept away. However, as it has been explained in the earlier text, all the strategies were not utilised because the floods came in the middle of the night and the communities were trapped in the floods. Local knowledge has also been used in various parts of Malawi to predict disasters such as floods and droughts. However, recently communities say that both the local knowledge and climate information from the meteorological department has not been adequate as a tool for weather forecast (Wellard et al., 2012). At the local level, there are no weather stations and some of the equipment in the weather station is outdated hence the unreliable weather forecasts (ibid). During focus group discussions, communities explained to me that 'we have lost trust with the climate information from the meteorology department, in 2010, they told us that we are going to experience floods, some people moved their belongings upland but instead, we experienced droughts. That year, our elders said we would experience drought and they were right' (FGD #6, TA Mbenje, May, 2022). Another woman said 'I remember hearing that we are going to experience floods during the day and we experienced the floods during the night. A car from the District Commissioner went around shouting that we should move because we would be experiencing floods. As usual, none of us moved, because we could not believe them. From what happened, I would suggest we use both our local knowledge and even the information from the Meteorology Department. If we had moved, some of us could have been better off by now '. Shaffer (2012) indicates that a mixture of indigenous knowledge and other scientific knowledge coupled with new technology is important and useful in helping communities to adapt to climate variability in their original homes. The realities of climate change effects is that they have become more complex and in some cases, difficult to predict using the traditional means of monitoring the river flow. The material presented in this chapter indicate that people who live in flood prone areas are motivated to live there because they monitor the floods locally and can therefore anticipate and deal with them. However, during the flooding event in 2019, their local knowledge on flood and local early warning system did not work effectively. This clearly illustrates that the early warning system needs to be integrated with modern ways of flood monitoring to enhance effective and appropriate flood preparation that would reduce loss to property and lives. There has been an effort from the Department of Climate Change and Meteorology, Department of Disaster Management Affairs and the Local Government through the District Council to disseminate climate information to those who live in flood prone areas for their safety during floods. The people are told to move upland to safer places to avoid being swept off by the floods. This had reduced vulnerability of the communities to floods in the flood prone areas since most of the times they were able to move upland before the floods occur and were able to save their belongings. During the 2019 floods however, the results indicated that flood occurrence is unpredictable and complex for the existing local, indigenous and scientific monitoring hence there is a need for a more comprehensive and sophisticated way of monitoring the floods that goes beyond recording amount of rainfall over a period. The traditional way of monitoring water levels with a stick in the flood prone area needs to be combined with reliable climate information from weather stations that have automated and more sophisticated weather monitoring equipment. In the long term, the results show that there is need for a comprehensive and more reliable early warning system from the service providers to build trust between the climate information service providers and those that need to utilise the information. My study corroborates another study that was conducted in the Philippines which indicates that communities do not trust the early warning system because the messages are not clear and do not provide convincing information to make people move upland (Neussner, 2014). This study shows that forced evacuation is necessary where people do not follow official evacuation advice and do not follow government warning (ibid). My research however suggests that insisting that communities should evacuate is totally impractical due to the affected people's livelihood, attachment to the area and other factors that are important but are ignored in the policy; some of which are political and costly in nature as will be discussed in the next section. --- Social and Cultural Factors in Living with Floods During the household survey, almost 80% of the communities indicated that they have a good social network with people living in upland areas. 'During floods our friends and relatives that live upland accommodate us in their homes, we keep our belongings there too and wait till our home is dry and we come back. During droughts, we offer them some piece of land where they can grow crops' (FGD 7#, TA Mlolo, June 2022). Another participant added that 'there are some people who let out some of the houses during floods. Based on the agreements, some charge in cash or number of bags of maize. The charge is very subjective and usually depends on the type of relationship that the people have' (KII #40, TA Mbenje, May, 2022). Another community member living in upland area indicated that 'We have a good relationship with communities that live in flood prone areas. During droughts many of us farm in the flood prone areas. Even in a normal year, we grow winter crops, including maize in the flood prone areas. We benefit from each other' (KII #45 TA Mbenje, June, 2022). The social networks and the relationships between the communities living in the flood prone areas and communities living in the upland communities suggest that land in flood prone areas is an asset in itself. This illustration presents a novel contribution in framing vulnerability to floods based on self-perceived vulnerability. communities living in the flood prone areas perceive their land in flood prone areas as an asset that supports their livelihood. Adding to the discussion earlier on that floods do not occur every year, the results suggests that communities are less vulnerable to climate variability in the flood prone areas. communities are therefore better off living in the flood prone areas than in upland areas due to droughts and limited space. In some cases, some communities that live in the flood prone areas are not happy that they are given names during floods. 'During floods, the communities that live upland mock us and give us various names that are associated with being displaced by floods even though during droughts, we do not give them names associated with droughts' (FGD #9, TA Mlolo). Another old man who has relocated 5 times indicated that 'we are called refugees in the same country, as if we do not have our own homes. This humiliation during floods has taught us that 'one is respected in their own land' hence we shall never settle in any man's land apart from our original land' (KII #11, TA Nyachikadza). Climate migration is another aspect of resettlement that have been discussed by various scholars as providing a degrading status for the relocated communities hence indicating how complex and dynamic the relocation process is as part of local adaptation (Addo & Danso, 2017). More details will be presented in the next chapter where I will be discussing the complexity of local adaptation and relocation. However, the public has different perception towards the communities living in the flood prone areas contrary to what was indicated by the communities. --- Adaptation to floods The results from the household survey reveal that 40% of the households mentioned that they were able to cope with the floods without external support indicating that they are able to survive on their own and recover after floods without any form of support from the government or humanitarian organisations. The other 60% of the farming households that participated in the research indicated that they cannot cope with the floods on their own without external support in terms of cash, shelter and food during the floods and about 5 months after the floods. Communities are used to the way they live in the flood prone areas. The social, cultural, political and livelihood interactions lead to local identities and convey a history of where they come from, how their ancestors have been surviving the floods and a unique livelihood which they are proud of. They are referred to locally as 'anthu akumadzi', meaning people who come from a place that is surrounded by water. This chapter therefore examines the opportunities that exist in flood prone areas that enable the communities and household to cope with the floods. This is important for adaptation studies and projects as it suggests that policy makers and project designers and should seek to support already existing strategies and practices. This would represent a less costly approach which would thus be particularly relevant for low-income countries. --- The Role of Assets in Adapting to Floods The farming households indicated that they have various assets that are useful in various ways during floods. The findings show that the valuable assets for the farming households in the flood prone areas are farmland, livestock, operational radio, operational mobile phone and a functional bicycle (Figure 7). A valuable asset, according to the research participants is an asset that communities identified as very important to their livelihood in the flood prone areas. The household survey revealed that over 40% of the farming households who are able to cope with the floods own a combination of assets, including farmland, livestock, operational radios, operational mobile phones and functional bicycle. These same households are able to cope without external support, therefore suggesting that in the flood prone areas, land, livestock, bicycles, radios and mobile phones are critical assets that support communities to be able to cope with the floods without seeking support from other people or institutions. Each asset will be discussed below in relation to the role it plays in facilitating adaptation in the flood prone areas. Farmland is one of the critical assets owned by those living in the flood prone areas. A middle-class participant in TA Nyachikadza explained that 'We depend on the land that was passed on to us from our ancestors, some of us have 4 acres. Many people have bigger pieces of land that is even not cultivated because of inadequate capacity to do so' (KII #11). In another interview, a very innovative participant in TA Mlolo articulated that 'We have huge pieces of land that even the government can utilise for agriculture to make the nation food secure. We feed Malawi, there are several trucks that collect maize during high peak of sales. Imagine if we were to grow maize on all the farmland that is available, Malawi would be food secure' (KII #36). Another smallholder farmer living in a village next to villages that flood claimed that 'communities living in the areas that flood is better off because they have huge farmlands which they grow crops and even some land is left idle. That case is very rare uplands because of population pressure, the land is not adequate for us and unfertile too resulting in low yields. That is why we go to rent farmland in the flood prone areas so that we should be food secure' (KII #43, TA Mbenje). Another participant relayed that 'Farmland is also a source of income for most of us, communities living upland, rent the farms from us during droughts and even during the normal rainy season. That way we are able to have money to buy soap, paying school fees, buy clothes, sugar, relish for example beef and eggs. Without the land we have as an asset, most of us would have been in trouble, no food, no extra cash' (KII #8, TA Ndamera). Whilst much emphasis is put on the economic losses of agricultural land and products in relation to levels of vulnerability and coping strategies in the flood prone areas of Anambra State in Nigeria (Enete et al., 2016), my study indicates that there is also an economic gain in the flood prone areas as well as economic losses. The land that is accessed and passed on from community's ancestors has created a form of security that enables them to sustain their livelihoods in times of droughts and minimal flooding. Participant explained that 'After the floods, we grow maize and sweet potatoes and they grow very well, as a result, we usually harvest enough to feed ourselves. We might lose our crops during the floods, but we are able to grow crops again using the moisture after the floods, that way, we do not suffer too much. We grow enough to even sell to communities that live in areas that do not flood, they still depend on us' (FGD #4, TA Mbenje). This suggest that these communities remain food secure after growing and harvesting their crops, even some are able to have surplus for sale, which enables them to have money to meet their daily needs as per the quotes in the key informant interviews. In some ways, floods enhance the communities' livelihood in the flood prone areas. It is crucial to understand the role of farmland in the flood prone areas. It reveals how the government and other development partners can learn from the existing strategies to develop strategies that ensure that the farmland in the flood prone areas is protected and that as part of recovery from floods, agricultural inputs should be included as part of the recovery support. communities clearly indicate that they are trying on their own to support themselves to cope with the floods. They also expressed that they would like the government to support the existing strategies to promote sustainable livelihoods rather than neglecting the relevance of their livelihood. communities living in the flood prone areas have been perceived as stubborn victims seeking sympathy from their governments and humanitarian organisations (Nyasatimes, January, 2015).. --- Crop Production The communities indicated that they grow various crops depending on the place and type of season. These farmers indicated that their main reliable crops are maize, rice and cotton. 'We usually grow rice in the flood plains because it is usually wet with more moisture which is not suitable for other crops such as maize. We also have vegetable gardens near river banks (in dambo areas) where we grow different types of leafy vegetables, tomatoes, onions, cucumbers, carrots, and sugarcanes amongst other crops' (FGD #4, TA Mbenje). Another participant shared that '...the soils we have here are also good for various fruits including mangoes, pawpaw, pears and bananas amongst the popular ones' (FGD # 2, TA Ndamera). According to the household survey, maize is one of the main staple food and is grown by approximately 95% of the farming households. Maize is grown twice in the year, during the normal rainy season and during winter through irrigation. During droughts, maize is still grown through irrigation. In addition, these farmers have been introduced to new farming methods that reserve moisture known as conservation farming in order to enhance crop production in the face of droughts and floods. This type of farming is being led by the Government through the Ministry of Agriculture and Food Security. Conservation Agriculture (CA), is the type of farming that is encouraged in areas that have been highly affected by environmental degradation and soil erosion and other types of farming that have resulted in loss of soil nutrients and soil cover (African Conservation Tillage Network, 2008). Conservation Agriculture (CA) aims to enhance agricultural production in areas that experience droughts and dry spells. Interventions, such as CA and Climate Smart Agriculture programmes are implemented across the whole of Malawi by the Ministry of Agriculture Government of Malawi ( 2012), without meeting specific and localised agricultural demands in some of the areas. In Nsanje district, about 80% of the targeted farming households revealed that agricultural production had reduced since the change of the old agricultural practices to CA whilst the other 20% of the farmers presented that they had noted an increase in the agricultural production. In addition, during FGD 1 in TA Ndamera, chiefs and some Civil Protection Committee Members indicated that CA is not suitable in the flood prone areas due to the nature of the soils and existing soil moisture levels. The findings on CA clearly indicate that national programs need to be tested at regional, local and even at a smaller scale, in this case, village level, in order to identify implementation challenges that need specific attention. The major highlight in this example is that such programs are not tested in the areas where it needs to be implemented before the actual implementation starts in order to check suitability and practicality of the intervention. There have been several criticisms of this type of farming as to whether indeed it increase the yields for farmers and reduces the workload for famers during weeding as it is claimed (Giller et al., 2009). In the Nsanje case, 40% of communities in the household survey suggested that conducting soil sample analysis would be appropriate to determine the type of soil that has accumulated over time through the flash floods to guide extension workers and farmers on the type of crops that would be suitable to be grown, which farming methods would be appropriate and the conditions that would be favourable to enhance crop productivity. A group village chief indicated that 'During the January 2019 floods, our farmlands have been affected badly. The soils have changed and the maize is no longer growing. We even tried potatoes, but they still have not grown, we do not know what to do now. The government and other organisations should do research and then advise us on what we should grow. We have been affected so much and we have run out of ideas' (KII #48, TA Mlolo). The current change of soil structure in some flood prone areas in TA Mlolo suggests that there is need for more agricultural research to identify suitable farming methods and appropriate crops to be grown. This also implies that the floods and unstable climate pattern will require more robust agricultural research to enhance agricultural productivity. Currently, in these flood prone areas, the change in soil structure due to floods in some areas and ineffectiveness of CA in other areas will exacerbate vulnerability of these smallholder farmers to floods and other effects of climate variability. In addition, my results indicate that the farmers who practice CA adopted it because they were advised by the agriculture extension workers to do so and that it was recommended by the government due to loss of soil nutrients and soil erosion. This is in line with Wall (2007) who suggests that successful adoption of CA depends on how well the farmers are aware of it and the problem they are facing in terms of low agricultural production. In this study, the farmers also indicated that they changed their ways of farming because they were having less yields and due to climate variability, they were experiencing droughts and lost most of their crops. However, for some farmers that have not experienced the benefits of CA, they indicated that they are hopeless because either way, there is a reduction in the crop yield and they said that they are waiting upon the government, through the Ministry of Agriculture to facilitate further research and come up with appropriate farming recommendations. Other farmers indicated that they are conducting various farming techniques on a trial and error basis and hope that one day they will discover for themselves what will work for them. This in a way indicates that some communities are being innovative by trying what they think can work for their agricultural land even though it is a long and frustrating process. As the climate is changing, with projected increase in rainfall intensity and drought events, it is becoming clear that new farming methods and types of suitable crops will be needed (NAPA, 2006). This therefore suggests the need for new studies on soil structures and textures in flooding sites to be able to provide adequate technical support to the communities. This also offers an opportunity for joint experimentation of agricultural research institutes and communities whereby new approaches to farming would be identified to enhance food security in the face of climate change. These farmers, through extension workers have been introduced to irrigation farming, drought tolerant crops and other transferable skills relating to agriculture. During floods, after their crops are swept away, they usually grow sweet potatoes. During drought and dry spells, they grow drought tolerant crops, mainly millet and sorghum. In terms of farming, agriculture extension workers play a major role in the type of farming these farmers are involved in. According to the farmers, the extension workers are government representatives hence are respected. However, it was observed that some communities followed what the extension workers taught them despite not understanding why they had to change the type of farming. Around 75% of the respondents when asked why it was important to follow new methods of farming, indicated that they had to follow what authorities say. This has various implications on the sustainability of the activities. The lack of ownership of interventions in communities' results in abandoning of the strategies when the projects end and there is no one leading the interventions. Some factors that have been attributed to as enhancing adaptation and development projects in Sub Saharan Africa are ownership, shared responsibility, transparency and community involvement therefore in cases when these factors are lacking, sustainability of project interventions is expected to be low (Ikejemba et al., 2017). CA interventions in Malawi are supported by international organisations such as the Food and Agriculture Organisation (FAO) through the Ministry of Agriculture. The findings that about 75% of the smallholder farmers implement CA activities because they are told to do so by agriculture extension workers suggests that if FAO changes focus and reduces support on CA, and if the Ministry changes its focus on another type of farming or intervention, it is very likely that CA will no longer be practiced by some farmers. This then reveals that there is inadequate and inappropriate training to the farmers to make them understand why they are doing what they do when introducing the interventions to ensure sustainability of the seemingly best agricultural practices. The results also illustrate that the farmers are not motivated to own the interventions and sustain them without the help of agriculture extension workers. There is lack of proper technical communication between the communities and agriculture extension workers which presents a lack of understanding on the new farming methods and technologies in farmers at the beginning of the interventions to ensure ownership and sustainability of beneficial interventions after the funding period is over. This raises concern on how sustainable some adaptation strategies are and how they will reduce vulnerability to climate variability in the long term. In areas where more people are poor, the adoption of CA and obviously some new technologies will be different and in other cases slower than anticipated (Cavanagh et al., 2017). In the flood prone areas, the findings illustrate that CA is not beneficial to all the smallholder farmers even though it is highly encouraged by the Ministry of Agriculture. The changes in the soil structure due to the fatal flash floods reveal that whilst other areas are benefiting from the fertile soils, communities in GVH Kadyampa in TA Mlolo have been negatively affected by the same flash floods. It is therefore evident that vulnerability contexts are different, unequally distributed and shaped by the way several physical and social factors interact, therefore adaptation strategies ought to incorporate the vulnerability context and address the specific challenges. In this case, the soil needs to be tested to identify the new type of soil that has been created due to the flash floods and therefore present the type of crops that can be suitable based on the soil characteristics and the physical environment. --- Post-harvest Handling In relation to post-harvest handling, all the research participants indicated that they currently store their maize in sacks designated for storing maize. These sacks are the regular bags for maize that are not treated and none of the research participants were aware of any new forms of storing maize. During conferences, I interacted with another Doctoral student working in a nearby community, who indicated that the Agriculture Research Institute in Malawi is planning to introduce the new forms of storing technologies to the smallholder farmers, such as treated empty sacks of maize and metal silos. These new storage technologies that will be introduced will be an entirely new concept for the farmers which as shown in my study has the potential of taking a long time to be accepted and established. In addition to that, some of the farmers do not produce enough for long term storage. This suggests that whilst communities are still struggling with how best they can produce more in some cases, the government is focusing on how to store the produce. According to the household survey, about 98% of the communities indicated that they do not suffer produce losses after harvesting, although this is possible during flooding and outbreaks of pests. However, it is important to note that I am not suggesting that the new storage technologies are not relevant in the areas, but rather that adaptation priorities need to be identified and understood. The discrepancies in the need and intervention by various programs indicate that there is inadequate dialogue between the communities that are affected by the floods and the service providers and donors. Older research participants indicated that they used to store their maize in Chikwa, which was a traditional form of silo. The Chikwa was handled with some cultural beliefs: only women and younger children were allowed to take maize from the silos because they were regarded as clean and not mischievous. Men and older children were not allowed because it was believed that if someone was mischievous in conduct, the produce would decrease. During that time, the maize would never go bad the whole time it was stored. Over the years, these traditional beliefs have been lost to new concepts of storage and all farmers have been advised to change. According to the agricultural extension workers, there was fear that the maize would be getting damaged by weevils and other pests hence the introduction of the new sack bags, whilst for some farmers, they indicated that the Chikwas used to be too big hence it was not necessary to use them with the amount of produce they harvested. Whilst agricultural extension workers indicate that they discouraged use of Chikwas because of weevils that attack the stored produce, over 90% of the farmers did not relate the change of storage due to that. This indicates that communication is a challenge between the extension workers and the smallholder farmers where the extension workers struggle to effectively pass on the technical and scientific explanation to the communities hence some communities do not know or understand why some changes are necessary. Even for a change in new farming technologies, most of the communities indicated that they heard from agricultural extension workers that they needed to change and they complied without understanding the reason behind it. The communication gap in this case could contribute to the slow adoption rate of adaptation strategies that might reduce vulnerability of the communities by enhancing their livelihood. During an in-depth interview with an extension worker, he indicated that it is true that sometimes they do not fully explain these scientific and technical issues to communities because they have huge pressure on their work and time is limited. In addition, he indicated that for some communities, they are not interested to know so due to a lack of adequate extension workers in the areas, they do not explain these things. This could be one of the reasons contributing to the death of projects after the development partners leave the area because there is no adequate ownership and understanding of the concepts within the local community. Apart from the farmland, crop production and post-harvest handling issues, I also examined the role of radios, mobile phones and bicycles as valuable assets that complement the livelihood of the communities in the flood prone areas. --- The Role of Livestock in Adapting to Floods Livestock also play an important role in the flood prone areas. Livestock is used as a source of prestige, exchange for shelter in the upland areas during flooding, and a source of income during seasons of crisis. 'In our community, owing livestock is a sign that someone is rich. Even though someone might not have cash readily available, the cash is tied to the livestock so we say those people are rich' (Community meeting during wealth ranking exercise, January, 2015). 'Those who have a variety of livestock such as cattle, goats, sheep, pigs, guinea fowl, ducks, chickens and doves are considered very rich. The others who only have chickens, guinea fowl, ducks and small livestock might not be considered very rich, but are considered to be better off. Those who have no livestock at all are considered poor in this community'. (KII #1, Nsanje District Council). An indication of wealth is subjective to the communities and district. During one of the wealth ranking studies I conducted in Chiradzulu in 2008 when I was working as a consultant, communities indicated that owing small livestock such as chickens, guinea fowls and doves was not a sign of being rich; being rich was associated with bigger livestock such as cattle, goats, pigs and sheep. In this community, the people who had chickens, guinea fowls and doves were considered to be poor because they were mostly beneficiaries of a livelihood project that target the ultra-poor. It is important to note that for the purpose of the livelihood of the communities living in the flood prone areas, the distinction between being rich, better off and poor is very important as it determines the adaptation strategies and survival during flooding. During the January, 2015 floods, poor smallholder farmers and those who had lost everything, including money lamented that those who had money, livestock or produce were rescued first by private boat owners who demanded payment instantaneously, leaving behind those who did not have any form of payment that was demanded by the boat owners. communities in Traditional Authority Mlolo indicated that 'During the floods, some people with private boats come to our rescue. They usually come with two boats, one to collect people and the other to collect a form of payment usually, bags of maize, rice and livestock, including chickens and goats. Those who do not have livestock or agricultural products are left behind' (FGD #11, TA Mlolo June, 2022). A relatively rich smallholder farmer explained that 'I gave the boat owners 2 goats and 3 chickens as a form of payment for me and my 5 children, the payment is subjective and it depends on how one negotiates the payment, it is subjective' (FGD #10, TA Mlolo, June, 2022. Another participant who belonged in the group that was perceived to be worse off in terms of economic status and therefore more vulnerable to floods lamented that 'During the January, 2019 floods, some of us who did not have money were left behind, I regretted to have been so poor. The boat owners refused to be merciful to me and just take me without any form of payment. They requested payment and said repeatedly that they were focussing only on those who had any form of payment. We ended up climbing the trees because there was no hope for us but we could not give up (KII #36 TA Mlolo, May, 2022). Another participant in the same group further lamented that that 'We had to climb trees until the government intervened. We were rescued by helicopters because everything was swept off, including our houses and everything was under water' (KII #40, TA Mlolo, May, 2022. The poor suffer more in the flood prone areas due to inadequate agricultural products and livestock that they could use to cope with the floods. Survival in the flood prone areas without government intervention thus depends on the type of assets a household has. The poorest of the poor, the elderly, orphans and those that could not afford to pay had to wait for the government arranged boats and helicopter, these are mostly the group that also survived the floods in trees whilst waiting to be evacuated by the government. During focus group discussions, the communities indicated that the communities lost their boats and did not know where the men with private boats came from. They emphasised that if they were from neighbouring villages, most of them would have negotiated to make the payments after the floods hence they could have been rescued without upfront payment. Communities who are perceived to be better off have two homes, one in the flood prone areas and another in the upland areas. Their livestock and produce is kept in both places unlike those that cannot afford to have another home in the upland area. During floods, those that have one home in the flood prone areas lose everything whilst those that have two homes have part of the livestock and produce in the upland areas hence, they sell part of that to recover after the floods. As indicated earlier on, the social networks are strengthened due to interdependency within the smallholder farmers. In the event where the communities in flood prone areas are destitute, the communities in upland areas do not keep them for free hence they have to live in evacuation camps until their land is dry. Communities offer livestock as payment for temporary shelter in the upland areas during the floods. 'During the floods, we either sell our livestock to get money that we use to pay for our rent in the upland areas or we give the tenants the livestock as a form of payment, it depends on the agreement' (FGD #7, TA Mlolo, April, 2022). 'Sometimes, the form of payment for rent in the areas that do not flood is both agricultural products such as bags of maize or rice and the livestock'. (FGD #1, TA Ndamera, February, 2022). Another participant indicated that 'the payment is subjective, for example in a month, one can pay 2 bags of maize and 2 chickens if the landlord wants the maize and the chickens. Sometimes, one can pay K20, 000.00 (approximately £20) it depends on the agreements and how close you are with the one offering the place' (KII #18, TA Mbenje, March, 2022). The results suggest that those who have livestock are more likely to survive without external support during the floods, since they can afford to negotiate the payment for temporary shelter, in this case, clearly suggesting that those with neither livestock nor surplus agricultural products cannot cope without external support. This further suggests that in times when the floods are unexpected and situations where the early warning system is ineffective, all the communities would not cope with the floods without external support since they would lose all their agricultural products and livestock. This clearly illustrates that the adaptation strategies that are implemented in flood prone areas such as irrigation farming and CA would only help communities to be food secure but do not adequately empower the communities to acquire assets that would enhance their adaptive capacity to cope with the floods, during and after the floods. Supporting the poor in the flood prone areas is not based on their needs hence does not meet the different demands of the poor in the various locations. A livelihood vulnerability study in Bangladesh indicates that the poor are more vulnerable to floods and depend on social networks and local authorities because they do have any asset that would enable them to cope with the floods (Toufique & Islam, 2014). There is need for the state to invest in various sectors including health, water and infrastructure in the flood prone areas (Ibid). There is no timely rescue plan at district (local) level to make certain that those who are poor and the vulnerable groups are pre-identified to facilitate the evacuation process. Timely evacuation from the flood prone areas during floods to established evacuation camps or to other relations and social networks is important to reduce vulnerability to floods that can be fatal if evacuation is delayed. It is clear that regardless of the efforts to stop the smallholder farmers from living in the flood prone areas by the government not providing all the necessary amenities in the areas, including boreholes and schools, these areas are still habitable and will continue to be habitable. It is clear that communities that are relatively poor will need adequate support from the government and other development partners to be able to cope with the floods and evacuate before the floods threated lives. In the next section, I will discuss the role of social networks in the flood prone areas as communities cope with the floods. --- The Role of Institutions and Organisations in Adapting to Floods According to Hodgson (2006), institutions are defined as systems of established and embedded social rules that structure social interactions. Institutions are there to support local adaptation at both the household, community, and district level hence enhancing local adaption to floods and droughts. In addition, churches, schools, households, communities, government department, civil society organisations, nongovernmental organisations have also supported the communities and households to cope with the floods. --- Household and Community Dynamics, and Adapting to Floods Only 10 out of the 227 households indicated that the responsibility to provide for the home lies between both the male head and the spouse. In this case,

A section of Nsanje District has been experiencing floods for over a decade and was declared a flood prone area the same time (GoM, 2006). Therefore, firstly, this research seeks to understand how the communities that live in the flood prone areas perceive their vulnerability to floods and how some of them have been coping with the floods since 1952, just over 6 decades. The overall aim of the research was to understand the impact of floods on sustainable development and how communities living in flood prone areas perceive their vulnerability to floods and climate variability, to investigate the power dynamics at household and community level and to explore the complexities associated with local adaptation programs in the flood prone areas. This research contributes towards framing vulnerability based on the perception of the different groups of people that are vulnerable to climate variability and have experienced environmental changes throughout their everyday lives. The communities living in the flood prone areas perceive floods as part of their livelihood. The communities have experienced floods for over 5 decades and have been implementing adaptive strategies to help them cope with the floods during this time. Temporary migration is one of the effective adaptive strategies that the communities living in the flood prone areas implement during floods.

the government not providing all the necessary amenities in the areas, including boreholes and schools, these areas are still habitable and will continue to be habitable. It is clear that communities that are relatively poor will need adequate support from the government and other development partners to be able to cope with the floods and evacuate before the floods threated lives. In the next section, I will discuss the role of social networks in the flood prone areas as communities cope with the floods. --- The Role of Institutions and Organisations in Adapting to Floods According to Hodgson (2006), institutions are defined as systems of established and embedded social rules that structure social interactions. Institutions are there to support local adaptation at both the household, community, and district level hence enhancing local adaption to floods and droughts. In addition, churches, schools, households, communities, government department, civil society organisations, nongovernmental organisations have also supported the communities and households to cope with the floods. --- Household and Community Dynamics, and Adapting to Floods Only 10 out of the 227 households indicated that the responsibility to provide for the home lies between both the male head and the spouse. In this case, they both participate in income generating activities and sometimes they share responsibilities where the woman's responsibilities are less than the man's responsibilities. 'The men usually make more money than the women so they are responsible to address more needs in the home whilst for women, it is usually supplementing what the men provide in the home' (FGD # 8, TA Mlolo). In this case also, it can be implied that men are still held responsible for the management of the home, where the man is alive and responsible and women only supplement what the men provide and when the men are not there. The household head is expected to meet the household need by virtue of their role as the head. Figure 8 shows who the responsibility for decision making for everything falls at household level amongst participants in the study, revealing that this mostly falls on men. The participants indicated that even in climate related decisions, men make the decision and women do not participate in the adaptation decisions in the home. This situation is another example of exclusion at household level where women do not take part in decision making even though they are also affected with the decisions that are made. During in-depth interviews with communities who had leadership positions and during focus group discussions, all the communities living in flood prone areas indicated that they consider floods as the main problem even though hunger is also amongst the challenges they experience (and was reported as the most important problem at the household level through the household survey). The contradiction in response could be explained by difference in community and household priorities. The results reveal that households focus more on individual and personal problems whilst in a group, the farmers will mention problems which affect them as a group, at the community level. It was also apparent that during in-depth interviews with some chiefs and members of the village committees, floods and droughts were mentioned as major problems affecting the communities, as community heads in this case. This indicates that chiefs and community leaders consider community problems more and are concerned with them personally unlike communities at household level whose concern is mainly on their household's welfare and livelihood. These findings illustrate that household members could be interested to participate in activities be it adaptation strategies or development activities that firstly addresses their household needs, otherwise they will always focus on how to deal with their personal, household problems first. This is an issue that demands attention when designing adaptation programs because reducing household vulnerability to floods and droughts require more collective strategies than individualism (Siegel and Alwang, 1999). Siegel and Alwang (1999) also indicate that for poor rural households, the ways of coping with either droughts or floods are usually not economically sustainable; they are short term thereby increasing household vulnerability to the climate variability they are trying to cope with. The farmers during my interviews indicated that when they are pressed with a need, they first of all consider doing piece work as a source of income, which will help them to address the current household need. If that fails, then they resolve into selling household assets such as livestock, bicycles, radios, phones and other agricultural products including the maize they store to take them to the next harvesting season. Now this in itself is not sustainable and can cause problems because like in the case of the farmers I interviewed, the price of the bags of maize they sell and whatever asset they sell when they are desperate for money is usually less than the amount, they will need to replace it. This in itself exacerbates their poverty and vulnerability during drought and floods. This strategy of selling assets for payment of services leaves the communities worse off because they never invest in other assets but instead, they spend the proceeds on consumables and perishables. This clearly illustrates that there is not any adaptation intervention that empowers the poorest of the poor economically to enable them to address their day-to-day problems; whilst also empowering them to be able to help themselves when disaster strikes. In addition, this suggests that because these poorest of the poor communities isolate themselves, they are likely to continue being more vulnerable when a disaster strikes. There are no deliberate efforts designed to encourage such communities to take part in community activities and also belong to certain social networks that would enable them to learn other transferable skills and empower themselves. The findings suggest that a deliberate social group at village level that would support the vulnerable groups at all times can be explored as a necessary aspect within the local adaptation strategy. --- The Role of Chiefs in Adaptation to Floods Community leaders and chiefs are approached when there is a community crisis or need unlike with a crisis at the household level where the head of that particular household is turned to. Out of the 227 households survey, 68% indicated that it is the village headmen who are approached when there is a disaster in the community, including floods and droughts (Figure 9). According to fieldwork discussions, the majority of the communities indicated that community leaders provide security and communities' short and long term solutions during a crisis, including a disaster such as flooding. The participant who indicated that they consulted no one during a crisis further indicated that 'we feel left out in the community, we are not regarded as part of the community when it comes to making decisions, our voices would not be heard at all. However during floods, we are rescued together with other affected people and during social cash transfers that are meant for the poor, we are also considered sometimes. Our chief is still responsible for us.' (FGD #5, TA Mbenje). This group according to the communities themselves included the most vulnerable households, the poorest of the poor, the aged and farmers who do not usually have social networks within and outside their living environment. This also verifies that there are various social groups within the affected communities that are left out in community activities because of their socioeconomic status and social group identification. However, 10 out of the 30 households that indicated not to consult anyone when a disaster strike indicated that they do so as a result of personal preference. Twenty households indicated that they feel that they are not welcome in other social groups, that they are stigmatised and hence withdraw themselves from such social groups. Apart from chiefs, there are church leaders, teachers and the police play a significant role during flooding. --- BARRIERS AND LIMITATIONS TO LOCAL ADAPTATION It is clear that gender and culture play a very important role in the lives of the smallholder farmers living in the flood prone areas. In this chapter, I will discuss how misconception of power dynamics and gender issues at both household and community level by policy makers, governments, the affected communities themselves and various institutions that support adaptation to climate variability exacerbates vulnerability and limits adaptation in the flood prone areas. In addition, this chapter illustrates that there are also some factors beyond the factors that are mentioned in the local adaptation frameworks, (Africa Climate Change Resilience Alliance (ACCRA), 2010; Engle, 2011;Gupta et al., 2010) that need to be adopted in order to comprehensively assess the adaptive capacity of households, communities and institutions. The factors that are highlighted in this chapter illustrate the factors that affect local adaptation in Nsanje and in Malawi therefore can be similar to other areas that have similar characteristics. However, this thesis is clear that vulnerability is differentiated, with some groups of people being more vulnerable than others. The adaptation strategies therefore should be different to suit and address the different levels of vulnerability and groups of people (Pelling et al., 2015). Finally, this thesis illustrates that some of the challenges to local adaptation are dynamic and more complicated hence present themselves beyond the categorised barriers and limits. Their manifestation therefore require more transformation in the way adaptation issues are strategised in order to effectively reduce vulnerability to climate variability. --- Gender imbalance Gender plays a hugely important role in creating vulnerability and has an opportunity to enhancing adaptive capacity amongst those living in flood prone areas. In terms of gender roles, the men are the head of the family and hence the decision maker in most households. Women are not given a chance to share their views. During focus group discussions, one woman indicated that 'even if there are rare cases when the men would listen to the women's view, their views are never considered and the men still do what they have decided' (FGD #11, TA Kunthembwe). According to the discussions and in-depth interviews, it was revealed that it is the men that make overall decisions in the home for various reasons. During a focus group discussion with male chiefs and community leaders, the men explained that 'we are men and household heads so we are supposed to make all the important decisions in the home, in actual fact, we are supposed to make all the decisions in the household' (FGD #1, TA Ndamera). Men are considered to be the head of the household therefore they feel they are supposed to make decisions in the home. The attitude of being a man and hence needing to make decisions in the home has contradictory consequences. In the positive, it gives the men the responsibility to make certain that the households are surviving in terms of availability of food and other household necessities. For any man to be considered responsible, these aspects could then mean that whilst the women are focusing on managing the home and farming, the men would focus on providing for the home and farming. However, the majority of women in the farming household and during focus group discussion indicated that this concept of men making decision in the household has led to women being passive. 'We are regarded as voiceless, powerless and not knowledgeable. In our homes, we cannot participate in any decision making. If we speak up, we are considered rude and stubborn, hence we just have to agree to everything that a man says' (FGD #7, TA Mlolo). In addition, one male farmer said 'a woman is always a woman, a man who bows down to what a woman says is considered to be a coward in our society. Where would they get the knowledge from? They know nothing!' (FGD # 6, TA Mbenje). Most of the men during the men focus group discussions also indicated the same which then implies that women are powerless and voiceless. During the focus group discussions with both men and women, women were respectful of the men and even chose not to argue with them accordingly when they were raising issues that were oppressive. The women only felt able to make unpleasant facial expressions and only made comments about it after I assured them that the issues that we were discussing were meant for studies and I would not expose them to community leaders and other people in their community. This idea reflects women are oppressed in their environments. There are many cases of injustice and unfair decisions that are made to the disadvantage of women and yet the women themselves are voiceless. This is one of the issues that is critical to enhance participatory decision-making at household level that would then lead to households being resilient to their own problems. In this case, strategic women's groups that would support and empower women and empowering women through education can partly inspire women to take part in meaningful discussions at household level. However, there are also challenges with this approach as will be discussed in the next section which suggests that transformation is required in order to reduce vulnerability to floods at all levels. --- Household Power and Capacity to Cope During a focus group discussion with women only, women articulated that 'If only these men were very supportive, we could have gone so far in terms of adaptation. It is difficult to effectively adapt to the floods and climate variability because the men pull us down at household level which also exacerbates household poverty. That is why we are failing to develop' (FGD #2, TA Ndamera). Another woman indicated that'many of us here are not happily married, our husbands have other girlfriends and they spend most of their time and money there. As women, we struggle on our own to provide for the home, food for us and the kids. We do small businesses and when we are coping and making profits, these men come and take the money from us. They say they have better plans with the money only to go and drink and go to other women and that's how we live' (KII #24, TA Mbenje). Another woman said that'men are very powerful here, sometimes, if you don't comply, they actually beat you up so to avoid being beaten, we surrender the money and we struggle again' (FGD #8, TA Mlolo). It is evident hereby that issues of gender and power dynamics at household level offer an opportunity of abuse for those that are weaker in the society. Abuse of power at household level and women not being given an opportunity to participate in household decision making reveals that unless adaptation strategies are deliberately inclusive of these specific gender issues, women will not benefit from adaptation interventions. Inclusive adaptation and deliberate strategies that incorporate women and encourage participation of women at both household and community level would eventually encourage women to actively take part in adaptation strategies that would eventually enhance household resilience to floods and other effects of climate variability. Dealing with attitude and breaking abusive and non-constructive cultural norms is necessary in flood prone areas to enhance the effective design of adaptation programmes. As Codjoe and Issah (2016) allude to in a culture and adaptation study in communities in Ghana, local cultural context is very important in the design of effective adaptation options. The authors found out that adoption and participation in adaptation strategies was mainly due to cultural norms in different communities hence suggesting that culture is a component that should be incorporated in the design and implementation of local adaption strategies. There are also cases where women themselves have low self-esteem, which also puts them at risk of being abused. 'We all need a man at some point and we are respected when at least people know we are married regardless of what the man does' (FGD #3 TA Ndamera). Many women agreed to this point and it has a reflection on how women feel they are so vulnerable and can never do anything on their own to make their lives better. This aspect of gender and power dynamics at household level has a negative impact on household adaptation hence exacerbates vulnerability of the household level. In addition, it is the women that are more stressed and under pressure because they work so hard to make ends meet whilst some men take advantage of them. The findings reveal that adaptation strategies are not comprehensive in addressing these issues that negatively affect adaptation at the household level. In addition, this suggests that women cannot be protected and empowered when they are living with abusive men. They fail to negotiate with their men to come up with collective solutions that will enhance resilience to floods and droughts in the household. Adaptation strategies do not address all aspects of household power dynamics therefore most women do not take part in climate related and livelihood decision at the household level. Climate related decisions at the household level are not done in a respectful and mutual understanding manner therefore fail to collectively address climate related challenges. Every individual in a household needs to be able to understand how they can reduce their vulnerability to floods and enhance local adaptation at household level to be resilient to climate variability, failure to which household vulnerability to floods and poverty levels will be exacerbated. My understanding in these findings is that if women are given much support, adaptation initiatives would be fruitful as they are practically home managers and responsible for household needs and events even though commonly, men are considered to be the head hence heading everything in the household. Women need to be supported and empowered to be able to negotiate with men to overcome obstructive traditions. --- CHAPTER FIVE SUUMARY, CONCLUSION RECOMMENDATION --- Introduction This final chapter summarizes the overall findings and contributions of the thesis, focusing on how the research questions have been answered in relation to self-perceived vulnerability to floods and local adaptation. The chapter presents significant findings on how complex vulnerability and adaptation issues are, including how they are generated and exacerbated in the contexts of the everyday lives of smallholder farmers living in the flood prone areas. In addition, the chapter highlights the disadvantages of permanent relocation process as a means of reducing vulnerability to floods in the flood prone areas versus the benefits of temporary migration as a commonly practised adaptation in the flood prone areas. The chapter then presents some of the implications for local, national and global adaptation to climate change plans and strategies, and suggests that the generic adaptation strategies are exacerbating vulnerability to climate variability. There is a need to contextualise vulnerability, understand how it is created, how the affected people perceive the vulnerability and how they are adapting to the environmental changes in order to meaningfully contribute towards reducing their vulnerability by enhancing their adaptive capacity in situ. --- Discussion of the Findings Environmental migration has been framed differently by various institutions based on their agendas. Ransan-Cooper et al., (2015) highlight the different terms and situations that define environmental migrants as adaptive agents, security threats, victims and political subjects. Furthermore, they highlight that the different frames have an implication of the how the policies that are formulated and thereby revealing power relations in the way adaptation projects are designed. The review of these various frames indicates how external actors and institutions perceive environmental migrants from particular viewpoints, whilst limited information and studies have been conducted to understand how the environmental migrants themselves perceive their vulnerability. It is evident that some groups of people in various regions of the world are more vulnerable to climate variability than other (Adger et al., 2003). Developing countries suffer the effects of climate change more than the developed countries because they do not have the financial and technical capacity to cope with the floods (UNFCCC, 2013). My research illustrates the contexts that shape and define the vulnerability to floods and climate variability in the flood prone areas and contributes to other scholarship on how the social, economic, political, cultural and natural factors interplay in various contexts to either exacerbate or reduce the vulnerability (e.g. Harrison and Chiroro, 2016;Tucker et al., 2015). Bryan et al., (2009) indicate that smallholder farmers will be vulnerable to climate variability more because they rely on agricultural products which will be directly affected by the extreme weather events. In another perspective, Dilling et al., (2015) reveal that vulnerability is dynamic and complex with various factors interplaying at the local level to shape and define vulnerability. The need therefore to understand the underlying causes of vulnerability and local adaptation in context is relevant to inform climate change policies and shape the design of climate change adaptation programs that seek to reduce vulnerability and enhance adaptive capacity of those that are heavily affected (Ribot, 2014). My research demonstrates that there is a discrepancy in the understanding of vulnerability to floods and local adaptation between the communities who experience the floods and the Government of Malawi, through the Department of Disaster Management Affairs. Firstly, this research contributes towards framing vulnerability based on the perception of the different groups of people that are vulnerable to climate variability and have experienced environmental changes throughout their everyday lives. The communities living in the flood prone areas perceive floods as part of their livelihood. The communities have experienced floods for over 5 decades and have been implementing adaptive strategies to help them cope with the floods during this time. Temporary migration is one of the effective adaptive strategies that the communities living in the flood prone areas implement during floods. This is contrary to how the government and the general public perceive environmental migration as an adaptation strategy, they suggest that permanent relocation is the best solution that would effectively address vulnerability to floods. Ransan-Cooper et al., (2015) presents a clear understanding of perceived vulnerability by various actors, including international organisations, politicians, governments and scholars that suggest that affected communities are helpless. Contrary to such understanding, communities living in flood prone areas engaged with in this thesis are aware that the environment is changing and have designed adaptive strategies that have enabled them to cope with the floods in a way that is beneficial to them. The communities spoken to during the study indicated that temporary migration during severe floods is normal for them and forms part of their livelihood. They are aware that they have to move during the flooding season and many smallholder farmers have set aside plans that facilitate the temporary migration. Social networks and inter-dependence between the communities living in the flood prone areas and those that live in the upland areas have made temporary migration during floods easier for the communities living in the flood prone arear. The communities living in the upland areas also relocate to flood prone areas during droughts, thus benefitting from agriculture through moisture in the flood prone areas. The strong relationship between these communities through social networks and interdependence clearly indicates that the smallholder farmers living in flood prone areas are not victims, neither are they security threats to neighbouring communities and country, Mozambique as suggested by some governments (Ransan-Cooper et al., 2015). However, it is evident that these communities are 'partially' adaptive agents and have become political subjects over time due to external power dynamics. These smallholder farmers are affected by both international and national disaster policies that present them as victims, severely exposed to floods and therefore they are helpless in their environment. The smallholder farmers in the case studies illustrated that permanent relocation to upland area would exacerbate their vulnerability to climate variability and that their livelihood is better managed and sustained in the flood prone areas. The communities indicated that they possess valuable assets such as agricultural land, livestock, bicycles, radios and mobile phones that sustain their livelihood in the flood prone areas. Secondly, communities exemplified that climate change adaptation strategies and policies in some areas will exacerbate vulnerability to climate variability and create new vulnerabilities that the affected communities will eventually fail to cope with. It is important to realise that the adverse effects on climate change can never be isolated from the political, social economic and cultural factors that interplay in the communities therefore suggesting that the same factors interact to reduce the impact of the disasters (Methsmann and Oels 2014). This thesis presents a unique and important social relation, revealing interdependence between the communities living in the flood prone areas and those that live in upland areas. The farming households in upland areas migrate temporary to the flood prone areas during droughts for farming purposes whilst the farming households in the flood prone areas migrate upland temporarily for shelter during floods. This in itself is a novel way of understanding how different communities affected by different climate hazard depend on each other as part of adapting to the hazard. The power of social relations and interdependence is usually undermined in adaptation programmes and policies. Selling of assets has been well articulated as an adaptation strategy (Enete et al., 2016) but this thesis presents evidence of a more sustainable way of utilising the land in flood prone areas through letting out part of the farm land to communities that are affected by drought. Adger (1999) explains that communities that are vulnerable to climate variability will find strategies within their localities within which they can use to adapt to climate variability. Land is an important and reliable asset for the communities living in flood prone areas because apart from yielding crops and produce from it, it is also a reliable source of income. The motivation to live in the flood prone areas mainly comes due to the possession of land which is used for agriculture and hiring out to communities that live in the upland areas during drought. The seasonality of floods, considering that they do not occur every year, presents a benefit to the communities because the upland communities are affected more with droughts than they are with floods. Overall, in cases where floods do not occur every year, temporarily, migration is an attractive adaptation strategy because the communities still have access to their farmland and possess it permanently which then enables them to rent it out during droughts as an alternative source of income in addition to agriculture. Thirdly, this research presents evidence that the complexity of local adaptation arises from the political, economic, social, cultural and institutional factors and processes that interplay within the households and communities. Sometimes, these same factors and processes work against effective local adaptation at both household and community level. However, Engle (2011) claims that adaptive capacity of a nation or community that is affected by climate variability is influenced by the institutions, management and governance. The findings presented in this thesis illustrate that at household level, women are not given a chance to contribute to household resilience to climate variability because of cultural values that suggest that women are subordinates and that therefore only men have the authority to actively participate in such developmental activities. For fear that women will become empowered and hence would be 'rude' and 'forget their household roles', men do not allow women to participate in activities that would empower them, thereby retarding household reliance to floods. Most importantly, both women and men are oppressed at household and community level even though due to culture, the oppressed accept the oppression as normal behaviour. Some cultural values work against efforts to make households resilient to floods thereby revealing that the need to address cultural values that households and communities hold to be able to address some fears and myths that work against effective adaptation to climate variability including floods. The need to understand gender issues and social groups should not be undermined as these shape the adaptive capacity of the households and communities (Madhuri, 2016). This study therefore exemplifies the dynamic and complexity of gender roles and expectation within the flood prone areas and women mostly are taken advantage of due to misconception of power and understanding of respect. In addition, politicians at all levels exacerbate the vulnerability of these communities to floods. It is evident using a Malawi case that politicians are not primarily interested in investing in disaster risk infrastructure, but rather votes. In contrasting evidence Obradovich and Zimmerman (2016) found that across Sub Saharan countries, with evidence from Malawi and South Africa, the voters are not interested with climate change policies and that they do not support the implementation of such policies. This evidence suggest that voters are interested in other things that constitute their welfare therefore politicians respond to those issues to be able to win votes. In the case of Malawi however, as illustrated in this thesis, politicians are actively involved in organising help in the moment when communities are affected by the floods even though they do not invest in longer term and more permanent disaster risk infrastructure. During voting period, the polling stations are positioned in the flood prone areas for voters to vote even though the government stopped providing for amenities in those areas after indicating that those areas are prone to floods and therefore not suitable for occupation. This research highlights a political aspect of vulnerability and local adaptation where politicians are influential but looking for the votes; and how the voters themselves are vulnerable to climate variability, they are not influential, but have the voting power. Fourthly, it is evident that an effective early warning system is desirable to reduce vulnerability to floods in the flood prone areas. A combination of indigenous knowledge, hydro-geology and sophisticated weather readings is essential to build an effective, reliable, meaningful and trusted early warning system. Flood management in developing countries remains a challenge due to economic, financial and technical challenges (Rahman and Di, 2007). Due to the intensity of the flash floods in Malawi in 2015, the indigenous early warning system there was not able to provide timely warnings to the communities of the flash floods. In addition, the way in which earth and rock movements persisted before the floods came and communities failed to realise that floods were coming explains how sophisticated the early warning system needs to be. Raju et al. (2016) explains the role of advanced remote sensing technology in predicting flash floods through depicting soil moisture content over time. This thesis suggest that a combination of advanced remote sensing, hydro-geological measurements and indigenous knowledge on the early warning systems needs to be encouraged in order to deal with the complex floods that are more becoming fatal than in previous years. Fifthly, In terms of adaptive capacity, with reference to the Adaptive Capacity Wheel (Gupta et al., 2010) the thesis illustrates that at both household and community level, leadership, resources and room for autonomous change are very crucial in enhancing the adaptive capacity of the communities It is evident however that processes in fair governance, variety and learning capacity are not adequately present in the areas. Communities and households only achieve half of the adaptive capacity wheel but are partly striving in the flood prone areas. This entails that communities are not homogenous in nature hence factors and processes that enhance their adaptive capacity will vary. With reference to The Local Adaptive Capacity Framework (ACCRA, 2011), however, the thesis provides evidence that the institutions are resource tight, not powerful enough to make decisions and provide sustainable solutions in the flood prone areas. Also, that entitlements are not clearly visible in many deserving smallholder farmers. However, at the household level, the asset base, innovation, knowledge and information, and flexible and forwardthinking decision making was evident in households that are able to cope with the floods. The findings in relation to adaptive capacity illustrate that household adaptive capacity is critical in flood prone areas and that community and structural responses to climate change are not reliable. An increase in the number of households that are able to cope with the floods on their own without financial support from the government and humanitarian organisation will enhance the adaptive capacity within the flood prone areas. Whilst community work is not totally discouraged, this research provides evidence that understanding households' adaptive capacity portrays a true reflection of vulnerability and adaptation challenges and provides foundation opportunities on how best climate change interventions can be designed. --- Areas for Further Research This thesis presents findings from research that took place at one moment in time, significantly either side of a major flood event. There is need for other similar studies in the same areas with the same communities in order to find out if there will be significant changes in terms of how these communities will be coping with the floods over time. Such studies would be important in order to highlight if there will be any changes in terms of the adaptive capacity and vulnerability context over a variety of time scales. There is need for research that would seek to find out to what extent women empowerment projects as an adaptation strategy would enhance the household adaptive capacity. Finally, I would recommend research that will examine, understand and explore the relationship between the state, politicians and voters in the disaster prone areas in order to explore opportunities and challenges that exist between these three groups of stakeholders that are also critical in reducing vulnerability to the effects of climate variability at village and community level. --- Chapter Summary This study provides evidence that floods have become more severe and fatal in some places, to which communities are having to deal with. In the effort to reduce vulnerability to climate variability, UNISDR suggests that communities that live in flood prone areas should relocate to safe areas to reduce vulnerability to climate variability. The Malawi Government declared some areas in Malawi flood prone areas and requested those that live in the areas to relocate to safer areas. There has been resistance by the affected communities to relocate upland.

A section of Nsanje District has been experiencing floods for over a decade and was declared a flood prone area the same time (GoM, 2006). Therefore, firstly, this research seeks to understand how the communities that live in the flood prone areas perceive their vulnerability to floods and how some of them have been coping with the floods since 1952, just over 6 decades. The overall aim of the research was to understand the impact of floods on sustainable development and how communities living in flood prone areas perceive their vulnerability to floods and climate variability, to investigate the power dynamics at household and community level and to explore the complexities associated with local adaptation programs in the flood prone areas. This research contributes towards framing vulnerability based on the perception of the different groups of people that are vulnerable to climate variability and have experienced environmental changes throughout their everyday lives. The communities living in the flood prone areas perceive floods as part of their livelihood. The communities have experienced floods for over 5 decades and have been implementing adaptive strategies to help them cope with the floods during this time. Temporary migration is one of the effective adaptive strategies that the communities living in the flood prone areas implement during floods.

description and Table S1. All of the datasets we consider describe the face-to-face proximity relations of the monitored subjects, with a temporal resolution of approximately 20 seconds 7,22. For every pair of individuals, the full sequence of individual interactions is resolved, with starting and final timestamps for every close-range proximity relation. These data can be represented as time-varying networks of proximity: nodes represent individuals and a link connecting two nodes indicates that the corresponding individuals are in contact, i.e., in face-to-face proximity of one another. --- Results Epidemic processes and activity clocks. We probe the temporal structure of the empirical networks with a simple Susceptible-Infected (SI) process. The population of nodes (individuals) is split into two compartments: susceptible nodes (S), who have not caught the ''infection'', and infected nodes (I), who carry the ''infection'' and may propagate it to others. In this simple epidemic model, infected nodes never recover. A node is randomly selected as the seed from which the infection starts spreading deterministically, through contacts between a susceptible node and an infected one (S 1 I R 2I). Transmission events are assumed to occur instantaneously on contact. We fingerprint the temporal network structure of the data by computing the times at which the epidemic process reaches the different nodes. Specifically, we focus on the probability distribution of arrival times for the SI process unfolding over the temporal network. In terms of wall-clock time, the arrival time at a given node is defined as the time elapsed between the start (seeding) of the SI process and the time at which the process reaches the chosen node. It has been shown 26 that the distribution of these arrival times is extremely sensitive to several heterogeneities of the empirical data, to the seeding time. In general, it displays strong heterogeneities due to the nonstationary and bursty behavior of empirical temporal networks that cannot be captured by simple statistical models. Thus, we shift to a node-specific definition of ''time'': each node is assigned its own ''activity clock'' that measures the time that node has spent in interaction or, similarly, the number of contact interactions that node has been involved in. The ''time'' measured by this clock does not increase when the node is isolated from the rest of the network. In the following, for clarity, we will indicate with ''time*'' the activityclock readings. The ''arrival time*'' of the epidemic process at a given node is defined as the increase of its activity clock reading from the moment the SI process is seeded to when it reaches the node. Arrival times* discard by definition many temporal heterogeneities of the empirical data and usually exhibit a well-defined distribution 26 that is robust with respect to changes in the starting time of the process and across temporal networks of human contact measured in different contexts. In the following we use activity clocks based on the number of contact events a node has been involved in. The arrival time* at a node, consequently, will be integer-valued and will measure the number of interactions each node was part of from the seeding of the epidemic until the node was infected. For each empirical time-varying network, we generate a hierarchy of synthetic temporal networks using both a top-down and a bottomup approach. The synthetic networks are designed to support our analysis by selectively retaining or discarding specific properties of the empirical data. Top-down approach: null models. We generate null models by applying to the empirical data randomization procedures that erase specific correlations 24. We keep the topology of the contact network unchanged. In the ''interval shuffling'' (IS) procedure, the sequences of contact and inter-contact durations are reshuffled for each link separately, while in the ''link shuffling'' (LS) procedure 24 the unaltered sequences of events are swapped between link pairs. Both procedures destroy the causal structure of the temporal network, but they both preserve the global distributions of contact durations, inter-contact durations, and number of contacts. The IS procedure also preserves, for every link, the total number of contact events and the cumulated interaction time, while the LS procedure does not conserve these quantities at the link level. We also consider a global time shuffling procedure (TS): we build a global list of the empirical contact durations and, for each link, we generate a synthetic activity timeline by sampling with replacement the global list of contact durations according to the original number of contacts for that link. While the global distribution of contact durations and of the number of contacts per link are conserved by construction, all temporal correlations are destroyed and the distributions of inter-contact times differs from the empirical one. Figure 1 illustrates the three randomization procedures defined above. All the procedures conserve the topology, the distribution of contact durations and the distribution of the number of contacts per link of the empirical networks. Table 1 summarizes the impact of the randomization procedures on different properties of the temporal networks. Bottom-up approach: generative models. We also define generative models for random temporal networks designed so that the resulting time-varying networks exhibit specific properties of the empirical data, in the spirit of the configuration model for static networks 29. We start by creating a static random Erdo <unk>s-Re <unk>nyi network with the same number of nodes and the same average degree of the temporally-aggregated empirical contact network. Then we assign to each link a sequence of synthetic contact events, according to different strategies. In the Inter-Contact Time model (ICT) we impose that the global distribution of inter-contact durations is the same as in the empirical data (see the Methods section for details). This is an important case to test against, as it is often considered in the literature that the distribution of inter-contact times plays an important role in determining and constraining spreading processes over temporal networks 26. Contact durations are fixed and equal to the average contact duration measured in the empirical data. In the Inter-Contact Time plus Contact-Per-Link model (ICT 1 CPL) we proceed as in the ICT case, but also impose that the distribution of the number of contact events per link must match the empirical one. In summary, in both models the topology and the contact duration distribution differ from the empirical ones. Table 2 summarizes the properties of the generated temporal networks that are constrained to match those of the empirical data. Arrival times measured with activity clocks. From each empirical network, we build synthetic networks according to each null and generative model. We simulate SI processes on both empirical and synthetic networks for different starting times and for different choices of the seed node. We then compute the distributions of arrival times* measured in terms of activity clocks. Figure 2 (panels a and b) compares the arrival time* distributions from the empirical data (HT09 conference and hospital datasets) with those yielded by the null and generative models. The results for the SFHH conference dataset are reported in the SI. In order to provide a quantitative assessment of the distribution similarity we compute the symmetrized Kullback-Leibler (KL) divergence 30 (see Methods) between the distribution of arrival times* for the empirical data and for each model. Given the relevance of large arrival time* values, which may be strongly influenced by causal constraints and in general by the peculiarities of the temporal structure of the network, we also compute the Kullback-Leibler divergence between the tails of the distributions. To this end, we only take into account arrival times* longer than a fixed threshold arbitrarily set to 10. We refer to this restricted Kullback-Leibler divergence as ''KL101'', and we have checked that our results are robust with respect to changes in the threshold. Table 3 and Fig. 2 report the symmetrized KL and KL101 divergences for the conference and hospital datasets we consider. Determinants of the arrival time distribution. Using KL and KL101 as guiding metrics we use the top-down approach to discard features that are unimportant in reproducing the arrival time* distribution and to narrow down a set of necessary features. We then use the bottomup approach to find the features that are sufficient to model the arrival time* distribution. The Interval Shuffling (IS) and Link Shuffling (LS) procedures lead to distributions of arrival times* similar to those of the empirical data. This indicates that the causal structure of the temporal network has a small impact on this distribution. Moreover, LS does not preserve the specific assigment of cumulated contact durations v AB and number of contacts n AB to individual links (see Table 1): we can therefore discard these as explanatory factors of the specific shape of the arrival times* distribution. According to Table 3 and Fig. 2 (panels c-f) the Time Shuffling (TS) procedure yields a very different distribution for the conference datasets, and a different tail for the hospital data. We know that, by design, the TS procedure does not preserve the distribution of intercontact intervals, which is directly related to the burstiness of contact activity. The failure to adequately model the arrival times* distributions stems thus from this feature and can be related to previous results 19,23 showing that burstiness plays an important role in spreading phenomena. Indeed, the distribution of inter-contact interval durations for the synthetic networks are quite different from those measured for the empirical networks (not shown, established by comparing the KL divergences between the corresponding distributions). In the hospital case, this difference between empirical and synthetic inter-contact interval durations is reduced, leading to the reduced difference in arrival times* distributions observed in panel d of Fig. 2 for the TS model. However, panels c-f of Fig. 2 also show that the distributions of arrival times* obtained for the ICT model, which is designed to preserve the distribution of inter-contact durations, exhibit KL differences that are similar or even larger than those of the TS case discussed above. The corresponding distributions in panels a and b of Fig. 2 are indeed much narrower than the ones obtained with the empirical dataset. This shows that correctly reproducing the distribution of inter-contact durations is not sufficient to adequately model the arrival time* distributions. In order to achieve that, we need to add to the ICT model the additional constraint of preserving the distribution of the number of contacts per link, i.e., to use the ICT 1 CPL model (see Table 2). This model captures the essential features of the data that are sufficient to reproduce the arrival time* patterns of the empirical data, especially for the tail of the distributions, as shown in panels e and f of Fig. 2. We remark that this model is quite parsimonious, as it does not retain the topology of the empirical network nor the distribution of contact durations or cumulated contact durations. Activity-correlated classes of links. Despite the success of the ICT 1 CPL model for the conference and hospital datasets, which we remark are quite different from one another, in the case of primary school data none of these models yields a distribution of arrival times* close to that generated from the empirical data, as shown in the panel a of Fig. 3 and Table 4. In particular the IS and ICT 1 CPL models both yield similar, narrower distributions. Compared to the datasets considered in the previous sections, the school dataset presents a few distinctive features. In the conference cases individuals mix in a rather homogeneous way, but most interactions occur at specific moments typically corresponding to social activities such as coffee breaks 7,22. In the hospital case, the interactions display characteristic role-dependent patterns, but contacts are distributed rather homogeneously during the day 28. The primary school dataset, on the other hand, exhibits both a strong community structure dictated by class membership, and correlated contact patterns across classes determined by the schedule of social activities 27. Contacts between children of different classes are possible during specific time intervals only, and strongly correlated during such periods, because the school schedule controls class-based activities rather than individual activities. To tease apart the respective roles of community structure and correlated activity of link groups we study the arrival time* distributions in the case of synthetic datasets exhibiting one or both of these for KL101) for the conference dataset (HT09, left column) and the hospital data (HOSP, right column). In each panel, IS, LS, TS, ICT, ICT 1 CPL stand respectively for Interval Shuffling, Link Shuffling, Time Shuffling, Inter-Contact Time model, and Inter-Contact Time plus Contacts-per-Link model. In the top row, ''data'' indicates the distribution of arrival times* obtained by simulating an SI process over the empirical temporal network (200 realizations with random starting times for each node of the network taken as seed of the epidemics). For each model, we consider 20 different realizations of the temporal network. For each of these realizations we run 20 different SI epidemics, each with a different random starting time. The arrival times* (top row) for all those runs are aggregated to yield the reported distributions. In the boxplots (middle and bottom row) the box extends from the lower to upper quartiles, and the line indicates the median value. The whiskers of the box correspond to the 95% confidence interval. features. These synthetic datasets are created using a toy model that generates temporal networks with tunable community structure (cs) and temporal correlations in the activity of inter-community links. To this aim, we impose a temporal modulation (tm) in the activity of inter-community links: contact events on these links can only occur during specific time intervals (see details in the Methods section). We subsequently compute arrival time* distributions for these synthetic datasets as well as for the corresponding ICT 1 CPL models. Panel b of Figure 3 shows that, when activity-correlated classes of links are introduced, the arrival time* distributions for the ICT 1 CPL case deviates significantly from that based on the corresponding synthetic dataset. This is similar to what we reported for the school data (even though the shape of the arrival time* distribution is different) and occurs regardless of the presence (or lack thereof) or a community structure in the synthetic dataset. When the synthetic dataset displays a community structure but no correlations between the activity of inter-community links, the same arrival time* distributions are indeed observed for both the synthetic network and the corresponding ICT 1 CPL model. --- Discussion The distribution of arrival times at various nodes of an epidemic process unfolding over a temporal network, when measured in terms of activity clocks, displays a behavior that is robust across very different settings and for different starting times of the process, despite the intrinsic heterogeneities and non-stationarities of the temporal network. The arrival time distribution expressed in terms of activity clocks thus represents an interesting tool for investigating the structure of temporal networks beyond their surface features. The burstiness observed in many real-world networks, indicated by a broad distribution of inter-event times, is known to be an important feature of temporal networks that influences dynamical processes taking place on them. Here we have carried out an analysis based on empirical networks of human interactions, measured in different social environments, and we have used suitably-designed null models and generative models for temporal networks to show that the burstiness of inter-event sequences is not the only essential property that needs to be retained when aiming at a realistic model of time-varying contact networks: the heterogeneity of the number of contacts per individual link also plays a fundamental role in determining the arrival times of the spreading process. Our results show that, in fact, it is possible to design parsimonious generative models of temporal networks, such as the ICT 1 CPL model, based on just the distribution of inter-event interval durations and on the distribution of number of contacts per link. The ensuing synthetic temporal networks adequately model the arrival time distributions of real-world networks measured in diverse settings. Interestingly, the behavior of the arrival time distribution expressed in terms of activity clocks is sensitive to complex features of the temporal network data such as the presence of activity-correlated classes of links, as exemplified by the case of the school temporal network, where the interplay of the community structure induced by classes and of correlated activity patterns due to schedule activities creates rich temporal structures in the data. We have shown that the presence of classes of links that are only active in a correlated fashion during specific time windows has an impact on the spreading time distribution and breaks down the ability to use parsimonious models such as the ICT 1 CPL one. Activity-correlated classes of links, which are arguably common in many real-world social systems, are difficult to uncover on the basis of simple statistical observables for the temporal network, and their impact on the dynamics of spreading process calls for more research. We have shown that arrival time distributions based on activity clocks are a precious tool in this respect as they have the ability to indicate the presence of such complex structures and correlations. Simple generative models, such as the ICT 1 CPL model, cannot possibly account for these complex structures and should thus be enriched, when necessary, by introducing additional features such as classes of links with correlated and temporally-localized activity. Here we have shown that toy models that minimally incorporate such features yield deviations in the arrival time patterns similar to those observed for the school temporal network. Overall, our results call for more work in the direction of both detecting and modeling complex temporal-topological structures in time-varying networks. Similarly, more work is needed to design minimal generative models that incorporate realistic features found in empirical data from real-world scenarios. --- Methods Definition of null and generative models. Here we describe the different shuffling procedures and generative models introduced in the main text. For the top-down approach, we start from the empirical temporal networks, on which we apply the following shuffling procedures: Interval Shuffling (IS). The sequence of contact and inter-contact intervals of each link is randomly shuffled. The original contact durations and inter-contact durations are thus preserved. Given a link (a, b) with n contact events, let us denote the contact intervals by (s 0, e 0 ), (s 1, e 1 ),..., (s n, e n ). The set of contact durations is thus given by (e 0 2 s 0 ), (e 1 2 s 1 ),..., (e n 2 s n ), and the set of inter-contact durations is (s 1 2 e 0 ), (s 2 2 e 1 ),..., (s n 2 e n21 ). We create a synthetic timeline for the link (a, b) by randomly shuffling the sequence of contact and inter-contact intervals, and then we randomly and uniformly translate the starting time s 0 0 of the link's new timeline within the remaining time interval T 2 (e n 2 s 0 ), where T is the full dataset time interval. Consistently, links with n 5 1 contact events have an empty set of inter-contact times and the single contact interval is simply randomly displaced in time. Link Shuffling (LS). Whole single-link event sequences are randomly exchanged between randomly chosen link pairs. Event-event and weight-topology correlations are destroyed. Time Shuffling (TS). Time intervals of the whole original contact sequence are randomly shuffled and reallocated randomly to each link retaining the distribution of the number of contacts per link of the original dataset. Temporal correlations are destroyed. The resulting shuffled network is built with a condition of no intersection between contact intervals in the same link. In the case of the generative models, we start by creating a static random network with approximately the same degree distribution, the same number of nodes, and the same number of links as the empirical network we want to study. Then, we build a temporal network by associating with each link a sequence of contact events, according to the following strategies: ICT. For each link we set the number of contacts per link equal to the average number of contacts per link of the original data. Each of these contacts is then generated with a duration equal to the average contact duration observed in the empirical data. The time between contact events is set by sampling with replacement the distribution of inter-event times measured in the data. ICT 1 CPL. The ICT 1 CPL model is based on the ICT model described above, with the additional constraint that for each link the number of contacts is not constant, but is set by sampling with replacement the distribution of the number of contacts per link of the empirical data. Symmetrized Kullback-Leibler divergence. The symmetrized Kullback-Leibler divergence is defined as: DIV s KL M D k <unk> <unk>12 X i M i <unk> <unk>log M i <unk> <unk> D i <unk> <unk> z X i D i <unk> <unk>log D i <unk> <unk> M i <unk> <unk>!,<unk>1<unk> where D(i) is to the distribution of (integer-valued) arrival times* in the empirical data and M(i) is the distribution yielded by the models. To assess the stochastic variability range of the Kullback-Leibler divergence, we generate several realizations of each null model or generative model, we compute the divergence between the distribution yielded by each realization and that of the original data, and we show a box plot summarizing the resulting values. Definition of a toy model with activity-correlated link classes. In order to understand which features of the school data make the distribution of arrival times* not reproducible by the synthetic networks of the ICT 1 CPL model, we introduce a toy generative model that produces temporal networks with some key features of the original school network, namely the community structure and the synchronization of the activity/inactivity patterns of some groups of links. We start by building a static network with a simple two-community structure: we consider N nodes and divide them into two groups of equal size. Within each group, two nodes are linked with a probability p 1. Nodes across the two communities are linked with a probability p 2 # p 1 (the case p 1 5 p 2 yields a random graph without community structure). This procedure defines the topological structure of the network. We build the temporal network by associating with each link a sequence of contact events. These activity sequences are all generated by sampling a Poisson process with a rate l 5 0.0056 s 21, which was chosen to yield an average number of contacts per link of the same order of the school data over the same global time T<unk>100,000 s. For the cross-community links we then remove all events outside of the interval [T/2(1 2 d), T/2(1 1 d)]. This last condition introduces a temporal modulation for the inter-community links, which are only active in the above time window. In the limit d R 1 we recover the non-modulated case. --- Author contributions --- Additional information Supplementary information accompanies this paper at http://www.nature.com/ scientificreports Competing financial interests: The authors declare no competing financial interests.

Dynamical processes on time-varying complex networks are key to understanding and modeling a broad variety of processes in socio-technical systems. Here we focus on empirical temporal networks of human proximity and we aim at understanding the factors that, in simulation, shape the arrival time distribution of simple spreading processes. Abandoning the notion of wall-clock time in favour of node-specific clocks based on activity exposes robust statistical patterns in the arrival times across different social contexts. Using randomization strategies and generative models constrained by data, we show that these patterns can be understood in terms of heterogeneous inter-event time distributions coupled with heterogeneous numbers of events per edge. We also show, both empirically and by using a synthetic dataset, that significant deviations from the above behavior can be caused by the presence of edge classes with strong activity correlations. T he field of complex networks has recently undergone an important evolution. Thanks to the recent availability of time-resolved data sources, many studies performed under the assumption of static network structures can now be extended to take into account the network's dynamics. Data on time-varying networks are becoming accessible across a variety of contexts, ranging from communication networks 1-6 to proximity networks 7,8 and infrastructure networks 9,10 . This avalanche of data is prompting a surge of activity in the field of ''temporal networks '' 11 . Data analysis has shown the coexistence of statistically stationary properties and topological changes, as well as the burstiness of interactions characterized by highly skewed distributions of interevent times 1-14 . These temporal features of networks influence the dynamics of network processes, just like the topological structure of static networks does 15 . As a consequence, and similarly to the case of static networks, simple dynamical processes such as random walks 16 , synchronization phenomena 17 , consensus formation 18 or spreading processes [19][20][21][22][23][24][25] can be used as probes to investigate the temporal and structural properties of timevarying networks. Previous works on the dynamics of spreading processes over complex networks have considered both the topological and the temporal structure of networks 11,19,23 , as well as the specific impact that the temporal structure bears on the spreading process. The quantities used to quantify the measured effects are typically network averages, such as the outbreak sizes of an epidemic or its prevalence. These average quantities, however, fail to account for important heterogeneities in the arrival times of the spreading process. Recent work 26 showed that the non-stationarity and burstiness of empirical temporal networks lead to noisy distributions of arrival times, and that shifting the perspective from a global notion of wall-clock time to a node-specific ''time'' based on node activity allows to expose a clear and robust pattern in arrival ''times''. Here we focus on the distribution of arrival times for spreading processes, based on a wide range of empirical data on time-resolved human proximity. In particular, we seek to identify the dynamic features of the temporal network that are responsible for the observed arrival time distributions. To this aim, we consider temporal networks of human contacts and we define hierarchies of null models and generative models that selectively retain or discard specific properties of the empirical data. We simulate simple spreading processes over these models and perform a comparative analysis of the arrival time distributions. Our results identify the most salient properties that characterize realistic models of human interaction networks, and highlight the properties that control the arrival time distributions, with applications to several domains such as opportunistic information transmission and epidemic spread and containment. We consider time-varying networks of human proximity measured using wearable sensors. The data were collected by the SocioPatterns collaboration (http://www.sociopatterns.org) in different social contexts: two conferences in Italy (HT09) and France (SFHH) 7,22 , a primary school in France (PS) 27 , and a paediatric hospital ward in Italy (HOSP) 28 . Details on the data collection methods are reported in the Supplementary Information

Study setting With the pandemic restrictions and the diverse location of our participants, J.K. conducted all interviews by way of Zoom or telephone, as per the participant's choice. The interviews were completed from Feb. 9 to Apr. 9, 2022, during the Omicron wave of the pandemic. Participants were given the option to interview in either English or French. Interviews lasted about 1 hour and were audio-recorded and transcribed verbatim. --- Data collection We conducted in-depth individual interviews with Black stakeholders with insights into experiences of Black communities during the pandemic. During the interview, we asked participants to discuss what they know about online disinformation among Black people in Canada and the effect of online COVID-19-related disinformation for Black Canadians, and to provide us with their perspective on evidencebased practices to address online COVID-19-related disinformation among Black Canadians. We collected data on sex from all participants and did not restrict demographic questions on sex to binary conceptualizations (Appendix 2, available at www.cmajopen.ca/content/11/3/E389/suppl/DC1). We disaggregated demographic data by age, or sex, place of origin, location in Canada, religion, vaccination status and role within the Black community. We also recognize that Black communities are a heterogeneous group. Thus, we strived for representation from African immigrants, Caribbean immigrants, members of historic Black communities in Canada and Black people from the United States. One author (J.K.) wrote field notes after each interview. --- Data analysis The data collection and analysis processes were iterative. We conducted an inductive content analysis, 15 drawing on analytical resources from intersectionality theory. 16 Intersectionality spotlights how the diverse elements of people's unique social identities could overlap to influence their experiences. 16 During the analysis, we considered issues related to age, sex, race, embedded inequalities and intersecting influences. Three authors (J.K., D.A.A. and A.O.A.) independently read 3 transcripts to familiarize themselves with the data and develop a coding framework. The advisory committee and the principal investigator (B.S.) reviewed the draft coding framework and provided input. Using NVivo 12 software, J.K. read all transcripts and applied the coding framework to complete the data coding and analysis. Preliminary results were shared with the advisory team members for their feedback to ensure the quality of data. Research team members observed reflexivity throughout the research process by maintaining subjective awareness of their multiple privileges, intentions and assumptions. --- Results We reached data saturation after interviewing 30 participants (20 purposively sampled and 10 recruited by way of snowball sampling), which included 16 males and 14 females. Our participants were 15 Black stakeholders from Alberta, 10 from Ontario, 2 from Nova Scotia, and 1 each from British Columbia, Manitoba and Saskatchewan. All participants chose to be interviewed in English. Details on the sociodemographic characteristics of participants are provided in Table 1. In reporting our results, where fewer than 5 participant views are presented, we have used the term some, and where more than 5 participants are presented, we have used the term most. We have made it clear for single participants' views (a representative example of a view from 1 participant). --- Main themes We identified 2 main themes: the nature of online COVID-19 disinformation and the facilitators of COVID-19 disinformation. Participant quotes supporting analysis of data are shown in Table 2. --- Nature of online COVID-19 disinformation and misinformation According to participants, online COVID-19 disinformation and misinformation was widespread in Black communities and typically included misconceptions of COVID-19 as a fallacy and SARS-CoV-2 vaccines as ineffective drugs with microchips. Some participants (<unk> 5 participants) indicated the belief that COVID-19 was a hoax at the beginning of the pandemic led to a slower acceptance of recommended public health interventions and delays in seeking health care (P024, female, 58 yr). Lack of knowledge on vaccine development led to misinformation about vaccine safety, given its fast development, as expressed by most participants (P008, male, 50 yr). Others believed a person could acquire COVID-19 by getting vaccinated and that the vaccines were not essential, given that vaccinated people were still susceptible to the SARS-CoV-2 infection (P024, female, 58 yr). Another participant explained that inconsistent information about vaccine dosages raised concerns about the efficacy of the vaccines within the community (P010, female, 40 yr). Some participants suggested that Black community members believed the SARS-CoV-2 vaccine contained microchips that Bill Gates and other Western leaders could use to track people who were vaccinated (P025, male, 34 yr), thus compromising their privacy and increasing their vulnerability to racial profiling. Hence, the SARS-CoV-2 workplace vaccination requirements and proof of vaccination to enter public places such as restaurants, further increased this suspicion and skepticism among the Black population as ex plicated by most participants (P029, female, 57 yr). These misconceptions contributed to vaccine hesitancy in Black communities and increased the risks of infection. A few mentioned that social media interactions among Black people were dominated by a lack of knowledge about the vaccines, which raised concerns about vaccine shedding, adverse effects on reproductive health, and infertility caused by the vaccines (P020, male, 40 yr). Other participants cited controversies surrounding previous mandatory vaccine programs in some African countries, linking vaccines with anti-fertility agents (P025, male, 34 yr). Moreover, some participants believed the vaccines were a scam to depopulate Black communities (P021, female, 40 yr). --- Facilitators of online COVID-19 disinformation and misinformation --- Miscommunication Participants identified a lack of credible information sources, distribution of unverified information and overwhelming conflicting information as facilitators of the spread of online COVID-19 disinformation and misinformation in Black communities. Most participants were concerned that credible information about COVID-19 was not readily available to Black communities, causing anxieties and panic within Black communities and pushing people to rely on social media, friends and family for information (P003, male, 32 yr). They indicated that credible information about the disease from governments and health authorities was not accessible to most Black community members. Most participants expressed that Black communities prefer face-to-face communication, but much of the credible information from these sources was distributed by way of electronic and print media, such as television and newspaper outlets (P009, male, 54 yr). This made credible information less accessible, especially to Black people who worked frontline jobs or multiple jobs and lacked time to access these sources (P017, male, 43 yr). Some participants asserted that Black people with limited digital literacy faced barriers to accessing credible information, causing them to depend on COVID-19 information relayed by others, which increased their risk of exposure to disinformation and misinformation (P025, male, 34 yr). Black communities also faced many conflicting messages from multiple sources, including social media, governments, health authorities, health care practitioners, expert opinions and international organizations such as the World Health Organization. Some participants indicated that the conflicting messages from these sources degraded the credibility of information from public health agencies (P004, male, 44 yr). This made Black communities susceptible to online COVID-19 disinformation and misinformation, including inaccurate messages that downplayed the seriousness of the pandemic and the efficacy of SARS-CoV-2 vaccines. Although most participants predominantly consisted of immigrants who use social media to maintain transnational ties with family or friends residing in their home countries, an overwhelming abundance of unverified COVID-19 information circulated within the networks of Black communities regardless of their country of origin. With the lockdowns, social media platforms such as Facebook and WhatsApp offered more accessible ways to interact and share The vaccine, they say, like I think when we got the vaccine, the first dose, and the second dose, and we thought like, "Oh, it's -that's going to be it.... And again, there is another, a third dose... and again, there's another fourth dose, so we don't know when this is going to end. Because we still don't know like what is happening." (P010, female, 40 yr) But, you know, the mis-spread of information on WhatsApp led some people to believe that this was some instrument that was being used to control the population by Bill Gates or, you know? Maybe the Western leaders... and for you to get around this, you have to get a chip, and like this chip, they would be able to monitor your activities... and that kind of raised some kind of fears among the community members. Yeah. (P025, male, 34 yr) Well, it has really created a lot of mixed reaction... Canada, it's supposed to be multicultural, and it's supposed to be a country that has a choice, you make your own choice. But during COVID, I don't think people were given that choice... they were kind of forced to take the vaccine, and they have been left with doubt.... They think the vaccine is intended to control the population, in the Black community. (P029, female, 57 yr) Uh, well, for females, there was like okay, there was a fear of if you caught the -if you caught the virus, you know, it could affect your fertility. (P020, male, 40 yr) Well, for starters, I think people, from my understanding, with the Africans... there was a time back then when at the point when polio was kind of rampant... so they have a fear that like some people who took the medication or the vaccinations were not able to have kids... and they think that the vaccines made them infertile, and it was a way of the West controlling them. (P025, male, 34 yr) Okay, so the misconceptions that I can think we find in our Black communities are first of all, that like it is something that was created by White people so that they can get rid of us Black people. (P021, female, 40 yr) --- Facilitators of online COVID-19 disinformation and misinformation --- Miscommunication The absence of timely, ongoing, trusted and connected sources of information where people can go and just really understand the fullness of the, you know, the virus or the vaccines, or what's happening in the community, in the absence of those kinds of thing, folks are -folks fill that void either with disinformation, assumptions, or, you know, general perceptions that -general perceptions or misconceptions that gets spread into the community. (P003, male, 32 yr) Another thing about the information was that it didn't consider how members of our communities get information. So, we like 1-on-1... so just targeting information in the media, I think it did not -it didn't reach members of our community. And again, we are very curious -I mean, we like to ask questions, okay? So, if you just put information out there without the means of engaging and feedback of that information, that will not help us. (P009, male, 54 yr) Yeah, for those that have a challenge to access Internet sources, yes, we can say that there is a gap. Because it is difficult to access information, for those that are struggling to survive, and that are working for 16 hours. (P017, male, 43 yr) But also, at the same time, we can't really blame our community members, because a lot of the people that came here maybe came as refugees and never had the opportunity to go to school, and this is something new to them, and they have no idea of what, you know, peer -peer review is, what kind of information might be coming from a peer-reviewed source, which information's reliable or not. They lack that, you know, that guidance. (P025, male, 34 yr) And like I said, they -the -yes, the government is talking, but sometime today they will say A, tomorrow they will say B. After, they will say C. So, it's like they don't know themselves, where they are going.... Now I think people get used with that. They don't even listen to them. So, myself, I stopped listening. (P004, male, 44 yr) Because everybody gets information from social media on various site sources... 1 community member has some information shared on 1 platform, and it's disseminated without verifying the source. (P020, male, 40 yr) --- Cultural and religious factors You know, while a lot of us living here, we have directly interfaced with what is happening back home.... So that link between here and back home has also played a key factor in people's perception, okay?.... Because the information flows not just online alone.... Because when they get something back home, they forward it to those of us who are here. And so that was also a means of misinformation, but also an opportunity for education that people didn't recognize here. (P009, male, 54 yr) Because even a lot of the misconceptions and the misinformation were fueled by some religious leaders who felt that the vaccine had something to do with spirituality and the anti-Christ and so many other things they practice.... And there's some churches that even told their members not to take, and some of their members did not, you know. (P009, male, 54 yr) A lot of us, or people of African and Caribbean or Black descent, believe in your body acquiring natural immunity, plus using natural protective, I guess, interventions, concoctions, whether it was [laughs] the famous ginger, lemon and whatever that was going around. (P030, female, 53 yr) I've also interacted with community members who had mentioned to me that they have herbs that could protect you from COVID-19, and as such do not need to wear masks, or take any prevention other than those teas.... And some of the sources are not verified sources, but, you know, the community consumes a lot of the information that is gotten through social media. (P001, female, 36 yr) Yeah, then some people also were thinking that with herbs you can kill the virus, that you don't need to take any vaccination. (P008, male, 50 yr) --- Distrust of health care systems So, there is a lot of broken trust between the ACB [African Caribbean Black] communities and the mainstream medical health care because of the medical history between the ACB communities and the system, the health system.... We could talk about the Tuskegee experiments.... And similarly, to look at other Black experiences in the health care that we see. (P001, female, 36 yr) We could also talk about the mental health, you know, looking at the trauma, you know, including medical PTSDs [posttraumatic stress disorders] for those who have experienced medical racism, directly or indirectly within the health care system. (P001, female, 36 yr) They just don't trust. Anything that they don't have control over, which is not much that Black people have control over it, they don't trust it. They don't trust the systems to me, because in their minds the systems always fail Black people, right? (P019, female, 44-54 yr) They feel disenfranchised, or they don't think they are part of the whole world, so to say. They feel that, you know, there's nothing, you know, in the system for them. They also go back to lack of trust of the system, and also the medical community, and I've heard even one of our own saying that, "You know, we don't trust you guys. We don't trust you doctors because you just, you know, you are just a part of the whole conspiracy." And then the historical perspective is very, very strong... they prefer to get their information online or they also have specific areas where they get their information from. (P024, female, 58 yr) But once that was done, and when it came to testing, distribution of testing kits, again, we were forgotten, you know? [laughter] You know, so it was quite incomplete, because you see, when it came to vaccines, yes, we were remembered, right, and those communities were used to disseminate the information that vaccines are available. And then when it came to testing kits, that would have helped people to kind of keep them safe and -right? (P018, male, 48 yr) --- Distrust of governments I know it sounds really hard and it sounds bizarre, but at the core of it I think is anti-Black racism. I think when you are a population that's the most hated population, when you're a population that faces the most disenfranchisement, when you're a population that has, next to the Indigenous, the highest number of people in child -in the number of Black kids in childcare, when you have the highest number of Black men incarcerated, I think it's all that. So Black people just don't trust. (P019, female, 44-54 yr [participant provided an age range]) I think very specifically, as it relates to the vaccine and as it relates to COVID-19, I think there is a proliferation of disinformation within that community that stems from challenges related to distrust of the state, rightful distrust of the state, right? But I think that it's kind of morphed into conspiracy theories that are unhelpful.... But nevertheless, they've kind of taken hold as an expression of that distrust, but I think that in the context of COVID-19, those can be really damaging and concerning (P014, male, 28 yr). Folks were told to physically distance, but there were some folks who had to go to work and, you know, were being called heroes, et cetera. And so they would be going to work early morning on the bus, but the buses were full. Folks asked for additional buses so that they could physically distance, and the official response from the government was that they weren't going to send more buses.... This is fertile ground not only for distrust and misinformation for some folks; it's also fertile ground for, um... those health inequities... you know, when you parse these things out 1 by 1, disinformation or, you know, vaccination and mandates, we're often not getting an understanding of how things coalesce or collocate to create the context in which people, like, think and act. (P003, male, 32 yr) Like for instance, we found out that during the COVID-19 spread, yeah, there have been significant experiences of discrimination among the Black people in -in -that's across the country. So that there was significant negative experiences in attempting to receive health care during the COVID-19 period.... So that can help them to accept misinformation if the health care system is not doing what they are supposed to do, to support their needs during a critical time of COVID-19. (P008, male, 50 yr) information about the pandemic. Therefore, some participants explicated that social media became a source of information fatigue through which COVID-19 disinformation and misinformation spread (P020, male, 40 yr). --- Cultural and religious factors Most participants explained how Black Canadians are highly attached to their countries of origin. Thus, rumours originating in their land of birth spread quickly in the diaspora community because of those strong ties (P009, male, 54 yr). Such culture-based disinformation included beliefs suggesting re ligious faith and natural health remedies were more effective than vaccines in combatting the spread of the SARS-CoV-2 infection. One participant illuminated that because most Black people are religious, they were readily susceptible to this kind of disinformation and misinformation, and even more so if such rumours were spread by religious leaders (P009, male, 54 yr). Although all participants reported being vaccinated, they observed that some Black community members opposed SARS-CoV-2 vaccines. In particular, some Black people viewed these vaccines and their developers as anti-Christ agents; misrepresentation of the scriptures created fear and contributed to refusal and a delayed acceptance of SARS-CoV-2 vaccines in Black communities (P009, male, 54 yr). Some participants eluded that consistent with traditional African worldviews, Black people were inclined to believe misinformation about herbal supplements being an effective preventive remedy or treatment for SARS-CoV-2 infection (P030, female, 53 yr). These participants explained that these beliefs mainly spread through social media (P001, female, 36 yr), bringing about a false sense of security that exposed more community members to the disease and discouraging their use of approved care supports (P008, male, 50 yr). --- Distrust of health care systems Some participants referenced the history of medical racism and continuous exploitation of Black people in the medical system as a factor that diminished trust in governments and health organizations. Participants referenced past injustices, such as the Tuskegee experiment (1932-1972), in which US Public Health allowed African-American men with syphilis to go untreated as a way of chronicling the progression of the disease (P001, female, 36 yr). Participants acknowledged the cumulative trauma to Black communities caused by racism in health systems that have continuously failed them (P001, female, 36 yr). These histories increased Black people's skepticism about COVID-19 and its vaccines, rendering them vulnerable to alternative truths about the disease (P019, female, 44-54 yr [participant provided an age range]). Systemic racism has consistently posed challenges to the health of Black people and exacerbated the health disparities they face. Participants highlighted how exposure to racism and discrimination has left the community feeling disregarded, reinforcing mistrust of health care organizations and diminishing efforts at increasing vaccine acceptance (P024, female, 58 yr). Participants suggested that many Black people would rather believe the information obtained from social media than messages communicated by health care professionals. This attitude served to undermine public health responses to the pandemic. Participants also indicated that inequitable distribution of the vaccines and testing kits reinforced the distrust Black people had with health care systems. Specifically, despite reporting a higher number of SARS-CoV-2 infections and deaths, Black communities were not prioritized when test kits were distributed; yet, they were being increasingly asked to get vaccinated (P018, male, 48 yr). Thus, addressing racism represented a more pressing need for Black communities besides adherence to public health interventions. --- Distrust of governments Participants described disinformation and misinformation within the Black community that occurred because of experiences of discrimination and differential treatment based on race and skin colour. Racism has perpetuated distrust of governments owing to the lack of commitment to addressing inequities faced by Black Canadians, especially in the areas of education, employment, housing, policing, child care and health care (P019, female, 44-54 yr). According to participants, Black people's perceptions of government influenced how they responded to public health interventions addressing the pandemic (P014, male, 28 yr). For instance, 1 participant explained that more Black Canadians, compared with other ethnoracial groups, worked in front-line jobs and used public transportation, where a lack of opportunities for physical distancing increased their risk of SARS-CoV-2 infection (P003, male, 32 yr). Thus, the Black communities' belief that governments are indifferent to their plight increased their skepticism of COVID-19 interventions, including vaccines. Some par tici pants expressed that governments should address anti-Black racism within institutions and should work on building trust with Black communities (P003, male, 32 yr; P008, male, 50 yr). --- Interpretation Social media, especially instant messaging platforms such as Facebook and WhatsApp, became a conduit through which COVID-19 disinformation and misinformation spread in Black communities. This form of information exchange also carried the dangers of disinformation and misinformation, affirming the risks of SARS-CoV-2 infections and poor health outcomes among Black Canadians. Most participants suggested racism and underlying systemic discrimination against Black Canadians promoted distrust of the government or health institutions and immensely catalyzed the spread of disinformation and misinformation in Black communities across Canada. Although our participants were all fully vaccinated, they noted that some members of their communities were vaccine-hesitant owing to distrust of vaccine manufacturers, health care systems and governments. They indicated that this hesitancy results from current and historical experiences of racist medical procedures. Most participants referred to the Tuskegee experiments when discussing health disparities and racism in health care in the Canadian context. It is apparent that the Black community in Canada strongly relates to the Tuskegee experiments and the health care system's history with Black communities in Canada, thus affecting how they respond to the SARS-CoV-2 vaccine. Some participants indicated that communication by health care authorities and the government during the beginning of the pandemic was not culturally appropriate to some Black people owing to conflicting messages and language barriers (besides their own experiences, participants were asked to provide rich information about the impacts of COVID-19 in their communities), prompting an overreliance on their community members over social media. Further, firm beliefs in cultural practices such as religious faith and natural health remedies contributed to disinformation and misinformation within Black communities. Our findings are similar to other published studies, indicating an upsurge in online disinformation and misinformation, exacerbated health inequities experienced by Black communities and undermined public health interventions to curb the spread and effect of COVID-19. 11 Research funded by the Rita Allen Foundation found that anti-vaccine groups weaponized Black people's historical encounters with health care systems to coerce Black communities into rejecting the SARS-CoV-2 vaccines. 17 Our results indicated that some Black people were hesitant to receive vaccinations owing to their religious beliefs, although previously reported findings showed some religious leaders were noted for encouraging their congregations to get vaccinated. 18 There is also evidence showing language barriers during the COVID-19 pandemic resulted in decreased adherence to public health directives and recommendations. 19,20 Other studies have reported issues relating to biases and lack of culturally appropriate care for Black people in Canada and the United States. 21,22 Our study shows the need to engage community-based health care clinics to promote SARS-CoV-2 vaccine uptake and eliminate language and cultural barriers to vaccine access. For instance, such a model in California indicated that communityengaged approaches rooted in principles of authentic partnership that include trust-building, power-sharing and co-learning are crucial for addressing public health crises, such as the COVID-19 pandemic. 21 This approach identified barriers to vaccine uptake that were then addressed by providing culturally appropriate care. 8 There is also a need to use an antiracist and anti-oppressive framework to guide health agencies in addressing inequalities within the health care system and understand how social and economic conditions, structural racism and systemic discrimination can engender qualitydriven trust. 23 Practices that stress continuity and communication result in a higher level of confidence in physicians, which is paramount in promoting the use of preventive services. 24 Further, religion plays a notable role in the Black communities, rendering the religious institutions essential resources for health care interventions within the Black communities. [25][26][27] --- Limitations Our inability to recruit participants from the Caribbean and some provinces (e.g., Quebec) with large Black populations is a limitation of our study. Further, overreliance on snowball sampling led to a largely homogenous sample in which most participants identified as Christian and cisgender. The participants were all vaccinated, a limitation since their accounts cannot fully represent those in the Black community who are vaccine-hesitant. We acknowledge that older adults are more at risk from COVID-19; however, the analysis of our results was dependent on the perspectives of the participants interviewed. We used purposive sampling and snowball techniques to recruit Black community leaders, Black-led organization leaders and Black service providers, resulting in a younger working age population. Future research should target older adults by recruiting through community-based organizations, senior adult living or long-term care homes. --- Conclusion Underlying systemic racism and related inequities in Canada created mistrust for public health authorities and contributed to Black people's preparedness for alternative truths about COVID-19. Therefore, there is a greater need to build trust and adopt collaborative approaches to addressing community concerns; for example, having genuine, respectful discussions of other topical issues affecting Black people, such as employment discrimination, medical racism and anti-racist workplace practices and policies. As such, we recommend supporting and steadily funding existing Black community organizations to develop culturally accessible health education material, including health information flyers and infographics. Funding community-based health centres may help disseminate and increase the uptake of health information and curb the spread of disinformation. --- Ethics approval The University of Alberta Ethics Board approved our study (Pro00114392). Affiliations: Faculty of Nursing (Kemei, Olanlesi-Aliu, Salami) and Health and Immigration Policies and Practices (HIPP) Research Program (Alaazi, Bukola Salami), University of Alberta, Edmonton, Alta.; Obstetrics and Gynaecology (Tunde-Byass), University of Toronto and North York General Hospital; Black Opportunity Fund (Sekyi-Otu), Toronto, Ont.; IMPACT Institute of Canada (Mohamud); Intersections of Gender Signature Area (Bukola Salami), University of Alberta, Edmonton, Alta. Contributors: Bukola Salami contributed to the conception and design of the work. Janet Kemei, Dominic Alaazi, Adedoyin Olanlesi-Aliu, Modupe Tunde-Byass and Bukola Salami contributed to the acquisition, analysis and interpretation of the data. Janet Kemei drafted the manuscript. All authors revised the manuscript critically for important intellectual content. All authors gave final approval of the version to be published and agreed to be accountable for all aspects of the work. Content licence: This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY-NC-ND 4.0) licence, which permits use, distribution and reproduction in any medium, provided that the original publication is properly cited, the use is noncommercial (i.e., research or educational use), and no modifications or adaptations are made. See: https://creativecommons.org/licenses/ by-nc-nd/4.0/ Data sharing: Research data are available on request to Bukola Salami. Funding: Funding was received from the Digital Citizen Contribution Program, Government of Canada. --- Supplemental information: For reviewer comments and the original submission of this manuscript, please see www.cmajopen.ca/content/11/3/ E389/suppl/DC1.

lack Canadians are disproportionately affected by the COVID-19 pandemic. 1 For example, as of September 2020, data tracking of COVID-19 cases in Toronto showed Black people accounted for 24% of positive cases despite constituting only 9.3% of the city's total population. 2 Black people and other racialized communities in Canada are also more likely to be admitted to hospital with COVID-19 than White and East Asian people. 3 Further, evidence shows a universal vaccine hesitancy across countries and subgroups. 4 Disinformation and misinformation about COVID-19 contributes to health disparities by posing a threat to the acceptance of the SARS-CoV-2 vaccine among Black Canadians. [5][6][7][8][9][10] The susceptibility of Black Canadians to online COVID-19 disinformation, vaccine hesitancy, infections and hospital admissions may be attributed to several individual-and structurallevel factors, including socioeconomic status, crowded living environments, cultural barriers, racial discrimination, poor access to health care and poor housing, that make it difficult for this population to adhere to public health directives. 11 Similarly, anti-Black racism and structural inequities within Canadian institutions expose Black people to socioeconomic vulnerabilities, which increase the burden of COVID-19 morbidity and death to this population, 12,13 thus increasing vaccine skepticism and hesitancy. On completing a scoping review of online disinformation among Black people, 14 we wanted to deepen our knowledge and situate the findings from our scoping review within the Canadian context. Despite the negative health behaviours promoted by disinformation, there is a paucity of qualitative studies examining online COVID-19 disinformation in Black communities in Canada. We aimed to describe the nature of online COVID-19 disinformation and misinformation among Black Canadians and identify the factors contributing to this phenomenon.

Introduction Almost all politicians in Western Europe todayfrom across the political spectrumapparently believe that some people are more entitled to inhabit particular places than others. Their belief is usually based on a form of 'primordial reasoning' ['we were here first'], where places are owned by 'native' groups who enjoy specific rights (prominent among them the right to feel at home) (Duyvendak 2011, p. 2). Research on children's ethnic, racial and anti-immigrant attitudes has focused on various factors and processes such as social-cognitive development, moral reasoning, in-group norms, group identification, intergroup contact, feelings of threat, and school composition and educational practices (see Levy and Killen 2008;Rutland and Killen 2015). What has received much less attention is the role of so-called lay theories for children's negative out-group attitudes. There is some work on the ways in which group essentialism beliefs (Diesendruck and Menahem 2015), beliefs about the malleability or fixedness of human attributes (Levy and Karafantis 2008), protestant work ethic beliefs (Levy et al. 2008), and shared conflict beliefs (Bar-Tal and Teichman 2005) justify children's negative attitudes. In addition to these lay theories, and similar to what is expressed in the quote above, children might reason that a place belongs to their in-group because they were there first and therefore that it is acceptable to exclude newcomers. In intergroup research, and following anthropological literature, this notion of primo-occupancy with the related ownership feelings has been labeled "autochthony beliefs" and these beliefs have been found to be important for adults' ethnic out-group attitudes (e.g., Martinovic and Verkuyten 2013;Smeekes et al. 2015). With three empirical studies conducted in the Netherlands the current research examines the importance of autochthony beliefs for majority group children's (grades 4-6) attitudes towards immigrants. The current aim is to introduce the novel construct of autochthony to the intergroup development literature (Rutland et al. 2010) and to empirically examine its relevance for out-group attitudes of ethnic majority group children. The importance of this lay theory was investigated while considering ethnic identification, and by taking perceived multicultural education (Studies 2 and 3), classroom composition (Studies 2 and 3), and feeling at home in the country (Study 3) into account. The general expectation that was tested is that, whereas ethnic identification is mainly relevant for in-group attitudes, autochthony beliefs with their sense of native collective ownership ("it is ours") are especially important for attitudes towards immigrant groups. More specifically, autochthony beliefs are considered to provide a justification for negative immigrant attitudes of majority group children, in particular when they at the same time do not feel at home anymore in their own country (see quote above). Thus, it was hypothesized that autochthony beliefs can serve to justify anti-immigrant attitudes and prevailing social inequalities (Jost et al. 2004). In short, this research breaks new ground in examining the importance of autochthony beliefs for children's evaluation of immigrant groups and in investigating when these beliefs are particularly important by considering the moderating role of a sense of feeling at home in one's country. In doing so we not only introduce the novel concept of autochthony to the literature but also make a contribution to the rather limited understanding of children's attitudes toward the increasingly important group of immigrants (e.g., Brown 2011;Brown and Lee 2015). Although young children (5 years) show tendencies to justify group advantages (Baron and Banaji 2009), the understanding and endorsement of lay theories that justify advantages appears later (Henry and Saul 2006;Levy and Karafantis 2008). It is only at around 8 years of age that children are able to use and weight different forms of information to assess and evaluate claims and rights (Smetana 2006). Furthermore, compared to objects, ownership of a territory might be a rather abstract issue for young children. Research has shown that children's knowledge and beliefs about countries as geographical territories develops from around seven years onwards (see Barrett 2007). In addition, longitudinal research in Western Europe has shown that early adolescence is a sensitive developmental period for the emergence of negative attitudes toward immigrants (Gniewosz and Noack 2015). Therefore, the current research focuses on late childhood (grades 4-6) and there were no reasons to expect meaningful age differences. Lay theories once learned instigate a distinct and stable pattern of evaluation and judgment with respect to the target group. For example, research on protestant ethic beliefs and on entity and incremental beliefs indicates that these beliefs tend to function the same in social judgments regardless of age (Levy and Karafantis 2008). --- Ownership and Autochthony Beliefs Perceived ownership is a pervasive notion that has profound implications for how individuals, think, feel and behave (Ye and Gawronski 2016). Ownership helps to organize the social and physical environment, regulates social interactions, and involves normative and moral privileges and responsibilities (Verkuyten and Martinovic 2017). Perceived ownership implies a bundle of rights such as the right to use what is owned and to decide whether to keep the target of ownership or not. Importantly, ownership also implies a 'gatekeeper right': the right to exclude others and to decide whether others are permitted or prohibited to use the object or have access to it (Merrill 1998) Considering these implications it is understandable that disputes over ownership of objects and places are among the most frequent and most intense intergroup conflicts (Toft 2014), also among children. Although ownership is not an obvious property of objects but rather abstract and imperceptible, young children already recognize it. Preschoolers have a basic understanding of ownership of physical objects and appreciate that owners are entitled to greater control over their property than non-owners (Kim and Kalish 2009;Rossano et al. 2011). By the age of 6 or 7 children's notions of ownership are also applied to ideas and intellectual property (Shaw et al. 2012), and to places (O'Neal et al. 1977). There are many situations in which groups of children make claims on a particular physical place, such as when children convert a site in their play area, club or hideaway (Factor 2004). Territorial behavior whereby an intruder is excluded or punished for invading 'our' play area has been found in observational and experimental research among children (O'Neal et al. 1977;Zebian and Rochat 2012). Children can infer ownership from seeing someone in possession of an object (Blake and Harris 2009), from verbal statements about who owns an object (Blake et al. 2012), from observing who decides on whether others can use it (Neary et al. 2009), and by using principles of past investment (creating or modifying an object), and ownership transfer (buying or giving) (Kanngiesser et al. 2014). In addition, children have been found to judge that an object belongs to the first person possessing it (Blake and Harris 2009;Friedman and Neary 2008). Older children and adults argue that the first person seen to possess a previously nonowned object is its owner (Friedman 2008;Friedman and Neary 2008), and the same has been found for the ownership of ideas (Shaw and Olson 2002). Similarly, being first at a particular place is information that children use to infer ownership. First arrival indicates one's presence at a place before anyone else and this in itself is an important basis for establishing ownership. Experimental research has demonstrated that children as old as eight years infer personal ("mine") and, importantly, also collective ("ours") territorial ownership from first arrival (Verkuyten et al. 2015;Verkuyten et al. 2015). In this research, first arrival on an island was found to be a consideration that undermines the notion of equal sharing which is a key moral principle for children (e.g., Fehr et al. 2008;Rochat et al. 2009) that is typically applied when there are no other considerations involved, such as social conventions, group norms, interests and personal benefits. These findings correspond with anthropological work on autochthony (Geschiere 2009) and "Sons of Soil" conflicts (Côté and Mitchell 2017) which demonstrates that primooccupants are considered as rightfully possessing an area. This is evident in the moral and legal claims on resources and territory made by indigenous groups and so-called "first nations" or "first peoples". The notion of autochthony suggest that first arrival determines place ownership with the related right to usage and to exclude others. The term autochthony can be traced back to ancient Greece and it means literally being "born from the soil" (Geschiere 2009). It is the belief that a country or a region belongs to its original inhabitants. This belief triggers self-evident notions of ownership and entitlements and thereby has an "implicit call for excluding strangers ('allochthons'), whoever they may be" (Ceuppens and Geschiere 2005, p. 386). In European non-settler countries, this notion has been used to reject immigrants and to justify prejudice towards immigrant groups (Ceuppens 2011;Geschiere 2009). An immigrant represents someone trying to become a member of one's national in-group and this typically elicits considerations of collective ownership and territorial belonging (Verkuyten and Martinovic 2017). Autochthony can function as a justifying belief (Jost et al. 2004) because it makes the more advantaged position of the native majority group understandable and just. Children as young as 5 appear to be sensitive to these sorts of justifying beliefs (Baron and Banaji 2009) and 10-year olds have been found to endorse them (Henry and Saul 2006;Levy and Karafantis 2008). Furthermore, lay theories can justify children's thoughts, feelings and behaviors toward minority out-groups (Levy et al. 2008). It was expected that majority group children who more strongly endorse autochthony beliefs will have a more negative attitude toward immigrant groups (Studies 1 to 3) and refugees (Study 2). Furthermore, in Study 3 the expectation was tested that the association between autochthony beliefs and out-group attitudes is especially strong among children who feel less at home in the Netherlands (Barrett 2007;Verkuyten et al. 2014). Social identity development theory (Nesdale 2008) proposes that negative out-group attitudes tend to emerge when majority members feel that their position or wellbeing is undermined in some way by members of ethnic out-groups (e.g., Nesdale et al. 2005;Nesdale et al. 2005). Children's attitudes and reasoning are influenced by social context and exposure (Killen et al. 2008) and research has shown that older children are aware of the societal debate on immigration and existing anti-immigration sentiments (Brown 2011). The Dutch immigration debate is typically framed in terms of what immigration means for the country, and the widespread sentiment is that the ethnic majority Dutch ("Nederlander" in Dutch language) feel estranged and no longer at home in their own country ("Nederland"): "The native Dutch, it is argued, have become like foreigners in their own country, feeling what foreigners should allegedly feel: not at home" (Duyvendak 2011 p. 98). Research shows that the ethnic Dutch do indeed increasingly (2006 to 2015) feel not at home in the Netherlands, and not feeling at home is higher among adolescents than adults (Huijnk and Andriessen 2016). This feeling can be expected to make autochthony beliefs more important for attitudes toward immigrant groups. Proprietary claims to a country that are accompanied by a sense of estrangement has been found to be associated with more negative ethnic out-group attitudes (Martinovic and Verkuyten 2013). Theoretically, and as indicated in the quote heading this paper, it is the combination of autochthony beliefs with the sense that one's ability to feel at home in one's own country is undermined, that in particular should be associated with more negative attitudes toward immigrant groups (Bobo 1999). Therefore, Study 3 tested the expectation that autochthony is stronger related to anti-immigrant attitudes among ethnic majority Dutch children who have a lower sense of feeling at home in the Netherlands. --- Ethnic Identification Autochthony and the related sense of ownership does not have to imply a sense of ethnic group belonging. Individuals can believe that their group owns a particular place because of first-arrival without having a sense of commitment and belonging to in-group members (Verkuyten and Martinovic 2017). Research among children has demonstrated that stronger in-group identification goes together with more positive attitudes toward the in-group, including the national in-group (Bennett et al. 1998;Pfeifer et al. 2007). A positive evaluation of the group to which one belongs provides a positive sense of self (Tajfel and Turner 1979). This means that it can be expect that in the three studies majority group children with higher ethnic identification will be more positive about their ethnic majority group. Higher in-group identification does not have to imply, however, that out-groups are evaluated more negatively (Cameron et al. 2001). Social identity development theory (Nesdale 2008) proposes that stronger feelings of in-group belonging lead to a stronger in-group orientation and commitment, but not necessarily to a rejection of outgroups. Higher compared to lower ethnic identifiers are predominantly focused upon and concerned about their ethnic in-group. Yet when concerns about the out-group come into play it is more likely that in-group identification is associated with negative out-group attitudes. According to social identity theory (Tajfel and Turner 1979), higher compared to lower identifiers are more sensitive to anything that could harm or undermine their feeling of in-group belonging. This means that it can be expected that the combination of autochthony beliefs with a sense of not feeling at home is particularly important for higher identifiers. Those who feel that the Netherlands is no longer their home and also consider being Dutch important to their sense of self are likely to base their negative attitudes toward immigrants on autochthony beliefs. This expected three-way interaction between autochthony, home feeling and ethnic identification was tested in Study 3. --- Current Study The general expectation tested is that stronger autochthony belief is associated with more negative immigrant attitudes and that this association is particularly strong for children who have a relatively low sense of feeling at home in the country but at the same time care about being Dutch. In two ways this research tried to provide clear empirical evidence for this expectation. First, in assessing the robustness of the associations found for autochthony it was examined whether the statistical effects of autochthony was similar for different age groups, boys and girls, level of ethnic identity, perceived multicultural education, and classroom ethnic composition. The presence of similar effects demonstrates that the role of autochthony for attitudes toward immigrants does not depend on these individual and classroom differences and thereby would underscore the general and robust importance of the construct of autochthony beliefs for children's immigrant attitudes. Second, given the increased importance of replication in research (e.g., Pashler and Wagenmakers 2012), it is important to ensure that the results could be replicated with different samples of ethnic majority group children and somewhat different measures and operationalizations of group attitudes. The expectations were tested in three separate analyses by using data from three different studies 1 on children's attitudes towards school and their academic engagement. In these studies questions on autochthony beliefs and attitudes toward the two most prominent and numerically largest immigrant-origin groups in the Netherlands (of Turkish and Moroccan origin, both around 400,000 people) were included. These two groups have a history of labor migration starting in the late 1960s, followed by a process of family reunification in the mid-1970s which means that nowadays majority group children are mostly exposed to second and third generation immigrants. --- Study 1 --- Method --- Participants and procedure Participants were 345 children (M age = 10.73, SD = 0.98; 49.7% girls) who had two ethnic Dutch parents and who self-identified as ethnic Dutch. These children were from 23 classes (grades 4-6: with on average 66.87% Dutch students, SD = 25.74, and 5.98% students with at least one parent born in Turkey or Morocco, SD = 10.65) in 8 schools in different parts of the country. Participation in the study was voluntary and anonymous and all children with parental permission participated. Together with their classmates, the children completed a questionnaire in their classroom under supervision of their teacher and a research assistant. Apart from the study variables, the questionnaire contained items on children's relations with their peers and teachers 2, and experimental vignettes related to diversity. Originally, the sample consisted of 347 ethnic Dutch students, but two cases that had missing scores on the dependent variables were deleted. Overall, very few scores on the items were missing (maximum 1.2% per item) and Little's MCAR test indicated that missingness was completely at random, <unk> 2 (53) = 63.90, p = 0.145. 1 The different studies involved three independent data collections. Two of the schools participated in Study 2 and Study 3. However, there was a two-year difference between both data collections, and for one of the schools the children were in respectively in grade 6 and grade 8 in the earlier and the later data collections. 2 In a previous research (Masked Reference), the immigrant attitude measures analyzed in the current Studies 1 and 3 and those from a prior wave of Study 2 were predicted from the quality of the studentteacher relationship. Likewise, two of these studies controlled for perceived peer acceptance. Student-teacher relationship quality and perceived peer acceptance were not included in the present studies as they were unrelated to autochthony. It was not theoretically relevant to include other variables from the datasets in the present analyses. --- Measures Autochthony was measured with two items adapted from previous research (Martinovic and Verkuyten 2013): 'The Netherlands belong to those who came here first' and 'Dutch natives can decide what happens in their country'. The response scale ranged from 1 (No!) to 5 (Yes!) and there was a medium-sized correlation between the items, r = 0.32. Ethnic identification was assessed with three items that have been successfully used in previous research (e.g., Sierksma et al. 2014). On a five-point scale ranging from 1 (No!) to 5 (Yes!) children indicated whether they liked being Dutch, and whether they were proud to be Dutch, and whether they found it important to be Dutch (<unk> = 0.57). Because the autochthony and identification measures had moderate internal reliabilities latent variables were used in the main analyses in order to correct for measurement error (see below). Immigrant group attitudes was assessed by using the'seven faces' scale developed by Yee and Brown (1992) which ranges from a big smile (coded as 7) to a big frown (coded as 1). Children used this scale to indicate their evaluations of, respectively, Dutch (in-group), and people of Turkish and Moroccan background. All scales were presented on the same page in the questionnaire. The evaluations of Turks and Moroccans were strongly correlated, r = 0.81, and therefore averaged into a single measure for out-group attitude. Children's in-group attitude (M = 6.80, SD = 0.58) was substantially more positive than their outgroup attitude (M = 4.49, SD = 1.68), t (344) = 25.04, p <unk> 0.001. --- Results To examine the distinction between autochthony and ethnic identification, and their unique relations with children's ethnic attitudes, structural equation modeling (SEM) in Mplus 7 was conducted. As there was no significant classroom variance in the dependent variables (Intraclass Correlations <unk>.06), and to retain an acceptable cases-tofree-parameters ratio (see Kline 2005), we did not include the classroom level in the analyses or examine the role of classroom ethnic composition. Four fit indexes were used: the comparative fit index (CFI), the Tucker Lewis index (TLI), the root mean square error of approximation (RMSEA), and the standardized root mean residual (SRMR). Model fit is considered good if CFI and TLI have values of 0.95 or higher, and RMSEA and SRMR are lower than 0.05. CFI and TLI values larger than 0.9 and RMSEA and SRMR values smaller than 0.1 are considered acceptable (Kline 2005). First, a confirmatory factor analysis (CFA) was performed to examine whether the autochthony and identification items loaded on two different but correlated factors. Results showed that the fit of the two-factor model was excellent, <unk> 2 (4) = 3.03, CFI = 1.00, TLI = 1.01, RMSEA = 0.00, SRMR = 0.02, and that the two factors were uncorrelated, r = 0.14, p = 0.15. 3Moreover, the two-factor model fitted the data significantly better than a model that included a single factor for both autochthony and ethnic identification, <unk> 2 dif (1) = 35.78, p <unk> 0.001. These findings demonstrate that for the majority group children, autochthony beliefs and ethnic identification were separate constructs. Next, a Stuctural Equation Model was specified in which children's in-group attitude and attitude toward immigrants were regressed on the latent factors for autochthony and ethnic identification. MLR as an estimator was used as the distribution of the in-group attitude was non-normal (skewness -4.62, SE = 0.13; kurtosis = 32.54, SE = 0.26; with 84.9% of the children gave the highest evaluation possible), and one factor for children's attitude toward the Turkish and Moroccan out-groups was specified. Age and gender were controlled for. The fit of the SEM model was satisfactory, <unk> 2 (16) = 23.31, CFI = 0.97, TLI = 0.93, RMSEA = 0.036, SRMR = 0.031. Results are shown in Table 1. As expected and indicated by the standardized effects, autochthony had a moderately strong negative relation with children's immigrant attitude but was unrelated to their in-group attitudes. Conversely, children's ethnic identification was positively related to their in-group attitudes but it was unrelated to their attitude toward the immigrant-origin groups. To examine the robustness of the effects of autochthony on children's out-group attitude, a SEM model was specified in which its interactions with identification, age, and gender was added. None of the interactions were significant (p > 0.66) which means that the link between autochthony and immigrant attitude was similar for lower and higher ethnic identifiers, younger and older children, and boys and girls. Lastly, because of the skewed distribution of the ingroup attitude measure, an additional analysis was conducted in which this measure was dichotomized and treated as a categorical variable. As in Table 1, there was a positive effect of ethnic ingroup identification but no effects of the other variables. --- Study 2 The first study demonstrated that stronger autochthony belief is related to more negative attitudes toward immigrant-origin groups. The association found was relatively strong and robust across age, gender, and ethnic identification. There were four reasons for conducting Study 2. First, to ensure that the results could be replicated with another sample of majority group children. Second, it could be argued that using two items to measure autochthony in Study 1 is limited and also that these items do not explicitly make the connection between being there first and having the right to decide. Therefore, in Study 2 a third item to the autochthony measure was added. Third, we wanted to examine whether the results are robust or generalize to a different measure of group attitudes and a different immigrant group. Therefore, in Study 2 a measure of trait evaluations was used and the participating children were asked not only about their attitudes toward the main immigrantorigin groups but also toward refugees. Youth tend to have more positive attitudes towards refugees because of feelings of sympathy and pity (e.g., Murray and Marx 2013). However, the importance of autochthony beliefs for outgroup attitudes might be similar for both types of outgroups. Fourth, perceived multicultural education as a meaningful factor for children's immigrant attitudes was considered and the role of classroom ethnic composition was examined as there was significant variation between classrooms (see below). Multicultural education tries to foster equality and inclusion which tends to improve intergroup attitudes (Schachner et al. 2016;Verkuyten and Thijs 2013). This could mean that the associations found are due to perceived multicultural education as a third confounding factor. Likewise, native majority students in more diverse, less segregated classrooms may be more open to ethnic diversity because they have more opportunities for intergroup contact (Thijs and Verkuyten 2014). Therefore, it was examined whether autochthony predicts immigrant attitudes independently of perceived multicultural education and classroom ethnic composition. --- Method --- Participants and procedure One-hundred-and-ninety-five native Dutch students (54.5% girls) from 22 classes (grades 4-6) in 15 schools in different parts of the Netherlands participated in this research. Together with their classmates, they took part in a larger short-term longitudinal study on teachers' classroom dealings with ethnic diversity. That larger study consisted of three waves and at all waves children anonymously filled in questionnaires in their classroom under supervision of their teacher or a research assistant. Apart from questions related to ethnic diversity, the questionnaires included items on children's experiences with their teachers and their peer relations. Again, children participated voluntarily and there was consent from their parents and the possibility to opt out. In the present study data from Wave 3 (June and July) contained questions on autochthony and were therefore used. However, the Wave 1 data (October and November) were relied on to select the participants because only these data included information about the ethnicity of the parents. At Wave 1, i.e. seven to nine months prior to completing the autochthony and group attitude measures, the children were on average 10.13 years old (SD = 0.82). To be included in the current analysis, the children had to self-identify as ethnic Dutch and to indicate that both of their parents were of Dutch origin. Originally, 212 participants were selected in this manner. Yet as few scores on the variables were missing (<unk>5%), and the pattern of missing values appeared to be completely at random, <unk> 2 (52) = 56.947, p = 0.30, listwise deletion was used. --- Measures Autochthony was measured with the same two items as in Study 1 plus one additional item, namely "The people who came to live here [the Netherlands] first may decide what might change". The items had a response scale ranging from 1 (No!) to 5 (Yes!) and yielded a Cronbach's <unk> of 0.82. Ethnic identification was measured in this dataset with two of the identification questions used in Study 1. Children were asked whether they liked being Dutch and whether they were proud to be Dutch, using response scales from 1 (Not all) to 5 (Very much). The correlation between both items was 0.54. Perceived multicultural education was assessed with three items taken from previous research in the Netherlands (see Verkuyten and Thijs 2013). Children were asked 'Does your teacher ever say that all cultures should be respected?', 'Does your teacher ever say that it is wrong to discriminate?', and 'Does your teacher ever say that people from all cultures are equal?'. The response scale ranged from 1 (Absolutely never) to 5 (Very often) with <unk> = 0.83. Classroom ethnic composition was operationalized as the proportion of Dutch children in each class (M %Dutch = 52.25, SD = 27.22). This proportion was strongly related to the proportion of students who self-identified as Turkish or Moroccan (r = -0.77) but considerably less skewed (M %Turkish/Moroccan = 16.81, SD = 24.84). Confirmatory factor analysis in Mplus with ML as the estimator showed that the items for the independent variables (autochthony, ethnic identification, and perceived multicultural education) loaded on three corresponding factors without cross-loadings or error correlations, <unk> 2 (17) = 30.007, CFI = 0.978, TLI = 0.964, RMSEA = 0.060, SRMR = 0.044. Immigrant attitude was assessed through trait evaluations of Moroccan and Turkish peers (see Brown and Bigler 2002) that have been successfully used in previous research in the Netherlands (e.g., Thijs 2017). More specifically, participants indicated whether they thought that most of the children in each group were, "honest", "fun to play with", and "eager to help you". The same traits were used to measure the in-group attitude. The response scale ranged from 1 (NO, certainly not!) to 5 (YES, certainly!). For each of the three groups the evaluations yielded a reliable scale: Cronbach's alpha was 0.79 for the evaluation of the Dutch in-group, and respectively, 0.89 and 0.88 for the Moroccan and Turkish out-groups. The attitudes toward the two outgroups were strongly correlated (r = 0.71). Therefore, and in keeping with Study 1, these two measures were averaged in one measure for immigrant attitude. Attitude toward refugees was measured with two items: "Some people think that there are too many refugees coming to the Netherlands and others don't. What do you think?" with a response scale ranging from 1 (Absolutely not too many!) to 5 (Far too many!), and "Some people think that refugees are helped too little and other people think they are helped too much. What do you think?" with a response scale from 1 (Far too little!) to 5 (Far too much!). The correlation between these items was 0.60. Items were recoded so that higher scores indicated a more positive attitude. Confirmatory factor analysis in Mplus showed that the items for the dependent variables loaded on different factors (without cross-loadings or error correlations) for the attitudes toward, respectively, the Dutch in-group, Moroccans, Turks, and refugees, <unk> 2 (38) = 89.907, CFI = 0.956, TLI = 0.936, RMSEA = 0.080, SRMR = 0.030. Moreover, there was negligible drop in model fit when a higher-order factor was specified for the attitude toward the two immigrantorigin groups, <unk> 2 dif (1) = 0.026, p = 0.87, which justifies the decision to examine children's attitudes toward Moroccans and Turks as a single out-group. --- Results The intercorrelations and means of the main variables are shown in Table 2. As in Study 1, children reported a more positive evaluation of their in-group compared to the immigrant out-groups, t (194) = 10.91, p <unk> 0.001. Compared to Study 1 the distribution of children's in-group evaluation was considerably more normal (skewness = -0.26, SE = 0.17, kurtosis = -0.91, SE = 0.35). To examine the unique contribution of children's autochthony beliefs on their attitudes, a multivariate regression model in Mplus was specified. Given sample size restrictions, observed variables were analyzed. Moreover, the multilevel structure was taken into account, as children's attitudes toward refugees differed systematically between classrooms (Intraclass correlation = 0.18). Because of the limited number of classrooms children's attitudes toward respectively immigrants and refugees and their in-group were analyzed in separate models. Whereas ML was used as the estimator for the first model, MLF was used for the in-group model due to estimation problems. In both models these attitudes were regressed on autochthony, ethnic identification, and perceived multicultural education on the individual level, and the proportion of Dutch students on the 3. Autochthony had a positive effect on the attitude toward the in-group, and moderate and large negative effects on children's attitudes toward, respectively, immigrant-origin children and refugees. Further inspection showed that, although the effect of autochthony was larger for refugees versus immigrants, the difference between the effects was not significant, <unk> 2 dif (1) = 1.553, p = 0.22. Next and similar to Study 1, children's ethnic identification was related only to their in-group attitude. Perceived multicultural education was positively associated with out-group attitudes, and, somewhat surprisingly, the proportion of Dutch students had a positive effect on children's evaluations of refugees. In an additional set of analyses it was tested whether autochthony interacted with ethnic identification, perceived multicultural education, classroom ethnic composition, age, and gender, and to avoid model under-identification, separate models for each dependent variable were estimated. None of these two-way interactions were significant, p s > 0.15, indicating the robustness of the effects of autochthony on the attitudes towards immigrants and refugees. Thus the findings in Study 2 are consistent with those in Study 1 and demonstrate that the importance of autochthony beliefs for immigrant attitudes is independent of ethnic identification, perceived multicultural education, and classroom ethnic composition, and generalizes to another sample of majority group children, another group attitude measure, and another type of migrant group (refugees). --- Study 3 The aim of Study 3 was to investigate whether feeling at home in the Netherlands as a country is a relevant condition for the association between autochthony and immigrant attitudes. It was expected that for majority group children with relatively low home feeling, autochthony is more strongly negatively associated with immigrant attitudes. The reason is that children are more likely to adopt a 'primordial reasoning' when their perceived ownership right to feel at home in their own country is undermined (Duyvendak 2011). Furthermore, this is especially likely for children who at the same time consider their Dutch identity an important aspect of their sense of self. According to social identity theory (Tajfel and Turner 1979), relatively low identifiers should be less concerned about their in-group and less inclined to justify negative out-group attitudes. Thus a three-way interaction between autochthony, home feeling and ethnic identification on attitudes toward immigrant was expected. --- Method --- Participants and procedure The sample consisted of 337 ethnic Dutch students from 36 classes (grades 4-6; M %Dutch = 41.48, SD = 26.98) in 16 schools in different parts of the Netherlands. These children (M age = 10.84 years, SD = 0.94; 52.8% girls) took part in a short-term longitudinal study on classroom dealings with ethnic diversity. The study consisted of two waves. Wave 1 was halfway through the school year (January-March) and Wave 2 was at the end of the school year (June and July). During both waves children were surveyed under similar conditions. They completed a questionnaire in their classroom, together with their classmates, and under supervision of their teacher and/or a research assistant. As in Study 2, the questionnaires included questions related to ethnic diversity and children's experiences with their teachers and peers. Participation in the study was voluntary and anonymous and all children with parental consent participated. 4 The children were selected based

Autochthony belief ("that a country is owned by its first inhabitants") can be an acceptable reason for claiming collective ownership of a territory and this claim can have negative consequences for newcomers. Children might reason that a place belongs to their in-group because "we" were here first and therefore have negative out-group attitudes. In three studies among Dutch majority group children (N = 879; M age = 10.13 to 10.84, SD = 0.82 to 0.98; 49.7 to 54.5% girls), the expected negative association between autochthony beliefs and attitudes was found for different measures of ethnic attitudes, and was robust across gender, age, immigrant target group, ethnic identification, perceived multicultural education and classroom composition. Additionally, the association was especially strong among ethnic majority children who felt less at home in their own country but at the same time cared about being Dutch. It is concluded that a focus on autochthony belief makes a novel and relevant contribution to the intergroup developmental literature and to our limited understanding of children's attitudes toward immigrant groups and newcomers more generally.

less inclined to justify negative out-group attitudes. Thus a three-way interaction between autochthony, home feeling and ethnic identification on attitudes toward immigrant was expected. --- Method --- Participants and procedure The sample consisted of 337 ethnic Dutch students from 36 classes (grades 4-6; M %Dutch = 41.48, SD = 26.98) in 16 schools in different parts of the Netherlands. These children (M age = 10.84 years, SD = 0.94; 52.8% girls) took part in a short-term longitudinal study on classroom dealings with ethnic diversity. The study consisted of two waves. Wave 1 was halfway through the school year (January-March) and Wave 2 was at the end of the school year (June and July). During both waves children were surveyed under similar conditions. They completed a questionnaire in their classroom, together with their classmates, and under supervision of their teacher and/or a research assistant. As in Study 2, the questionnaires included questions related to ethnic diversity and children's experiences with their teachers and peers. Participation in the study was voluntary and anonymous and all children with parental consent participated. 4 The children were selected based on their ethnic self-labeling at Wave 2 but the additional criteria were used that this self-labeling should be similar to that of Wave 1 and that both parents of the children should be born in the Netherlands. Originally 359 children could be selected in this way but because there were few missing values on the variables (<unk>2%) and missings appeared to be at random, <unk> 2 (40) = 39.506, p = 0.49, listwise deletion was used. --- Measures All measures were collected during Wave 2, except children's perception of multicultural education which was assessed at Wave 1. Autochthony was measured with the same three items and response scales used in Study 2 (<unk> = 0.71). Ethnic identification was measured with the same three items as in Study 1 (<unk> = 0.66). Home feeling was assessed with three items that have been successfully used in previous research among children in the Netherlands (Verkuyten et al. 2014). On a five-point scale ranging from 1 (No!) to 5 (Yes!) children indicated whether they felt at home in the Netherlands, whether they were proud of the Netherlands, and whether they liked it in the Netherlands (<unk> = 0.86). Perceived multicultural education was measured with the same three items that were used in Study 2 (<unk> = 0.74). Confirmatory factor analysis in Mplus (on n = 359, and with ML as the estimator) showed that these four measures corresponded to four different factors with independent error terms and no cross-loadings, <unk> 2 (48) = 146.738, CFI = 0.930, TLI = 0.903, RMSEA = 0.076, SRMR = 0.057. The fit of this model was substantially better than that of a three-factor model in which ethnic identification and feeling at home loaded on one factor, <unk> 2 dif (3) = 104.601, p <unk> 0.001. Further, because the items for home feeling had a skewed distribution, CFA was also conducted with MLR as an estimator. Model fit was acceptable after allowing an error correlation between the first and the last item for home feeling, <unk> 2 (47) = 107.602, CFI = 0.946, TLI = 0.924, RMSEA = 0.060, SRMR = 0.053. Attitude towards immigrants was again measured with the'seven faces' scale (Yee and Brown 1992) in relation to people of Turkish and Moroccan background. As these evaluations were strongly related (r = 0.77) these measures were combined into a single out-group attitude score. In this study the attitude toward refugees was not assessed and ingroup attitude was measured with the same'seven-faces' scale. --- Results The intercorrelations and means of the main study variables are shown in Table 4. As in the previous two studies, children reported a more positive attitude toward their ingroup than toward the immigrant out-groups, t (336) = 23.07, p <unk> 0.001. As in Study 1, the distribution of the ingroup attitude was non-normal (skewness -3.99, SE = 0.13; kurtosis = 21.55, SE = 0.27). To examine the unique contribution of children's autochthony beliefs on their attitudes a series of multivariate regression models in Mplus was specified. To account for the non-normal distributions of in-group attitude MLR was used as an estimator. Moreover, given sample size restrictions, observed variables were analyzed and the multilevel structure was taken into account as there were substantial differences between classrooms in children's evaluations of the immigrant groups (Interclass correlation = 0.17). Prior to the analyses, all continuous variables were standardized (z scores) at the individual level. In the first model, children's in-group attitude and immigrant attitude were regressed on age, gender, and the classroom proportion of Dutch students, as well as on autochthony, ethnic identification, home feeling, and perceived multicultural education. The results of the model are shown in Table 5 (Model 1). Autochthony had a small positive effect on the evaluation of the in-group and again a medium-sized negative effect on the evaluation of the immigrant-origin groups. These effects were independent of ethnic identification which was positively related to the evaluation of the in-group, and in this study also negatively to the evaluation of the immigrant out-groups. Home feeling was positively related to both in-group attitude and the attitude toward immigrants. Additionally, boys were more positive about the majority in-group, and similar to Study 2, perceived multicultural education was positively related to children's attitude toward immigrants. There were no effects of the proportion of Dutch students. In the second model, the two-and three-way interactions between autochthony, ethnic identification and home 5) none of the interactions were significant for the attitude toward the ethnic in-group. However, for the attitude toward immigrants the two-way interaction between autochthony and home feeling, and the three-way interaction between autochthony, ethnic identification and home feeling were both significant. Simple slope analyses were conducted to decompose these interactions. The two-way interaction between home feeling and autochthony was first examined by calculating the effect of autochthony at low (1 SD > M) versus high (1 SD > M) versus levels of home feeling. As shown in Fig. 1, this effect was stronger in the former case, b = -0.49, se = 0.07, versus b = -0.28, se = 0.08, p s <unk> 0.001. Next, it was investigated whether this two-way interaction was different for children with relatively high (1 SD > M) versus low (1 SD > M) levels of ethnic identification. This showed that the two-way interaction was positive and significant for the high identifiers, b = 0.26, se = 0.07, p <unk> 0.001, but not for the low identifiers, b = -0.04, se = 0.06, p = 0.45. Figure 2 shows the effects of autochthony on children's immigrants attitude for different levels of ethnic identification and feeling at home. The effect of autochthony on immigrants attitude was strongest at high levels of ethnic identification combined with relatively low levels of home feeling, b = -0.66, se = 0.12, p <unk> 0.001, and weakest at high levels of ethnic identification combined with high levels of home feeling, b = -0.14, se = 0.09, p = 0.12. Moreover, the attitude was clearly most negative among children who endorsed autochthony, did not feel at home in the Netherlands and considered their Dutch identity an important part of their sense of self. Two further models (not in Table 5) examined whether autochthony interacted with the remaining predictors (respectively, with perceived multicultural education and the proportion of Dutch students in Model 3a, and with age and gender in Model 3b) by adding these two-way interactions to the previous model (Model 2). None of these twoway interactions were significant (p s > 0.10). Finally, in light of its skewed distribution, a logistic regression was performed on dichotomized version of the --- FH low Fig. 1 Effects of autochthony on the attitude towards immigrants (standardized) for low (1 SD > M) versus high (1 SD > M) levels of home feeling (FH) in-group attitude variable (just as in Study 1). This analysis failed to replicate the positive main effect of autochthony (Table 5), OR 1.373, p = 0.29 which means that positive main effect should be interpreted with care. The findings of Study 3, again, show that autochthony is negatively associated with attitudes toward immigrantorigin groups. However, this association is stronger for those children who have a lower feeling of being at home while at the same time self-identify as Dutch. This suggests that autochthony functions as a justifying belief for native children who care about being Dutch but do not feel at home in the Netherlands. --- Discussion The notion of ownership is widespread, pervasive, and applied to a range of objects including places in children's lives at home, in school, their neighborhood and the country they live in (e.g., Barrett 2007). Furthermore, the belief that a territory belongs to those who arrived first is often selfevidently used to claim rights (e.g., "first nations, indigenous people") and to exclude outsiders and newcomers (Ceuppens 2011;Geschiere 2009). Those who were somewhere first are typically considered to own the place with the related rights of usage and "gatekeeping" which provide justified reasons to exclude others (Merrill 1998). Yet, research on children's intergroup attitudes has not paid systematic attention to these issues and the current research is the first one that has examined the importance of autochthony beliefs for children's attitudes toward immigrant groups. In three studies, it was found that autochthony belief and ethnic identification were empirically distinct constructs, and that autochthony was not strongly and consistently associated with in-group attitude. These findings suggest that first arrival is a basis for inferring ownership and not so much for a sense of group belonging or in-group liking (Verkuyten et al. 2014). Furthermore, the findings demonstrate that ethnic majority group children with stronger autochthony beliefs had more negative attitudes toward immigrant-origin groups and refugees which suggests that both groups of newcomers were perceived in similar ways. This expected association was found among three different samples, by using different measures of ethnic attitudes, and by controlling for ethnic identification, perceived multicultural education and classroom composition. Furthermore, the importance of autochthony belief for out-group attitudes was robust across gender, age, level of ethnic identification, and level of perceived multicultural education. Thus there was clear and substantial evidence (small to medium effect sizes) that majority children who believe that their in-group was here first and therefore owns the country, have more negative attitudes toward immigrant groups. Additionally, Study 3 showed that this association was especially strong among children who felt less at home in their country. This finding is reminiscent of the widespread discourse in many Western countries that argues that immigration makes majority members feel like foreigners in their own country (Duyvendak 2011). Although they were here first and therefore have the right to feel at home in their own country (see quote heading this paper), they do not feel at home anymore and therefore should take back control, the argument goes. This discourse is particularly appealing to people who identify with the majority group (Martinovic and Verkuyten 2013). Children are aware of the societal debate on immigration and anti-immigration sentiments (Brown 2011) and similar to the prevailing discourse, it was found that especially for children with high ethnic identification, the combination of autochthony beliefs with not feeling at home was associated with negative attitudes. Interestingly, in the three studies the importance of autochthony beliefs for the attitudes did not depend on the level of ethnic identification. This indicates that autochthony belief is not a more relevant justifying belief for higher compared to lower identifiers. However, when at the same time there is a sense of not feeling at home, the level of identification seems to matter. This finding can be interpreted in terms of social identity development theory (Nesdale 2008) which proposes that ethnic prejudice appears when children who identify relatively strongly with their group feel that their in-group position or well-being is undermined in some way by members of ethnic out-groups. As with all studies, there are limitations to our research that provide possible directions for future research. First, although the findings are consistent and robust across the three studies, future research could try to use more extensive measures and try to assess additional constructs (e.g., feelings of outgroup threat, moral reasoning) which could Fig. 2 Effects of autochthony on attitude towards immigrants (standardized) depending on ethnic identification (EI) and feeling at home (FH), with 'low' and 'high' denoting, respectively one standard deviation below and above the means increase the explained variance. For example, the measure of country belonging did not ask whether children felt less at home in their country because of the arrival and presence of immigrant groups. Considering the societal discourse (Duyvendak 2011) and research findings (Huijnk and Andriessen 2016) this is very likely and the pattern of findings is in agreement with this interpretation, but future research could assess this directly. Future research could also examine the associations longitudinally or using an experimental design. There is experimental evidence that children consider first arrival a legitimate reason for claiming personal as well as collective territorial ownership (Verkuyten et al. 2015a(Verkuyten et al., 2015b). Yet, in contrast to research among adults (Martinovic et al. 2019), there is among children no experimental evidence that collective ownership based on primo-occupancy influences out-group attitudes. Second, the role of autochthony beliefs for attitudes toward immigrant groups in the Netherlands was examined. Similar to most European countries, in the Netherlands there is a large native majority population. This is different from immigration countries such as the United States, Canada and Australia that were inhabited first by indigenous groups of Native Americans and Aborigines. This could mean that for the White majority population in these countries collective ownership based on autochthony beliefs is less useful for justifying anti-immigrant attitudes. However, a sense of ownership with the related entitlements might also be derived from being earlier than later immigrants and from having invested in and developed the land (Verkuyten and Martinovic 2017). First arrival might be disregarded when later arrivers think that they are the ones who have made the land prosper. There will be many situations in which place ownership inferences are not based on the first arrival assumption alone, or at all. A lay belief can be interpreted and used in more than one way across situations: when one of its interpretations is not useful or appropriate in a particular context, another interpretation can be invoked (Levy et al. 2008). Future research could examine these possibilities in different national contexts, including in contexts of territorial disputes, such as in the Middle East (Zebian and Rochat 2012). Third, autochthony belief was examined in relation to the country but there are many other places in relation to which youth makes first arrival ownership claims, such as when children convert a place in their home or outside in their private playing area (Factor 2004), and in gang behavior among youth (Childress 2004;Kintrea et al. 2008). Having a sense of collective ownership ("ours") can involve many different settings with similar exclusionary consequences. Hence, it could be examined whether the present findings also apply to neighborhoods, schools, play areas and other places in which children feels that they own the place because they were there first. It is likely that a sense of collective ownership based on primo-occupancy is a more general process that has implications for children's prejudices and forms of social exclusion in a range of settings. Fourth, concern with ownership is already evident in young children but the evidence relates primarily to ownership of objects and ideas and therefore it is unclear how young children reason about place ownership and how this develops. We focused on older children and future studies could use a longitudinal design to examine, for example, at what age children develop an understanding that land, or a particular place, can be owned and the type of information that they use to infer place ownership and the related entitlements. With such a design it can also be examined how autochthony beliefs develop and whether and how these beliefs, for example, relate to the ethnic and national identity development of youth. Furthermore, such a design would be useful for examining where children's autochthony beliefs come from and the roles that parents, teachers and peers play in developing these beliefs and using them to justify negative attitudes towards immigrant groups. --- Conclusion This research has tried to make a novel and first contribution to the intergroup developmental literature (Rutland et al. 2010) by focusing on the importance of autochthony beliefs. This literature has examined out-group attitudes in relation to, for example, identification processes, threats, group norms and moral reasoning, and there is work on lay theories and justifying beliefs. Extending this literature to the important field of perceived ownership (Nancekivell et al. 2013), we showed that being here first is a relevant consideration for children's out-group attitudes. This corresponds to research among adults that has demonstrated that notions of autochthony are central in "sons of soil" conflicts (Côté and Mitchell 2017) and that people use these notions in territorial disputes and in exclusionary behavior and negative feelings towards outsiders and immigrants (Ceuppens and Geschiere 2005;Geschiere 2009). For children, being here first can be an acceptable reason for claiming collective ownership of a place and this claim can have negative consequences for newcomers. The reception and accommodation of immigrants and refugees is one of the major social challenges of our times. Understanding how majority group children think about these newcomers is important for trying to address their feelings and concerns. Hopefully, the current research adds to a further understanding of children's out-group attitudes and thereby helps to evaluate and further think about initiatives to improve their attitudes. --- Authors' Contributions All authors developed the questionnaire for this study. MV wrote the introduction and the discussion, and provided input on the analyses; JT performed the analyses and wrote the method and results sections. Both authors read and approved the final manuscript. --- Data Sharing and Declaration The dataset generated and the analysis of the current study are not publicly available but are deposited at the safe storage facility of the University and they are available from the second author on request. --- Compliance with Ethical Standards Conflict of Interest The authors declare that they have no conflict of interest. Ethical Approval All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. Informed Consent Passive parental consent was obtained, and all students included in the study participated voluntarily. Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://crea tivecommons.org/licenses/by/4.0/), which permits use, duplication, adaptation, distribution, and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.

Autochthony belief ("that a country is owned by its first inhabitants") can be an acceptable reason for claiming collective ownership of a territory and this claim can have negative consequences for newcomers. Children might reason that a place belongs to their in-group because "we" were here first and therefore have negative out-group attitudes. In three studies among Dutch majority group children (N = 879; M age = 10.13 to 10.84, SD = 0.82 to 0.98; 49.7 to 54.5% girls), the expected negative association between autochthony beliefs and attitudes was found for different measures of ethnic attitudes, and was robust across gender, age, immigrant target group, ethnic identification, perceived multicultural education and classroom composition. Additionally, the association was especially strong among ethnic majority children who felt less at home in their own country but at the same time cared about being Dutch. It is concluded that a focus on autochthony belief makes a novel and relevant contribution to the intergroup developmental literature and to our limited understanding of children's attitudes toward immigrant groups and newcomers more generally.

socialism'. 3 Back in 2003, Lynne Haney considered it a 'prevailing mythology' to perceive 1989 as a 'grand historical marker', a rupture in terms of the welfare systems. 4 Instead, it has become the dominant agenda of recent historical scholarship to broaden the scope and examine the historical rupture of '1989' as a reflection of a long process of transformation. 5 Herein lies the 'ambivalent nature of 1989', which manifests itself in the fact that it is understood and examined 'both as an event and as a convenient label for the roughly decade-long process of political, social and economic changes that followed the demise of the late-socialist regimes'. 6 In that vein, Florian Kührer-Wielach and Sarah Lemmen suggested to focus on those processes of adaptation rather than solely on the dialectic relationship between continuities and discontinuities. 7 Recent research attempts to capture the diverse individual responses to the postsocialist transformation. In 1999, Katherine Verdery and Michael Burawoy called this period an 'Uncertain Transition', 8 that was 'not a unilinear one of moving from one stage to the next [...] but a combined and uneven one having multiple trajectories'. 9 Twenty years later, Jill Massino speaks of 'Ambiguous Transitions', 10 stressing the importance of capturing people's 'complex and ambiguous' responses to the postsocialist transformation, which were'shaped by their experiences of both past and present and the disconnect between expectation and reality '. 11 Keeping this complexity in mind, children's encounter with 1989 can also serve as a magnifying glass onto the longer period of political rupture and transformation. Beyond serving as a container for individual and collective experiences, the label '1989' captures well the simultaneity of rapid political and social change 12 as well as the longer process of political and social transformation of political systems and social practices. In this double function '1989'as an abrupt caesura and a reflection of long-term social negotiations-manifested itself in the everyday life of children. Massive privatization, high unemployment, new housing and living conditions/standards, migration to the West, and new pedagogical ideas of children's care and education brought about fundamental changes to children's upbringing. But how unique was century affected the everyday lives of children. Our incentive is that the study of political transformations throughout the twentieth century has grown extensively, 13 but the experiences of children and young people in the turmoils remain less researched. 14 Against this backdrop, the contributions to this double special issue offer to shed light on three major political ruptures and the longer term transformations these entailed in the twentieth century, including their impact on notions and everyday realities of childhood and youth: World War I and its aftermath, World War II and its following years, and '1989' as the end of the Cold War and the beginning of postcommunism. While Martin Conway considered a comparison of 1989, 1945, and 1918 a 'rather awkward triple jump across the 20 th century', 15 he and his co-editors still ventured to provide an entangled comparison of these three watersheds. He felt that such a comparison could be historically volatile as'modern European wars [...] have ragged ends, and to seize upon a specific endpoint of these conflicts risks neglecting the after-wars that were such a prominent element of all three conflicts'. 16 Taking up Conway's proposal, this special issue goes one step further and approaches the historical moments 1918, 1945, and 1989 not as isolated political ruptures or caesuras. Instead, it explores these key moments as symbolic markers of longer historical processes of political and social transformation. Against this backdrop, the contributions scrutinize how these abrupt ruptures and the longer processes of political and social transformation could become visible in children's lives. We approach the two World Wars, the Cold War, and their respective aftermaths as long(er) transformation periods that could affect and/or alter children's everyday lives. The articles assembled here encourage us to study what it meant for children that the 'transition from war to peace' and from state socialism to postsocialism 'was rarely clear-cut'. 17 One Polish Jewish child survivor of the Holocaust felt that 'the war began for me after the war'. 18 Joanna B. MIchlic argued that for Jewish child survivors in Poland, the 'early postwar period was still very much a turbulent era full of magnitude of messy and confusing events.' These events were so decisive for determining 'the future of these children and how they were to develop, both as individuals and as members of national and cultural communities.' Children's social identities were in the early postwar period still 'fragile, fluid and exposed to many shifts'. 19 Not only the children's identities were fluid in the postwar, but so was the transition from war to peace. With this in mind, this double special issue pursues three major aims. First, it proposes to take up the contention that not only the two World Wars and the Cold War 'provided a defining framework for the shape of the century'; their aftermaths and the subsequent 'transitional periods', during which important 'new structures of power and of organizations became apparent, also had similar effects'. 20 This allows us to not only juxtapose the impact of the two World Wars and the Cold War on childhood but also integrate the aftermath of these conflicts into a diachronic perspective on longer processes of political and social transformation. Second, it aims to test the usefulness and applicability of the notion of 'transformation' for describing not only the processes that accompanied the postsocialist period but also those that surrounded the watersheds of the World Wars. As 'radical system or regime changes', which occurred in 1918, 1945, and 1989, 'only take place in a crisis after the prevailing weakening of the ruling regime' 21, an analysis of political transformation needs to consider the preceding wars and/ or the Cold War, the regime changes during 1918, 1945, and 1989, and their long aftermaths. We also engage with the question as to how the two World Wars, 1989 and their aftermaths brought about a social transformation, which left an imprint on the everyday lives of children. Here we examine how these periods of transformation became tangible in children's lives. We are curious to investigate how they transformed family ties and children's relationships, children's upbringing and education, and children's welfare. And third, this double special issue uses the lens of childhood to provide a diachronic exploration of the relationship between political and social transformation processes and their implications for children's lives in the twentieth century. Judith Szalai argued in a United Nations International Children's Emergency Fund (UNICEF) report in 1992 that 'One can hardly find more sensitive indicators of the well-being of a given society than those measuring the living conditions and future prospects of the children in that society'. 22 As modern notions of childhood considered children to be innocent, vulnerable, pure, dependent, and ignorant, children were to be protected, cared for, and should enjoy special freedoms. For that reason, as the childhood historian Paula Fass stressed it, modern states were 'committed to withdrawing them from certain adult spheres like war and sexuality'. 23 But if, despite such utopian visions, wars and political crises invaded children's private lives, their everyday lives reflect well the gravity and scope of political and social ruptures as well as of political and social transformation. As children's everyday concerns were (and still are) fundamentally tied to and dependent on the overall functioning of societies, their lives mirror how political crises and transformations affected societies in general. In so far as states also considered children a 'political tabula rasa' 24 or a 'blank slate' 25 that could be used to revolutionize contemporary and envision future societies, children also came to gain particular value in processes of transformation. Children's new role and reinvented education, their new position vis-à-vis the state, and their altered everyday lives literally embodied the respective systemic change and the longer transformation process. This double special issue offers a range of thematic contributions that tackle selected historical case studies of (post)war and (post)socialist childhoods. Whereas employing the case study approach offers the opportunity to detail childhoods in their specific historical settings, gathering them in a double special issue allows readers to detect similarities and differences among children across time and space and support a thinking of the twentieth century through the lens of 'transformation'. The papers combine a top-down perspective on children (focusing on child policies and on how adults treated children) with a meticulous unraveling of how children experienced their upbringing in times of political rupture(s) and political and social transformation. Just like the lives of adults, the lives of children were influenced by such events and changes. Indeed, children not only experienced situations differently from adults but also often faced other horizons of experience. Young children, for example, were often less politically informed and, as a result, did not understand or share the opinions of adults. 26 As children are 'historical actors with their own ideas, intentions and identity-forming experiences [...] [and not only] [...] mere objects of adult design' 27, we aim to engage with children's own subjectivities and their social agency, 28 their actual reactions, their own wishes, and their resilience. Julia Grant accentuates children's agency with the following words: 'Children are unruly, unpredictable, complex and cannot easily be encapsulated in a historical narrative'. 29 Zsuzsa Millei observed that the socialist period of children's everyday life 'could not be easily reduced to the dichotomies of official and unofficial, the public and the private, the state and the people'. 30 Taking children's perspective into account, we have assembled exemplary case studies that explore displaced, orphaned, uprooted, unaccompanied children, borderland children, young survivors of forced migration, Jewish child survivors of the Holocaust, and children of the postsocialist transition. Through examples of how these children were treated, as well as the experiences of these children, we engage with changing conceptions and practices of children's upbringing and institutional care in orphanages and in displaced persons' camps. Engaging with case studies from Lithuania, Russia, Poland, France, Hungary, Ukraine, and Estonia, the authors investigate various notions and everyday realities of childhood through the prism of transformation. --- World War I and its aftermath: transforming childhoods The Great War and its often long and troublesome aftermath had a decisive influence on the way in which children were approached by adults in the political entities that are discussed in this double special issue, as well as on how they experienced childhood. However, these turning points took on different forms for different children. 31 Many families and children were facing the absence and death of soldier fathers, the loss of one's home and house, migration and displacement, and their new responsibilities. Even though studies often only address childhood during war, 1918 as the war's official end did not end children's encounter with this violent conflict. The postimperial condition of Central and Eastern Europe translated itself in children's everyday lives in the following ways: it caused high infant mortality rates, the spreading of diseases and epidemics, the decline in birth rates, the interruption of children's education and care, and the massive abandonment, neglect and displacement of children. While children's experiences with the war and its troublesome aftermath were as complex as the conflict itself, they were instrumental in paving the way toward the international recognition that children should be better protected. Before the war, children's protection was mostly dependent on 32. R. Barbry the family and on religious associations. As the war and its aftermath exhibited children's particular vulnerability, it brought about the evolution and expansion of children's protection and of child welfare institutions. Child relief also became an issue of international concern and activism. 32 In the postwar period, new international relief organizations, such as Save the Children, were established and posited children as the 'principal victims of war' who were meant to be rescued and whose relief was considered the 'best way of investing in peace and promoting international cooperation'. 33 Save the Children was instrumental in pushing for the recognition of children's rights. The 1924 Geneva Declaration on the Rights of the Child was the internationalist response to the suffering that the war and the postwar years had brought to Europe's children. The declaration demanded that the hungry must be fed, the orphan must be sheltered, the child shall be the first to receive help, and it should receive an education to help itself. 34 With that in mind, Andrea Griffante focuses on the way that Lithuanian state formation was experimented with through child welfare during the World War. The political rupture of the German occupation established in 1915 yielded a transformation in child relief initiatives. The Lithuanian elite saw an unprecedented chance to include the lower social strata into its ranks and proliferate its ethno-national discourse. Not only did their initiatives now privilege displaced, destitute, and/ or abandoned children above other social groups, such initiatives were also organized on a massive scale for the first time. Lithuanian elite members believed that isolating the young into children's orphanages-hostels was the appropriate solution to execute hygiene control and ensure consistent education in the Lithuanian language and what were considered to be Lithuanian norms and values. Interestingly, the hosteled children remained largely unaffected by the sanitary catastrophe that led to the death of 9% of the city population of Vilnius in 1917. The transformation in child welfare initiated during World War I could, however, not be implemented fully due to the inadequate supply of the necessary means, such as foodstuffs and medical equipment, and the insufficient capability of the elite to negotiate the everyday practices in orphanages-hostels with the children. Griffante's analysis of children's voices provides us a deeper insight into the quality of that transformation, by bringing to the fore that the young protested against the living conditions, the isolation from social life outside the hostels, and even refused to speak Lithuanian. As a result, when Lithuania appeared as an independent state on the map of Europe following a next political rupture, the wartime hostels became the model for how to 'discipline and normalize prototypical Lithuanians' (Griffante). However, given the fact the disciplining and normalizing ideals had already been hypothetical during the war, these were not converted into a convincing set of real practices once that war came to an end, leading to the longterm repercussions of the transformation being rather moderate, in comparison with, as the author indicates, neighboring Poland. To conclude, by shedding light on child relief initiatives, practices, and discourses during the war and its immediate aftermath, Griffante's article points to the crucial role that the Lithuanian elite envisioned children to play for its nation-building, as well as that elite's unfamiliarity in negotiating a modus vivendi with the latter. Also Friederike Kind-Kovács's article concentrates on emerging child welfare during and after World War I, but the lens of her analysis is put on the Hungarian part of the Austro-Hungarian Monarchy and on the constitution of the independent Hungarian postwar state. She points to the emergence of 'war orphans' as a new category of victims, and, similar to Griffante, to the transformation of child welfare through its gradual professionalization as well as the installation of an infrastructure, which remained in place once the war had come to an end. War orphans became publicly visible as threefold symbols. Through their destitute or physically harmed appearance, they embodied the violence of the war. In addition, they personified a transformation in their society's attitude. Their parents were no longer blamed for inappropriately bringing up their children, but acknowledged for their participation in the war, and the societal need to take care of their children was recognized. At the same time, children were considered to symbolize the future, in which they as young adults would carry their country's reconstruction on their shoulders. This way of thinking provoked the emergence of wartime welfare initiatives in which one experimented with how a future postimperial Hungarian nation-state would and should look like. Although gathering war orphans from different backgrounds was meant to overcome ethnic nationalism, tensions between Hungarianspeaking, German-speaking, and Jewish children still filled the papers in 1917. A deep contrast also arose in the depiction of female war orphans and orphans' mothers. Whereas beauty contests were organized to provide girls with a scholarship and remind'readers of the country's well-being and future' (Kind-Kovacs), mothers of (half-)orphans were negatively stereotyped as being motivated by greed. The fear was articulated that the mothers' transformation into breadwinners within the wartime economy would have long-term repercussions on gender roles in postwar society, a fear that nourished conservative pronatalist policies in the years after the Great War. Machteld Venken's article concentrates on the transformation of elite education once the state borderlines of the European continent had been redrawn in the aftermath of World War I. In Polish Upper Silesia, the change in state sovereignty from German to Polish meant a rupture in the political scene, but not all the implications for elite education turned out to be disruptive. Venken finds the transformation phase to have been characterized by highly contested power relations and frequently changed school curricula. She also uncovered that radical changes in the schools' clientele were a feature of Polish-speaking schools in the area, while German-speaking schools could continue to educate the traditional bourgeois strata of society as long as the League of Nations executed supranational control. The author aims for a better understanding of the multidimensional interwar everyday life in Polish Upper Silesia, which in recent historiography has often been analyzed with the help of the analytical category of national indifference. By dissecting the practices of local decision-makers within a sequence of transformative events that inevitably led to a compliance with social norms, and within a space considered to be of crucial interest to Polish nationalists, the article demonstrates that placing oneself outside of the imagined national community, or dictating its meaning, was no longer an option. --- World War II and its aftermath: rehabilitating childhoods While World War I and its aftermath had invaded family lives extensively, Tara Zahra judged World War II as having 'destroyed the family as completely as Europe's train tracks, factories, bridges and roads.' 35 However, the end of the war 'did not mean that children's lives ceased to be uprooted. Many had lost close relatives during the war and would grow up within other family setups than before' as Machteld Venken and Maren Röger observed. 36 far greater toll on civilians than the Great War', 37 but also the war's aftermath continued to cast, for many children, a more intense or longer shadow over their further lives. Yet, the postwar period also captured a new international moral consciousness and the dedication to the public protection of children as one particularly vulnerable social group. When comparing relief in Central and Eastern Europe after the World Wars, we can detect a shift of attention from children's bodies to their souls. The period after World War II was invested in the social and psychological recovery of those children that were harmed by the trauma of the war and the Holocaust. The Universal Declaration of Human Rights, Article 25, claimed in 1948 that'motherhood and childhood are entitled to special care and assistance. All children, whether born in or out of wedlock, shall enjoy the same social protection'. 38 Here we also observe a new public obligation to provide long-term social protection to all those children and their mothers who were in need. Set against this background, two articles analyze the lives of children on the move during World War II, while placing spatial reconfiguration center stage. The children being brought to France through Kindertransport, Laura Hobson Faure argues, 'established their own chronologies of war' (Hobson Faure). Through an analysis of the diaries written by Jewish refugee children, the author demonstrates how political and personal rupture could relate in different ways and puts forward the thesis that a transformation of children's everyday reality took place through subsequent political and personal ruptures. Political and personal rupture could follow each other, such as in the case of child diary writer Heinz Löw lamenting the separation from his loved ones when his train departed in the spring 1939, on which his grandmother had put him after the occupation of Austria in 1938. Personal rupture could also coincide with watersheds on the global scene, such as the outbreak of World War II, which incited this young diary writer to articulate his feelings in writing, while his friend illustrated the diary for him. Afterwards, more Jewish refugees arrived in France, and children with different ideological profiles were put together in children's homes, a development that provoked, sharpened, or transformed their political awakening. When the German army invaded France in 1940, the Jewish refugee children were transferred southwards to France's unoccupied zone. They found shelter in a predominantly French-speaking environment for the first time. Later, when the Vichy regime started raids against Jews, the children had to flee or go into hiding. These frequent political ruptures influenced the work of the left-oriented educators who had founded progressive Children's Republics in children's homes upon the arrival of the Jewish refugee children. While the majority of the children survived Nazi persecution, others perished. By bringing us closer to the inner world of the Jewish children experiencing the turmoil of war, Hobson Faure points to the adaptability of children to take over the role of a parent in times of distress. Another group of children on the move, as Katharina Friedla postulates, were 'Polish-Jewish children, born or raised in Germany, who survived World War II in the Soviet hinterland (...)' (Friedla). Like the Kindertransport children, the journey of these children started before the outbreak of the war. In the autumn 1938, they were transported to Poland after the Polish state had revoked their Polish citizenship. When the German army occupied the Western part of Poland, many of these refugees escaped or were transported to the East, eventually reaching Siberia. In 1941, when the Polish-Soviet agreement was signed, many embarked on a journey to Central Asia, and up to 1000 children later joined the Anders' Army to Palestine. After the World War, many returned to Poland, but almost all left and traveled overseas within the following years. Like Hobson Faure, Friedla sees a profound transformation of children's lives emerging as a result of the rapid sequence of political ruptures. By the time they had reached Central Asia, these children had taken up the responsibility for their own survival. The author also found testimonies of former children who admitted that they had left their parents and had joined an orphanage to receive food. Other than Hobson Faure, Friedla also pays attention to the 'emotional transition' of the children caused by the pain and trauma of their migrations, which remained with them later in life as they found it difficult to articulate their memories (Friedla). On one hand, child survivors refused to identify as survivors of the Holocaust and compete with Holocaust memory because their stories 'only' consisted of forced migration, humiliation, and starvation, instead of mass extermination. On the other hand, since child survivors only remembered fragments of what they had experienced or witnessed through children's eyes, they held the generation of their parents responsible for constituting the collective memory of the Polish Jews who lived, were born, or raised in interwar Germany. The article of Olga Gnydiuk concentrates on displaced unaccompanied minors and their caregivers in the early aftermath of World War II. The author puts her lens of analysis on the fate of Polish-Ukrainian displaced children in the post-1945 world, who saw themselves cut off from their home grounds when these shifted sovereignty from Polish to Soviet sovereignty. This rupture was accompanied by 'political transformations', as the author defines: 'a gradual process that embraced the continuities of the war period but also reached out in time to have longer-lasting effects (...)' and took the form of 'a change of policies, humanitarian approaches and social practices, as well as crumbling diplomatic relations' (Gnydiuk). Gnydiuk's article provides an example of how child welfare activities after the World War II did more than provide children material support. Out of a belief that war had caused emotional damage, welfare workers were concerned with restoring victims' social and psychological health. While welfare workers followed the official principle of international relief, that is, to act in 'the best interests' of children, that principle, as Gnydiuk argues, appeared malleable enough to provide the opportunity to ignore the repatriation dictate to the Soviet Union included in the Treaty of Yalta in 1945, and thus led more children emigrate to Western countries. That child protection takes precedence over the rights of an individual country over 'its children' was an idea developed and practiced in the early postwar period, an idea that would pave the path to further developments of children's rights later. --- The long transformation of '1989': individualizing childhoods The third political rupture and longer transformation process that we center our attention on and which left a particular imprint on children's lives, especially in Central and Eastern Europe, accompanied the year '1989'. From its very inception, state socialism had pushed for an increasing shift from parental care to institutional care. Children's everyday lives had turned into a territory of intervention by state authorities, and institutionalization was believed to help bring up ideal versions of future generations. This conviction was fundamentally questioned by the political rupture in 1989. The collapse of state socialism translated itself in processes of liberalization, deregulation, and privatization. As most of children's care had been outsourced to state institutions, children had been highly dependent on the quality of care in kindergartens, orphanages, and childcare institutions, which had not always been a given. It had become apparent that the socialist type of comprehensive child welfare had not only been too expensive, but that it was often lacking in quality. The economy of shortages had visibly affected childcare provisions and institutions. The year 1989 witnessed the beginning of a process of deinstitutionalization; yet it also witnessed with the signing of the Convention on the Rights of the Child, an international agreement demanding that all states'shall ensure that the institutions, services and facilities responsible for the care or protection of children shall conform with the standards established by competent authorities'. 39 In contrast to the earlier child right declarations, this declaration called upon the signing states to secure the quality of public childcare. What followed in 1989 was an increasing 'individualization of life situations contra the prior ideal of collectivism and clear-cut conception of a normal biography'. 40 This also comprised a shift from a collective to a more individualized conception of childhood and children's upbringing and care. In the early postsocialist years, the state increasingly withdrew from children's public childcare. Yugoslavia's last generation experienced the postsocialist transformation as'marked by a deep sense of uncertainty and instability, as everything could be, and oftentimes was questioned'. 41 Beyond the top-down political transformation, the period was marked by an 'inner dynamics of transformation', and this 'transformation of the official rhetoric, politics and practices' could be studied from 'within the institutional youth sphere' where it also 'unfolded'. 42 The various postsocialist countries shared these experiences of an uncertain transition, which was shaped by the tension between socialist legacies and neoliberal policies. Much of what children had considered as eternal before 1989 either vanished into thin air or was abruptly devalued. 43 The change of school curricula and pedagogic reforms, the closure of state (and factory) kindergartens and nurseries, as well as parents' loss/change of their workplace, housing insecurities, and processes of increasing social stratification triggered insecurities, lack of social integration, and feelings of loss. This special issue contains two articles breaking new ground on the long-term impact of the collapse of the Eastern Bloc on (former) children and their caregivers. Raili Nugin examines how the communicative memory of what she coins the 1970s age cohort in Estonia presents the collapse of communism as a process rather than a political rupture. The author investigates people who enjoyed childhood during Soviet times and experienced their coming of age during the 1990s in Estonia. Other than their parents, who considered the collapse of communism a political rupture, the age cohort born in the 1970s experienced the late 1980s and 1990s as a period of transformation in which a change happened 'on both personal and social level' that was accompanied by 'a certain sense of moving forward and asking questions about identity' (Nugin). The author argues that the way in which individuals of this age cohort communicated their memories of coming of age in the public sphere provoked the development of a generational consciousness. They indicated that their Soviet memories mobilized them later in life to value freedom and independence, and bestowed them with a specific cultural capital (such as the capability to understand certain jokes) and the knowledge of hardship. This symbolic resource formed the basis for the cohort's generational discursive capital. In addition, Nugin analyses the interesting discourses of those who were too young to have been potentially able to play a decisive role during the Singing Revolution, due to which they discursively stretched the time period of the emergence of political culture in Estonia to the first decade after the country's independence. In this way, they molded their personal memories into a discursive tool enabling them to (falsely) claim a crucial social role for themselves during this process of change. Anna Kozlova builds the argument of her article around the observation that former counselors of the two most prominent Soviet pioneer children's centers did not remember the exact time when the Soviet Union collapsed, and stressed the continuity of the centers' activities instead. Since the Thaw in the second half of the 1950s, counselors in these pioneer camps had learned to detach rituals from their propagandistic meaning, and to bestow these with a self-appropriated understanding that was derived from humanistic pedagogy. During the later stagnation, inspectors representing the Soviet regime were not capable of erasing these seeds of progressive ideals from the minds and practices of individual camp counselors. This meant that the system's change in the early 1990s did not significantly affect the camps' working. By that time, indeed, these camps had come to function as semi-autonomous institutions constituted by the agency of their counselors, who gladly reactivated or intensified humanistic practices. On the basis of oral history interviews, the author unravels individual counselors' transformation, that is, how they supported and developed humanistic pedagogy by means of their daily practices, and in this way tried to give meaning to how they had co-shaped the camps' profiles and activities from the late 1980s into the 1990s and up until today. --- Exploring twentieth-century transformations through childhood Gathering a number of exemplary case studies of children's experiences in periods of transformation, this double special issue recommends an understanding of the twentieth century that reaches further than a central focus on the political ruptures of both World Wars, as well as the collapse of the Eastern Bloc in 1989. First, the contributions follow Tony Judt's observation that the century was characterized by the rise of state involvement. 44 Second, they put children, who had been posited as a distinct social group for receiving support from the state throughout Europe in the nineteenth century and the beginning of the twentieth century, at the center of the historical analysis. In this way, they aim to demonstrate how these three political ruptures were intertwined with political and social transformation. Taken together, the contributions engage with the various ways in which political rupture and political and social transformation related to one another in different time periods and how they impacted children in different or similar ways. The various articles detail these implied personal rupture(s), transformation(s) of the care provided to children, and changes in the practices, thoughts, and emotions of (former) children. What emerges from these readings is that an understanding of transformation cannot be reduced to a clearly defined period after a political rupture. Instead, a political and/or social transformation could start before (e.g. Nugin, Friedla) and continue during and after a political rupture, and that its duration depended on the historical context (e.g. Venken). Throughout the twentieth century, thus, political and social transformations were contingent, possessing a time frame that could shrink and expand depending on the circumstances at the time. With this investigation of the policies for and practices of children and their caregivers in a variety of local settings during political ruptures and political and social transformations, the special issue includes more experiences and voices in the study of our contemporary past.

In the past years, the so-called East German 'Children of the Transition' [Wendekinder] have come to raise the question of how 1989 and its aftermath affected children's lives in the past and how their memories still shape their individual and collective biographies up to today. 1 This new perspective on the years of postsocialist transformation allows for the reading of the history of '1989' through a historical exploration of children's past experiences. As Jan Kubik and Amy Linch pointed out in 2016, it is no longer of much use to approach the 'postcommunist transformations as a straightforward, uniform, linear, and easily generalizable "progress" from totalitarianism to democracy and from command economy to market economy', as it was initially proposed by the so-called transitologists of the 1990s. 2 Neither is it any more helpful to continue centering our attention on single years like '1989' as markers of abrupt and dramatic political and economic change. Historians such as Gregor Feindt have 'challenge [d] 1989 as a clear-cut break', and proposed instead to 'study the long transformation of state

Introduction The COVID-19 pandemic has renewed scholarly interest in the effects of work from home 1. Interestingly, studies have documented both positive and negative effects of WFH on employees' work and home lives (Allen et al., 2014;Kniffin et al., 2021;Massar et al., 2023), but have failed to provide a comprehensive explanation of why employees experience WFH differently. For example, employees who work from home report difficulty mentally unplugging during breaks and downtime, or being interrupted by their partners (Dunn, 2020). However, employees also report that WFH allows them to spend breaks with loved ones and return to work with renewed energy (Gohmann, 2020). Related research suggests that spillovers at the work-home interface may be the mechanism explaining why employees experience WFH differently (Huyghebaert-Zouaghi et al., 2022). The work-home interface encompasses four types of inter-role spillovers, based on the directionality (work-to-home and home-towork) and quality (conflict vs. enrichment) of the interaction: workto-home conflict and home-to-work conflict are negative spillovers, and work-to-home enrichment and home-to-work enrichment are positive spillovers. Although the four spillovers are theoretically and empirically distinct, they share some common predictors (Frone, 2003;Greenhaus and Powell, 2006;Huyghebaert-Zouaghi et al., 2022). Building on boundary theory (Ashforth et al., 2000), we first theorize that WFH intensity-which is the number of hours worked from home-can trigger all four types of inter-role spillovers. This view implies that WFH intensity has positive and negative effects on the work-home interface as well as positive and negative downstream consequences. Second, we develop a contingency view to examine the role of situational and personal moderators: work equipment, room conditions, and domain centrality. Specifically, we propose that the quality of one's work equipment can mitigate the work-to-home conflict, whereas room conditions can mitigate the home-to-work conflict. In addition, we propose that the centrality of the work and home domains can enhance the relationships between WFH and enrichment in the respective other domain. Third, we build on the domain specificity (Frone et al., 1992) and source attribution models (e.g., Shockley and Singla, 2011) to examine the downstream influences of spillovers on employees' satisfaction with their romantic relationships and jobs. Figure 1 depicts our theoretical model. To test our theory, we use a unique longitudinal data set that we collected in a two-step random sampling procedure during the COVID-19 pandemic in the city of Munich/Germany. We test our hypotheses with moderated regression analyses and use multiple mediation models to analyze whether negative and positive spillovers exist in parallel, and what their individual and joint effects on satisfaction outcomes are. Our work contributes to the literature in three ways. First, we theorize and test how WFH intensity affects all four dimensions of the work-home interface. WFH literature-before (for a review see Allen et al., 2015) and during (e.g., Kossek et al., 2021;Leroy et al., 2021) the pandemic-has primarily studied negative spillovers, whereas positive spillovers have received only scattered mentions (e.g., Rothbard et al., 2021) and less empirical scrutiny (e.g., McNall et al., 2009;Huyghebaert-Zouaghi et al., 2022). The joint consideration of negative and positive spillovers has theoretical and practical relevance. Theoretically, boundary theory predicts that WFH intensity can be a stressor and a resource simultaneously, triggering negative and positive spillovers through role integration (e.g., Ilies et al., 2009;Kossek et al., 2012;McNall et al., 2015). A one-sided focus on negative spillovers does not do justice to the complexity of the theory and might lead to incomplete practical considerations. Second, we test whether domain centrality moderates the relationships between WFH intensity and spillovers, thus contributing to an ongoing debate in boundary theory (Capitano and Greenhaus, 2018;Lapierre et al., 2018), which has competing views on whether domain centrality is more strongly related to positive spillovers in the receiving domain or in the originating domain. Our focus on workplace characteristics as moderators further provides actionable insights about how the negative consequences of WFH can be mitigated through organizational interventions (Kniffin et al., 2021). Finally, we theorize and test how WFH intensity affects the satisfaction of employees with their jobs and their romantic relationships, an important home-related outcome. So far, the effects of WFH on the private lives of employees are rather unknown, both to research and organizations. By examining employees' relationship satisfaction, we respond to calls to study WFH outcomes that are important but not directly relevant for the profits of organizations (Kramer and Kramer, 2021). --- Theoretical background and hypotheses In research and practice, boundary theory (Nippert-Eng, 1996;Ashforth et al., 2000;Clark, 2000) has been widely used to understand the dynamics of the work and non-work domains as well as their interactions. Specifically, boundary theory posits that roles are surrounded by psychological, physical, and temporal lines of demarcation (i.e., boundaries). Individuals navigate the boundaries and manage the transitions and interactions between the roles and domains to maintain a balance between the work and non-work domains. We use boundary theory as the overarching theoretical framework of our research. In what follows, we first explore how and why WFH can blur the boundaries between the work and the home domains, with negative consequences for employees' work-home interface and their work and home lives. --- Work from home intensity and negative spillovers Work from home requires employees to integrate their work and home roles, because they engage in work activities in physical spaces that are normally dedicated to their private lives. In such a situation, work roles are necessarily-and often unintentionally-enacted during times that are normally devoted to the home domain, and vice versa. This is embodied in the concept of boundary control (Kossek et al., 2012;Lapierre et al., 2016). With decreased control over one's boundaries, interruptions from partners, children, and other household members are more likely and can detract employees from work (Leroy et al., 2021), just as home domain activities (e.g., household duties, homeschooling) are more likely to interfere with one's work (Shockley and Allen, 2015), resulting in home-towork conflict. Employees engaged in WFH can also experience work-to-home conflict. The absence of face-to-face communications in the office can foster an "always connected" work culture. With an abundance of digital technologies and communication tools available, role transitions take place faster and more frequently, which in turn increases the likelihood of employees engaging in work activities during times that were traditionally dedicated to the home domain (Ferguson et al., 2016). In line with this logic, Lapierre et al. (2016) found that mandatory WFH was positively related to work-to-home conflict in a sample of financial sales professionals, and Leroy et al. (2021) reported that, since the onset of the pandemic, the number of work and non-work interruptions has increased, which applies equally to work-to-home conflict and home-to-work conflict. Therefore, WFH intensity-defined and operationalized as the number of hours worked from home-can create negative spillovers. WFH intensity will often lead to a stronger WFH concentration, such that single (or multiple) days of the week are completely devoted to WFH. Whereas there are some obvious advantages (which we discuss below) of intensive WFH, it comes along with a loss of boundary control. The loss of boundary control increases both the potential of work-to-home conflict and home-to-work conflict. For example, when complete days are worked from home, employees might find it increasingly difficult to switch off mentally after work, such that their non-work relationships can suffer. Likewise, with more hours worked from home it becomes increasingly difficult to avoid the regularly occurring interruptions of the home domain, such as children leaving to and returning from school, or household duties like cooking or house cleaning. Consequently, we pose the following hypotheses: Hypothesis 1: WFH intensity increases work-to-home conflict. Hypothesis 2: WFH intensity increases home-to-work conflict. --- Boundary conditions of the WFH intensity-negative spillover relationships Although it carries the potential for negative spillovers, WFH intensity is neither necessarily bad for everyone nor under all circumstances. In fact, meta-analyses prior to the pandemic have reported a small negative effect of WFH intensity on negative spillovers (Gajendran and Harrison, 2007;Allen et al., 2013). One plausible explanation is self-selection. Most employees who self-selected into WFH prior to the pandemic saw primarily the advantages of it, whereas many employees who worked (or had to work) from home during the pandemic adapted to the new circumstances only gradually and suboptimally. A critical factor determining whether employees see primarily the upsides or downsides of WFH are their working conditions at home, including work equipment (e.g., laptop, broadband infrastructure) and spatial situation (e.g., a separate office room at home). If employees use low-quality work equipment at home (e.g., a couple sharing the same computer or a slow internet connection), this is likely to impede the proper execution of work tasks and can trigger stressful reactions (Chong et al., 2020;Wang et al., 2020). With increasing WFH intensity, stress and negative emotions are becoming more likely, leading to more negative spillovers to the home domain (e.g., Vaziri et al., 2020). In contrast, high-quality work equipment at home (e.g., an own laptop, a fast broadband internet connection) facilitates employees' job activities so that they can be performed smoothly and seamlessly, which mitigates negative spillovers. We propose: Hypothesis 3a: The quality of one's work equipment at home moderates the positive effect of WFH intensity on work-to-home conflict, such that the effect is attenuated at high levels of work equipment quality (relative to low levels). A separate office room at home allows individuals to spatially separate from the home sphere, which is a common boundary management strategy (Kreiner et al., 2009;Fonner and Stache, 2012). When employees retreat to this dedicated workspace, a boundary is drawn and interruptions from the home domain are mitigated or completely avoided (Sonnentag et al., 2010;Allen et al., 2021;Bundeskriminalamt, 2021). We hypothesize: Hypothesis 3b: The spatial situation at home moderates the positive effect of WFH intensity on home-to-work conflict, such that the effect is attenuated when employees have a separate home office room (relative to not having one). --- Work from home intensity and positive spillovers More recent developments in the boundary theory field have highlighted that role integration can also facilitate positive spillovers (e.g., Ilies et al., 2009;Kossek et al., 2012;McNall et al., 2015). For example, Ilies et al. (2009, p. 100) note that "by integrating work and family domains employees can magnify the benefits of the positive features of work. " We underline this notion by arguing that WFH intensity can also have positive spillover effects, for two main reasons. Just as WFH intensity increases the probability of interruptions of one's work tasks at home, we argue that it can also increase the frequency of positive resource transmissions between the work and home domains. Some of the interruptions that can cause conflict can actually also provide respite from taxing roles, as suggested by the Frontiers in Psychology 04 frontiersin.org micro-break literature (Bennett et al., 2020) and found by Wu et al. (2021). Micro-breaks can be used to solve pressing issues that would otherwise lead to mental absence (Ashforth et al., 2000;Kossek et al., 2012). In addition, when employees are working from home, they and their loved ones can notice cues of challenging work-related situations and can help each other (e.g., provide emotional support). Giving and receiving care feels good (Feeney and Collins, 2001) and creates positive emotions that can spill over from one domain to the other. WFH intensity can also increase the intensity of positive spillovers. Face-to-face interactions transmit affect and information more immediately and effectively than digital, text-based interactions or phone calls (Judge and Ilies, 2004;Schiffrin et al., 2010). Positive experiences from an enjoyable lunch with one's significant other (Bosch et al., 2018) remain accessible during a work meeting when this takes place in the next room rather than miles and hours away. Similarly, good experiences from work remain fresh and pure when communicated immediately to an accessible partner who can reinforce the positive experience. We hypothesize: Hypothesis 4: WFH intensity increases work-to-home enrichment. Hypothesis 5: WFH intensity increases home-to-work enrichment. 2.4. Boundary conditions of the WFH intensity-positive spillovers relationship Greenhaus and Powell (2006, p. 86) propose that resources and affect emerging from role A are more likely to "promote high performance in Role B when Role B is highly salient than when it is not highly salient. " Again, we follow boundary theory (Ashforth et al., 2000), which takes the opposite perspective: individuals are highly motivated to enact highly central roles while being engaged in other roles, such that enrichment is most likely to happen from role A to role B if role A is highly central to the individual. For instance, if role A (e.g., family) is highly salient, individuals are motivated to create permeable boundaries around role B (e.g., work), such that role A-related permeations into role B are permitted. For example, one might encourage his or her partner to call anytime while being at work. This "enactment effect" (Capitano et al., 2017;Capitano and Greenhaus, 2018), as posited by boundary theory, has received meta-analytic support (Lapierre et al., 2018): work centrality is more strongly related to work-to-family enrichment (<unk> = 0.38, 95% CI [0.24, 0.52]) than to family-to-work enrichment (<unk> = 0.05, 95% CI [-0.18, 0.07]), and family centrality is more strongly related to family-to-work enrichment (<unk> = 0.21 95% CI [0.07, 0.36]) than to work-to-family enrichment (<unk> = 0.07 95% CI [-0.03, 0.16]). The longer individuals work from home, the longer both roles are simultaneously accessible and thus increasingly more possibilities exist to enrich one another. For example, employees with a strong work centrality who work from home may be able to benefit longer from the positive work-related emotions because they are motivated to share such positive experiences with their family members (which is an example of work-to-home enrichment). We hypothesize: Hypothesis 6a: Work centrality moderates the positive effect of WFH intensity on work-to-home enrichment, such that the effect is strengthened at high levels of work centrality (relative to low levels). Hypothesis 6b: Home centrality moderates the positive effect of WFH intensity on home-to-work enrichment, such that the effect is strengthened at high levels of home centrality (relative to low levels). --- The relevance of inter-role spillovers for satisfaction outcomes in the home and work domain Operating through opposing spillover paths, WFH intensity can also influence employees' home-and work-related satisfaction. Relationship satisfaction is an important construct representing the home domain. Romantic relationships were particularly important during the pandemic when individuals could not easily visit friends or relatives and spent (or had to spend) more time with their partners than usual, with both positive and negative consequences. For instance, the German Federal Statistical Office (Statistisches Bundesamt, 2021) reports a slightly reduced divorce rate in 2020 (as compared to 2019), but the Federal Criminal Police Office (Bundeskriminalamt, 2021) registered a sizable increase of cases of domestic violence. Relationship satisfaction is a good indicator of such dynamics, as it is often "the final pathway that leads to relationship breakdown" (Fincham et al., 2018, p. 422). It is also well suited for our purpose because it can be measured independent of one's broader family and household configuration (existence of children, parents, etc.). Relationship satisfaction, an outcome that is not directly related to organizational performance parameters, further allows us to better understand how far-reaching the influence of work policies can be (Kramer and Kramer, 2021). Job satisfaction, representing the work domain, is probably the most studied work-related construct of the social sciences (Spector, 1997;Judge et al., 2001) and has important behavioral and organizational consequences, such as extra role behaviors, in-role performance, and withdrawal (e.g., absenteeism and turnover). How do inter-role spillovers affect relationship satisfaction and job satisfaction? According to the domain specificity model, proposed by Frone et al. (1992), spillover effects originate within the source domain and transpire to the receiving domain, where they materialize. For example, work-to-home conflict (work-tohome enrichment) originates in the work domain (e.g., by work conditions) and manifests in the home domain as decreasing (increasing) relationship satisfaction. An alternative lens, the source attribution model (e.g., Shockley and Singla, 2011), suggests that individuals blame (are grateful to) the source domain, where the spillover effect originates. From this perspective, individuals blame (are grateful to) their job if they experience work-to-home conflict (work-to-home enrichment). Therefore, conflicting (enriching) spillovers can also lead to decreasing (increasing) job satisfaction. We integrate both streams and suggest both positive --- Materials and methods --- Sample and procedure We conducted our study in Munich, Germany. Munich is home to approximately 1.5 million inhabitants (more than 6 million in the metropolitan area), and to the headquarter of seven of the 40 German DAX companies, as well as to a large number of global firms of all sizes.2 As a reaction to the pandemic, 10,000 of jobs were moved to home offices in and around Munich, immediately and directly affecting the lives of many employees and their respective households. We aimed to collect representative data from employees who had a romantic relationship at the time of the survey. To reach the target group, we randomly approached Munich households through a two-step sampling procedure. In the first step, 50 of the 755 Munich voting districts were randomly selected, with the selection probability weighted by the number of inhabitants per district. In the second step, a random geographic point within each selected district was located. Between May 12 and May 15, 2020, research assistants took randomly determined routes from these points, inviting every fifth household to participate in the study by dropping a letter in the mailbox. Within each selected district, 100 letters were distributed, delivering basic information about the study, a unique QR code, and a link to an online survey (T1). Later, on July 20, 2020, we invited participants who had provided an email address to participate in a second wave (T2). In the letter, in reminders, and in the survey itself, we encouraged respondents who lived in a household with family members or cohabitants to share the survey link with these other individuals, specifically with a significant other (if there was one). As an incentive, all participants took part in a raffle for vouchers with a total value of €1,500. Five thousand households were initially approached, and 800 individuals from 580 households completed at least 50% of the first survey (T1), which translates into a household response rate of 11.6%. We excluded individuals who responded to less than 50% of the T1 survey, one individual who responded to the T1 survey twice, and one individual who provided implausible responses. We further restricted the sample to employed individuals who had a romantic relationship. This left us with a final sample of 545 individuals from 375 households in T1, of which 301 individuals from 229 households responded to the repeat survey at T2 (i.e., the individual attrition rate was 44.7%, and the household attrition rate was 38.9%). The following descriptive statistics are based on the T1 sample. 53.21% were female. The average age was 44.04 years (SD = 12.03); on average, the participants worked 22.59 (SD = 18.18) hours per week from home, and the average weekly hours worked, including overtime, was 32.09 (SD = 16.20). 24.49% were supervisors/managers; most employees worked in the service industry (57.06%). On average, the participants had 0.43 (SD = 0.83) children under 18. 56.51% of the participants were married; 84.77% lived together with their partner. See Appendix A for further details on the representativeness of our sample and attrition analyses. --- The context Figure 2 displays the COVID-19 situation in Germany around the data-collection period. The T1 wave of our data was collected about 2 months after the first Bavarian state of emergency. Restrictions were still in place, but the registered COVID-19 cases had already plummeted. The T2 data collection occurred prior to another surge and the second state of emergency. At this time, normality was largely established, contact restrictions were relaxed, and schools, daycare centers and shops were open again. Therefore, our data collections did not occur during extreme situations or periods of volatile adjustments; they rather set in when the work arrangements had already been changed and when individuals had adapted (or were adapting) to them. In this context and at this time, decisions on who would work from home, and when WFH would happen, were made regardless of personal preferences (Chong et al., 2020), thus reducing the selfselection bias inherent in pre-pandemic data and meta-analytic results of WFH. --- Measures --- Work from home intensity Following recommendations (Allen et al., 2015), we asked individuals to report the number of weekly WFH hours at both points in time. At T1, we also asked individuals to report their average weekly WFH hours from before the pandemic. See Appendix B for additional descriptive analyses. --- Negative spillovers We measure work-to-home conflict and home-to-work conflict with two four-item German short scales from the German Family Panel (Pairfam Group, 2020). The scales were originally suggested by Carlson et al. (2000) and later translated into German and validated by Wolff and Höge (2011) like" (work-to-home conflict); and "Conflicts in my personal life reduce my work performance" (home-to-work conflict). Individuals responded to the items on 5-point response scales, ranging from strongly agree (1) to strongly disagree (5). 3 --- Positive spillovers To measure work-to-home enrichment and home-to-work enrichment, we translated and back-translated (Brislin, 1986) the two three-item scales suggested by Kacmar et al. (2014). The translations were performed with the help of two bilingual German native speakers. Disagreements were resolved via discussions between the authors. We replaced "family" from the original scales with "household"/"household member" or "relationship"/"partner, " depending on whether participants indicated that they lived with many others (potentially including their partners) or only with their significant other. Example items are: "The involvement in my household [relationship] puts me in a good mood and this helps me be a better worker" (home-to-work enrichment); and "The involvement in my job makes me feel personally fulfilled and this helps me be a better household member [partner]" (work-tohome enrichment). --- Satisfaction outcomes We measure relationship satisfaction and job satisfaction with the well-accepted single-item measures from the German Socio-Economic Panel study, a prime research resource in the German speaking area (Kantar Public, 2019; "How satisfied are you currently with your relationship?" and "How satisfied are you currently with your job?"). Single-item satisfaction measures often correlate strongly with multi-item measures of the same constructs and can have strong reliability and validity metrics. Research has demonstrated that both single-item measures of relationship satisfaction (Fülöp et al., 2020) and job satisfaction (e.g., Wanous et al., 1997) have adequate psychometric properties, so we are confident in the benefits (low cognitive tax, face validity) of the measures. Both satisfaction measures have ten-point response scales, ranging from very dissatisfied (1) to very satisfied (10). 3 Unless indicated otherwise, all item measures have the same response scale. --- Workplace characteristics at home We measure the two workplace characteristics at home with the following items: "At home, I have optimal equipment [e.g., internet connection, computer] to perform my work tasks" (work equipment); and "At home, I have a room where I am not disturbed when I work" (spatial separation). --- Centrality of the work and home domains We measure work and home centrality using the two two-item scales by Kossek et al. (2012). An example item is "I invest a large part of myself in my work" (work centrality) and "People see me as highly focused on my family/relationship" (home centrality). --- Common method bias To deal with common method variance, we follow the recommendations of Podsakoff et al. (2012). Specifically, we employ different response formats, such as an hour raw count of WFH intensity and varying Likert-type response scales for the spillovers and satisfaction outcomes, which should effectively reduce artifactual inter-construct correlations. In addition, the satisfaction measures were collected at T2, 8 weeks after T1, to create temporal separation. --- Analytical strategy We first use confirmatory factor analysis (CFA) to establish the discriminant validity of the measures. To test hypotheses 1-2 and 4-5, we estimate the direct effects of WFH intensity on inter-role spillovers using structural equation modeling (SEM). To test the conditional effects (hypotheses 3a-3b and 6a-6b), we include the moderators: work equipment and room conditions, and home centrality and work centrality. 4 To test the influence of spillovers on relationship 4 We predict a single boundary condition for each main effect. However, our proposed boundary conditions might moderate the relationships between WFH intensity and other inter-role spillovers as well. We test a complete model (with all possible moderations) and report the results in the Appendix E. Frontiers in Psychology 07 frontiersin.org satisfaction and job satisfaction, we conduct (moderated) multiple mediation analyses. Multiple mediation analysis examines the indirect effects of WFH intensity on satisfaction through inter-role spillovers, and can test whether the spillovers (while controlling for each other) simultaneously mediate the indirect effect (Preacher and Hayes, 2008). All analyses include robust standard errors clustered at the couple level to account for the non-independence of observations where both couple members answered our survey. Following Graham (2009) and Newman (2003Newman (, 2014)), missing values in all inferential analyses are handled with a full information maximum likelihood (FIML) procedure. The (moderated) multiple mediation analyses follow the procedures outlined in Preacher and Hayes (2008) and Preacher et al. (2007). Due to the non-normal distribution of the products of coefficients, we calculate confidence intervals for the indirect effects using a parametric resampling (Monte Carlo) approach (MacKinnon et al., 2004) with 20,000 repetitions. All estimations were conducted in the statistical environment R (v3.6.1; R Core Team, 2019), using the lavaan (v0.6-6; Rosseel, 2012) and semTools (v0.5-2; Jorgensen et al., 2020) packages. --- Control variables To alleviate concerns of omitted variable bias and to investigate alternative explanations, we estimate our models both with and without several control variables. We include as controls: Individual segmentation preferences using the German version of the four-item measure developed by Kreiner (2006) and translated into German by Janke et al. (2014). An example item is: "I do not like work issues creeping into my home life. " Preferences for segmentation (vs. integration) of the work and home domains can cause people to spend less (more) time working from home (Ashforth et al., 2000), and they are probably also related to spillovers (McNall et al., 2015;e.g., Lapierre et al., 2016). Following previous studies of inter-role spillovers and WFH (e.g., Delanoeije et al., 2019), we control for sex (1 = female), individual monthly income after taxes (1 = less than € 1,000; 5 = more than €4,000 EUR; at intervals of 1,000 EUR), number of children younger than 18 living at home, age (in years), marital status (1 = married), and whether the respondent lived together with his or her significant other (1 = living together). We control for the respondents' household composition (relationship/partner vs. household/household member; 1 = household), and for employment characteristics, namely whether the participant was a supervisor (1 = supervisor) and whether he or she was full-time or part-time employed (part-time is defined as working less than 35 h per week) (1 = part-time). We further include two dummy variables to indicate whether the participant was selfemployed (1 = contract: self-employed) or had some "other" type of employment (1 = contract: other), with regular employment as the base category. We use three industry dummies for the primary and secondary industry (reference category), tertiary industry, and the public sector. We also control for pandemic-related task changes: "To what extent did your daily work activities change due to the Corona crisis?" using a 5-point response scale ranging from my activities did not change at all (1) to my activities are completely different now (5), and include a dummy for individuals on a Government sponsored reduced working hours scheme (German "Kurzarbeit") (1 = Kurzarbeit). Finally, we include the number of hours that individuals had worked from home before the pandemic, capturing the individuals' familiarity with WFH. 5 Our results did not qualitatively change when the controls were included. The changes in the standardized coefficients of interest (for example for H1-H2 and H4-H5) were smaller than 0.1, the threshold recommended by Becker (2005). We therefore follow recent advice (Becker, 2005;Carlson and Wu, 2012;Butts et al., 2015) and present our final models without the control variables, but include the models with the full set of control variables in the Appendix (Supplementary Table D.1). --- Results --- Confirmatory factor analysis and correlations The CFA tests whether the four inter-role spillover measures, individual segmentation preferences (a latent control variable), and work centrality and home centrality (two hypothesized moderators) are conceptually different. Because the work-to-home conflict and home-to-work conflict measures have two sub-dimensions each (i.e., stress and time), we modeled the sub-dimensions first and the superordinate work-to-home conflict and home-to-work conflict factors second. The model had a good fit to the data (<unk> 2 [185] = 384.1, comparative fix index [CFI Robust ] = 0.959, root mean square error of approximation [RMSEA Robust ] = 0.046, Tucker-Lewis index [TLI Robust ] = 0.949). Table 1 presents the means, correlations, and standard deviations of all variables, as well as <unk> (omega) reliabilities for all multi-item constructs. All constructs showed acceptable to good reliabilities (<unk> Min-Max = 0.69-0.89). --- Results of hypothesis tests Hypotheses 1-2 and 4-5 suggest that WFH intensity is positively associated with work-to-home conflict, home-to-work conflict, workto-home enrichment, and home-to-work enrichment. In line with Hypotheses 1 and 5, we find a significant and positive effect of WFH intensity on work-to-home conflict (b = 0.008, p <unk> 0.01) and home-towork enrichment (b = 0.005, p <unk> 0.05). Hypotheses 2 and 4 do not receive support, as both coefficients are not significant at conventional levels (Table 2). Next, we tested Hypotheses 3a and 3b, predicting that the work equipment (3a) and room conditions (3b) moderate the positive effect of WFH intensity on work-to-home conflict and home-towork conflict, respectively. All continuous variables were meancentered to avoid bias due to nonessential collinearity between the predictors and the multiplied terms (Dalal and Zickar, 2012). In support of H3a, we find a significant and negative moderation effect of the quality of one's work equipment on work-to-home conflict (b = -0.005, p <unk> 0.05; Figure 3 depicts the simple slopes). Specifically, we find that a 1 h-increase of WFH increases the employees' 5 When this control variable is included, the WFH intensity effect becomes a change in WFH intensity effect (on all outcomes). Bölingen et al. 10.3389/fpsyg.2023.1191657 Frontiers in Psychology 08 frontiersin.org. <unk> reliabilities are in the diagonal (multi-item constructs). To simplify the table, industry (2 dummies), contract-type dummies (part-time dummy, and two dummies indicating self-employment and "other" with employed as base category), and a dummy indicating the point of reference of the WHE and HWE items (household composition) are not included but used as controls in all analyses. 10.3389/fpsyg.2023.1191657 Frontiers in Psychology 09 frontiersin.org work-to-home conflict by 0.017 units (95% CI [0.010, 0.025]) if they have low-quality work equipment (-1SD), whereas the effect is not statistically significant for employees with high-quality work equipment (+1SD) (b = 0.002, 95%

Why do employees experience work from home (WFH) differently? We draw on boundary theory to explain how WFH influences employees' work-home interface. WFH intensity increases negative spillovers (i.e., work-to-home conflict and home-to-work conflict) and positive spillovers (i.e., work-tohome enrichment and home-to-work enrichment) between the work and home domains. Negative spillovers can be mitigated through high-quality work equipment and beneficial spatial conditions at home. Domain centrality predicts who can benefit from increased WFH intensity. We test our theory with a sample of 545 employees, obtained through a two-step random sampling procedure in the city of Munich/Germany during the COVID-19 pandemic. We find that WFH intensity increases work-to-home conflict and home-to-work enrichment, affecting employees' relationship satisfaction and job satisfaction. High-quality work equipment mitigates the detrimental effects of WFH. Employees with a high family centrality can reap benefits of more WFH because they experience more home-to-work enrichment. The simultaneous desirable and detrimental effects of WFH intensity can partly explain why studies have found heterogenous WFH experiences among employees. KEYWORDS work-home interface, job satisfaction, relationship satisfaction, work from home, technology 1 We use the term "work from home" for reasons of readability and consistency but acknowledge and include in our review the general literature on telework and telecommuting, which can be performed in other places than at home (for a review of definitions see Allen et al., 2015).

H increases the employees' 5 When this control variable is included, the WFH intensity effect becomes a change in WFH intensity effect (on all outcomes). Bölingen et al. 10.3389/fpsyg.2023.1191657 Frontiers in Psychology 08 frontiersin.org. <unk> reliabilities are in the diagonal (multi-item constructs). To simplify the table, industry (2 dummies), contract-type dummies (part-time dummy, and two dummies indicating self-employment and "other" with employed as base category), and a dummy indicating the point of reference of the WHE and HWE items (household composition) are not included but used as controls in all analyses. 10.3389/fpsyg.2023.1191657 Frontiers in Psychology 09 frontiersin.org work-to-home conflict by 0.017 units (95% CI [0.010, 0.025]) if they have low-quality work equipment (-1SD), whereas the effect is not statistically significant for employees with high-quality work equipment (+1SD) (b = 0.002, 95% CI [-0.005, 0.009]). We do not find support for Hypothesis 3b, though. Hypotheses 6a and 6b proposed that work centrality and home centrality moderate the positive effect of WFH intensity on work-tohome enrichment and home-to-work enrichment. We do not find a significant moderation of the WFH intensity-work-to-home enrichment relationship and therefore reject H6a. In line with hypothesis 6b, we find that home centrality moderates the positive effect of WFH intensity on work-to-home enrichment (b = 0.006, p <unk> 0.01). Specifically, we find that employees, who report their home to be less central (-1SD) experience little increase in home-to-work enrichment when WFH-hours increase (b = 0.002, 95% CI [-0.005, 0.009]). Employees who report their home to be more central (+1SD), however, can benefit from WFH. On average, their hometo-work enrichment increases by 0.012 units with each additional hour worked from home (95% CI [0.005, 0.019]). Figure 4 plots the effects. The analyses reveal that WFH intensity has adverse and beneficial effects. The positive effect of WFH intensity on work-to-home conflict implies a negative effect of WFH intensity on relationship and job satisfaction, whereas the positive effect on home-to-work enrichment implies a positive effect on both satisfaction outcomes. We use multiple mediation models (Preacher and Hayes, 2008) to examine whether these significant paths exist above and beyond each other. We do not include home-to-work conflict and work-to-home enrichment, because we did not find direct effects of WFH intensity on these inter-role spillovers. 6 Specifically, we reject hypotheses 7b, 7c, 8b and 8c. Due to non-significant a-paths, the indirect effects of WFH 6 We also do not find indirect effects of WFH intensity on relationship satisfaction and job satisfaction via home-to-work conflict (H7b, H8b) and work-to-home enrichment (H7c, H8c) in mediation analyses. The insignificant effects of WFH intensity on home-to-work conflict and work-to-home enrichment are not moderated by the spatial situation (H3b) or work centrality (H6a). intensity on relationship satisfaction and job satisfaction via work-tohome conflict or home-to-work enrichment are statistically non-significant. To test hypotheses 7a, 7d, 8a, and 8d, we estimated a multiple mediation model in which we included relationship satisfaction and job satisfaction as outcomes, and the significant mediation paths from our main analyses, i.e., work-to-home conflict and home-to-work enrichment. Figure 5 presents the estimated paths, and Table 3 reports the indirect effects of the multiple mediation analysis. We find that work-to-home conflict mediates the negative effect of WFH intensity on relationship satisfaction (IND = -0.005, 95% CI [-0.011, -0.001]) and job satisfaction (IND = -0.009, 95% CI [-0.016, -0.003]). An indirect effect tells us "how much two cases that differ by a unit on X are estimated to differ on Y as a result of X's influence on M, which in turn influences Y" (Hayes, 2017, 95). For example, if two employees have zero versus 22.59 WFH hours per week (the sample mean), our model predicts that they will differ by approximately 0.20 job satisfaction units, based only on the mediation through work-to-home conflict and neglecting other potential pathways. This represents about a tenth of a standard deviation of job satisfaction, a small but noteworthy effect. The indirect effects of WFH intensity via home-to-work enrichment on relationship satisfaction (IND = 0.004, 95% CI [0.0004, 0.007]) and job satisfaction (IND = 0.003, 95% CI [0.0001, 0.006]) are positive and statistically significant. Including the previously statistically significant moderators work equipment and home centrality into a moderated multiple mediation model confirms the findings; individuals can alleviate the adverse effects of WFH intensity on satisfaction outcomes through work-tohome conflict by higher-quality work equipment; likewise, individuals who regard their home as more central have weaker negative effects and can even enhance positive effects through increased home-towork enrichment (see Figure 5). --- Discussion Building on boundary theory (Ashforth et al., 2000), this paper argues that WFH intensity results in negative and positive spillovers, i.e., work-to-home conflict, home-to-work conflict, work-to-home enrichment, and home-to-work enrichment. For each effect, we propose a boundary condition. We hypothesize that workspace characteristics attenuate the relationship between WFH intensity and work-to-home conflict and home-to-work conflict. In addition, we propose that higher home centrality and work centrality enhance the positive relationship between WFH intensity and home-to-work enrichment and work-to-home enrichment. We test the relevance of these effects for satisfaction outcomes, namely relationship satisfaction and job satisfaction. Using a two-step random sampling procedure in Munich, Germany, we obtained a diverse sample of 545 employees which we surveyed twice. In line with our hypotheses, we find a significant positive effect of WFH intensity on work-to-home conflict (H1) and home-to-work enrichment (H5). Multiple mediation analyses reveal that these effects exist simultaneously and have negative (via work-tohome conflict, H7a, H8a) and positive (via home-to-work enrichment, H7d, H8d) implications for employees' relationship satisfaction and job satisfaction. Surprisingly, we do not detect a significant effect of WFH intensity on home-to-work conflict (H2) and work-to-home enrichment (H4). We further find that the positive effects of WFH intensity on work-to-home conflict and home-to-work enrichment can be attenuated through high-quality work equipment (H3a) and are stronger for individuals with a high home centrality (H6b), respectively. We contribute to several current discussions in the literature. First, our finding of opposing effects of WFH intensity on the workhome interface confirms that WFH is a work stressor and a workrelated resource (Huyghebaert-Zouaghi et al., 2022). This aligns with boundary theory (Ashforth et al., 2000), which predicts positive and negative spillovers as a consequence of role integration. The heterogeneous consequences of WFH intensity warn against a one-sided focus on WFH. Second, we contribute to the ongoing debate about positive spillovers by analyzing the role of domain centrality as a moderator candidate. Following Greenhaus and Powell (2006) home-to-work enrichment would be more likely to result from WFH intensity when the work domain is central. Enactment-effect logic (Capitano et al., 2017;Capitano and Greenhaus, 2018) takes a different perspective, according to which individuals seek to enact highly salient roles in other roles. Therefore, individuals with high home centrality should experience higher home-to-work enrichment when the WFH intensity increases. Our findings support the latter line of reasoning, and are in line with recent meta-analytic results (Lapierre et al., 2018). Third, we fill a gap and study how WFH intensity relates to relationship satisfaction. Research has mostly neglected the non-work outcomes of work arrangements, taking an organizational perspective (Kramer and Kramer, 2021). However, spillovers from the work to the home domain are increasingly determining how employees experience their jobs and select their employers, making it clear that seemingly non-work outcomes have both individual-level and organizationallevel relevance. Given the negative views on WFH as portrayed in the media (e.g., Dunn, 2020), it is important to understand whether and how WFH affects the home domain. Fourth, we add to current research that examines whether the pre-pandemic WFH literature can be generalized to the new, postpandemic era of WFH (Kniffin et al., 2021). In line with other pandemic-related research (Chong et al., 2020;Leroy et al., 2021), we find a robust and significant positive relationship between WFH and work-to-home conflict, leading to a negative indirect effect of WFH on job satisfaction, whereas pre-pandemic WFH research has found an opposite effect (e.g., Gajendran and Harrison, 2007;Allen et al., 2013). A key difference between pre-and post-pandemic WFH research seems to be that current WFH arrangements do not always increase employees' flexibility (probably due to lower self-selection autonomy); rather, they seem to decrease the employees' boundary control (consistent with Lapierre et al., 2016). Surprisingly, we failed to find a significant effect of WFH on home-to-work conflict, which contradicts much of the public debate (e.g., Dunn, 2020) but aligns with the pre-pandemic literature (Allen et al., 2013). --- Practical implications As companies plan to reduce office space and fixed costs, WFH is here to stay (Aksoy et al., 2023). Practitioners argue that more WFH intensity is not exclusively beneficial for companies, but has also positive implications for employees' work and private life (e.g., Bundeskriminalamt, 2021). Indeed, meta-analyses have found small but positive effects of WFH intensity on the work-home interface and Effects of WFH intensity on work-to-home conflict at lower-and higher-quality levels of work equipment. --- FIGURE 4 Effects of WFH intensity on home-to-work enrichment at lower-and higher-levels of home centrality. Frontiers in Psychology 11 frontiersin.org job satisfaction before the pandemic (Gajendran and Harrison, 2007;Allen et al., 2015). Our findings suggest that this may be true for some employees; more importantly, however they suggest that reality is more complex. On the one hand, higher WFH intensity increases employees' home-to-work enrichment, which correlates positively with job satisfaction and relationship satisfaction. However, increased WFH intensity also triggers work-to-home conflict. Since work-tohome conflict impairs central work and family outcomes, it is necessary (for employers and employees!) to reconsider for whom the positive aspects outweigh the negative aspects, and how to manage employees who work from home. In this challenging situation, our study suggests some promising interventions. Our findings indicate that employees' specific situations matter. Employees who have high family centrality are more likely to reap the benefits of WFH. To minimize the harmful effect of WFH intensity on the work-home interface, organizations can provide employees with improved work equipment. We encourage companies and policymakers to study further key mechanisms that could crowd out the negative and crowd in the positive consequences of WFH. Some companies have already taken steps in this direction by designing software and IT-tools that enable individuals to better manage their work-home boundaries (e.g., Liu, 2020), or by providing employees with a budget for office tools and equipment so that they have what the need to continue working from home (e.g., Kelly, 2021). --- Limitations Our study has limitations. First, our model is static, with predictors and mediators measured at one point in time, and outcomes measured 8 weeks later at another point in time. This data structure is not sufficient to adopt a dynamic view of inter-role spillovers, a call we echo (Allen et al., 2019). Second, we conducted our study at the beginning of the COVID-19 pandemic. Although this setting helped us to control for some empirical issues (e.g., self-selection, reverse causality; e.g., Maxwell and Cole, 2007), we cannot rule out all potential biases. For example, omitted variables may bias the results. For example, employees' resilience and coping skills may affect both WFH intensity during the pandemic (assuming that employees have some autonomy to determine their WFH hours, which is not clear, however) and negative spillovers. Reverse causality can also be an issue. For example, individuals suffering from high workto-home conflict might be less inclined to choose WFH (again assuming that employees have a choice). A quasi-experimental design, with initial measures taken before the pandemic and follow-up measures taken during/after the pandemic, would have been preferred. Third, the contingency view we developed focuses on employeebased moderators, but neglects organization-based and occupationbased contingencies. We echo calls to examine how organizational and occupational characteristics influence the experience of WFH (Kossek and Lautsch, 2018). Finally, generalizability may be an issue. Our study was conducted during the first months of the pandemic in Munich, Germany. The timing of the study allowed us to test the effect of WFH at its peak, but the longitudinal stability of the effects remains unclear. In addition, the inhabitants of Munich can be described by the WEIRD acronym (i.e., they represent a Western, Educated, Industrialized, Rich, and Democratic society; Henrich et al., 2010). More studies are needed to obtain a broader, cross-cultural picture of the role integration demands that the pandemic has triggered. --- Conclusion Collecting data from a diverse sample of 545 working employees who were in a romantic relationship, we found that WFH intensity impacts individuals' work-home interface, opening the door for negative and positive spillovers and ultimately influencing employees' satisfaction with their home and work domains. We find that individuals with high a high home centrality experience higher hometo-work enrichment with increasing WFH, thereby providing sound empirical evidence for the enactment-effect (Capitano et al., 2017;Capitano and Greenhaus, 2018). We further found that the negative effects of WFH intensity through work-to-home conflict could be mitigated through higher-quality work equipment, an actionable insight for both managers and WFH employees. In summary, these findings propose answers to our initially presented research questions: The consequences of WFH are multifaceted with beneficial and adverse effects on employees' workhome interface. In addition, employees' experiences with WFH depend on boundary conditions such as the work environment and personal differences. --- Data availability statement The datasets presented in this article are not readily available because the raw data supporting the conclusions of this article will be made available by the authors upon request. Requests to access the datasets should be directed to boelingen@lmu.de. --- Ethics statement Ethical review and approval was not required for the study on human participants in accordance with the local legislation and institutional requirements. The patients/participants provided their written informed consent to participate in this study. --- Author contributions All authors listed have made a substantial, direct, and intellectual contribution to the work and approved it for publication. --- Conflict of interest The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. --- Publisher's note All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher. --- Supplementary material The Supplementary material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpsyg.2023.1191657/ full#supplementary-material

Why do employees experience work from home (WFH) differently? We draw on boundary theory to explain how WFH influences employees' work-home interface. WFH intensity increases negative spillovers (i.e., work-to-home conflict and home-to-work conflict) and positive spillovers (i.e., work-tohome enrichment and home-to-work enrichment) between the work and home domains. Negative spillovers can be mitigated through high-quality work equipment and beneficial spatial conditions at home. Domain centrality predicts who can benefit from increased WFH intensity. We test our theory with a sample of 545 employees, obtained through a two-step random sampling procedure in the city of Munich/Germany during the COVID-19 pandemic. We find that WFH intensity increases work-to-home conflict and home-to-work enrichment, affecting employees' relationship satisfaction and job satisfaction. High-quality work equipment mitigates the detrimental effects of WFH. Employees with a high family centrality can reap benefits of more WFH because they experience more home-to-work enrichment. The simultaneous desirable and detrimental effects of WFH intensity can partly explain why studies have found heterogenous WFH experiences among employees. KEYWORDS work-home interface, job satisfaction, relationship satisfaction, work from home, technology 1 We use the term "work from home" for reasons of readability and consistency but acknowledge and include in our review the general literature on telework and telecommuting, which can be performed in other places than at home (for a review of definitions see Allen et al., 2015).

Introduction In the past 40 years of reform and opening-up, China has undergone rapid economic development accompanied by massive internal migrants. The results of China's 1% population sample survey show that the number of migrants was 246 million in 2015 [1]. This number is still growing, with the latest census data showing that the number of the migrant population has grown to 375 million in 2020 [2]. Report on China's Migrant Population Development 2018 shows that the percentage of working migrants aged 16-59 reached 84% in 2015, and they played an important role in increasing urban labour force participation rate and promoting sustained urban economic growth [3]. However, with the ageing of earlier migrants and the characterized family migration, the number of older migrants continues to grow. The number of elderly migrants in China increased from 5.03 million in 2000 to 13.04 million in 2015 [4]. They are mainly composed of four groups, namely migrant workers, disabled migrants, healthy retired migrants and migrants following their children [3]. The interaction between health and migration is complex and dynamic, and migration can have an impact on the physical and mental well-being of migrants [5]. Lower healthcare utilization for disease prevention such as screening services and oral health check-ups, among migrants compared to nonmigrants was found in Europe, suggesting inequitable access to preventive health services for migrants [6,7]. Such results were observed in both developed countries and developing countries in which migrants have fewer medical treatment options and uncertainty about healthcare opportunities and health outcomes. Rural-urban migrant children were less likely to be fully immunized than urban nonmigrants and the general population in China, India and Nigeria [8]. Migrant workers face difficulty accessing medical assistance or delay seeking healthcare during illnesses in Malaysia and Bangalore city [9,10]. This problem may be more common among older migrants who face physical, psychological, and economic cost barriers caused by spatial transformation, accompanied by the characteristics of increased health risks as they age, low education levels, low income, and heavy burdens of medical expenses. The usage of health services is currently limited, although these older migrants have high health needs [11,12]. Migration includes the types of international and internal migration, where international migrants are defined as "individuals who remain outside their usual country of residence for at least one year" around the world, facing more common cultural and linguistical barriers, while internal migrants are defined as "individuals move within the borders of a country, usually measured across the regional, district, or municipal boundaries, resulting in a change of usual place of residence" [13]. Due to the unique household registration system, internal migration is more common in China. Therefore, the participants are the internal migrants in this study. According to the hukou-China's household registration system-the population was divided into household registered and non-household registered based on whether the actual place of residence coincides with the place of hukou registration. For ease of understanding, they are referred to below as nonmigrants (residents with hukou) and migrants, respectively. This household registration system has been reported to be a major barrier preventing internal migrants from enjoying equal rights, such as access to housing, education, social and medical services, to nonmigrants in the areas where they live [14]. Studies from China have reported on the obstacles of the healthcare-seeking behaviours among these migrants. Employed migrants had the lowest recent physician consultation rate and the lowest annual hospitalization rate, which might be related to 'healthy worker effect' [15], and migrants were more likely to utilize private rather than public services for general healthcare and delivery care [16]. Currently, studies mainly focus on employees or maternal and child populations, and most of them are studies on the utilization of health-care services [17]. Considering the important role of physical examination in increased chronic disease recognition and treatment, risk factor control, preventive service uptake, and improved patient-reported outcomes [18], it is necessary to analyse the annual physical examination of migrants. Besides, Guo reported a low rate of medical behaviour among older Chinese migrants [19], but whether there is a difference in annual physical examination between older migrants and nonmigrants in China remains unknown. Although researchers have analysed the specific differences between migrants and locals in seeking healthcare services when illnesses occur in a particular area [20,21], evidence of annual physical examination among older migrants still lacks support from a wider range of data. This study aims to explore how different the annual physical examination of older migrants than that of older non-migrants in China is by using a coarsened exact matching method, and to provide some policy implications for improving older migrants' health by exploring what factors are associated with annual physical examination of them in China. [22]. The CMDS conducted a special survey for older migrants only in 2015 and collected data on 13,043 older migrants (aged 60 and above) from 9242 households. The CHARLS collected data on 20,284 respondents from 11,797 households in 2015, from which we obtained 7385 registered households over 60 years old that were nonmigrants. Detailed descriptions of the databases are available on the official websites https://chinaldrk.org. cn/wjw/#/data/classify/population/topicList (accessed on 20 August 2021) and https: //charls.charlsdata.com/pages/Data/2015-charls-wave4/en.html (accessed on 20 August 2021). As our study used secondary data in the public domain, approval from institutional review boards (IRBs) was not required. After handling missing values, the sample size in the study was 17,242, including 11,622 older migrants and 5620 older nonmigrants. --- Materials and Methods --- Data --- Measures Annual physical examination was measured by the question "Have you had a physical examination in the past year?" for older migrants in CMDS 2015. For older nonmigrants in CHARLS 2015, whether they had a physical examination was based on the response to "When did you have your last physical examination?" Only respondents who reported a time within the past 12 months were defined as having physical examination in the past year. Older migrants were defined by three conditions: (1) people aged 60 years old and above, (2) people who remained outside their usual hukou location for at least one month and (3) people without hukou for their current living place. In contrast, older nonmigrants were those aged 60 and above who lived in the location of their hukou. The covariates included in the study were age (60-69,70-79,80-89, and 90 and above; an ordinal variable); gender (male and female; a nominal variable); marital status (married and others; a nominal variable); educational level (primary school and no formal schooling, middle school, and high school and above; an ordinal variable); employment (yes and no; a nominal variable); hukou (urban and rural; a nominal variable); household economic status (in quintiles: the poorest from 2 to 500 yuan, poorer from 500.4 to 750 yuan, medium from 750.2 to 1000 yuan, richer from 1001.6 to 1500 yuan, and the richest from 1502.2 to 60,039.7 yuan; an ordinal variable), which was measured by the per capita monthly expenditures of the family; health status (good and poor; a nominal variable) based on self-evaluated health; and health insurance (yes and no; a nominal variable). In addition, to explore the factors associated with annual physical examination among older Chinese migrants, a set of variables was included. The main source of income is divided into self-derived (including labour income, savings, pension, living security, and rent), family members and others (self-derived, family members, and other; a nominal variable), type of migration (rural to urban, urban to urban, rural to rural, and urban to rural; a nominal variable), range of migration (interprovincial, intraprovincial but intercity, and intracity but intercounty; a nominal variable), years of migration (less than 3 years, 3-9 years, and 10 years and above; an ordinal variable), willing to stay in current location (yes, no, and unsure; a nominal variable), having health records in local community (yes, no, and unsure; a nominal variable), and the number of local friends (none, 1-4, 5-14, and 15 and above; an ordinal variable). --- Statistical Analysis Categorical variables were described using rates and examined by the chi-squared test. Considering potential differences between groups from two different sampling databases, coarsened exact matching (CEM) was used to reduce imbalances in the estimation of the differences in annual physical examination between older migrants and nonmigrants. CEM is a Monotonic Imbalance Constraint (MIB) class of matching method that reduces the number of potential matches for a given covariate by a set of "coarsened" potential confounders that are matched simultaneously in order to increase the number of matches achieved [23]. It is preferable to other matching procedures in terms of reducing model dependence, and estimation error [24]. It has been shown to produce good covariate balance between groups and reduce the impact of confounding in observational causal inference [25]. To control for some potentially confounding influences of pretreatment variables, the covariates on which a match is made are first stratified (divided into discrete categories), and then the exact matching algorithm is used to match the objects in each layer according to the empirical distribution of the sample. Finally, the matched research objects were retained, and each older migrant (treatment observations) was matched to an older nonmigrant who had the same age, gender, and other covariates. In the matched data, all analyses were conducted by incorporating matched weights. After matching, the balance between covariates is evaluated by the L 1 statistic, which is based on the difference of all pretreatment covariates in the treated group and those in the control group, ranging from 0 to 1. A good match would result in a lower L 1. Details of this monotonic imbalance-reducing matching method can be found in other literature [26]. A logistic regression model was used to analyse the factors affecting older migrants' annual physical examination in the unmatched group. Odds ratios (ORs) and 95% confidence intervals (95% CIs) were reported in the regression results. The regression results were two-tailed with a significance level of 0.05 (p <unk> 0.05). All statistical analyses were performed using STATA 14.0. --- Results --- The Characteristics of the Participants Table 1 shows the characteristics of the sample. The sample was mostly under 70 years old (72.8%), male (51.6%), possessed primary school and no formal schooling education (66.5%), rural (72.4%), married (81.4%), employed (51.6%), low-and middle-income (64.8%), possessed good health status (82.5%), and possessed health insurance (91.7%). There were significant differences between older migrants and nonmigrants in age, gender, education, hukou, employment, household economic status, and health status (all p <unk> 0.05). Compared with older nonmigrants, older migrants were more concentrated as under 70 years old (75.9% vs. 66.3%, p <unk> 0.001), male (52.2% vs. 50.5%, p = 0.036), middle school and above education (40.4% vs. 19.1%, p <unk> 0.001), urban (33.0% vs. 16.3%, p <unk> 0.001), unemployed (49.5% vs. 46.1%, p <unk> 0.001), possessing poorer to richer household economic status (66.4% vs. 44.1%, p <unk> 0.001), and possessing good health status (88.9% vs. 69.3%, p <unk> 0.001). As shown in Table 1, there was no statistically significant difference between older migrants and nonmigrants in annual physical examination (35.6% vs. 35.7%, p = 0.90). --- Difference in Annual Physical Examination between Older Migrants and Nonmigrants Table 2 shows the difference in annual physical examination between older migrants and nonmigrants. A total of 4139 (35.6%) older migrants and 2007 (35.7%) older nonmigrants reported having had physical examinations. Migration was not associated with higher annual physical examination in the unadjusted analysis (OR = 0.99, 95% CI: 0.93 to 1.06). Using a logistic regression model after coarsened exact matching, 10,347 older migrants and 5341 older nonmigrants were matched, and the L 1 decreased from 0.431 to nearly zero (6.267 <unk> 10 -15 ). We identified that older migrants had less annual physical examination than older nonmigrants (OR = 0.91, 95% CI: 0.85 to 0.98). This persisted in the multivariable logistic regression model (OR = 0.92, 95% CI: 0.86 to 0.99). The multivariate imbalance measure L 1 statistic before and after CEM is reported in Table 3. Notes: the mean is labelled in parentheses and reports the difference in means. --- Factors Associated with Annual Physical Examination Table 4 shows the results of the three multivariate regression models. The first model incorporated the demographic and socioeconomic characteristics of the older migrants, the second model incorporated the migration and social integration characteristics of the older migrants, and Model III included all characteristics. In Model I, middle school education was related to higher use of annual physical examination (OR = 1.196, p <unk> 0.01), but it was lower for those with a high school and above education (OR = 0.789, p <unk> 0.01). Higher annual physical examination was associated with urban migrants (OR = 1.274, p <unk> 0.01), good self-rated health (OR = 1.392, p <unk> 0.01), and health insurance (OR = 1.467, p <unk> 0.01). For employed individuals (OR = 0.786, p <unk> 0.01) with richer and richest household economic status (both p <unk> 0.05), annual physical examination was lower. Compared with older migrants whose main source of income was themselves, those whose incomes were mainly derived from family members and others presented lower annual physical examination (both p <unk> 0.05). In Model II, annual physical examination was higher among urban-urban migrants and lower among rural-rural migrants compared with rural-urban migrants (urban to urban: OR = 1.163, p <unk> 0.05; rural to rural: OR = 0.750, p <unk> 0.01). Older migrants with a closer migration range have a higher utilization (all p <unk> 0.01), and those whose migrations have lasted 3-9 years have higher utilization (OR = 1.136, p <unk> 0.01). Having health records in local community was significantly positively associated with annual physical examination (OR = 4.375, p <unk> 0.01) and having more local friends (all p <unk> 0.01). The results of model III were similar to those of Model I and Model II, with only one exception: there was no significant difference in annual physical examination between older migrants whose income was mainly derived from others and whose income was mainly derived from themselves. --- Discussion We conducted secondary analysis using large and nationally representative datasets to explore the differences in annual physical examination between older migrants and older nonmigrants, and to investigate the factors associated with annual physical examination among older migrants in China. We aim to show the gap in their utilization and to provide a reference for realizing the health equity of older migrants. The study found that the annual physical examination rate of older migrants was 35.6%, which is slightly lower than the report from Guo (39.64%) [19]; and the rate of older nonmigrants was 35.7%. There was no difference in the reported annual physical examination rates between older migrants and nonmigrants. However, when confounding factors were taken into account, the results of both multivariable logistic regression and CEM suggested a lower annual physical examination rate among older migrants. Chinese researchers reported the "healthy migrant effect" and convergence of health status during China's internal migration process, that is migrants tend to have better health than nativeborn residents initially, but this physical health advantage diminished significantly with their increasing length of residence [14]. Due to poor living conditions, insufficient health knowledge, and a lack of social support and social integration [27], the migrating population is susceptible to various psychological and physical health problems. In addition, migrants are often at a disadvantage as outsiders [28]. Not only are they generally disadvantaged to residents in terms of social resources and human capital, but they also face inadaptability in social culture and customs [29]. The above factors have caused a higher need for health services and more difficulties in the decision-making process of health-related issues. Due to China's household registration system, studies have shown that access to public health services outside the place of hukou is often restricted [30]. In response to this problem, the social welfare reform in China has taken measures such as encouraging qualified migrants to apply for local hukou, but it is still questionable whether older migrants with low education meet the restrictions on settlement clauses. In addition, some policies aim to expand social welfare, such as free physical examinations for all older people in a community [31], but whether they effectively improve the annual physical examination for older migrants needs to be verified by the latest data. The results of Model I showed that education, employment, and household economics significantly affected the use of annual physical examinations for older migrants. This is consistent with other studies that showed that socioeconomic status, such as income and education, is an important factor influencing healthcare service utilization [32]. Surprisingly, we observed that although the elderly with middle school education had higher annual physical examination than those with primary school and no formal schooling education, for those with high school and above education, annual physical examination was reduced. Other Chinese researchers provide similar findings [33]. We consider the reason for this difference to be explained by the mediating role of health literacy in the relationship between education and health [34]. The elderly with higher health literacy were significantly less likely to have risky behaviours, and more likely to undergo health examinations regularly, report good self-rated health, and access sufficient health information from multiple sources [35], but health literacy is not just the result of formal education [36]. Findings from a comparative study reported that literacy skills in Bermuda better capture health-related knowledge/skills/behaviours than formal education [37]. This study found that higher household economic status was negatively correlated with annual physical examination. The main reason is that we used expenditures to measure household economics. Compared with disease treatment, physical examination is generally considered to be cheap, which partly explains the negative association between family expenditures and physical examination. From the perspective of the main source of income of older migrants, annual physical examination was lower among those with income coming from family members and others, which can reflect that income is an important factor in the use of healthcare services in China [38]. The relationship between employment and annual physical examination showed that employed older migrants had lower utilization. This could be attributed to the lack of time for employed migrants, which is also confirmed in other Chinese studies as an important factor preventing migrant workers from choosing medical treatment when getting sick [39]. Older migrants with urban hukou were associated with higher annual physical examinations compared to the rural group. The potential residential gap occurred in the case of preventive care utilization [40], in health knowledge and in the utilization of health services in China [16]. This gap could be an example of how rural migrants are disadvantaged in their access to healthcare. The annual physical examination of older migrants with good health was significantly higher than that of migrants with poor health, which is closely related to self-care awareness. In addition, health insurance participation has stimulated awareness of annual physical examination among older migrants, which contrasts with previous Chinese research [40]. Although the existing social health insurance does not include preventive health services such as physical examination, tumour screening, or vision or hearing examination, it addresses the health service needs of the insured by providing medical expense compensation when they visit a doctor, thereby promoting insured older migrants to adopt active health strategies. Based on the above analysis of demographic and socioeconomic factors, we propose formulating plans to improve health literacy and to improve individuals' positive awareness of physical examination based on health education and cultural interventions. In addition, it is also necessary to formulate health policies to protect older migrants, especially to ensure that migrant workers enjoy the right to physical examinations for employees and to promote the realization of social equity by exploring various means, such as increasing health insurance benefits and providing transfer payments to older people without fixed incomes. In Model II, we included features of migration and social integration. Consistent with our discussion above, migrants who came from rural places were more inclined to use less annual physical examination, which was reflected in the higher utilization of urban-urban migrants than rural-urban migrants. In addition, regarding destination, for older migrants with rural hukou, utilization is lower in those who migrated to rural areas than for those who migrated to urban areas, suggesting that there may be differences in healthcare access between urban and rural areas in China [41]. This observation is also shown in the utilization of urban-rural migrants, although no statistical significance was found. Compared with interprovincial migrants, those with closer distances showed higher utilization. This is consistent with the results of other Chinese studies [42], as the latter are only influenced by regional segmentation with fewer sociocultural differences [16]. Older migrants with a longer duration were more likely to use annual physical examination, as migrants' social integration improves along with their years of residence in their destinations [43] and provides them with a better understanding of local medical service facilities [44]. The utilization of migrants with health records in the local community was significantly high, indicating the importance of social support for older migrants. Consistent with other Chinese studies [45], we found that migrants with more local friends were more likely to use annual physical examinations than migrants without local friends. Local friends could help migrants rebuild relatively rich social networks [44] and provide local and useful health information [46], which are important for older migrants to use healthcare services [46,47]. In Model III, we included all features of older migrants. The OR value with the results of factors were similar to the Model I and Model II when the additional factors were included. These results have important policy implications for the healthcare system and local governments to promote social equity. Policy-makers need to consider how to formulate targeted social and health policies and establish a more complete rural health delivery network to increase accessibility. Better information on services should also be provided to migrants, thereby improving their overall healthcare access and opportunities for health services. In addition, in the current primary healthcare reform, more resources and attention need to be placed on the migrant population, especially those who lag far behind their local peers in the use of services [48], which should not only help them establish health records in the local community to ensure that they can enjoy the same health policy as the locals, but also promote social integration by giving them spiritual care and organized community activities. There are some limitations to this study. First, the data for our study came from 2015. This is because the CMDS conducted a special survey for older migrants only in 2015. To the best of our knowledge, this is the largest and most representative survey data on older migrants in China. Other than the CMDS 2015, the CMDS does not currently release updated data on the older migrants. To be consistent with CMDS, CHARLS 2015 was used for older nonmigrants in the study. Thus, the updated large-scale survey is needed to this issue. Second, our data were sourced from two different databases, although the matching method was used to reduce the variation in the sample, it may still lead to biased results. Besides, like other cross-sectional studies, we could not determine the cause-and-effect relationship. In addition, measures of annual physical examination were self-reported and could suffer from recall bias. To capture the use of older nonmigrants, the judgement of annual physical examination used the time of their last physical examination. Some older migrants may be excluded because of uncertain reporting times, which may lead to underestimation. Moreover, due to the availability of data, some factors that may affect annual physical examination for older migrants are not included, such as family income, which was measured by family expenditures in the analysis. Other factors, including the role of the family living mode and activities of daily living, need to be further explored once the data are available. --- Conclusions The annual physical examination needs of older migrants in China have not been fully met. Older migrants are likely to be unable to enjoy the same annual physical examination as older nonmigrants. This requires the government and public departments to adopt active countermeasures to improve the fairness of the use of annual physical examination for older migrants so as to avoid leaving this vulnerable group behind. --- Data Availability Statement: The data in this study are available from China Migrants Dynamic Survey 2015, which were used under license in this study, and China Health and Retirement Longitudinal Survey 2015, which could apply from institution on the website of https://charls.charlsdata. com/pages/Data/2015-charls-wave4/en.html (accessed on 20 August 2021). Data of China Migrant Dynamic Monitoring Survey 2015 are available from the authors upon reasonable request and with permission of the Migrant Population Service Center of the Chinese National Health Commission on the website of http://www.chinaldrk.org.cn/wjw/#/application/index (accessed on 20 August 2021). --- Conflicts of Interest: The authors declare no conflict of interest.

It has become a top priority to ensure equal rights for older migrants in China. This study aims to explore how different the annual physical examination of older migrants is compared to that of older nonmigrants in China by using a coarsened exact matching method, and to explore the factors affecting annual physical examination among older migrants in China. Data were drawn from the China Migrants Dynamic Survey 2015 and China Health and Retirement Longitudinal Survey 2015. The coarsened exact matching method was used to analyse the difference in the annual physical examination of older migrants and nonmigrants. A logistic regression was used to analyse the factors affecting annual physical examination among older migrants. The annual physical examination of older migrants was 35.6%, which was significantly lower than that of older nonmigrants after matching (Odds ratios = 0.91, p < 0.05). It was affected by education, employment, hukou, household economic status, health, health insurance, main source of income, type of migration, range of migration, years of migration, having health records in local community and number of local friends among older migrants in China. Older migrants adopted negative strategies in annual physical examination compared to older nonmigrants. Active strategies should be made to improve the equity of annual physical examination for older migrants in China.

Introduction Food insecurity and intimate partner violence (IPV) are major health issues in the United States (U.S.) [1,2]. Food insecurity, experienced by an estimated one in nine households in 2018 [2], refers to having limited or uncertain availability of adequate food [3]. Black, Latinx, low-income, and female-headed households are among those that bear the greatest burden of food insecurity [2]. Women living with HIV (WLHIV) are also significantly more affected, with approximately 40% estimated to be food insecure in the U.S. [4]. Lifetime IPV prevalence-36% for women overall [5]-is disproportionately higher for women of color and WLHIV [1,6]. IPV can include physical (e.g., beating, choking), sexual (e.g., rape, sexual coercion) or emotional violence (e.g., stalking, threats of violence) perpetrated by a current or former intimate partner [7]. High rates of food insecurity and IPV, particularly among HIV-affected populations in the U.S., have significant public health implications. Both food insecurity and violence are wellestablished determinants of HIV risk behaviors such as condomless sex, transactional sex, and substance use, [8][9][10][11] as well as HIV-related morbidity and mortality [12][13][14][15][16]. For example, food insecure individuals may increase their risk of HIV exposure by engaging in high-risk sex, such as condomless or transactional sex, to access food [10]. IPV in the form of coerced or forced condomless sex can also directly increase transmission risk [17]. Food insecurity can lead to poor HIV control by undermining antiretroviral therapy (ART) adherence if people living with HIV (PLHIV) avoid medication to prevent adverse effects from taking ART on an empty stomach (e.g., nausea, vomiting) [14]. Additionally, PLHIV may need to make tradeoffs between spending limited resources on food instead of ART or transportation to clinic appointments, negatively effecting disease management [14,18]. Reduced engagement in HIV care and treatment can likewise arise from fear of or experiences with IPV. Women who anticipate a violent reaction from their partner upon learning of their HIV status are less likely to link to and remain in HIV care and treatment [19][20][21][22][23]. WLHIV who experience violence are also less likely to adhere to ART [19,21,[24][25][26][27][28][29][30][31][32][33][34] and achieve viral suppression [25,[35][36][37]. While a large body of research has demonstrated that food insecurity and IPV are independently associated with worse HIV outcomes, few studies have explored how intersecting experiences of food insecurity and IPV shape HIV risk behaviors and treatment adherence. Past research has found that food insecurity and IPV are associated [38][39][40][41][42], with some evidence of a bi-directional relationship [43,44]. For example, in a longitudinal study among women in the U.S., Conroy and colleagues found that past and current experiences of food insecurity predicted current experiences of sexual or physical and emotional violence [43]. Another longitudinal study of mothers in the U.S. found that experiences of IPV predicted an increased risk of household food insecurity two years later. [44] A small but growing body of evidence suggests that food insecurity and IPV may be linked through pathways of poor mental health and relationship stress and conflict [44][45][46][47]. Specifically, research suggests that depression executive committee. This is a requirement of cohort IRB approvals ensuring secure, timely, and ethical sharing of the cohort's data. Data access is managed by the MACS/WIHS Combined Cohort Study (MWCCS), and data requests can be made by contacting aMWCCS@jhu.edu. mediates the relationship between food insecurity and IPV [44], and that food insecurity may increase IPV through elevated household stress and related conflict [44][45][46][47]. However, limited research has explored the unique ways in which food insecurity and IPV intersect in the context of HIV, or the mechanisms through which food insecurity and IPV may shape HIV risk and treatment behaviors. Such research is critical given that WLHIV are disproportionately affected by both food insecurity and IPV, and each have been shown to be independently associated with HIV risk behaviors and suboptimal engagement in HIV treatment. It also aligns with recent calls for syndemic approaches to understanding the intersecting experiences and cumulative effects of food insecurity, mental health and domestic violence [48]. Accordingly, we conducted a qualitative study to explore how food insecurity and violence intersect and shape HIV risk and treatment behaviors among women living with and at risk for HIV in the U.S. --- Theoretical framing This study draws upon Singer's syndemic theory, which describes the co-occurrence of two or more disease clusters at the population-level, whose interaction exacerbates the negative health impacts of each disease involved [49]. Syndemics can involve health conditions of all types, including infectious and non-communicable diseases and mental health conditions [50]. Importantly, this framework acknowledges that the clustering of such epidemics arise from harmful social and structural contexts, such as poverty, racism, and stigma, which contribute to health disparities [49,50]. The most widely studied syndemic is the substance abuse, violence and AIDS (SAVA) syndemic, in which substance abuse, violence and HIV/AIDS cooccur and mutually reinforce each other in a manner that magnifies worse health outcomes [35,49]. Food insecurity remains an understudied element of syndemic theory [51], despite the critical role it plays in the lives of marginalized and low-income populations, including WLHIV. It is important to note that although food insecurity and socio-economic status are related, food insecurity is a distinct construct that warrants its own consideration and examination through a syndemic lens. Indeed, not all low-income individuals in the U.S. experience food insecurity [52]. Furthermore, studies have found that individuals living in wealthier households can experience food insecurity in the context of unequal household food allocation, drought, or poor access to healthful foods in their neighborhood [53][54][55]. Food insecurity has also been found to be associated with poor health outcomes above and beyond the effects of socioeconomic status [56]. Given the evidence, outlined above, that food insecurity and IPV are associated with each other and independently associated with poor HIV outcomes, it is possible that they could be syndemic factors working together to worsen HIV-related health. As such, we qualitatively explored the circumstances in which food insecurity and IPV cooccur and the mechanisms through which their convergence influences HIV risk and treatment behaviors among women living with and at risk for HIV in the U.S. --- Methods --- Research design and setting We conducted a qualitative study nested within the Women's Interagency HIV Study (WIHS), now part of the Multicenter AIDS Cohort Study-WIHS Combined Cohort Study (MWCCS). WIHS was a multicenter prospective cohort study of women living with HIV and demographically similar HIV-negative controls with a history of sexually transmitted infections or behavioral or demographic characteristics that increased their risk of acquiring HIV, in nine sites across the U.S. [57]. Survey, clinical, and biological data were collected semi-annually from WIHS participants. In this qualitative study, we aimed to explore how experiences of IPV and food insecurity intersect and influence HIV risk and treatment behaviors among women enrolled in the San Francisco, California and Atlanta, Georgia WIHS sites. Women at the San Francisco site lived in the city itself and the urban and suburban settings of the greater Bay Area. Women at the Atlanta site lived in the city and in the surrounding suburban and rural areas. We refer to these sites as Northern California and Georgia to account for these geographic differences within each site. These two sites were chosen due to their geographic diversity and differences in social welfare provision; despite high poverty rates in both locations, California has above average per capita welfare spending, while Georgia has one of the lowest per capita welfare spending in the country [58]. The large number of WIHS women experiencing both food insecurity and IPV at these sites also made our ability to recruit the sample more feasible. --- Sampling and recruitment We recruited women from the Northern California and Georgia WIHS sites from January through March 2019 who met our inclusion criteria of reporting: 1) physical, sexual or emotional violence from an intimate partner in the past year at their most recent WIHS visit; and 2) marginal, low or very low food security, as defined by the Household Food Security Survey Module [3,59], at or before their most recent WIHS visit. Participants were also purposively sampled so that half were living with HIV. This approach allowed us to explore whether experiences of IPV and food insecurity differed by HIV status and how their intersection may influence HIV risk behaviors as well as engagement in HIV care and treatment. WIHS staff from each site contacted eligible cohort members over the phone to assess their interest in participating in the study. If interested, the study interviewer contacted them to schedule an interview. --- Data collection A semi-structured interview guide was developed by the first author, with input from SDW, AAC and PCT. The guide included broad, open-ended questions that asked about experiences with violence and food insecurity and perspectives on how such experiences influence HIV risk and treatment behaviors. For example, to explore experiences of food insecurity, participants were asked to describe what they ate on a daily basis, their perception of the quality of food they ate, and any challenges they faced in accessing the amounts and types of food they desired. The guide also included questions asking about the circumstances under which women experienced physical, sexual and emotional violence from partners. The guide included probes to explore the ways food insecurity and IPV might overlap, such as experiencing violence in the context of arguments over lack of food or feeling unable to leave an abusive relationship due to reliance on one's partner for food. Finally, the guide included questions designed to explore how experiences with food insecurity and violence influenced women's emotional well-being and their ability to manage their health including using condoms, attending clinic appointments and taking medication, including ART. Participants were also asked about how experiences of food insecurity and violence might influence their engagement in other HIV risk behaviors including condomless sex and transactional sex. Questions in the guide were informed by a literature review and our prior research on these topics [13,45,54,60,61]. A trained female Master's-level researcher with qualitative interviewing experience conducted the interviews in English in a private office at the local WIHS site. Interviews lasted between 60-120 minutes each and were audio-recorded and transcribed verbatim. Participants were assigned unique identification numbers and compensated $55 at the end of each interview for their time. This study received human subjects research approval from the Institutional Review Board of the University of California San Francisco (UCSF) (site lead for the Northern California WIHS), and by the WIHS Executive Committee. Emory University (site lead for the Georgia WIHS) waived approval because the Georgia data was collected by UCSF researchers. All participants provided written informed consent before starting the interview. --- Data analysis Data were analyzed using thematic content analysis methodology [62] following an inductivedeductive approach [63]. An initial codebook was developed by AML and JS, based on a priori codes informed by prior research. The codebook was iteratively revised during the analytic process by adding new inductive codes to reflect emergent themes identified during transcript review. The researchers met regularly to discuss emergent themes and codes and their application until they reached consensus around a final codebook. JS coded the transcripts, using memos to elaborate upon the codes and their application. AML reviewed the codes of approximately one-quarter of the transcripts, and JS and AML discussed disagreements in coding and came to a consensus. JS single coded the remainder of the transcripts. Once all transcripts were coded, AML and JMZ analyzed the data. Coded segments related to our research question, describing how violence and food insecurity intersect and influence HIV risk and treatment behaviors were iteratively reviewed. Key themes were developed and refined with descriptions of common and divergent experiences and perspectives and illustrative quotes representing both common and divergent perspectives. The key themes were reviewed by authors, with discrepancies resolved through discussion and consensus. These results are summarized below, with quotes anonymously identified by pseudonym, site, HIV status, race/ethnicity, and age. --- Results We interviewed 12 participants from the Northern California site (n = 7 HIV-positive; n = 5 HIV-negative) and 12 participants from the Georgia site (n = 5 HIV-positive; n = 7 HIV-negative). All women were cisgender, and the majority reported very low food security, meaning that they experienced disruptions in their eating patterns or reduced food intake because they lacked the resources to access food [64] (Table 1). The most common forms of violence were physical violence, emotional violence, and concurrent experiences of physical and emotional violence (Fig 1). Nearly half the women described experiences of child abuse (n = 10). The majority of participants reported relying on social safety net programs such as Supplemental Security Income, Social Security Disability, and Supplemental Nutrition Assistance Program (SNAP) (i.e., food stamps) to access income and food. All participants described these services as insufficient to meet their daily needs. Half of the sample were living with HIV, and most participants revealed that they had other chronic conditions such as diabetes mellitus and hypertension. A dominant theme that emerged at both sites was one of co-occurring and mutually reinforcing experiences of violence, food insecurity, poor mental health, and substance use. Specifically, most participants described how intersecting experiences of violence and food insecurity caused or exacerbated poor mental health outcomes and spurred their use of illicit substances as a coping strategy. At the same time, poor mental health and substance use amplified risks for violence and food insecurity. The converging experiences of violence, food insecurity, poor mental health and substance use were described to increase HIV risk behaviors and undermine optimal health management, including HIV care and treatment among WLHIV. Findings did not differ by site. --- Intersecting experiences of violence and food insecurity/material deprivation Many women described how, in the context of food insecurity and other material need insecurities (e.g., housing, income), they often needed to rely on intimate partners or strangers to meet their daily living needs. Women noted how this reliance created unequal power dynamics that left them vulnerable to abuse. Some women said men coerced them to engage in unwanted sexual acts in exchange for food, both in the context of intimate partnerships and sex work. We did have conflict over food sometimes because he would bring food into my house and then if I don't have sex with him or something he would want to take it back. So, that happened several times that he would get mad and say oh, I'm taking my food. (Harriet, 40 years old, African American, HIV-negative, Georgia) Women who depended on intimate partners to meet their material needs also described feeling unable to leave abusive relationships, particularly if they wanted to ensure consistent access to resources for their children. We'd get just in a bad situation where we think the guy's going to help us out and help us with food and our children and stuff.... And instead it turns into an abusive relationshipsome women just...let it happen because these day and time it's hard to...get a meal. (Doris, 55 years old, African American, HIV-positive, Northern California) Some women revealed that their experiences with food insecurity began after leaving an abusive partner whom they relied on to access food. For example, Barbara, a 53-year-old HIVnegative white woman from Northern California described how leaving her abusive partner after a particularly violent encounter sent her down a path of food insecurity and homelessness: "I never went back after that. So that's what started my life like being really homeless and hungry and everything". Another theme was stress over lack of food as a cause of conflict and violence in relationships. For example, several participants described conflict arising when their partners would eat most or all the food available for the week, leaving nothing for the rest of the family. It kind of pissed me off 'cause the way I had to calculate everything... it's enough just for me and my kids. Then when my baby daddy back in the day was staying with me...he'll try to eat up everything... But he didn't buy the food. And then if he get mad at me and shit, he gonna take my shit out the freezer and shit and sell it, and he ain't thinking about the kids. (Candice, 52 years old, African American, HIV-positive, Northern California) Candice revealed that this behavior caused her to fight with her partner, in some cases resulting in severe physical violence. After one such conflict, she described her partner beating her to such an extent that "he tried to kill [her]". --- Violence, food insecurity, poor mental health outcomes and substance use: A vicious cycle The majority of women described how co-occurring experiences of food insecurity and violence contributed to feelings of low self-esteem, psychological trauma, and depression. To cope, many women described using substances to "get rid of the pain". At the same time, poor mental health and substance use further perpetuated food insecurity and violence. Mental health experiences. The majority of women noted how experiences of violence and food insecurity were associated with feelings of low self-worth and "failure". Sofia, a 51-year-old woman living with HIV in Georgia, shared that, "you'll be like, man, nobody care, give a shit about me". When reflecting on how experiences with food insecurity made her feel, another woman said: It has made me feel unworthy, you know, that I can't get food. It makes me feel less of a person...I'm not worthy of getting the basic stuff that I need. (Brianna, 49 years old, African American, HIV-negative, Northern California) The experiences of food insecurity, violence, and substance use caused women emotional pain and long-term mental duress that they struggled to get past. For example, Barbara, who became homeless and food insecure after she left her abusive partner, and then had to exchange sex for food while living on the streets, described how her experiences affected her mental health in the following way: It's taken a toll on my psyche, period, because sometimes I-and this is the PTSD. I can smell something or go past a certain place, and I relive everything.... It just keeps me on pins and needles, and I constantly relive it over and over and over again...It has taken a toll on my psyche. It makes me paranoid. (Barbara, 53 years old, White, HIV-negative, Northern California) Some women who experienced severe physical discomfort from extreme hunger also remembered the feeling vividly even years later, causing them ongoing distress. Several women described that the poor mental health effects noted above often led them into bouts of what they described as depression in which they withdrew from life, did not see other people, and stayed in bed or slept all day. When asked about how difficulties accessing food made her feel, one woman with very low food security said:...it deepens my depression, definitely, and I already have some days where I really don't even want to get up and get out of bed and do anything. And it definitely doesn't help my energy or anything like that not to be able to go in the kitchen and fix a nutritious meal...-Some days I might just lay there unless I have something specific to do. I'll sleep most of the day off. So I'm not thinking about food or whatever else. (Serena, 37 years old, African American, HIV-positive, Georgia) In extreme situations, some women felt driven to suicidal thoughts or attempts. Another woman with marginal food security and a history of violence and substance use said:...I tried to commit suicide a few times.... [B]ecause I didn't want to do anymore drugs. I was tired of being on the streets, I was tired of being abused, I was tired of being homeless, I was just tired, you know, and I felt like if I would just take some pills and just sleep and never wake up. (Dorothy, 49 years old, African American, HIV-negative, Georgia) Substance use as a coping mechanism. Many women spoke of using drugs and alcohol as a way to cope with the mental health effects of food insecurity and violence. Shirley, who experienced physical, emotional, and sexual violence from intimate partners and engaged in sex work and begging to procure food while living on the streets, reflected on the role of drugs in dealing with trauma from these experiences: [M]ental illness, which is all that trauma...and drugs go hand-in-hand. They go hand-inhand because mental illness most of the times, well, for me, is what caused me to use dope...I wanted to suppress all the shit that had happened, and I didn't know no other coping mechanism but to get high. (Shirley, 57 years old, African American, HIV-positive, Northern California) Dorothy described how she used drugs to cope with her experiences of abuse and her feelings of low self-worth after exchanging sex for food and other resources:...a lot of times, I dealt with the drugs because I didn't want to deal... that was my...relief. You know, like, I didn't want to feel no pain, I didn't want to feel no hurt, I didn't want to feel...the feelings that I was feeling about myself... And sometimes... sucking a man's dick outside, around the corner in the bushes, you know, just like bent over, it was just degrading where I took myself. (Dorothy, 49 years old, African American, HIV-negative, Georgia) Perpetuating the cycle of food insecurity and violence. Poor mental health outcomes and substance use were described to perpetuate the cycle of food insecurity and violence. Some women said that the mental health impacts of abuse and food insecurity rendered them emotionally vulnerable to abusive and controlling men. A woman from Georgia with a lifelong history of abuse who experienced very low food security said, "You know the parts of me that felt unloved and... low self-esteem allowed me to be in abusive situations..." (Molly, 29 years old, White, HIV-negative, Georgia). Several women also described how substance use placed them at risk of violence and exacerbated their food insecurity. Some women noted how they or their partners would use all of their money on drugs or alcohol and would not have any money left to purchase food, leading to conflict. There was a time where food wasn't available because of... bad habits like drugs... and it became an argument about what happened to the food?...Somebody exchanged food for money to get [drugs] instead of what the family wanted or needed (Margaret, 44 years old, African American, HIV-negative, Northern California) Conflict over lack of food could become particularly violent in the context of substance use. One time I was hungry and both of us was [high], but I put the blame on him and it caused an argument...We had violence a couple of times...We fought one time when we was clean, but it wasn't as violent as it was when we was on drugs. (Harriet, African American, 40 years old, HIV-negative, Georgia) --- Effects on health management including HIV risk and engagement in HIV care and treatment Women described how intersecting experiences of food insecurity, violence, poor mental health and substance use increased their HIV acquisition risk and undermined optimal health management including HIV care and treatment. Increased HIV risk. Women described experiencing sexual violence when they exchanged sex for food and/or money, thereby increasing their risk for HIV acquisition. Substance use was noted to amplify these risks by increasing the frequency of sexual exchange for drugs or drug money and by creating dangerous situations in which sexual and physical violence or coercion could occur. One woman with a history of exchanging sex for food and money said: I think people that's got drugs and alcohol, any kind of that substance... are more capable of getting infections like that [HIV] because they're constantly kind of like numb...So I think that it made me more easy to get caught up in situations. And especially the guys, because they feel real loose and they will rape you and bunches take turns (Margaret, 44 years old, African American, HIV-negative, Northern California) Poor mental health outcomes resulting from the intersection of abuse and food insecurity were described to increase the likelihood of future substance use and of risky drug or sexual behaviors due to low self-regard or "giving up". As noted in the following quote, injection drug use carried its own risk of HIV acquisition. I used to get so messed up behind these men. Let me tell you what I did. This is how I got it (HIV). I went and shot dope with somebody I knew to have full-blown AIDS, and I didn't care. I wanted to die because of how this man was treating me and kicking me around. I didn't know what to do. (Shirley, 57 years old, African American, HIV-positive, Northern California) As described earlier, poor mental health status also made it more difficult for some women to leave abusive relationships, maintaining the cycle of violence and/or food insecurity that increased HIV acquisition risk. Medication non-adherence. Several women shared that they missed their medication doses during periods of extreme depression when they didn't leave their house (or even their beds). Low appetite and desire for food during these periods also undermined their ability to take certain medications that required food. The stress and emotional upheaval of food insecurity and especially of violence also made it difficult for women to take their medication as prescribed. Substance use further undermined medication adherence and increased the chances of an empty stomach since women often described forgoing food during periods of intense substance use. These factors are captured in the following quote:...sometimes if I know I haven't eaten, I might not take a certain medication...If I'm not eating, I don't want my stomach to feel really bad...sometimes if I'm going through those violent situations, to block all that or to counteract that, I might start using. And then if I'm using, I'm definitely not eating or taking the medication. (Serena, 37 years old, African American, HIV-positive, Georgia) Another woman living with HIV described how emotional violence, poor mental health and substance use prevented her from taking her ART. They get to call me a bunch of names. We get to fighting. I go get high, and guess what? [I don't take my medication]... My viral load and everything. It affected me to the point where my doctor told me this is my last regimen. She says she ain't got no more combinations. This is my last one. (Shirley, 57 years, African American, HIV-positive, Northern California) While ART non-adherence was more frequently described, many study participants also struggled with medication adherence for treatment of pre-existing mental and chronic health conditions (e.g., clinical depression, bipolar disorder, diabetes, hypertension). In the following quote, a woman describes how her lack of food and poor mental health stemming from experiences of violence leads her to withdraw and miss her medications for arthritis, hypertension, depression, anxiety and other ailments. If I don't eat well I have less energy. I'll get a headache. My stomach starts bothering me...It just affects me wanting to get up and take care of myself sometimes. I just want sometimes to just lay down and just sleep all day or something like that...I take ten pills a day, so I have to put something into my system. (Brianna, 49 years old, African American, HIV-negative, Northern California) Missed clinic appointments. Finally, women described being more likely to miss clinic appointments when they isolated at home during periods of depression due to violence and food insecurity. Many remarked that if they were actively using substances or hung over after a period of intense use, they were more likely to skip an appointment in order to hide this from the doctor. Well, my ex-husband, we'd get in a fight or something, and he'd get to call me a bunch of maggots and all that kind of stuff. Then I want to go get high, and if I get high, I'm not going to the doctor. I'm not going. (Shirley, 57 years old, African American, HIV-positive, Northern California) Women also avoided doctors out of embarrassment or shame when they had physical signs of abuse. Sofia, a 51-year-old woman living with HIV from Atlanta, said, "I couldn't even go to the doctor 'cause I had black eyes. So, that's-it's very embarrassing". Other participants described how in the context of food insecurity and hunger, they chose to skip clinic visits to access food at food banks, or to save the money they would have used on transport to the clinic to purchase food. And even with my doctor's appointment if it's like in the...morning time, I have difficulty getting there if I haven't eaten anything... So, I'm hungry and I have to choose and pick what's more important. So, I choose food. (Brianna, 49 years, African American, HIV-negative, Northern California) Women experiencing extreme hunger described sometimes lacking the energy to physically get to clinic appointments. For example, when asked how experiences of food insecurity influenced her ability to attend clinic visits, Beth, a 71-year-old African American woman living with HIV from Northern California experiencing very low food security said that she sometimes felt "too weak" from lack of food that she would stay in bed and miss her appointments. --- Discussion In this study, we aimed to explore how experiences of food insecurity and IPV intersect and shape HIV risk and treatment behaviors among women living with and at risk for HIV in two geographic areas of the U.S. Our findings revealed that converging experiences of food insecurity and violence led to or exacerbated poor mental health conditions, and the majority noted turning to substances to help them cope. Poor mental health outcomes and substance use then continued to perpetuate the cycle of food insecurity and violence. Ultimately, the co-occurring and mutually reinforcing interactions between food insecurity, IPV, poor mental health and substance use contributed to worse health outcomes by increasing HIV risk behaviors and undermining engagement in HIV care and treatment. In line with syndemic theory, all participants in this study described lives shaped by structural disadvantage, including unemployment, poverty, unstable housing, food insecurity, poor overall health status, and life-long histories of violence. Given that the majority of participants in this study were African American, it is also important to acknowledge that the socio-structural context described by participants is largely a product of structural racism-policies implemented throughout U.S. history that have systematically denied African Americans equal access to social, economic and health resources [65,66]. Our findings suggest that this constellation of structural disadvantage gives rise to the clustering of epidemics of IPV, food insecurity, poor mental health, and substance use among this population of low-income, largely African American women. The synergistic effects of these four epidemics is perhaps best illustrated by the study participant Shirley, who described acquiring HIV by knowingly injecting heroin with someone she knew to have AIDS because of her suicidal thoughts driven by intersecting experiences of food insecurity and violence. Her use of substances to cope with her poor mental health condition led her to consistently miss ART doses and fail HIV treatment so many times that she is currently on the last line of HIV treatment available. This syndemic and its negative impact on health outcomes may be even more pervasive during the COVID-19 pandemic. Emerging evidence suggests that rates of food insecurity, IPV, poor mental health and substance use have risen substantially since the onset of the COVID-19 pandemic and associated social distancing policies [67][68][69]. Thus, it is possible that these factors may also synergistically intersect at higher rates among vulnerable women, such as those included in the present study, to exacerbate negative health outcomes, including HIVrelated morbidity and mortality. Indeed, recent evidence suggests that engagement in HIV care and treatment has declined among people living with HIV since the onset of the social distancing policies [70]. Our findings build on past research that has documented a relationship between food insecurity and IPV, and their independent effects on substance use, mental health, and HIV outcomes. Food insecurity and IPV are independently associated with poor mental and physical health outcomes including depression, anxiety, trauma, low self-esteem, and suicidal ideation [60,[71][72][73][74], substance use [74,75], HIV risk behavior [8][9][10][11] and poor engagement in HIV care and treatment [12][13][14][15][16]. Substance use and poor mental health, in turn, are also associated with each other [76,77] and independently with HIV risk behaviors [78][79][80] and worse HIV outcomes [81][82][83]. Findings from the present study extend this past research to suggest that IPV, food insecurity, poor mental health and substance use co-occur and synergistically interact to exacerbate HIV risk behaviors and worsen engagement in HIV care and treatment. Our results also expand upon the extensive research into the SAVA syndemic [35,49], which has demonstrated SAVA's impact on increased HIV risk behaviors, poor mental health outcomes, and worse HIV care and treatment outcomes among U.S. women [35,84,85]. Specifically, our results highlight that food insecurity and poor mental health conditions also synergistically interact with violence, substance use, and HIV/AIDS, leading to increased HIV risk behaviors and worse engagement in HIV care and treatment. Findings from this study highlight the myriad ways in which these factors interact to impact HIV risk and treatment behaviors that quantitative studies have not uncovered to date and suggest that simple associations between two variables may be insufficient to understand the complex lived experiences of women living with or at risk for HIV. Future quantitative studies should employ a syndemic approach to confirm the intersections uncovered in this paper. --- Strengths and limitations Our study was limited to WIHS participants, which includes cisgender and largely heterosexual women who have participated in the cohort for several decades, and thus may not be generalizable to all women living with or at risk of HIV in the U.S. Our findings are strengthened, however, by the use of two distinct geographic locations that may offer important insights into common patterns in experiences, with potential implications for the broader U.S. context. We have provided detailed descriptions of our study context, methods, and sample to assist readers in determining whether and how our findings apply to other contexts. Future studies should continue to explore the intersection of food insecurity and IPV among other important populations, including undocumented immigrants, who face citizenship-related barriers to accessing government social services such as food aid [86] and violence support services [87]. Research is also needed

Food insecurity and intimate partner violence (IPV) are associated with suboptimal HIV prevention and treatment outcomes, yet limited research has explored how food insecurity and IPV intersect to influence HIV-related behaviors. To fill this gap, we conducted a qualitative study with women living with or at risk for HIV in the United States.We conducted 24 in-depth interviews with women enrolled in the San Francisco and Atlanta sites of the Women's Interagency HIV study (WIHS). Participants were purposively sampled so half were living with HIV and all reported food insecurity and IPV in the past year. Semistructured interviews explored experiences with food insecurity and IPV, how these experiences might be related and influence HIV risk and treatment behaviors. Analysis was guided by an inductive-deductive approach.A predominant theme centered on how food insecurity and IPV co-occur with poor mental health and substance use to influence HIV-related behaviors. Women described how intersecting experiences of food insecurity and IPV negatively affected their mental health, with many indicating using substances to "feel no pain". Substance use, in turn, was described to perpetuate food insecurity, IPV, and poor mental health in a vicious cycle, ultimately facilitating HIV risk behaviors and preventing HIV treatment adherence.Food insecurity, IPV, poor mental health and substance use intersect and negatively influence HIV prevention and treatment behaviors. Findings offer preliminary evidence of a

to impact HIV risk and treatment behaviors that quantitative studies have not uncovered to date and suggest that simple associations between two variables may be insufficient to understand the complex lived experiences of women living with or at risk for HIV. Future quantitative studies should employ a syndemic approach to confirm the intersections uncovered in this paper. --- Strengths and limitations Our study was limited to WIHS participants, which includes cisgender and largely heterosexual women who have participated in the cohort for several decades, and thus may not be generalizable to all women living with or at risk of HIV in the U.S. Our findings are strengthened, however, by the use of two distinct geographic locations that may offer important insights into common patterns in experiences, with potential implications for the broader U.S. context. We have provided detailed descriptions of our study context, methods, and sample to assist readers in determining whether and how our findings apply to other contexts. Future studies should continue to explore the intersection of food insecurity and IPV among other important populations, including undocumented immigrants, who face citizenship-related barriers to accessing government social services such as food aid [86] and violence support services [87]. Research is also needed to explore this syndemic among queer individuals, particularly transgender ('trans') women, who are disproportionately affected by HIV and IPV [88,89] and experience extremely high rates of food insecurity, poor mental health outcomes and substance use due to stigma and discrimination related to their gender identity/ expression [90][91][92]. The sensitive and retrospective nature of some of our interview questions may have made our findings less representative of and faithful to the participants' lived experience [93]. To mitigate this and to increase participant comfort, we utilized a sex-matched interviewer trained in qualitative and trauma-informed methods, including neutral probing and strategies to address participant discomfort, and conducted the interviews at WIHS offices already familiar to participants. Nevertheless, the similarity and frequency of themes we found among participants both living with and at risk for HIV across both study sites underscores the strength of our findings. --- Research implications and conclusions To our knowledge, this is one of the first studies to describe a syndemic of food insecurity, IPV, poor mental health and substance use and its negative influence on HIV prevention and treatment behaviors. Quantitative research is needed to understand how common this syndemic is and to elucidate the synergistic effects of these multiple factors on HIV outcomes. It will also be important for research to examine the social and structural drivers of this syndemic and how policies, such those designed to increase food access through the Biden-Harris Administration's newly enacted American Rescue Plan, [94] can address this syndemic and improve HIV prevention and treatment outcomes. Further, while we focused on HIV as the core health issue among these women, our findings also suggest that HIV is just one of many health issues affected by the syndemic of food insecurity, IPV, poor mental health and substance use. As WLHIV age and develop other comorbidities-a rising problem in the U.S. [95], future research should explore how syndemics like the one presented in this paper can impact long-term health and wellbeing. Findings from this study also point to the need for multi-level and trauma-informed interventions that address the social and structural context in addition to individual-level health behaviors [50]. For example, interventions could include some of the following components: providing food and economic support, conflict resolution skills, mental health and substance use support, and support for HIV prevention and engagement in HIV care and treatment. Such an approach could also include policies that increase and expand eligibility to SNAP, which has been shown to reduce food insecurity and poverty and improve health outcomes, including mental health and medication adherence [96]. A trauma-informed approach may also be particularly useful to address this syndemic. Trauma-informed care provides a framework to increase providers' understanding that exposure to trauma, such as IPV and food insecurity, impacts all aspects of survivors' lives and shapes coping strategies, such as substance use and health care avoidance [97]. It also enhances providers' ability to recognize the signs and symptoms of trauma and guides their responses to include support provision and helping women develop a safety plan when they feel they are at risk of violence. Trauma-informed care has been shown to improve mental health and substance use outcomes and has been identified as a promising approach to improve HIV prevention and treatment outcomes [98][99][100]. Ultimately, our findings suggest that addressing the syndemic of food insecurity, IPV, poor mental health and substance use among women living with and at risk for HIV in the U.S. may be critical to improving HIV prevention and treatment outcomes among this population. --- Data are from the Women's Interagency HIV Study (WIHS), now part of the MACS/WIHS Combined Cohort Study --- Ethics --- Author Contributions Conceptualization: Anna M. Leddy, Amy A. Conroy, Phyllis C. Tien, Sheri D. Weiser.

Food insecurity and intimate partner violence (IPV) are associated with suboptimal HIV prevention and treatment outcomes, yet limited research has explored how food insecurity and IPV intersect to influence HIV-related behaviors. To fill this gap, we conducted a qualitative study with women living with or at risk for HIV in the United States.We conducted 24 in-depth interviews with women enrolled in the San Francisco and Atlanta sites of the Women's Interagency HIV study (WIHS). Participants were purposively sampled so half were living with HIV and all reported food insecurity and IPV in the past year. Semistructured interviews explored experiences with food insecurity and IPV, how these experiences might be related and influence HIV risk and treatment behaviors. Analysis was guided by an inductive-deductive approach.A predominant theme centered on how food insecurity and IPV co-occur with poor mental health and substance use to influence HIV-related behaviors. Women described how intersecting experiences of food insecurity and IPV negatively affected their mental health, with many indicating using substances to "feel no pain". Substance use, in turn, was described to perpetuate food insecurity, IPV, and poor mental health in a vicious cycle, ultimately facilitating HIV risk behaviors and preventing HIV treatment adherence.Food insecurity, IPV, poor mental health and substance use intersect and negatively influence HIV prevention and treatment behaviors. Findings offer preliminary evidence of a

Introduction Community supported Agriculture (CSA) is an agricultural model that reflects growing concerns with the conditions of food production, especially the increasing market pressure on producers and their lack of autonomy to opt for climate and environment friendly practices (Birtalan et al. 2021;Zoll et al. 2018). In North America, Europe and Japan, where many studies have been conducted, CSA has been described as direct, local and long-term partnerships between producers and consumers. CSA is often depicted as a community-based organization of members who share the entire costs of a farm. Varying between different CSA groups, a share of the harvest can be prepaid for several months, a farming season or a whole year (Forbes and Harmon 2008;Urgenci 2019). CSA is thus a risk-sharing model in which producers benefit from advance payments as well as commitment of members (Birtalan et al. 2021). The main motivations for consumers to participate in CSA include their wish to support local farmers, knowing where food comes from, addressing environmental concerns and access to a wide range of fresh and often organically produced food (Farmer et al. 2014;Forbes and Harmon 2008;Haney et al. 2015;Galt et al. 2017). Concerning the transformation of food systems through agroecology, Gliessman (2016) stresses the importance to, "re-establish a more direct connection between those who grow our food and those who consume it". CSA has thus been identified as a way to support agroecology, defined as a science, a set of agricultural practices and a social movement, supporting the emergence of a more holistic vision of the food system (Wezel et al. 2018;Francis et al. 2003). CSA has mostly been directed towards just income for farmers. This is encapsulated in the names used to describe CSA as an agricultural model. For example, in Germany, CSA is called solidarity-based agriculture. In Belgium, the chosen term is Solidarity Purchase Groups for Peasant Agriculture (GASAP). In France, they are Associations for Maintaining Peasant Agriculture (AMAP). More recently, the issue of accessibility from the consumers' perspective has gained more prominence. For instance, the third principle of the French AMAP charter states that "each AMAP seeks to broaden the accessibility of such food to all" (MIRAMAP 2014). Similarly, the British CSA network sets as its 3rd pillar fairness, solidarity and reciprocity: "CSA farming [...] see [s] healthy food as a right and work [s] towards equity and sovereignty in our food systems" (CSA Network 2021a). In the United States, the CSA Innovation Network emphasizes equity: "we must first recognize and address the systems of injustice that weaken it" (CSA Innovation Network 2020). Even though justice plays a role, currently, most CSA members in the Global North belong to a homogenous group with above-average education and income (Galt et al. 2017;Matacena 2016;Renting et al. 2012). Only a small fraction comes from low-income backgrounds (Hanson et al. 2019;Vasquez et al. 2017). Low income has been identified as a major barrier to participation in CSA (Farmer et al. 2014;Forbes and Harmon 2008;Galt et al. 2017;Urgenci 2019). Therefore, CSA has been criticized for not adequately addressing inequalities in the food system (Matacena 2016;Mert-Cakal and Miele 2020;Renting et al. 2012). Few CSA programs have been designed to include lower-income members (Quandt et al. 2013). As low-income consumers have limited access to local food (Sbicca 2012), some CSA groups provide opportunities for them, for example, by offering unclaimed shares or selling discounted shares (Urgenci 2021). The literature identified reasons why lowincome households are less likely to become CSA members (Farmer et al. 2014;Forbes and Harmon 2008;Galt et al. 2017) and outline time, transportation, as well as food habits and preferences as potential obstacles to direct producer-toconsumers relationships (McGuirt et al. 2019(McGuirt et al., 2020;;Garner et al. 2021). Cotter et al. (2017) also identified discomfort of having to pay in advance without knowing the content of the box, as well as a lack of knowledge about the way CSA functions as barriers. For low-income families, these uncertainties are difficult to bear as they are constantly on a tight budget and may not be able to afford a backup option (Cotter et al. 2017). In the United States of America, low-income households joining CSA through government-supported programs reported improved diets, increased vegetable consumption, a reduction in time spent shopping, and less money spent on food. For example, Basu et al. (2020) suggest that " [...] CSA-based interventions may be cost-effective for improving diets among low-income persons". Moreover, lowincome households "place a slightly higher monetary value on their share [...] despite their lower incomes, and they are just as interested in food-related activities (...)" (Galt et al. 2017). Thus, the inclusion of low-income households in CSAs can also be of advantage for farmers. Pole and Grey (2013) demonstrated that low-income households could be more prone to share the risks with the producers. Even though social support actions in CSA have received some attention recently, their limits and potentials have rarely been documented. Moreover, a consolidated typology is missing to assist CSA groups in their efforts to diversify their membership. This research is a contribution towards filling these gaps. Our main objective is to systematize and appraise social support actions of the CSA movement. Taking the CSA principles as a starting point, our main research question asks: How do social support actions in CSAs operate in terms of social inclusion and what obstacles and challenges are associated with them? The article is structured as follows: first we outline the theoretical background of strategic action fields. We describe CSA as moving from the strategic action field of food sovereignty, to the broader field of food justice, both concepts will be defined in our theoretical section. We then describe our methods and data: building on an initial list of support actions, we interviewed seven initiatives spearheading social support actions to come up with a classified typology of support actions. Our results systematize the strategies to diversify membership in CSA to address social inclusion and food justice. We also discuss the challenges encountered while implementing these strategies. In our discussion, we reflect upon the shift in strategic action fields. To conclude, we offer pathways for future research and action. --- Theoretical background: strategic action fields of CSA Strategic action fields are "the fundamental units of collective action" (Fligstein and McAdam 2012). SAFs offer an analytic frame to investigate how collective actors in social movements try to gain strategic advantage in and through interactions with other groups. According to Fligstein and MacAdam (2012), social life is structured by a complex web of embedded strategic action fields. Each social actor is framing her / his action in a field where everyone follows common rules (Suckert 2017). Depending on the field and where actors position themselves, they might be incumbents, holding a dominant position, or challengers, opposing power holders and carrying an alternative vision of how the field should be organized. There are also governance units, "charged with overseeing compliance with field rules and, in general, facilitating the overall smooth functioning of the system" (Fligstein and McAdam 2011). In social movements, it is typical for SAFs to develop around specific issues or concerns. SAFs are dynamic and the order of the field might change with regard to how collective action is organized or embedded. Fligstein and McAdam (2012) identify four characteristics of SAFs: first, an understanding of the issue at stake (i.e. viewpoints on social inclusion in the transition towards sustainable food systems); second, a variety of actors involved with different degrees of power (e.g. farmers, consumers or other stakeholders); third, the actors involved have a joint understanding how the field operates, the rules of the game (i.e. how CSAs operate); fourth and last, there is an overarching interpretive frame, which might be contested from the various actors involved. The analytic approach of SAFs has recently been applied for the first time to the German CSA movement in terms of the culture of cooperation (Degens & Lapschiess 2023). We build on this work and focus on the consequences for SAFs when CSAs shift their focus from just income for farmers to include a more diverse range of members. CSA had until recently been framed by CSA networks as part of the food sovereignty movement (Réseau des Gasap 2011; Miramap 2014). The Declaration from the Food Sovereignty Forum held in 2007 in Mali defines food sovereignty as "the right of peoples to healthy and culturally appropriate food produced through ecologically sound and sustainable methods, and their right to define their own food and agriculture systems" (La V<unk>a Campesina 2007). Food sovereignty is used to reclaim the right for small-holders to self-determine the conditions for food production, distribution and consumption. According to Holt-Gimenez (2010;2011), food movements can be placed along a spectrum stretching all the way from a neoliberal to a radical vision of food systems. The SAF of food sovereignty focuses on a radical systemic change. In this field, farmers are incumbents that are praised in a narrative with agrarian undertones. The contribution of CSA to the SAF of food sovereignty is clear: CSAs are offering community support to family farmers applying ecological practices and are aiming at restoring the autonomy of small-scale farmers in the food system. Food sovereignty tends to operate with mechanisms and policies that empower various actors in the food system. In CSA, the particular focus lies on supporting farmers. In contrast to food sovereignty, the contribution of the CSA movement to the more encompassing SAF of food justice is still largely uncharted. Yet, there are some developments in the CSA movement, which aim to broaden the focus of CSA not only towards farmers, but also towards low-income consumers. Food justice advocates perceive the food system as characterized by transnational corporations dominating food production and distribution. They are interested in economic pressures and power imbalances in the food system and aim at creating access to appropriate food irrespective of the consumer class, race or gender (Alkon and Agyeman 2011;Romer 2014;Sadiku et al. 2018). Yet, the interpretative frame of the field of food justice as well as the interpretation how it operates is different: for Gottlieb and Joshi (2010), food justice is an intersectoral normative and egalitarian model aiming at removing all kinds of injustice, as for example racial or class injustice. Food justice aims to ensure "that the benefits and risks of where, what and how food is grown and produced, transported and distributed, and accessed and eaten are shared fairly" (Gottlieb & Joshi 2010:6). The food justice movement also acknowledges the underlying power structures that are perpetuating oppression and food inequalities and tries to address them. Here it is where both, low-income households and marginalized producers come into play. For example, food justice advances the opportunity for the most marginalized in the food systems to make choices about their food by establishing fair relationships with local producers, or by growing their own food (Sadiku 2018). Thus, the analytical approach of SAFs assists in scrutinizing the opening of the CSA movement towards the broader issues of food justice provides important insights regarding the dialectics between the obstacles to social support actions and the solutions found to overcome these obstacles in the respective SAFs. We included a variety of initiatives: ones with a long and well-recognized experience in social inclusion, like Rock Steady Farm in the US; as well as other initiatives which are less vocal about this topic, but who stated in their replies to the Covid-19 survey that they had been implementing social support actions since their creation, like the German Solawi and the French Saint-Denis based AMAP. We also included voices from the Global South by talking to CSA Demétria in Brazil. One interview was conducted with a CSA network, Paniers marseillais (PAMA), with a unique experience of organic solidarity shares. This additional network was identified as a particularly advanced initiative, a governance unit according to the theory of SAFs and used to validate some of the results of the previous interviews. --- Approach A case study approach has been adopted (Yin 2018) to build a typology of social support actions. The case we focus on is the implementation of social support actions in CSA, opening up towards a more just food system. Seven semistructured qualitative interviews were conducted (Misoch 2019, see overview in Table 1). Each interview consisted of guiding questions (see below) on the social support actions implemented by the CSA, on the barriers that prevent from becoming members and on the reasons that prevent members from making use of the social support actions. The case study approach helps us, on the one hand, to classify the support actions, but also assists in reflecting upon how the CSA movement is moving from the SAF of food sovereignty towards the SAF of food justice. It enables us to synthesize, for each of the four characteristics of the theorization of SAF, the main limits faced by the CSA movement, and to study how the change in the interpretative frame in term of issues at stake, power imbalances, acknowledging how the food system operates as well as offering an interpretive frame, which seeks to address exploitative and unequal issues in the industrial food system. --- Materials and methods In order to systematize and appraise the social support actions implemented in the strategic action fields of the CSA movement, we build upon an initial list of social support actions. This list was established in the frame of a report about the response of CSA and other local solidarity-based partnerships during the Covid-19 pandemic (URGENCI 2021). This initial list is based on an online survey with 328 replies and an additional 40 interviews conducted with CSA network coordinators mainly in Europe, North America, Japan, China and Brazil. Seven social support actions were identified: 1) Logistical support to marginalized and vulnerable people (e.g. home deliveries); 2) Food donations to marginalized and people in need; 3) Food donations to soup kitchens, pantries, charities; 4) Integration of migrants, asylum seekers in the group; 5) Discount shares in poor neighborhoods; 6) Home deliveries to vulnerable and disabled people; and. 7) Cooperation with social projects and social organizations: e.g. join ad hoc solidarity group. For this paper, we first identified CSA groups implementing some of these solidarity mechanisms and contacted them via email. Seven initiatives located in North-and South America as well as in central Europe agreed to participate: six CSAs and one local CSA network. Although the sample size is limited, it comes with a certain scope, both geographically and in terms of the experience and roles taken by respondents in their respective groups. Hence, it than a decade at the international level and has experience and knowledge on the topic based on numerous projects and exchanges during field visits. --- Results: towards socially more inclusive CSA The analysis of the qualitative interviews resulted in the identification of two focal areas that could assist the strategic action field in which CSAs operate to become more socially inclusive. First, we appraise social support actions (SSA) differentiating between emancipatory social actions and those that are rather motivated by charity. Second, we outline challenges faced while implementing these actions. --- Typology of social support actions Based on the interviews, we classified the earlier identified social support actions (URGENCI 2021) into two categories: the 'emancipatory support actions' (see Fig. 1 below, more towards the center) empower low-income households and lead into reconsidering some of the fundamental rules of CSA. The other category, 'charity support actions' (see below, towards the outside) support actions that provide short-term relief that might not affect the way CSA functions. Emancipation is understood as the process through which (groups of) individuals gain agency by being freed from someone else's control, whereas Charity is understood as the voluntary assistance of those in need. We further distinguish between support actions which are either of financial or of non-financial nature (see Fig. 1). By financial support, we mean that an action is affecting the contribution paid by the members, should it be the amount paid or the calendar of payment. By non-financial support, we mean towards food justice has consequences regarding how these limits can be addressed. --- Protocol A consent form was signed by all interviewees. The interviews were carried out in April and May 2021. Each interview was held via video call and recorded. The length of the interviews varied between 28 and 156 min. The semi-structured interviews were fully transcribed, and the qualitative data analysis was supported by the MAXQDA<unk> software. A content structuring analysis grid was designed in order to generate the codes by a combination of deductive and inductive approaches (Kuckartz & Rädiker 2023). The obstacles to join a CSA and the social support actions offered by the CSA were all predefined, whereas the typologies of obstacles and social support actions and challenges have been derived based on the interviews. Below are the guiding questions that were asked during the interviews. • What mechanisms and social support actions do you implement in your initiative? Are there factors that are crucial for the successful implementation of social support actions? • What barriers still exist that might prevent people from becoming members of your initiative and members from making use of social support actions? We analyzed the interviews through coding along the main categories (Kuckartz and Rädiker 2023). We complemented the interview data with a variety of sources ranging from individual CSAs to CSA networks' websites and recordings of a series of webinars on the topic. In addition, one of the authors has been involved in the CSA movement for more stay in the CSA or not in order to reduce obstacles for joining CSA. The purpose of trial memberships is more general than the sole integration of low-income households. It can be used as a strategy to increase membership, without taking into account their social background. Nevertheless, our case study stresses flexible payment plans as an efficient tool, whereby potential members, regardless their social background, can learn "how to do CSA". Furthermore, the respondents from Colombia, Brazil and France remark that food from a CSA is not necessarily cheaper than food from the supermarkets, but it is cheaper than organic food from other markets. The respondent from CSA Demétria emphasizes: The respondent from CSA Agromandala stresses the specificity of CSA in setting up prices over the whole year, thus providing an exceptional stability both to producers and consumers. This allows members to plan their expenses. "If --- Charitable non-financial support actions CSA principles consider an uncollected share as unclaimed and lost. In some CSA, these unclaimed shares are given to a nearby charity, rather than being given to another member. Another option for members is to leave some food from their shares in a basket at the distribution site. This allows members to swap food they do not like or food that is exceeding their consumption capacities. There are also other forms of donations. CSA Bel Aire connects members via an Emaillist, where they can share clothes or furniture, but also apartments or language courses. Similarly, CSA Agromandala initiated a chat to share recipes for unfamiliar food. The respondents indicate that food donations are not only accessible for members of the CSA, but also for individuals from the neighborhoods or farm workers. The farmers are able to donate surplus food instead of selling it on other markets, because the CSA offers them financial security. Another non-systemic support action is voluntary work on the farm, mentioned by the respondents from the two German CSA groups. Even if assisting on the fields is a requirement, members decide on their own when, how often and how long they would like to work. There is no connection between the amount of the financial contribution and the number of voluntary working hours. In many CSA groups, the financial contribution is anonymous anyway, so that no that an action is not affecting the contribution paid by the members, but instead results in in-kind social support. The distinction between charity and emancipatory social support actions is of analytic nature. Social support actions might move between these poles, depending on the CSA and how they have implemented the support action. Through the lens of SAFs, CSA actors who strive for emancipatory actions are embedding the CSA movement into the larger strategic action field of food justice. --- Charitable financial support actions Several social support actions do not address the emancipatory ambition of the CSA movement, but provide direct and fast help without necessarily implementing pathways towards a more just food system. One key principle of CSA is upfront payments. However, the payment frequency varies: members might pay on a monthly basis, a whole season, several months or even a year in advance. The aim is to ensure a sufficient budget for the whole growing season and simple accounting for the producer. Sometimes, however, members can be late on their payment plans. The interviewee from CSA Ortenau emphasizes that financial difficulties should not be a reason to leave: "I also had an individual case once, (...) that someone signed off during the course of the year and said: <unk>'ve lost my job and can no longer afford to do this. And then I agreed with that person to continue and to pay a lower rate now, and when that person has a job again, to pay more. That was accepted and worked out well." (Interview 6, pos. 20). Members of Rock Steady Farm CSA have the possibility to sign up for a flexible payment plan. Members are asked to pay a 20% deposit at the beginning of the season and to pay the rest one or two months later (Rock Steady Farm 2021a). In general, members are encouraged to get in touch with the CSA, if they are unable to pay upfront. All respondents stressed that suitable solutions can be found. Several CSA groups offer trial memberships to lift the obstacle of signing up for CSA membership. This targets individuals who are interested in joining, but would like to get familiar with the system first. After a couple of weeks or months, they can decide whether to stay or quit. CSA Bel Aire even increased accessibility by offering a discount for new members. A similar approach is provided by CSA Agromandala, in which members who are under contract for at least three months receive a discount. This is a way to ensure a stable income for the farmers without forcing subscribers into long contracts. Moreover, some CSA groups allow members to decide each month if they would like to Steady Farm, every member chooses the payment level based on their income and wealth. Guidelines are available on the CSA's homepage to help members make their choice. These guidelines take property, occupation, health status and social and financial security into account. Members can choose between four different payment levels. The lowincome level includes a 20% discount. The middle price, called "market price point" is the break-even price, at which all costs of the farm are covered. The upper level requires an additional 20% increase and the contributor level 40%. Table 2 illustrates the breakdown of contribution levels of CSA Rock Steady Farm in 2020. CSA Rock Steady Farm members also have the option to receive a smaller share at a lower price. If the average contribution ranges below the price necessary to cover the costs, the farm has to make up for the gap by using food access funds or by appealing for donations. If the CSA ends up with more money than is necessary to cover all the costs, they add the surplus to their reserve to be able to finance the following seasons. Thanks to the sliding scale, about 57% of the production was shared with low-income households in 2020. This proportion was reached through a mix of sliding scale, solidarity shares and wholesale produce for food pantries. The farm had about 310 paying members in 2020. Additional 130 households received solidarity shares, subsidized by 40-100%. These are meant for individuals who cannot afford to pay the lower sliding scale level. The farm finances such solidarity shares via individual tax-deductible donations, grants and sponsorships (Rock Steady Farm 2021). The farm collaborates with a Health Center serving HIV/AIDS patients with additional health conditions and LGBTQIA + communities. The members facing high medical costs can register for solidarity shares. As for the sliding scale, the solidarity share classification is also based on trust. No proof of income is required. The CSA provides information to assign oneself to one of the payment levels, but the members decide on their own which level they feel they belong to. CSA Demétria offers a single solidarity share for a person who is facing an insecure situation. In order to provide the person with a harvest share, all members have agreed to pay a slightly higher contribution. Another differentiated contribution model is the bidding round (also called pledging round). The respondent from CSA Trier describes the procedure of the bidding rounds as follows: At the beginning of the season, the annual expenses of the farm are calculated and a general meeting is organized. Based on the calculated costs and the number of members, an average monthly contribution is announced to provide guidance. In the first round, members anonymously place their bid: they state how much they would like to pay. If the costs of the season are not covered, a second, even one is keeping track of working hours committed by each member. The respondent from CSA Trier said: "(...) We also explicitly do not want a lower contribution to the bidding round to be linked to working on the field, because not everyone is able to do that. And solidarity, as we see it, means that everyone contributes according to their possibilities. This also means that if you do not have money and time, because you are in a precarious employment situation or are a single parent, you still have the opportunity to participate." (Interview 5, pos. 36). Voluntary work by CSA members can be considered as a social support action in an indirect way, because if many individuals work voluntarily, the labor costs and thus the membership fees can be reduced. Thus, members who are physically able to work in the fields, and who have the time and resources to do so, relieve the burden of those who don't have the same opportunity. By so doing, volunteers structurally support the CSA initiative and its increased openness. The respondent from CSA Trier adds: "So, people who do not want to be part of the garden community, but who only want to get the vegetables, are also allowed to participate." (Interview 5, pos. 52). Food donations and voluntary work are support actions that, although not emancipatory, can facilitate access to CSA membership at an operational level. --- Emancipatory financial support actions Besides stable prices, the possibility to choose between differentiated contributions is another strategy towards financial support for CSA members. Several social support actions fall in the category of differentiated contributions, e.g. sliding scales, solidarity shares, bidding rounds and self-assessment, which we describe in the following. Sliding scales are strategies to mitigate economic barriers. Shares are priced lower for lower-income households and higher for higher-income households. For example, at CSA Rock Transparency in a bidding round is the basis for mutual trust. Talking about income also helps middle class members to appreciate their financial privileges. CSA Ortenau follows a slightly different principle in their bidding round: a "non-binding guiding value" is determined to assist the members in their decision on how much to contribute. The aim is to reach the annual budget without further bids. However, in some cases, the costs of production are not fully covered. The members are then asked to slightly increase their contributions. Neither lower limit nor proof of income are required. The respondent from CSA Ortenau states that often members with fewer resources are also quite willing to pay more. Figure 2 illustrates the distribution of the financial contributions of CSA Trier and CSA Ortenau in 2021. --- Emancipatory non-financial support action Emancipatory non-financial support actions include collaborations with other organizations to achieve greater reach. Our interviews and findings of other studies show that collaboration with other organizations, such as social services, health centers or food pantries, helps to reach populations that otherwise would not know about CSA (Urgenci 2019). Especially collaboration with governmental aid programs have proven to be helpful: this is a way for CSA to delegate to professionals the high responsibility to ensure that those who need the CSA shares the most eventually receive them. CSA Rock Steady Farm also supports the project "CSA is a SNAP" in order to make CSA shares more accessible to Supplemental Nutrition Assistance Program -beneficiaries. Finally, it is important to emphasize webinars and trainings organized by CSA networks on the topic. Rock Steady Farm and Paniers marseillais have both been playing an active role in learning from experienced food poverty organizations, sharing their experience and linking with food aid beneficiaries through webinar series (Rock Steady Farm in some cases a third, bidding round might be necessary. A majority of members pay more than the average and others pay significantly below the average price. This model is illustrated in Fig. 2 (below). In 2018, an additional table was collectively created to provide guidelines on how to self-assess one's contribution. The table displayed monthly recommended contributions based on the net income of a single household. For example, for a person with an income of 1,524 Euros, the guiding value contribution of 95 Euros corresponds to 6.23% of the net income. For somebody earning a minimum wage, the CSA contribution would then be only 73.80 Euros. Thus, the table also has the important function of showing members that it is acceptable to contribute less due to low financial resources. The bidding round does not require a minimum bid value: the only rule is that all the farm's financial needs must be covered before the farming operation can start. The interviewee from CSA Trier reflects on the principle of the bidding round: "It is exciting and thrilling every time, but everyone has the same interest. So, everyone wants it to continue. So yes, everyone pulls together and adds another two Euros. And if more people with lower incomes would join them that would probably also work out somehow." (Interview 5,pos. 46). Transparency is crucial for a successful implementation. The respondent of CSA Trier notices: "We have to stop not talking about money <unk>, because I find it very difficult how a society is supposed to change towards more solidarity with bidding rounds, when you don't know at all how much I earn and whether the contribution is a lot or it is a small amount." (Interview 5,pos. 62). clash of food cultures is that some of the solidarity shares' beneficiaries are actually dropping out, as in the case of Agromandala: --- "[Members] don't get what they want, what they think they want in their heads. They don't understand that's what we grow and that's what nature gives to you and it is not enough to explain to them that it's not avocados all year [...]". (Interview 1,pos. 88) The respondent from Brazil points to the wider context: "[...] For the people here, it is status, for example, to drink Coca Cola. We have hundreds of varieties of fruits. Wonderful juices and so on [...]. --- But you can't get to anybody in the house that's gonna offer you lime juice, you know? Cause that's like "hey what's the matter, you don't have money to buy a Coke?" [...] You have to have these things to gain status, to show that you have some kind of status". (Interview 2, pos. --- 44) The respondents underline that some new members are not willing to adapt their eating and cooking patterns to what they get from the CSA. In addition, the share also provides members with vegetables that they are not used to eat. CSA is a strategic action field with rules that are hard to understand for newcomers. Interviewees see this as a major hindering factor for low-income households. Another barrier is reaching the pick-up stations. It requires time and additional transportation to pick-up shares of the harvest. The respondent from CSA Rock Steady Farm explains: "We have fruits, eggs, you know, we have a diverse box, but it <unk>s always when someone really is strafed on time, then they just got to go to the grocery store once they get everything they need as cheap as possible." (Interview 3, pos. 12). Lastly, our respondents also mention a lack of public awareness as an obstacle: "One barrier is that many people still don <unk>t know us. During the first few years after we founded the CSA, we did a lot of information sessions and events, but of course you never reach all sections of the society". (Interview 6,pos.50) Often, however, there is no effort on communication as CSA groups are not able to accommodate more members. The example from CSA Ortenau is interesting: in its early 2021; CSA Network 2021b; Miramap 2020). Solawi Trier also conducted a seminar about income and payment levels in CSA. During this seminar, the table showing the levels of differentiated contributions based on the income was presented and discussed collectively. --- Challenges for CSA to implement social support actions Our analysis reveals key challenges and limitations of social support actions. The socio-economic backgrounds of members of the CSA groups participating in this study are quite heterogeneous. They range from wealthy areas with middle-class members up to one of the poorest suburbs of Paris, even though this latter CSA also predominantly consists of individuals with an academic background and a high social status. In this latter case, low-income households from the neighborhood do not join the CSA, although it is close to their homes. In other CSAs the financial status of the members is mixed, but low-income members are hardly represented. A set of factors mentioned by our respondents are eating habits, consumption patterns and expectations of size, shape and immaculacy of food. Even if these are general barriers, affecting everyone potentially becoming a member, they might be particularly challenging for low-income households. The lack of choice was mentioned as a major obstacle. In most of the CSAs, although diverse fruits and vegetables are offered, the content of the share is the same for all. This can come as a surprise for consumers who would prefer to choose food following their personal tastes: "I think sometimes it can be a little bit overwhelming if someone doesn <unk>t actually eat that many vegetables. So, we have two different sizes, but it <unk>s amazing some people say that even the smaller sizes are too much." (Interview 3, pos. 37). Expectations about preferred foods in CSA shares can be both culturally influenced and dependent on the educational background. This cultural or educational gap makes it challenging to meet expectations. Beneficiaries would like to have a choice; they don't like to have the composition of the share "imposed". The farmer who initiated the Agromandala project in Colombia pointed out that members often do not know how to use and prepare food from the CSA because they lack the knowledge to do so. This leads to sharing advice with members on how to add unfamiliar vegetables or salads to their daily diet or other cooking methods that go beyond frying. She emphasized that individuals with low incomes not only lack the money for healthy food but, above all, the knowledge about it. The result of such a Thus, the simple fact that being a member is a requirement can be a limit preventing the generalization of social support actions. Another aspect that limits the use of social support actions is dignity. For a lot of potential beneficiaries, there is a high risk of being exposed, stigmatized as a

Community Supported Agriculture (CSA) seeks to address injustices in the food system by supporting small-scale farmers applying agroecological practices through a long-term partnership: a community of members covers the cost of production and receives a share of the harvest throughout the season in return. Despite an orientation towards a more just and inclusive food system, the existing literature points towards a rather homogeneous membership in CSA. A majority of CSAs tends to involve (upper) middle-class consumers with above average education and income levels. Low income is still a major obstacle in joining a CSA. Membership diversification through social support actions is one possible way. Our main objective is to systematize and appraise social support actions of the CSA movement. Taking the CSA principles as a starting point, our main research question is: How do social support actions in CSAs operate in terms of social inclusion and what obstacles and challenges are associated with them? The theory of strategic action fields assists in describing how the CSA movement is positioning itself as an actor in and across neighboring strategic action fields. The CSA movement is clearly positioned in the Food Sovereignty field. By shifting the focus from justice to farmers to justice for members, the CSA movement is now also exploring the Food Justice field. Indeed, the CSAs' contribution to the food justice movement is still largely uncharted. In our results, we identify both social support actions that are already implemented in the CSA movement in different countries, and the challenges that are associated with these actions. We pinpoint a classification of social support actions implemented by CSA organizers to increase access to their initiatives. We make a distinction between the emancipatory actions that empower beneficiaries and contribute to a systemic change, and punctual, charitable interventions that neither affect the structure of a CSA nor the food system.

but it <unk>s amazing some people say that even the smaller sizes are too much." (Interview 3, pos. 37). Expectations about preferred foods in CSA shares can be both culturally influenced and dependent on the educational background. This cultural or educational gap makes it challenging to meet expectations. Beneficiaries would like to have a choice; they don't like to have the composition of the share "imposed". The farmer who initiated the Agromandala project in Colombia pointed out that members often do not know how to use and prepare food from the CSA because they lack the knowledge to do so. This leads to sharing advice with members on how to add unfamiliar vegetables or salads to their daily diet or other cooking methods that go beyond frying. She emphasized that individuals with low incomes not only lack the money for healthy food but, above all, the knowledge about it. The result of such a Thus, the simple fact that being a member is a requirement can be a limit preventing the generalization of social support actions. Another aspect that limits the use of social support actions is dignity. For a lot of potential beneficiaries, there is a high risk of being exposed, stigmatized as a person in need of assistance. The respondent from Brazil reports a case in which a person was supported by a solidarity share. The person was uncomfortable with the support offered and was even upset that they talked about the financial situation. An interviewee from Germany puts this challenge as follows: "[The beneficiaries had] the feeling of not being an equal member of the community or exploiting us as gardeners in case they pay less". (Interview 5,pos. 28) This quote shows that members would often find it difficult to pay less than the recommended amount of contribution, and thus often prefer to drop out instead of being a perceived "second class member". --- Discussion The objective of this paper was to categorize and appraise social support actions implemented in the CSA movement. The results of our analysis emphasize two types of social support actions in the current CSA movement. First, support actions which seem to have the same limitations as other food donations: the beneficiaries are put in a passive position, they have neither the right nor the power to define the content of their shares. These donations perpetuate altruism on the one hand and indignity on the other hand and may lead to dependency of the receivers (Allahyari 1999; Rock Steady Farm 2021). The second type of social support actions consists of emancipatory actions, which can result in revisiting the underlying principles of CSA in order to associate more closely the beneficiaries and ensure their active participation in the partnership. We also identified numerous challenges for the social support actions implemented by CSA. In this discussion, we contemplate how moving from the field of food sovereignty to the field of food justice can increase the potential for emancipatory support actions. The four characteristics of SAFs assist in thisreflection. The first characteristic of any SAF is the shared understanding of what is at stake. The challenge in CSA is that social support actions assisting low-income households does not necessarily resonate with supporting smallholder farmers. The hindering factors of social support actions are in line with previous research about limitations of access to CSA: the income and the social background are years, it focused a lot on advertisement in the region, particularly in the countryside. Even if this CSA is still offering guided farm tours and educational work in cooperation with schools, shares are already booked so that they do less publicity. This results in less opportunities to know about the existence of the CSA for individuals outside of the small circle already participating. The respondent from the US even considered the ethnic origin as a core cultural element explaining the difficulty for social support actions to be fully endorsed by potential beneficiaries. About the CSA concept she commented: "[...] it's the States, European, it's not like global in a way. So, people might just be like: <unk>whatever that's just not my thing <unk> you know <unk>not my culture, not my background I am not interested <unk>." (Interview 3,pos. 37). Another challenge is that the introduction of social support actions requires additional effort. Members of CSA groups appreciate guidance to choose the level of their contributions, as they still struggle to assess their financial privileges. Coming up with such classifications to guide members in their contributions requires an important effort from the core CSA group. Moreover, in order to reach individuals of concern, easy access to informative content deems to be important. Prospective members need to know which payment options are available. Low-income individuals will not see CSA as beneficial if they do not know about the flexible payment options. In order to work efficiently, including the organization of polls to know what was appreciated and what was not, and in order to prepare a regular newsletter, the respondent from the US estimates that 25% of the work on the farm is used for communication purposes. CSA Bel Aire core group members decided not to work with governmental programs to avoid becoming dependent and overloaded with administrative work. The length of the contract term and the requirement to collect the shares from the pick-up may be perceived as temporal and spatial barriers. Adding to this, many newcomers feel unfamiliar with the responsibilities that go along with a CSA membership: "We had people who wanted to buy something here and when they found out that you have to be a member, it scared them off. [...] By the way, I had the same problem at the beginning. I also found it a bit strange. One is then suspicious: "what kind of cult is that?" (...) Here you have to be active. And active means being a member and paying membership fees. And that's the first step from passivity to activity." (Interview 4,pos. 86). (Miramap 2020;Paturel and Ndiaye 2020). Their flagship policy proposal is the creation of a social security for food. The French CSA movement seems to be passing an alliance with challengers in the field of food justice. The third characteristic of SAFs relates to the rules of the game. We witness a lack of recognition of the "rules of the game" by low-income households. Committing on a permanent, long-term contract with a CSA appears challenging and somewhat risky for consumers who do not have a stable income. The membership fee is a frequently mentioned reason why persons with a low income are excluded from CSA, as the long-term commitment of the membership exceeds their planning horizon. Moreover, the mere requirement to become a member in order to be eligible for a share seems to be repelling many, as it is interpreted as a sign of a closed community. Within the new SAF of food justice, the CSA movement is encouraged to revise the agreed rules of the game and to revisit how CSAs are functioning, in order to enhance the participation of low-income households. Our results show interesting examples such as the bidding round system developed in German CSAs, or the sliding scales where "higher priced shares subsidize the lower priced shares" (Forbes andHarmon 2008, 2017) in Rock Steady Farm, as well as the integration of CSA in governmental food justice programs. These actions are all transforming the structural functioning of CSA. Shifting from food sovereignty to food justice implies revisiting some of the ways of doing CSA, in particular by involving other organizations, getting public support, involving food justice organizations in the recruitment of members, or asking all members to self-assess their wealth in order to opt for the right income scale. The fourth and last characteristic of SAFs is the overarching interpretive frame. Confronted with the limits of their social support actions, CSA coordinators and organizers have been embedding the CSA movement into the broader SAF of food justice. This can be interpreted as a major change regarding the overarching interpretative frame. Indeed, within the interpretative frame of food justice, not only the support to producers is at stake, but a more inclusive vision of the community in CSA and the overall food system. The Table 3, summarizes our discussion. It shows the difficulty, identified by our respondents, for lowincome households to recognize what is at stake in a CSA, to find their place as challengers among the incumbents (CSA organizers and farmers), and to align with the "rules of the game". Can CSAs do better, or at least different, than food donations? By moving to the field of food justice, CSAs are activated as a learning environment: the stories told are not only about doing the right thing, but also about gaining experiences and skills. The theory of SAFs underlines discriminatory elements (see Cotter et al. 2017;Forbes and Harmon 2008;White et al. 2018). As McGuirt et al. (2020) demonstrate, the sociodemographic context impacts the participation in CSA programs. This is also confirmed with research previously conducted among CSA farmers: "A well-functioning CSA requires members who are aware of the responsibilities that go along with membership. This applies equally to people of all income levels" (Sitaker et al. 2020, 105). Our interviews emphasize the openness of CSA members and producers to embark on a new strategic action field: they want to learn what is at stake from actors already positioned in the field. They are gaining literacy about food poverty. A series of online webinars and training sessions were organized on the topic in the US, in France as well as in Germany. They provide interesting examples: CSA group coordinators and CSA network coordinators are willing to be trained in order to learn and to improve their understanding of what is at stake in another field. This observation seems to be supporting previous descriptions of CSA as a social knowledge-sharing space (Piccoli et al. 2020) or a self-managed research and grassroots innovation movement participating in the co-construction of knowledge (Anderson et al. 2021). The second characteristic of SAFs considers the variety of actors with differentiated power. Our results demonstrate contention in the field of food sovereignty, with, on the one hand active CSA members (incumbents), and, on the other hand low-income households. Some social support actions can come indeed across with a whiff of paternalism. Lowincome households' consumption patterns are understood by respondents as self-determined, which can be summed up as: they get what they believe they need to get, and they thus choose unhealthy diets although CSA tries to educate them. This seems to ignore the fact that low-income households are tight on money and time. Even if they are interested in joining a CSA, they simply cannot afford to risk a meager or even empty share. In contrast, middle-class members can afford to get an empty share every once in a while, and make up for it with additional food shopping. Thus, it seems CSA cannot really welcome low-income households as empowered actors in the strategic action field of food sovereignty. There is a need to move to a different field, where the actors are different and include low-income households as key protagonists. As a response to this challenge is the recognition of the need to build equal partnerships with organizations who have a long experience in the field. For example, within the French CSA network's accessibility working group, the collective analysis underscores that food justice has been used by a constellation of local initiatives, and even by state institutions. However, the current system is still failing to restore food precarious citizens' dignity and agency research. One way forward would be to study the reaction of beneficiaries instead of the CSA coordinators: how do they relate to the social support actions? Another, more ambitious, project would be to compare the social composition of a variety of CSA groups, those with social support actions and those without, to see if those implementing social support actions are more diversified than others. This would require designing objective criteria of diversification and gathering a large sample of initiatives both with and without social support actions. --- Conclusion Our main research question asked how social support actions in CSAs operate in terms of social inclusion and what obstacles and challenges are associated with them? We suggest distinguishing between charitable and emancipatory actions within the responses to the issue of access to CSA developed by the groups we studied. We categorize the mechanisms with differentiated contributions, including sliding scales, as emancipatory type of action, since they closely involve all members and beneficiaries in the design of their contribution, often based on a self-assessment of their wealth, which includes many different parameters. This is similar to equal partnerships with community leaders and organizations representing low-income households. In contrast, food donations, flexible payment plans, trial subscriptions and stable prices are unilateral support actions that do not seem to contribute to restoring low-income households' agency. Voluntary work on the farm could be considered as emancipatory in some cases, but we have no evidence that this social support action has been used for individuals who otherwise would not be able to be part of a CSA. The broader implications of our findings for practitioners are mostly articulated around the necessity of an increased recognition of the needs of low-income households. A thorough assessment of the potential participants' demand that developing emancipatory support actions for food justice requires a capacity to orient oneself in a new field. CSA farmers and members willing to address the issue of access to healthy organic local food are learning by doing. CSA groups are answering a "living lab", "do it yourself" logic, rather than basing their actions on well-cut methodologies (MIRAMAP 2020). As shown in the interviews, the approach of CSA groups is often spontaneous and might be lacking preparation. CSA members initially think they can easily open up to new populations by offering discounted or even free shares, but they then realize that a process of acculturation, of learning about the actors, the issues and the common rules is necessary. The use of the theory of SAFs sheds a new light on the CSA movement. It offers a promising avenue that could be used in future research. Up to 16 different "propositions" (Fligstein & MacAdam 2012) are presented to describe different dynamics across SAFs. One of the propositions states that "SAFs are generally destabilized by external shock originating from other SAFs, invasion by other groups of organizations or large-scale crises." Further research would be needed to be able to distinguish between the role of the Covid-19 crisis, which was the initial context of the first typology of social support actions (Urgenci 2021) implemented by CSA partnerships. The role of food justice actors who started challenging the traditional definition of food sovereignty is worth further investigation to restore the position of small-holders in the food system. A further limitation of this study is that we did not interview low-income households involved in CSAs. Such interviews would be needed to complete the picture. The perspective of low-income households that have been involved in the past, that are currently involved or that intend to become involved might draw up a more differentiated picture. Yet, the advantage of our approach is that it illuminates settings from very different angles providing indepth and personal insights into the work and responsibilities within CSAs. The field of access to CSA is unstable and subject to major changes. There is a lot of space for further --- Food justice Table 3 Exploring a new field: towards food justice and more emancipatory social support actions from a Strategic Action Field perspective, considering the 4 main characteristics of SAFs. to social support actions and the consequences of embed should be conducted. Each member should be recognized as a partner of the farm in an equitable relationship. Research and action should thus emphasize the emancipatory aspect of CSA, which lies in a strong relationship between producers and consumers. Moreover, in any move they make for social integration, CSA groups should not work in isolation: they should rather build strong alliances and be part of political coalitions asking for systemic change. Ideally, the initiative for social support actions should come from low-income households themselves rather than being a process driven by the CSA groups to try to attract low-income consumers. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons. org/licenses/by/4.0/. --- Philipp Weckenbrock is a researcher at the professorship for organic agriculture at the Justus Liebig University Giessen, Germany. There, and with the research organization "Die Agronauten", he has worked on agroecological production and sustainable food systems. Judith Schryro holds a Master's degree in Nutritional Sciences and Home Economics. She has been working at the chair of Food Sociology at Justus Liebig University Giessen University, Germany. Currently she works as a consultant for food and consumer education at the Consumer Rights Agency in Berlin.

Community Supported Agriculture (CSA) seeks to address injustices in the food system by supporting small-scale farmers applying agroecological practices through a long-term partnership: a community of members covers the cost of production and receives a share of the harvest throughout the season in return. Despite an orientation towards a more just and inclusive food system, the existing literature points towards a rather homogeneous membership in CSA. A majority of CSAs tends to involve (upper) middle-class consumers with above average education and income levels. Low income is still a major obstacle in joining a CSA. Membership diversification through social support actions is one possible way. Our main objective is to systematize and appraise social support actions of the CSA movement. Taking the CSA principles as a starting point, our main research question is: How do social support actions in CSAs operate in terms of social inclusion and what obstacles and challenges are associated with them? The theory of strategic action fields assists in describing how the CSA movement is positioning itself as an actor in and across neighboring strategic action fields. The CSA movement is clearly positioned in the Food Sovereignty field. By shifting the focus from justice to farmers to justice for members, the CSA movement is now also exploring the Food Justice field. Indeed, the CSAs' contribution to the food justice movement is still largely uncharted. In our results, we identify both social support actions that are already implemented in the CSA movement in different countries, and the challenges that are associated with these actions. We pinpoint a classification of social support actions implemented by CSA organizers to increase access to their initiatives. We make a distinction between the emancipatory actions that empower beneficiaries and contribute to a systemic change, and punctual, charitable interventions that neither affect the structure of a CSA nor the food system.

INTRODUCTION According to Dr. A.P.J. Abdul Kalam Azad "Youth below the age of 25 are the most powerful resources on the earth, under the earth and above the earth. We have to empower them through value-based education and leadership." Agriculture, along with its allied sectors, is the largest source of livelihood in India, particularly in rural areas. It also contributes significantly to the country's GDP. Sustainable agriculture, which includes ensuring food security, rural employment, and environmentally friendly practices such as soil conservation and biodiversity protection, is crucial for holistic rural development [1,2]. India has witnessed revolutions in agriculture, dairy, poultry, and fisheries, indicating the importance of these sectors. This section provides detailed information on agricultural products, machinery, research, government policies, schemes, loans, market prices, animal husbandry, fisheries, horticulture, sericulture, and more.Although agriculture remains the primary occupation for over half of India's population, the involvement of youth in this sector is declining [3][4][5][6]. As the youngest nation in the world, India has a massive youth resource that could contribute significantly to the agriculture sector. However, the sector is facing emerging challenges such as land fragmentation, labor shortage, depleting natural resources, climate change, low profitability, market competition, and nutrition and gender issues [7][8][9]. To address these challenges, the agriculture sector needs new ideas, innovations, rebranding, and revitalization, which can be provided by the energy and vitality of the youth. However, there is a decreasing trend in youth participation in agriculture [10-14]. Given the current scenario and the changing dynamics of agriculture in India, it is essential to leverage the potential of youth to bring prosperity to the sector and improve the lives of young people. This raises the question of whether there is enough scope to involve youth in the agriculture sector and harness their innovative mindset, energy, and enthusiasm to uplift the sector. (Rajarshi Roy Burman et al., 2019). It is observed that India has the relative advantage at present over other countries in terms of distribution of youth population. India's advantage in young population is also evident when it is compared with other Asian countries. In Comparison to China and Indonesia, the two major countries which determine the demographic features of Asia, India is seen to remain younger longer then the other two. These three countries together accounted for 68 per cent of the population of Asia in the year 2010 and the share of Asia itself is about 60 per cent in world population. The proportion of females in the youth age bracket is generally lower due to the fact that females tend to live longer than males. (UNESCO, 2017). This gender difference is more pronounced in developed regions, but in India, the gender differentials are less significant compared to other countries. As the population ages and shifts to higher age groups, the share of the 15-34 age group decreases, indicating increased longevity. This widening gender difference reflects the aging population and the fact that females typically live longer, resulting in a lower proportion of youth among females. Interestingly, in India, the similar share of youth among males and females suggests a healthy fertility rate and population growth. Education plays a crucial role in addressing development challenges in rural areas, as it provides a direct link between food security and the education of rural children. Basic numeracy and literacy skills have been shown to improve farmers' livelihoods. Access to knowledge and information is essential for rural youth to address the challenges they face in agriculture and to influence agricultural policies that affect them, such as access to markets, finances, green jobs, and land. However, in both developed and developing countries, rural youth often lack access to even the most basic formal education, especially in developing countries where educational institutions are underdeveloped. Formal primary and secondary education can equip young people with basic skills in numeracy, literacy, and business management, as well as introduce them to agriculture. Non-formal education, including vocational training and extension services, and tertiary agricultural education can provide more specific knowledge related to agriculture. Unfortunately, access to information and education is often worse in rural areas of developing countries compared to urban areas, and this disparity is evident as early as primary school. Many rural children in developing countries face hunger and lack the energy to attend school or absorb information. During peak agricultural seasons, labor shortages may lead parents to prioritize their children's involvement in household and agricultural activities over attending school. Additionally, rural schools often have poor infrastructure and lack classroom materials, while accessibility can be a challenge as schools are located far away from rural communities (FAO.2009). Overall, improving access to education and information is crucial, particularly in rural areas of developing countries, in order to empower rural youth, address agricultural challenges, and promote sustainable development. Social media offers a unique opportunity for farmers and ranchers to connect with a diverse audience and share their stories. It allows them to build relationships, share information, and address any concerns or misconceptions about modern farming practices. In India, the agricultural sector is facing challenges, such as a decline in the number of farmers and a growing need for food security. Encouraging young farmers to get involved and addressing the challenges they face can play a crucial role in ensuring a sustainable future for Indian agriculture. India is losing more than 2,000 farmers every single day and since 1991, the overall number of farmers has dropped by 15 million [15] Globally, there is a growing interest in finding ways to engage youth in agriculture, and social media can be a powerful tool in achieving this goal. (IFAD, 2012; Paisley, 2013). Considering the above facts the present study was conducted with the objective -<unk> To study the training needs of Rural youth. <unk> Suggestions of rural youth to promote their participation in agricultural activities. --- MATERIALS AND METHODS The study was conducted in the year 2020-2021 in Lucknow district of Uttar Pradesh of India selected. Exploratory Research Design was used for the study. There are four Tehsils in Lucknow namely Bakshi Ka Talab, Lucknow, Malihabad and Mohanlalganj. Out of these four Tehsils Mohanlalganj Block was selected for the study. For present study 5 villages were selected namely Atrauli, Daudnagar, Bhadesuwa, Hasanpur Kaneri and Barvaliya. From each village 21 respondents were selected for the study and information were collected regarding their agricultural activities. A total number of 105 respondents were selected through random sampling from eight sample village panchayats. Data was collected through personal interview with a semi-structured schedule from farming and non-farming youth. The reliability and validity of the schedule was checked. Percentage, cumulative frequency and rank were used for calculation and drawing the inferences. --- RESULTS AND DISCUSSION It is noticed from the Table 1 that, majority (62.86 per cent) of rural youth need training in cashew and mango production, followed by identification of pest and diseases (58.09 per cent) and post harvest technology (56.19 per cent). Further they need training in vermicompost/compost making (54.28 per cent), integrated nutrients management (52.38 per cent), organic vegetable production (51.42 per cent) and mixed farming (50.47 per cent), while less than half of them need training in spices and medicinal plant (49.52 per cent), intercropping (47.68 per cent), mechanization in agriculture (46.66 per cent), poultry/dairy (44.77 per cent), goat farming (42.86 per cent), mushroom production (40.95 per cent) and Nursery management (40.00 per cent). This indicates that rural Youth are fascinated towards mango and cashew production. The probable reason could be that as Lucknow district is Mango belt so if they get training in mango production they can earn more income. Also as cashewnut is plantation crop if they get proper training for cashewnut production and processing they can sell it at better rate and earn better income and can raise their standard of living. --- CONCLUSIONS Study found that a significant issue among the farming youth is their lack of awareness about new technologies, schemes, and policies. This lack of exposure to innovative production techniques indirectly affects the productivity of crops. Additionally, the farming youth are generally uninformed about government policies aimed at attracting and retaining them in the agricultural sector, and the benefits associated with these policies. To address this issue, awareness programs, camps, and demonstrations should be organized by KVKs, SAUs, and NGOs to make the farming youth aware of these initiatives. Furthermore, many farming and non-farming youth have limited education when it comes to handling smartphones. Therefore, the government should promote adult education specifically through platforms established by KVKs. It is crucial to strengthen agricultural policies to retain the youth's interest and involvement in farming. --- COMPETING INTERESTS

India's population is predominantly rural, with approximately 68.9% residing in villages. Therefore, rural India truly represents the essence of our country. It is crucial to empower the rural youth by offering them opportunities for growth and well-being that align with their aspirations. However, the modern world demands skilled workers, and industries are increasingly reliant on technological advancements, particularly in information and communication technology (ICT). This rapid pace of technological change necessitates individuals to possess diverse cognitive skills. Consequently, it is imperative to address these challenges and equip India's rural youth with the appropriate education and skills they need to navigate the demanding global landscape. While the government

Introduction Like other nations in the world, China is experiencing a rapidly aging population and the population aged 65 and over has reached 13.50% in the latest census [1]. Chinese women have a longer average life expectancy (80.5) than Chinese men (74.7); hence, elderly 2 of 14 women are significantly more likely to be widowed than elderly men [2,3]. In 2010, China's sixth census revealed that of the 47.74 million elderly widows, 33.45 million were female. Such characteristics of female-dominant elderly widows are also reflected in reports from other countries [4,5]. It is estimated that the number of elderly widows in China will reach 94.49 million by 2050 [2]. Studies revealed that widowhood in older women is associated with poorer mental health and reduced coping resources compared with those living with their partners [6][7][8]. Older female widows living alone in rural and remote areas even face more stressors and fewer coping resources in their daily lives compared with their counterparts in urban areas [9][10][11]. However, studies on the prevalence of and factors associated with their psychological well-being for this most vulnerable population are scarce. This study addresses the gaps in the research. It is well-researched that elderly widows experience numerous changes including cognitive decline [12], sleep problems [13], and decreased life satisfaction [14]. While those changes gradually declined in the majority of elderly widows, some of them developed persistent and negative psychological symptoms such as loneliness, depression, and anxiety that can lead to increased morbidity and mortality rates among older widows [15][16][17][18], main factors affecting society to achieve the WHO and the UN healthy aging and quality of life for older people [19,20]. Loneliness is defined as a psychological condition resulting from the discrepancy between an individual's existing and desired social relationships [21]. The prevalence of loneliness among elderly widows in China is 40.7% [22]. Studies show that nearly 70% of elderly widows find loneliness is the most difficult aspect of daily life to deal with [23]. Anxiety is defined as apprehensive anticipation of future danger or misfortune, accompanied by feelings of unease or physical symptoms of tension [24]. Depression is often characterized by persistent sadness and a lack of interest or pleasure in previously valuable or enjoyable activities [25]. The prevalence of depression is about 10-45% in elderly widows worldwide [8,10,26]. The social production function theory developed by Lindenberg explains the likelihood of negative psychological symptoms in elderly widows living alone from a sociological perspective [27,28]. Lindenberg believes that people are goal-oriented and seek to produce psychological or emotional well-being [29]. A person's psychological well-being is determined by physical and social well-being. The achievement of physical and social well-being depends on the achievement of instrumental goals, such as comfort to produce physical well-being and affection to produce social well-being. A lack of resources can prevent people from generating well-being and induce depressive symptoms. Being married and living with a spouse is a great resource for generating all the instrumental goals that contribute to well-being. On the contrary, widowhood and living alone are associated with a lack of resources, and affection to engage in meaningful physical and social activities; thus, contributing to poor well-being and negative psychological symptoms in this population. Studies identified that long periods of widowhood, declining income, or managing multiple chronic illnesses in elderly widows were associated with these negative psychological symptoms [30,31]. However, the mental health of elderly widows attracted less attention in China, as evidenced by a relatively small number of studies [10,[32][33][34]. Such a situation also indicates that many elderly widows remain undiagnosed and untreated in China. Remaining in this situation could increase the burden on China's healthcare system and hinder the goal of achieving equitable access to healthcare. In China, the huge disparities in economic development between urban and rural areas cause a large proportion of China's young and middle-aged rural labor force to migrate to urban areas for employment opportunities, while most rural elderly widows live alone [35]. Such internal migration is more prevalent in remote areas where employment opportunities are even less compared with rural areas [36,37]. Chinese remote elderly female widows living alone may experience multiple risk factors of poor health and well-being that are widely reported in the literature including older age [38,39], female gender [8,31,40], living arrangements [41], and rural and remote region [10,42]. However, few empirical studies have carefully investigated the prevalence of and factors associated with negative psychological symptoms among them. Therefore, a cross-sectional study was conducted in remote regions of southeastern China to determine the prevalence and risk factors of negative psychological symptoms, including loneliness, depression, and anxiety. --- Materials and Methods --- Ethics This study has been reviewed and approved by the Ethics Committee of Xiangya School of Nursing, Central South University (Project Number 2019021). Participants finished the questionnaires anonymously. The questionnaires were then coded uniformly by the investigators. Participants who scored above the cut-off value on the self-rating scales of loneliness, depression, and anxiety were offered psychological counseling by investigators within 48 hours. Investigators assisted with referrals to medically designated institutions or mental health professional institutions for those whose negative psychological symptoms did not improve after psychological counseling. --- Study Design A cross-sectional study was conducted to investigate the loneliness, depression, and anxiety of remote elderly widows living alone in China. This study was conducted from December 2018 to March 2019. --- Study Setting This study was conducted in Longshan County, Hunan Province, China. Longshan County is located on the northwestern edge of Hunan Province and is one of the most remote counties in Hunan. The per capita Gross Domestic Product (GDP) of Longshan County in 2021 (RMB 20, 409) is still below China's per capita GDP (RMB 80, 976) and the per capita disposable income (RMB 17, 560) is also below the country's average (RMB 35, 128) [43,44]. Longshan county has a relatively diverse population, with ethnic minorities accounting for 81% of the country's total population [45]. The proportion of people aged 60 and over in Longshan County (18.99%) is similar to the national level (18.70%). In addition, the sex ratio (the ratio of males to females, using females as 100) of 104.66 for the population of Longshan County is also very close to that of the country (105.07) [46,47], predicting a similar proportion of remote elderly widows living alone as in China. Given the same social status and traditions of remote elderly widows living alone in China, the survey conducted in Longshan County should be representative. --- Participants and Recruitment Based on the list of elderly women registered with the village committee, and combined with recommendations from the committee and villagers, the investigators invited elderly widows to participate in the study if they met the inclusion and exclusion criteria. The inclusion criteria were as follows: (1) female <unk>60 years old; (2) not remarried after widowhood; (3) living alone due to widowhood and not accompanied by children or other partners; (4) having clear awareness and communication skills; and (5) voluntary participation in the study. Individuals who met one of the following criteria were excluded: having a severe psychiatric illness, severe acute or non-communicable disease, or having undergone major surgery within the last year. In this study, the required sample size was calculated using the logistic regression formula from the textbook: N = (Z 1-<unk>/2 +Z <unk> )2 [P 1 (1-P 1 )]b 2, <unk> = 0.05 [48]. P 1 is the rate of positivity when x is taken as the mean value, and b is the estimated value of the regression coefficient corresponding to x. P 1 and b were identified from the pilot study of participants. A sample size of 230 was required, taking into account a 15% missed visit rate. Trained investigators explained the purpose and process of the study to remote elderly widows living alone by telephone. They were informed that they were at liberty to decide whether to participate in the study without any consequences. Appointments were made for potential participants who had agreed to participate in the study for an in-home survey. After participants completed the informed consent form, investigators instructed them to complete the questionnaires. The contents of the questionnaires were verified in time after completion to eliminate invalid questionnaires. 2.5. Measures 2.5.1. Social-Demographic and Widowhood-Related Information Social-demographic data, including age, race, education, occupation, acute and chronic medical conditions, and the number of children, were collected through a self-designed questionnaire. Widowhood-related information including the duration of widowhood, age at death of the spouse, marital happiness, living alone period, cause of death of the spouse, length of the marriage, being the primary caregiver of the spouse before widowhood, and anticipation of the death of the spouse was also collected. --- The University of California at Los Angeles Loneliness Scale (ULS-8) The ULS-8 was used to evaluate the loneliness of the participants [49]. The scale consists of eight items and uses a four-point Likert scale: 1 "never", 2 "rarely", 3 "sometimes", and 4 "always". The total score of the ULS-8 ranged from 8 to 32. The higher scores, the higher levels of loneliness. The Chinese version of the ULS-8 demonstrated good reliability and validity. The Cronbach's <unk> coefficient is 0.741. Exploratory factor analysis showed that the cumulative variance contribution was 55.39% when extracting one common factor from the Chinese version of the ULS-8 scale, which was consistent with the theoretical conception of the original scale [50]. The cut-off value for ULS-8 in this study used the 24 points determined in previous studies [51,52]. --- Short-Form Geriatric Depression Scale (GDS-15) The GDS-15 was used to assess participants' depressive symptoms during the last week. It includes 15 items. The response options of "yes" or "no" are adopted. Items 1, 5, 7, 11, and 13 were negatively worded and scored in reverse in a certain order. Higher scores indicate more severe depressive symptoms [53]. The Chinese version of the GDS-15 has good reliability, with a Cronbach's alpha coefficient of 0.793. The Activity of Daily Living Scale assessed significant differences in GDS-15 scores among elderly people with different self-care abilities, demonstrating the GDS-15 scale's good discriminant validity. The cut-off value for the Chinese version of the GDS-15 was 5 [54]. --- Generalized Anxiety Disorder Scale (GAD-7) Participants' anxiety in the past two weeks was measured by the GAD-7. It consists of seven items whose scores range from 0 "not at all" to 3 "nearly every day". A higher total score represents the more serious anxiety symptoms [55]. The cut-off value of the Chinese version of the GAD-7 scale score was 9. The Cronbach's <unk> coefficient of the scale was 0.898. The criterion validity of the Chinese scale was high, with a sensitivity of 86.2%, a specificity of 95.5%, and a kappa value of 0.825 [56]. --- Data Analyses Epidata 3.1 (The EpiData Association, Odense, Denmark) was used for data entry. The data were analyzed by SPSS 26.0 (Statistical Product and Service Solutions, IBM Corp., Armonk, NY, USA). All measurement data were assessed for normal distribution using the Kolmogorov-Smirnov test. The composition ratio, mean and standard deviation are used for statistical description. A chi-square test and correlation analysis were used in univariate analysis to test for differences in loneliness, depression, and anxiety symptoms across socio-demographic factors and widowhood-related factors. Binary logistic regression was used to predict factors contributing to the three negative psychological symptoms. The significance level was set as p <unk> 0.05. --- Results --- Socio-Demographic Characteristics of Remote Elderly Widows Living Alone A total of 271 remote elderly widows living alone were enrolled in the present study. Additionally, 234 valid questionnaires were returned (valid response rate = 86.3%). The mean age of the participants was 73.29 <unk> 7.37. The majority (69.7%) of participants had acute or chronic medical conditions. The vast majority (97.3%) of participants had been married for more than ten years. Participants who had been widowed for more than three years amounted to 78.7%. Almost eighty percent (80.8%) of widows were satisfied with their marriage. Approximately 60.3% of the participants were their spouse's primary caregivers before they were widowed (Table 1). --- Prevalence of Loneliness, Depression, and Anxiety The prevalence of loneliness was 8.1%. The mean ULS-8 score for the lonely participants was 24.68 <unk> 1.00. Almost half (44.0%) of widows had depression symptoms. Participants' mean score of GDS-15 was greater than the cut-off value (5.29 <unk> 3.88). The mean GAD-7 score was relatively low (4.50 <unk> 4.10), and the proportion of widows with anxiety was 16.7% (Table 2). --- Comorbidity between Loneliness, Depression, and Anxiety Loneliness and depression are comorbid in 6.0% of the participants. Anxiety and loneliness have a 3.8% comorbidity. Additionally, 15.0% of participants experienced anxiety and depression. There are 3.8% of participants with all three of these negative psychological symptoms (Figure 1). was 16.7% (Table 2). --- Comorbidity between Loneliness, Depression, and Anxiety Loneliness and depression are comorbid in 6.0% of the participants. Anxie loneliness have a 3.8% comorbidity. Additionally, 15.0% of participants experience iety and depression. There are 3.8% of participants with all three of these negativ chological symptoms (Figure 1). --- Associated Factors with Loneliness, Depression, and Anxiety Univariate analyses were conducted to test the differences in socio-demograph widowhood-related factors in loneliness, depression, and anxiety symptoms. Ac chronic medical conditions, marital happiness, being the primary caregiver befor owhood and anticipating the death of the spouse were associated with negative p logical symptoms (P <unk> 0.05) (Table 3). --- Associated Factors with Loneliness, Depression, and Anxiety Univariate analyses were conducted to test the differences in socio-demographic and widowhood-related factors in loneliness, depression, and anxiety symptoms. Acute or chronic medical conditions, marital happiness, being the primary caregiver before widowhood and anticipating the death of the spouse were associated with negative psychological symptoms (p <unk> 0.05) (Table 3). Loneliness, depression, and anxiety symptoms were each examined as the dependent variable in a binary logistic regression analysis. The univariate analysis's significant variables served as the independent variables. Marital happiness (OR = 0.283, p <unk> 0.05) was found to be the protective factor against loneliness. Being the primary caregiver of the spouse before widowhood was a risk factor for loneliness (OR = 7.101, p <unk> 0.01). Marital happiness was also the protective factor for depression and anxiety (OR = 0.332, p <unk> 0.01; OR = 0.194, p <unk> 0.001, respectively). Acute or chronic medical conditions are a risk factor for depression (OR = 2.257, p <unk> 0.01) (Table 4). --- Discussion Remote elderly widows living alone in China are a vulnerable population. However, there are few comprehensive assessments of this population's loneliness, depression, and anxiety prevalence. In this study, the prevalence rates for these negative psychological symptoms were 8.1%, 44.0%, and 16.7%, respectively. Furthermore, comorbidities of these symptoms are frequently present. The high prevalence of loneliness, depression and anxiety among remote elderly widows living alone suggests that this population has substantial mental health care needs. Based on previous research, elderly widows living alone face numerous challenges and are prone to negative psychological symptoms [8,41]. Remote elderly widows living alone lack a supportive environment and adequate resources that can help maintain their physical and mental capacities [57][58][59]. The prevalent negative psychological symptoms lead to a decrease in the quality of life of elderly widows and an increase in the risk of hospitalization, which is detrimental to global healthy aging [19]. In China's traditional family-oriented culture, the elderly place a high value on the daily companionship of their spouse [60]. As a result, elderly widows in China experience greater symptoms of loneliness than widows in other countries [61]. Besides, in remote areas, the willingness of elderly widows to remarry is weak due to the constraints of children's opposition, public opinion, and traditional attitudes [62,63]. Few elderly widows had remarried, which may also contribute to the high prevalence of loneliness among remote elderly widows living alone. Local government officials should positively encourage remarriage by shaping public opinion to activate the spouse's role in providing moral support. Additionally, other support systems should be bolstered to improve spiritual companionship. The government should encourage children to provide financial support or care to elderly widows and foster a harmonious and supportive neighborhood. As a result of providing care for their partners, older women are more prone to suffering from poor mental health [30]. After the death of their spouse, they have far less daily contact and communication with their intimate partner, which increases their risk for mental health issues [64,65]. The caregiver role in this study predisposed the experience of loneliness, contrary to the results of earlier research [29]. Our finding indicates that the relationship between the pre-widowhood caregiver role and negative psychological symptoms is inconclusive and requires further research. The participants' pervasive depression and anxiety may result from living alone, having no one to discuss their negative feelings about widowhood, and the lack of recreational facilities and activities in remote areas [66]. Additionally, remote elderly widows always neglect their mental health and lack the knowledge to utilize the social security system to improve their physical and psychological health [67]. Village Councils need to strengthen social support for this vulnerable group by engaging them in self-care activities, for instance, regular seminars on mental health, self-assessment of mental health, and seeking health from health professionals for negative psychological symptoms. Secondly, healthcare workers in remote rural areas should receive relevant training to enhance their ability to identify and treat negative psychological symptoms in elderly widows. They also need to frequently visit elderly widows at risk of developing negative psychological symptoms, for instance living alone and having acute or chronic diseases, and encourage them to actively seek help from family members and close friends. Participants who have more acute or chronic medical conditions are suspectable to depression. Therefore, primary care health professionals need to monitor the mental health conditions of this group, take preventive measures and provide timely interventions to this group of elderly windows. Regarding medical care, regular free medical checkups for remote elderly widows living alone should be added to public health service programs to reduce the risk of negative psychological symptoms in this population. The types of illnesses covered by medical insurance for elderly widows living alone should also be expanded by covering mental health problems. We found that marital happiness was a protector for elderly widows from developing negative psychological symptoms in this study. The finding is similar to previous studies which identified that marriage could provide the elderly with emotional satisfaction and physical health benefits [29,68,69]. Nonetheless, widowhood dissolves the marriage and deprives the spiritual support of the spouse, resulting in negative psychological symptoms [70][71][72]. One possible factor contributing to our finding might be that the vast majority had been widowed over 3 years. Therefore, they might have adapted to a new life without their partners. We suggest that future studies need to add a qualitative study component to enhance an understanding of the influence of marital happiness on the adaptation to widowhood in the elderly. Research evidence strongly suggests that negative psychological symptoms were due to the difficulty of adapting to the change in widowhood (e.g., the decline in income, coping with the chores of life alone) after being married for a relatively extended period [30,73,74]. Village Councils should assist elderly widows in maintaining contact with friends or family members and meeting their spiritual and material needs. Moreover, early interventions should be developed to aid them in adjusting to widowhood sooner. Group-based complicated grief therapy and mindfulness techniques [75,76], as well as individual writing-based emotional expression therapy [77,78], are effective in adapting to widowhood. These evidence-based interventions should be considered as strategies to prevent remote elderly widows who live alone from developing mental health problems. Szczepanska-Gieracha and colleagues identified that older women's sense of responsibility for their health through goal-focused group psychotherapy subsequently improved their depression symptoms [79]. Their studies provided new sight into the treatment of negative psychological symptoms in elderly widows living alone. The goal-focused group psychotherapy aims to establish group relationships among older women in similar situations and encourages them to discharge their negative feelings. The therapists then help to model older women's right attitudes and motivate them to become more independent in their lives. This kind of psychotherapy can significantly reduce depressive symptoms in elderly women and improve their sense of well-being. Moreover, Szczepanska-Gieracha and colleagues also identified that the Virtual Therapy Garden provided for older women helped them better recognize and use relevant mental health resources [80]. This kind of Virtual Therapy includes stories and interactions in the Virtual Therapy Garden by which older women realized that there is much they can do to improve their mental health and take responsibility for their future. Virtual Therapy demonstrates a significant reduction in anxiety and depression. These evidence-based interventions should be considered as strategies to improve the mental health problems of remote elderly widows who live alone. To our knowledge, this is the first study to examine depression, anxiety, and loneliness in remote elderly widows living alone. Most previous studies focused on elderly widows and explored only a portion of the three negative psychological symptoms [8,41,81]. Besides, participants in this study were well represented. Longshan County is a typical remote area with backward economic development [82], and the age and gender structure of remote elderly widows living alone are extremely similar to national levels. Some limitations need to be acknowledged in the current study. Our research is a cross-sectional study, thus causal inferences about the correlation cannot be made. A longitudinal study should be considered in the future. Second, participants' self-rated loneliness, depression, and anxiety may have been somewhat influenced by recall bias. Finally, the results of this study may only be generalized to areas with similar economies and cultures. Based on the poor mental health status of this population and the great health care needs, there is a request for future research on a wider population. --- Conclusions Our research revealed that remote elderly widows living alone in China frequently experience loneliness, depression, and anxiety. Marital happiness was the protective factor against three negative psychological symptoms. Elderly widows who have been the pri-mary caregiver for their spouses and suffering from acute or chronic illness are more likely to feel lonely or depressed. There is an immediate need for individualized psychosocial intervention programs and related services to relieve these symptoms and improve their life outcomes, contributing to the achievement of universal healthcare equality in China. --- Data Availability Statement: Data for this study are available from the corresponding authors upon reasonable request.

1) Background: Research indicates that most elderly widows are at a high risk of experiencing negative psychological symptoms. It is common for elderly women in rural and remote areas to live alone without family support to cope with stress due to the mass rural-to-urban migration of China's youth labor force. Such a situation further worsens their psychological health and well-being. However, the prevalence of and risk factors associated with negative psychological symptoms (loneliness, depression, and anxiety) among remote elderly widows living alone in China are currently unclear; (2) Methods: A cross-sectional study was conducted in Hunan Province, China. The loneliness was assessed through the University of California at Los Angeles Loneliness Scale (ULS-8). The depression and anxiety were assessed with the Short Form Geriatric Depression Scale (GDS-15) and Generalized Anxiety Disorder Scale (GAD-7), respectively. The Chi-square test and correlation analysis were conducted to identify factors associated with negative psychological symptoms. Logistic regression was performed to predict risk and protective factors contributing to loneliness, depression, and anxiety symptoms. The significance level was set as p < 0.05; (3) Results: A total of 271 remote elderly widows living alone were enrolled in the present study. Additionally, 234 valid questionnaires were returned (valid response rate = 86.3%). The prevalence of loneliness, depression and anxiety was 8.1%, 44.0%, and 16.7%, respectively. Acute or chronic medical conditions, marital happiness, being the primary caregiver before widowhood and anticipating the death of the spouse differed significantly in the distribution of negative psychological symptoms. Logistic regression analysis predicted that participants who were satisfied with their marriage had a lower likelihood to experience loneliness, depression, and anxiety (p < 0.05). Being the spouse's primary caregiver before widowhood was more likely to have symptoms of loneliness (p < 0.01). Those with various acute or chronic medical conditions were more likely to suffer from depression (p < 0.01); (4) Conclusions: Remote elderly widows living alone in China are prone to loneliness, depression, and anxiety symptoms. Being the primary caregiver before widowhood and having many acute or chronic medical conditions are risk factors for loneliness and depression, respectively. Marital happiness is the protective factor against negative psychological symptoms. To accomplish the goal of equitable access to mental health care in China, evidence-based policy and resource development to support psycho-social interventions that prevent and manage negative psychological symptoms for remote elderly widows living alone are urgently needed.

Introduction In China, the degree of aging is much higher in rural areas than in urban areas. According to the Seventh National Population Census (Nov. 26,2021), the proportion of older adult people over 60 years of age was 23.81% in rural areas while only 17.72% in urban areas. 1 China has a tradition of filial piety for thousands of years, family old-age care has long been dominant in the urban and rural old-age care systems, especially in rural societies where the social old-age care system is not well developed (1), and children's support plays an important role in family old-age care. Dependence on family for old age is a Chinese tradition, children's support plays an important role in the mental health of older people, but current research has come to different conclusions. One view is that children's support is beneficial to the older adult in maintaining their mental health. Scholars have found that children's financial support can increase the social participation of the older adult (2,3), and that emotional interactions and caregiving can reduce the negative emotions of the older adult (4-7), thus enabling them to maintain their mental health. In addition, children's support can increase the level of subjective well-being of the older adult (8), and enhance their spontaneity, willpower and integrity (9). Another viewpoint suggests that intergenerational supportive behaviors of children may have a negative impact on the mental health of older adults. Excessive support from children fails to maintain respect for older people (10), and violates their personal privacy (11). Daily caregiving by children may reduce the ability of older adults to live autonomously, ultimately decreasing their life satisfaction and increasing their depression (8,(12)(13)(14). Furthermore, older persons receive financial support from their children, which may violate their traditional role as breadwinner and perceive themselves as a burden to their children, leading to an increased sense of powerlessness and psychological burden on older persons (15,16). Since the reform and opening up of China, a large number of young and middle-aged rural workers have moved to the cities to work and do business, and the incomplete nature of large-scale urban-rural migration has prevented rural family members, especially parents, from moving with their adult children (17), increasing the physical distance between generations (18). The separation of parents' and children's living space makes the traditional style of family eldercare no longer realistic, destroying traditional Chinese family structures and endowment arrangements (1), making the older adult more prone to depression (19), and jeopardizing their physical health due to lack of care (20,21). In rural China, the migration of children has made the burden of agricultural labor on the older adult heavy, widening the gap between the financial capital of the older adult and that of their children, and damaging the selfesteem of the older adult (22,23). On the other hand, children's migration also has favorable effects on parents. Migration allows older people to spend more time with friends and to participate actively in social activities (24). Adult children are able to provide other forms of support to their parents, for example, the economic and knowledge transfers of migrant children can expand household budgets (1), promote risk management strategies, and increase access to health care (25,26). Migrating children significantly increase the willingness of parents left behind to participate in mutual support for the older adult and promote the development of new models of old-age care (27). Throughout the existing literature, it can be seen that established studies generally recognize the role of children's support on the mental health of the older adult, but related studies have not reached a consensus conclusion on the impact of children's support on the mental health of the older adult. And the mass migration of children has evolved rapidly in response to social change, which has caused an impact on the function of family care for thousands of years in China, and it is especially true in rural areas, and the impact of children's support on the mental health of rural older adults in this context, there is a paucity of relevant research, so this paper intends to use the data from the 2014 China Longitudinal Aging Social Survey to analyze the impact of children's support on the mental health of rural older adults with different migration status of children. --- Methods --- Data sources This paper uses data from the 2014 China Longitudinal Aging Social Survey (CLASS). The survey targets Chinese citizens aged 60 or older and covers 28 provinces (autonomous regions and municipalities directly under the central government) in mainland China. The questionnaires were collected through household interviews, and a total of 11,511 resident questionnaires were completed. The individual questionnaires for the older adult specifically collect information on the marital status, health status, retirement planning, economic status, family and children of the older adult aged 60 or older. This study mainly uses the information of older adult personal data in CLASS 2014, firstly screening the older adult population living in rural areas, and then screening the data according to the level of depression in the older adult, loneliness, children's support, and other important research content of this paper, and finally get the effective sample data of 4,085. In order to study the impact of intergenerational support on the psychological health of rural older adults in the context of children's migration, we defined migrant children as children whose residence are not in the same township as that of his or her older adult parents, and categorized older adults into two main groups based on their status of having or not having migrant children, namely, have not migrant children and have migrant children, with sample data of 1,047 and 3,038, respectively. --- Variable setting 2.2.1. Dependent variable: mental health The dependent variable studied in this paper is the mental health of older adults, which consists of two main dimensions: depression level and loneliness. The CES-D (Center for Epidemiological Survey-Depression Scale) is a common tool widely used around the world to screen for depressive symptoms in the general population, and has good reliability and validity (28,29). Nine questions from the CES-D were included in the CLASS questionnaire, by which depression levels were measured, the responses were assigned a value, and the scores of each question were summed up, with higher scores representing a more severe tendency to depression in older adults. Russell developed the third edition of the UCLA(University of California at Los Angels) Loneliness Scale (UCLA-3) in 1996, and it was found that the UCLA-3 had good internal consistency, test-retest reliability and discriminant validity (30,31). The CLASS questionnaire includes three UCLA-3 sub-questions, which were used to measure loneliness in this study, scored in the same way as the depression scale, and the scores for each question were summed, with higher scores representing higher levels of loneliness in older adults. According to the test, the Cronbach's alpha of the depression scale and loneliness scale were 0.917 and 0.918, respectively, with good reliability (Table 1). --- Independent variables: children's support Children's support is usually analyzed from the perspective of social support, which can be divided into economic support, labor force support and emotional support, economic support is the monetary support given to the older adult by their children, labor force support is the life care of the older adult by their children, and emotional support is the emotional comfort of the older adult by their children (32). Therefore, this paper categorizes children's support into three dimensions: financial support, life support, and emotional support. In the CLASS database, ask for information on up to five children of older persons. The older adult received intergenerational support from multiple children, but the questionnaire does not cover the depressive level and loneliness of the older adult when receiving support from a specific child. These two variables represent the overall psychological well-being of the older adult, and this paper examines older adults' holistic perceptions of intergenerational support from their children. If samples are paired with the older adult and a specific child, it would result in different children's support all pointing to the same mental health status of older adults, which is inconsistent with existing research findings and departing from the integrity of older people's cognition, so in this paper, children's support is summed to measure the intensity of support for the older adult. Financial support was examined and valued using the questionnaire question "In the past 12 months, has a child given you any money, food, or gifts, and what was the total value of these items?", life support was measured and valued by the question "How often in the past 12 months has a child been able to help you with household chores?", and the emotional support was measured and valued by the question "Do you feel that this child does not care enough about you?". The questionnaire asked up to five children about their support for the older adult, and the support of each child was summed to get the scores for the intensity of the children's financial support, life support, and emotional support, respectively. --- Control variables Based on research needs, the control variables in this paper cover the main socio-demographic characteristics variables of the older adult, including age, gender, and marital status, the socio-economic status variables, including education level, and income-generating jobs, and the health status of the older adult, including self-assessed health status, chronic diseases, and the basic activity of daily living (BADL). The following table shows the scores of all variables (Table 2). --- Research methods The study selected the cross-sectional data of CLASS database in 2014, and through the variance inflation factor test, the VIF values are all less than 5, which means that there is no serious multicollinearity relationship among the variables, and the regression analysis can be performed by multiple linear regression model. In this study, the empirical analysis was conducted through a linear regression model with multiple independent variables, with depression level and loneliness in old age as dependent variables, and children's financial, life, and emotional support as independent variables. In the regression analysis, the multiple linear regression model was used to analyze the linear correlation between the dependent and independent variables, and the model is as follows: 0 1 1 2 2 n n Y X X X = <unk> + <unk> + <unk> +...+ <unk> + <unk> In the above equation, Y is the dependent variable, X 1, X 2......X n are the independent variables, <unk> is the random error term, <unk> 0 is the regression intercept, <unk> 1, <unk> 2......<unk> n are the regression coefficients. This study first analyzes the differences in mental health between the full sample and the sub-sample of rural older adults with and without children's migration in terms of depression level and loneliness using descriptive statistics to preliminarily determine the impact of child relocation on the mental health status of rural older adults. Then through the full sample multiple linear regression analysis of the mental health of rural older adult to explore the impact of children's support on the psychological status of rural older adult, and finally through the sub-sample multiple linear regression analysis to further explore whether there is a difference in the impact of children's support on the mental health status of rural older adult with or without children's migration. This paper uses Stata16.0 software for regression analysis. --- Results --- Descriptive statistics In terms of depression level and loneliness, the scores of the older adult without children's migration are higher than those of the older adult with children's migration. In terms of children's support, the children's financial support of the older persons without children's --- Regression analysis Specifically, from the regression results of the full sample of rural older adult, children's financial support has a significant negative effect on the level of depression and loneliness of the older adult population (p <unk> 0.01), with a greater effect on the level of depression. For every unit increase in children's financial support, the level of depression and loneliness of the older adult population decreases by 0.106 units and 0.038 units respectively, namely, the higher the number of children's financial support received by the older adult, the better the mental health of the older adult. In terms of children's life support, children's life support has a significant negative effect on loneliness in the older adult population (p <unk> 0.05). For every unit increase in children's life support, the loneliness of the older adult population were reduced by 0.017 units, namely, the higher the intensity of children's life support received by the older adult, the more the loneliness of the older adult was weakened. And children's emotional support had a negative effect on both depression levels and loneliness in older adults, but it is not statistically significant. In terms of control variables, male rural older adults have better mental health, rural older adults with higher education levels have better mental health, older adults with better self-assessed health and no chronic diseases have better mental health, and basic activity ability has a non-significant effect on the mental health of older adults (see Table 4). --- Sub-sample regression analysis of children's migration As the level of social support and medical insurance for older people in rural China is relatively poor compared with that in urban areas, children are often the main providers of support for their old age, and large-scale urban-rural migration has resulted in an increase in the separation of adult children from their older adult parents, which may disrupt this old-age care arrangement and affect the psychological well-being of older people in rural areas. Therefore, this paper constructs a grouped comparative linear regression model of the impact of children's support on the mental health status of the older adult population, to study in depth the factors related to the mental health status of the older adult population with migrant children. Regarding the effect of children's support on the mental health of rural older adults, the regression results showed that for older adults with children who migrated, financial support had a significant negative effect on the depression level of older adults (p <unk> 0.01), while for older adults without migrated children, financial support did not have a significant effect on the depression level of rural older adults, and the effect on loneliness was also no more significant than for older adults with migrated children. Life support has a significant negative effect on loneliness only for the rural older adult with children migrated (p <unk> 0.01). Emotional support from children has no significant effect on the depression level and loneliness of the older adult with or without children's migration, which is consistent with the regression results of the full sample (see Table 5). --- Discussion According to the results of the above empirical analysis, the higher the level of financial support provided by children, the more favorable it is for the older adult to maintain their mental health. Because rural China has long been characterized by a family care model for old age, with low levels of social security and support, the financial support of children has a greater impact on the quality of life of the older adult. In addition, the rural older adult have a stronger sense of raising children to provide against old age and increased emphasis on children giving back to their families, thus the more financial support they receive from their children, the more the older people's sense of self-esteem and fulfillment is fulfilled, which reduces their negative emotions and makes it easier for them to keep their moods happy, the result that is in line with the previous study (33). The effect of children's financial support on mental health was not as significant in the older adult without children's migration as in the older adult with children's migration, which may be since rural older adult without children's migration receive lower levels of material support from their children, resulting in a weaker contribution of financial support to the health of the older adult. The higher the level of life support from children, the higher the level of mental health of older persons. Older adult people's physical fitness and self-care ability gradually decline due to physiological reasons, and they urgently need the help of their children, and their children's care is more capable of satisfying their life needs compared to outsiders, thus the higher the level of psychological health of the older adult, and previous studies have also arrived at similar results (34). But in analyzing the heterogeneity of older adults according to the presence or absence of child migration, it was found that receiving life support from children effectively reduced the loneliness of older adults with migrated children, while it did not have a significant effect on the mental health of older adults without children's migration. On the one hand, it may be because the level of children's life support is higher for the older people without children's migration, and children's life care for the older people to some extent reduces the strength of the financial support provided by the children, which makes its effect on the mental health of the older people not obvious. On the other hand, in terms of emotional reasons, it may be because older people with children's migration receive life support from their children, which means that the children may return home to visit their parents, and the meeting with their children makes the rural older people feel happy, and thus the life care has a more significant impact on the mental health of the older people. The effect of children's emotional support on the mental health of the older adult is not significant. This may be since older people living in rural areas have a restricted social circle and have a stronger need for emotional support from their children, but the "sense of loss" caused by the discrepancy between subjective needs and objective support may substantially reduce the role of emotional support in lowering the level of depression and diminishing the sense of loneliness. Based on the results of the above analysis, it can also be seen that older people's own characteristics have a greater impact on their levels of depression and loneliness than the impact of children's support on their mental health, and that individual characteristics explain most of the variation in older people's negative emotions. The strength of this study lies in the large size of China's rural population and the representativeness of research on the mental health of older adults. The limitations of this study mainly lie in the following: firstly, this paper only uses the simple method of multiple regression analysis to analyze the correlation between children's support and mental health of rural older adult, and future research will try to use other methods to do a more comprehensive analysis. Secondly, although control variables were added where possible, there may still be omitted variables, and future research should test for potential confounders or mediators. --- Conclusion Based on data from the 2014 CLASS, this study used multiple linear regression to investigate the impact of intergenerational support from children on the mental health of older adults in rural China and analyzed the heterogeneity for older adults with different children's migration status, and the following findings were derived from the empirical analysis: (1) The higher the level of financial support provided by children, the more favorable it is for rural older adults to maintain their mental health, which is more pronounced among the older adults with children's migration. (2) The higher the level of children's life support, the higher the mental health of rural older adult, but there is no significant effect of children's life support on the mental health of the older adult without children's migration. (3) Individual characteristics of older people have a greater impact on their mental health, the mental health of the rural older adult is higher among those who are male, have good marital status, high education level and better physical health. According to the results of the above research, this paper draws the following results: Firstly, to better play the role of rural family care, rural grass-roots governments should actively publicize the culture of filial piety, forming a good atmosphere of love and respect for the older adult, and encouraging their children to provide support to meet the needs of the older adult. Secondly, the rural social security system should be improved, the level of rural old-age insurance should be raised appropriately, the level of medical services in rural areas should be raised, and village collectives should be used as the basis for introducing modes of onsite service and collective old-age care to satisfy the personalized needs for older adult care of the rural older adult, and to promote the sustainable development of both collective and family old-age care. Thirdly, social forces should actively utilize their advantages in resource allocation and combination to provide the rural older adult with cultural and recreational activities, legal counseling, mental health counseling, and other forms of services tailored to different needs, to ensure that the older adult have a sense of worthiness and enjoyment, and to continually improve the level of mental health of the older adult in rural areas. --- Data availability statement Publicly available datasets were analyzed in this study. This data can be found at: http://class.ruc.edu.cn/index.php?r=index/ index&hl=en. --- Author contributions JL: conception and design, funding acquisition, and preparation. MJ and JL: methodology, writing-original draft and supervision. MJ, JL, ML, and AW: writing review and editing. All authors contributed to the article and approved the submitted version. --- Conflict of interest The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. --- Publisher's note All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher. --- Supplementary material The Supplementary material for this article can be found online at: https://www.frontiersin.org/articles/

Background: Family old-age care is dominant in Chinese rural society, and children's support is an important force in family old-age care. However, the migration of a large number of young and middle-aged rural laborers has undermined the traditional arrangements for old-age care in rural areas and affected the psychological health of the older adult. Methods: 2014 China Longitudinal Aging Social Survey targets Chinese citizens aged 60 or older and covers 28 provinces in mainland China. In this paper, the database of the CLASS was selected for empirical analysis to explore the impact of children's support on the depression level and loneliness of rural older adults through multiple linear regression, and was divided into two groups according to children's migration to analyze heterogeneity. Results: Children's financial support facilitates the maintenance of mental health among rural older adults. Children's support promotes mental health among rural older adults, but this association does not exist among older adults without children's migration. Individual characteristics of older people have a greater impact on mental health. Discussion: Our study firstly compares the differences of children's migration status between children's support and mental health among the older adult in rural China. In order to improve the mental health of the older adult, it is necessary to create a favorable atmosphere of love and respect for the older adult, improve the social security system in rural areas, and give full play to the strengths of the social forces, so as to ensure that the older adult have a sense of worthiness and enjoyment in their old age.

Introduction Caring for elderly relatives represents a serious issue facing caregiving family members in an aging society. Although caregiving has also a positive aspect for the family members (1,2), the burden of informal care underscores the need for new methods of enacting sustainable long-term care. Further, in addition to the burden of care experienced by family caregivers due to the care receiver's disease, such as impaired eyesight (3), pressure ulcer (4) and underlying disease (5), the relationship between care receivers and caregivers has also been shown to play a crucial role in the caregiving process (6)(7)(8)(9). Informal care dynamics are complex in the social context of a family (10). Family dynamics, the interactions between family members, and the varied relationships that exist within a family involved in long-term care have been suggested to be influential on the care of frail elderly people; however, these parameters remain difficult to measure (11). Therefore, we previously developed a validated scale (the Index of Family Dynamics for Long-term Care; IF-Long) to describe the relationship between primary caregivers and care receivers before caregiving (12). Japan is the world's fastest aging society, with one-quarter of the entire population aged over 65 years in 2013 and a particularly high proportion aged over 75 years (12.3%) (13). Rapid shifts in demographics have created a situation in which many elderly people receive long-term care from family members who themselves are elderly, termed 'elder-to-elder care' (14). The nursing-care insurance system was publicly established in 2000 to support this situation (15). In this scheme, the municipality review board assesses the physical and cognitive function of older people to determine their eligibility and the degree of care required on an individual basis, which by definition defines the extent to which they can avail themselves of formal care services, including but not limited to home support care, personal care, nursing care and respite care (16). Although this system has reduced the burden on family members to some degree (17), informal family caregiving is still expected, partly because the number of formal care services provided by the nursing-care insurance system is not generally sufficient to cover all the care needs of frail elderly people living at home and partly due to the deep-seated desire among many older people to receive care from their family members instead of professionals. This rising demand for long-term care in an aging society and the limitations of resources within the field of longterm care have underscored the global importance of the increased burden on family members in caring for their elderly relatives. With their holistic perspective on the situation, primary care physicians play an important role in designing sustainable longterm care systems. Given the increasingly important role of family members in caring for elderly relatives in an aging society, future long-term care systems should account for the relationships between family members and their charges. Here, to clarify the influence of family dynamics between primary caregivers and their frail elderly family members on the burden of care experienced by the primary caregiver, we measured these dynamics using our validated scale of IF-Long score (12) and examined its association with burden of care, adjusting for the characteristics of care receivers and caregivers. --- Methods --- Study design We conducted a multicenter cross-sectional study at six primary care clinics between July and September 2011. Data used in the study were obtained from multiple sources, as follows: (i) Primary caregivers answered self-administered questionnaires about the nature of caregiving and their burden of care, (ii) The care receiver's primary care physician assessed the family dynamics and the receiver's functional and cognitive status and (iii) The care receiver's care managers provided information on the utilization of formal care services. --- Participants and setting We sequentially selected caregivers who mainly cared for their 'frail elderly' family members receiving outpatient or home treatment due to chronic disease at any of six primary care clinics. We excluded caregivers who cared for family members being treated as inpatients or living in a nursing home. 'Frail elderly' in this study was defined as either individuals who were approved to be eligible for nursing-care insurance by the municipality review board, or other older people aged <unk>65 years who either needed some mobility assistance at outpatient clinics or had been diagnosed with cognitive impairment by at least two physicians. --- Main exposure: family dynamics Our main exposure was family dynamics, as measured by IF-Long (12). This scale was initially developed for use by clinicians, generally primary-care physicians, with a long-standing relationship with the care receiver and caregiver, as it assesses problems or undesirable situations that might be alleviated via 'intervention through a family system', such as family consultations or family meetings. The IF-Long comprises four items inquiring about the existence of certain situations between the primary caregivers and care receivers, as follows: 'Disengagement (a situation in which the caregiver feels no sense of fulfilment or accomplishment from caregiving, or feels no joy from the patient's gratitude)', 'Scapegoating (a situation in which the caregiver believes that the family would be happy if only the caregiving were easier),' 'Transfer of problems across generations (a situation in which the care receiver compares the present with his or her past experiences as a caregiver, and, on that basis, is unnecessarily demanding of the caregiver)' and 'Undesirable behaviour patterns (a situation in which the care receiver's daily life has become very dependent on the caregiver and can be sustained only by the caregiver, and in which the caregiver similarly feels dependent or reliant on the care receiver)'. Response options for these four items ranged from 0 (not present) to 3 (definitely present), with IF-Long scores calculated as sums of response values to the four items (possible score range: 0 to 12). Higher IF-Long scores indicated poorer relationships between care receivers and their caregiving relatives. We tested the reliability (0.73) and validity (construct validity and concurrent validity) of the IF-Long scale in a previous study. Care receivers in the present study were divided into three categories based on IF-Long score tertile (best <unk>2, intermediate 2 to <unk>5, worst <unk>5). --- Main outcome: burden of care for caregivers Our main outcome was burden of care for caregivers, which was evaluated using the burden index of caregivers (BIC) (18). The BIC comprises 11 items covering five sub-domains (Time-dependent Burden, Emotional Burden, Existential Burden, Physical Burden and Service-related Burden) and one domain of total care burden. BIC scores were calculated as the sum of the response values to the 11 items (possible score range: 0 to 44). The BIC was assessed via a selfadministered questionnaire for the caregivers. --- Other factors Information on the characteristics of the caregivers and caregiving (age, gender, employment status, frequency of hospital visits by the caregiver, time required for care and duration of care) was obtained from the caregivers via questionnaire. Care receivers' degree of cognitive impairment was evaluated using Functional Assessment Staging (FAST) (19), and activity of daily living was assessed via mini Functional Independent Measure (mini-FIM) (20) by the primary care physicians. Care managers, who create the care plan for care receivers and their family, provided further information about usage of formal care services. --- Data management and ethical issues We obtained written informed consent from the care receivers and their family members. Because we used an anonymous questionnaire, we printed unique identification numbers on the questionnaire paper and then used the number to link data obtained separately from caregivers, primary care physicians and care managers. Researchers in charge of statistical analysis were blinded to care receivers' clinical data. This study was approved by the institutional review board of Kyoto University. --- Statistical analysis We described caregiver's characteristics and care receivers' mini-FIM and FAST scores based on the three IF-Long score categories. We then calculated the mean BIC score for each IF-Long score category, and trend tests were performed to examine the dose-dependent association between IF-Long score and BIC score. We constructed a linear regression model to estimate adjusted mean difference of BIC score between categories of IF-Long score. The reference category of IF-Long score was the best IF-Long group (score <unk>2). We included caregiver age and gender, duration of care, time required for care and care receiver's mini-FIM score and FAST score in the model to examine the association between categories of IF-Long score and BIC score after adjusting for potential confounders. In our initial sensitivity analysis, we defined two categories of IF-Long score (Good <unk>5, Bad <unk>5) and estimated probability (propensity score) of being in the bad IF-Long group. We then used the inverse probability of treatment weighting method (IPTW) (21) to adjust for potential confounders. In a subsequent sensitivity analysis, we used the multiple imputation method using chained equations (22) to impute missing values of 'duration of care' (n = 5) and 'time required for care' (n = 13). All analyses were performed using STATA software (version 14.0; STATA, College Station, TX), with two-sided significance set at 0.05. --- Results --- Characteristics of caregivers and care receivers We asked 249 caregivers to answer the questionnaire and received 199 answers (collection rate, 80%). Tables 1 and2 show characteristics of caregivers and care receivers overall and by categories of IF-Long score, respectively. Among all caregivers, mean age was 63.2 years, with 40% of caregivers aged over 65 years and 19% of caregivers aged over 75 years. A total of 79% of caregivers were female, and 64% required regular visits to the hospital themselves. Mean duration of caring for a family member was 5.2 years, and mean time required for care was 32.3 hours per week. A total of 43% of caregivers cared for their family for more than 3 hours a day. Time required for care was greater in the worst IF-Long group than in the best IF-Long group. In terms of care receivers' characteristics, mean mini-FIM score was 33.9, and 20.1% of care receivers had severe cognitive impairment (FIM score <unk>6). IF-Long score varied between 0 and 10 points, with a mean of 2.4 (SD, 2.1). The mean scores of the four items of the IF-Long showed that the worst/best IF-Long group indicated the worst/best in all of the items. Burden of care by categories of family dynamics --- Sensitivity analyses We re-defined two categories of IF-Long score (good and bad group) and examined the association between group assignment and BIC score using IPTW to adjust for potential confounders more appropriately. Associations between IF-Long score and BIC score in this instance were found to be similar to original results, and the mean difference between the two groups was 2.70 (95% CI, 0.12-5.29). In our second sensitivity analysis using the multiple imputation method, associations between IF-Long score and BIC score remained similar to original results after imputing missing data. Mean differences were 1.52 (95% CI, -0.62 to 3.66) and 4.36 (95% CI, 1.19-7.53) in the intermediate and worst IF-Long groups compared to the best IF-Long group, respectively. --- Discussion In summary, we found that family dynamics strongly influence burden experienced by primary caregivers, regardless of the carereceiver's degree of cognitive impairment. Severe issues straining the family dynamics (Worst IF-Long score) were found to exacerbate the burden of care (high BIC score). We also detected a dose-dependent association between IF-Long score and BIC. Taken together, these findings suggest that the relationship between care receivers and their primary caregivers in place before caregiving should be carefully considered when deciding on a care plan for care receivers. In the present study, we found that caregiving family members tended to be elderly themselves, with 40% of caregivers aged over 65 years. Notably, the mean age of caregivers was similar to national figures for Japan (23), indicating that our findings were not unique to our study or contaminated by outliers. In addition, approximately two-thirds of caregivers themselves visited the hospital regularly to receive treatment for chronic diseases. Given this situation in which elderly individuals care for other elderly individuals, the burden of care is clearly a key factor for consideration when developing a sustainable care system in an aging society. --- Implications for practice and education Measuring family dynamics using the IF-Long score may help primary care physicians identify issues involving these dynamics when developing long-term care plans, thereby allowing physicians to support family members through education and family conferences involving all immediate family, relevant extended family, significant friends and members of the patient's support network. Developing sustainable long-term care in an aging society will require greater emphasis on 'family-oriented care', in which physicians describe the framework of the family and analyse what is happening to the family with respect to their roles in caring for other family members/ themselves. Physicians should also evaluate the relationship between family members with respect to the family life cycle and identify potential relationship problems among members. Through such family-oriented care, physicians can determine family dynamics and design specific interventions including target family members and specialized approaches to communication (11). Of note, the adjusted mean difference of BIC between the worst IF-Long group and best IF-Long group was similar to that between the normal cognitive group and severe cognitive impairment group as defined by FAST score. Previous studies have shown the impact of FAST on care receivers' adverse events of mortality (24) and unplanned hospitalization (25). These results may therefore suggest that family dynamics are as influential as severity of care receiver's cognitive impairment on family members' burden with long-term care. --- Strengths and limitations Our study had several major strengths. First, we used our validated IF-Long score to quantitatively evaluate family dynamics, an important but difficult-to-measure concept in 'family-oriented care'. Second, we conducted this study at primary care clinics in Japan-a rapidly aging society. This setting ensures that our results can be easily extrapolated to other developed countries with similar aging societies. Third, we obtained consistent results even after using IPTW to adjust for confounding. However, several limitations to the present study also warrant mention. First, given the cross-sectional nature of our study, we can't eliminate the possibility of reverse causality. Specifically, a poor family relationship may not be the cause of heavy burden of care, but instead the result. Indeed, we observed that the time required for care was longer for caregivers in the worst IF-Long group than among those in the best group. However, the IF-Long questionnaire asks physicians, who can determine a familial relationship based on information obtained from the care receiver and their caregiving family during outpatient care, to describe the relationship before caregiving. Second, we lacked precise data on what kind of care was given to care receivers. Although we adjusted for the measured covariate of 'time required for care', marked differences in the contents of care between the worst and best IF-Long group may confound the association between IF-Long score and BIC. However, we believe that adjusting for care receiver's characteristics (ADL and cognitive impairment) may reduce the effect of residual confounding. In addition, results remained consistent even after adjusting for those confounding factors using IPTW. Given the above limitations, interpretation and generalizing our results should be done with care. --- Conclusion We found that family dynamics strongly influences the burden experienced by caregiving family members, regardless of the care-receiver's degree of cognitive impairment. These results may underscore the importance of evaluating family dynamics between care receivers and their family when discussing burden of care. --- Conflict of interest: none.

Background. Long-term care for the elderly is largely shouldered by their family, representing a serious burden in a hyper-aging society. However, although family dynamics are known to play an important role in such care, the influence of caring for the elderly on burden among caregiving family members is poorly understood. Objective. To examine the influence of family dynamics on burden experienced by family caregivers. Methods. We conducted a cross-sectional study at six primary care clinics, involving 199 caregivers of adult care receivers who need long-term care. Participants were divided into three groups based on tertile of Index of Family Dynamics for Long-term Care (IF-Long score), where higher scores imply poorer relationships between care receivers and caregiving family: best, <2; intermediate, 2 to <5; worst, ≥5. The mean differences in burden index of caregivers (BIC-11) between the three groups were estimated by linear regression model with adjustment for care receiver's activity of daily living and cognitive function. Results. Mean age of caregivers was 63.2 years (with 40.7% aged ≥ 65 years). BIC-11 scores were higher in the worst IF-Long group (adjusted mean difference: 4.4, 95% confidence interval: 1.2 to 7.5) than in the best IF-Long group. We also detected a positive trend between IF-Long score and BIC-11 score (P-value for trend <0.01).Our findings indicate that family dynamics strongly influences burden experienced by caregiving family members, regardless of the care receiver's degree of cognitive impairment. These results underscore the importance of evaluating relationships between care receivers and their caregivers when discussing a care regimen for care receivers.

Research has established that housing is a health issue and that the worse the housing situation, the worse the impact on health. Those experiencing homelessness, particularly chronic homelessness, for instance, experience high rates of physical health issues, chronic diseases, substance abuse, and mental health challenges. Furthermore, health issues cause as well as are exacerbated by homelessness. 1 Housing that is poor quality or in disrepair increases the incidence of injuries, asthma, cancer, cardiovascular disease, and other health issues. Living in unhealthy housing also worsens mental health, in part due to the stress of living in unsafe and unhealthy environments. Conversely, affordable housing reduces negative health outcomes, because a lower proportion of income is dedicated to housing and greater proportion is available for food and health care. 2 Accessing healthy living conditions is an economic issue for those most likely to be living in poor housing conditions that negatively impact health. To improve health and housing conditions for extremely low-and low-income families requires addressing both the availability and the access to healthy housing. According to the Housing Alliance Delaware report, 3 Delaware is short more than 21,000 additional affordable housing units to meet the needs of extremely lowincome families. Second, affordable housing must be affordable. Low income and extremely low-income Delawareans need to work more than two minimum wage jobs to afford a Fair Market Rent apartment. Even if this was reasonable, the path to attaining affordable and quality housing is interrupted by the benefits cliff. The benefits cliff, in simple terms, is the point on the economic mobility path where those dependent on governmental benefits and entitlements increase income but not enough to cover the additional expenses associated with the loss of benefits. The benefits cliff emerged as an unanticipated consequence of implementing the Personal Responsibility and Work Opportunity Reconciliation Act of 1996 (PRWORA), commonly referred to as Welfare to Work. The welfare reform legislation aimed to reduce dependency on government assistance and to increase self-Doi: 10.32481/djph.2023.06.010 sufficiency. Initial studies on the impact of PRWORA showed significant increases in employment among welfare recipients; however, the employment was unstable and temporary. 4 Studies looking at the impact of individual benefit programs on work show that food stamps 5 and housing 6 have negative effects on incentives to transition from welfare to work. On the other hand, several studies show that although childcare assistance appears to have the most positive impact, it also has the largest cliff effects. 4 An extensive study in Colorado shows how families dependent on government benefits "strategize" to avoid the benefits cliff and identify the most reasonable and least risky path toward self-sufficiency. 7 Prenovost and Youngblood documented the stress created trying to navigate the benefits cliff. 8 Significant attention has been given to identifying the point where work pays off, referred to as the effective marginal tax rate (EMTR). 4,9 The EMTR is the percentage of any additional earned income that a household pays in taxes or loses in government benefits. As Dolan notes, the EMTR creates a poverty trap where the risks and gains of working become disincentives and barriers to self-sufficiency and economic mobility. 9 EMTR analyses have produced various models of economic mobility out of welfare to self-sufficiency, including modeling combining various types of benefits. 4,9 Typically, however, EMTR modeling fails to take into account factors associated with the readiness for work. As Hauan and Douglas point out, most welfare recipients face at least one barrier and many encounter multiple barriers preventing readiness for stable and sustainable work. 10 They identified the most common barriers, in order of most frequently identified, as not having earned a high school diploma or GED, unmet childcare needs, mental health issues, transportation issues, limited work experience, unstable housing, and physical health problems. This research suggests that navigating the benefits cliff requires welfare recipients meet a set of criteria considered necessary for stable and sustainable work, a status necessary to access healthy housing. Assisting residents dependent on government assistance become eligible and prepared to cross the benefits cliff to become economically self-sufficient and eligible for housing that does not negatively affect health has become a critical issue in comprehensive revitalization efforts. Clearly, building new, healthy housing is not enough. Extremely low-and low-income families need significant assistance becoming economically eligible for the housing. REACH Riverside's experience demonstrates the interconnected relationship among poverty, health, and housing. The REACH Riverside Development Corporation was formed in 2017 to remove the barriers to prosperity and well-being facing residents of Wilmington's Riverside neighborhood, a community experiencing the devastating impacts of decades of racial discrimination and disinvestment. REACH is the managing partner of The WRK Group, a group of three nonprofits (REACH Riverside, Kingswood Community Center, and The Warehouse) working together to implement the Purpose Built Communities model, a holistic approach focused on housing, education and health initiatives. REACH is leading a comprehensive revitalization of the Riverside community in NE Wilmington, which includes creating 700 units of mixed-income housing (Imani Village) along with education and health oriented initiatives. Phase one of Imani Village (74 units) was completed in December 2022. Construction of Phase two was finished in May 2023, bringing another 67 units on stream. Five hundred fifty two more rental units are planned in Phases three through seven, along with about 100 homes for ownership. In support of this transformation in the built environment, REACH works closely with service partners Kingswood Community Center and The Warehouse to provide social service programs for all Doi: 10.32481/djph.2023.06.010 ages. This has included creating a cradle-to-college/career pipeline between Kingswood's Early Learning Academy, EastSide Charter School, and The Warehouse as well as providing on-site medical services to the community through a partnership with ChristianaCare. As with all the WRK Group efforts, the new housing plan prioritizes current residents living in Wilmington Housing Authority (WHA) housing. In assessing eligibility for the new housing, approximately 40% of the WHA families were not eligible primarily as the result of past due rent and utility bills. REACH worked with Kingswood Community Center to enroll residents in Kingswood's EMPOWER program. EMPOWER assists residents of Riverside and surrounding neighborhoods obtain economic self-sufficiency. At intake, EMPOWER clients with assistance from their navigational coach complete a comprehensive assessment. A 2023 analysis of intake assessment data showed that over half of the EMPOWER clients' income situation, employment situation, and/or food situation were assessed as in crisis or vulnerable. The credit situation for one third of the clients was in crisis or vulnerable and one fourth were living in housing that was assessed as in crisis or vulnerable. The large majority of EMPOWER clients are WHA residents and for them in crisis or vulnerable housing situations indicates that they are at risk of eviction. The intake assessment data underscores the multiple programming needs of extremely low-and low-income families. One of the biggest barriers faced by EMPOWER staff and clients in developing plans to achieve economic self-sufficiency to meet the inter-related and multiple needs is the benefits cliff. EMPOWER clients, like many extremely low-and low-income families, cannot simply work more or obtain better paying jobs to become economically independent because they risk losing benefits, most significantly those related to childcare and healthcare. This barrier to economic stability and self-sufficiency also creates barriers to accessing the new healthy housing designed specifically for Riverside residents. The EMPOWER staff and the REACH leadership are actively exploring options for bridges across the benefits cliff but these bridges are, in part, dependent on policy changes. Programmatic developments without complimentary and supportive policy changes will be ineffective and possibly infeasible. Policies to avoid the benefits cliff or mediate its negative effect on the transition to selfsufficiency are limited and vary greatly by state. The National Conference of State Legislatures (NCSL) 11 and the Federal Reserve 12 produced reports defining and describing the benefits cliff. The Alliance for Strong Families and Communities categorizes policy efforts into four categories: 1) policies modeling or adapting the Earned-Income Tax Credit; 2) policies ensuring continued access to benefits; 3) policies eliminating some assets, such as cars and savings accounts, in deciding benefits levels; and 4) policies excluding portions of income in benefits calculations. 13 The NCSL report provides an overview of state level efforts addressing the benefits cliff. The report notes that 16 states enacted legislation to mitigate the impact of the benefits cliff. Delaware was not one of those states and to date has not proposed benefits cliff mitigation legislation. The most comprehensive initiative, undertaken by five New England states, piloted both policy and program level changes to mitigate the negative effects of the benefits cliff. 14 Most state level efforts continue to address welfare policies individually when the reality is recipients must navigate the benefits cliff considering all sources of assistance. Dolan 9 and Richardson and Blizard 4 suggest that the current benefits system, rather than policies, is structured in a way that discourages participants from becoming fully self-sufficient, prohibiting economic mobility. Dolan highlights the need to create individualized approaches because there Doi: 10.32481/djph.2023.06.010 is a matrix of benefits, income sources, and personal factors that must be considered for a family to traverse the benefits cliff. 9 Richardson and Blizard are bolder in their recommendations, suggesting terminating all social programs as an option. 4 Other options included creating a longer glide path, universal basic income, and private sector creation of non-taxable fringe benefits, such as childcare assistance. Addressing the housing crisis in Delaware as well as improving health and economic stability demands that we address the benefits cliff looming on the horizon for extremely low-and lowincome families struggling toward self-sufficiency. Effective policies and interventions, at a minimum, need to be informed by the experiences of those struggling to navigate the benefits cliff and achieve economic mobility. In other words, policies need to data driven. And, effective policies and interventions must be integrated. Siloed policy development will not effectively address the benefits cliff preventing economic self-sufficiency and access to affordable, healthy housing. There is not a single pathway from dependence on government assistance to selfsufficiency. Similarly, there is not a linear continuum of housing options paralleling a linear continuum of economic mobility. Multiple entrance ramps with varying lengths across the benefits cliff need to be created in collaboration with housing, social services, healthcare, educational, workforce, government, and private sector partners. Given our size and promising comprehensive revitalization efforts, Delaware has an opportunity to model integrated policy changes that bridge the benefits cliff and support extremely low-and low-income Delawareans on their journey toward economic self-sufficiency, health, and affordable, quality housing. Dr. Dillard may be contacted at ddillard@desu.edu.

This Commentary introduces the reality of the benefits cliff into the housing and health relationship. Improving both housing and health requires addressing the benefits cliff facing extremely low-and low-income people. Those living below the poverty line are at greatest risk of poor, inadequate, and unhealthy living conditions. They are also the group most at risk for losing benefits and economic supports on their journey toward economic self-sufficiency. Addressing the housing crisis in Delaware demands that we recognize the benefits cliff looming on the horizon for extremely low-and low-income families struggling toward self-sufficiency. Effective policies and interventions, at a minimum, need to be informed by the experiences of those struggling to navigate the benefits cliff and achieve economic mobility and need to be integrated across housing, economic benefits, and health care.

Introduction Branded as the 'black swan' event (Renjen, 2020), the COVID-19 pandemic has inflicted substantial harm not only on the healthcare sector but also on the broader economy. Within the global economy, encompassing primary, secondary, and tertiary sectors, the pandemic's disruptive impact has been felt to varying degrees, each contingent on the market's demand and supply dynamics. Among these, the tourism industry, falling within the tertiary sector, stands out as one of the hardest hit due to the ongoing crisis. Consequently, the turmoil in the tourism industry has cascaded to affect a multitude of stakeholders, with social media travel influencers prominently among those bearing the brunt. This research article aims to address two primary research questions: (i) What were the consequences of the pandemic on Indian Instagram travel influencers? and (ii) What strategies did Instagram travel influencers in India employ to navigate the global crisis? The significance of this research is multifaceted. Despite the tourism industry being severely impacted by the pandemic, there exists a dearth of literature that specifically addresses the issues faced by its stakeholders, particularly influencers. In its light, this study serves as a platform to raise awareness among both the public and government about the numerous challenges faced by the Indian travel influencers during the global crisis. The study also illuminates innovative strategies employed by travel influencers, offering valuable insights for other influencers and individuals considering a similar career path but who may be hesitant due to pandemic-related uncertainties in the future. The article is divided into x sections. The first section illuminates the impact of the global crisis on the tourism industry, including the travel influencers. It walks the readers through Peter Burn's stakeholder map (2013) and the stakeholder theory by Freeman (2009). The second section explicates the methodology, including the construction of the social survey, followed by data collection and analysis procedures. In the third section, the ramifications of the pandemic on the travel influencers are discussed, followed by the distinct coping strategies used. The fourth section suggests a way forward, focusing on some innovative coping strategies as adapted from the existing literature, followed by the final section underscoring the directions for future research. --- Impact of COVID-19 on Tourism industry In the backdrop of COVID-19, which further divided the already polarised world and accentuated the various social inequalities, the work-from-home regime coupled with the social distancing norms, became the new normal. With over 59 million infected cases globally by the end of November 2020, of which nearly 9.1 million cases belong to the second-largest pool of coronavirus, India (Worldometer, 2020), the present impediment appeared to have an unknown expiry date. This was when the pandemic was at its peak, and numerous industries faced a setback, including the tourism industry. The tourism industry which accounted for 10.3% of the global Gross Domestic Product (GDP) and 6.6% GDP in South Asia in 2019 (Economic Impact Reports, 2020), was amongst the hardest-hit ones in 2020 (Nicola et al., 2020). The World Travel and Tourism Council extrapolated a loss of nearly 50 million jobs in the tourism industry, all attributed to COVID-19, with Asia being the largest sufferer (Faus, 2020). Also, the United Nations (UN) predicted a 60-80 % loss in international tourism due to the turmoil ("The Economic Times," 2020). The blow faced by Indian tourism on account of COVID-19 cloud is in sync with the global trend, as is evident through the systemic risk survey, which mentioned it to be one of the five adversely hit sectors, with over 90% of the respondents estimating its stagnant recovery in the next six months (Reserve Bank of India, 2020). National tourism, which generated 8.1% of the total employment (Moolya Foundation, 2020) and contributed 9.2% of India's GDP in 2018, the third among the other 185 countries (Impact of COVID-19 on Indian Tourism Industry, 2020), was forecasted to incur a monetary loss of 75% in the Financial Year (FY) 2020-21 in their business (BW Online Bureau, 2020). These statistics throw light on the profound impact of COVID-19 on the tourism industry. The deep wounds on India's tourism tend to perpetuate the pain to its stakeholders, who can easily be identified through Peter Burns's stakeholder map (Burns, 2013), as given in Figure 1. --- Travel Influencers: A stakeholder in the Tourism industry From figure 1, we find that the travel influencers occupy a place of primary stakeholders in the tourism industry, both in the form of developers and tourism business owners and managers. Their experiences and reviews are believed to be trusted more by the people in comparison to other online reviews (Influencity, 2018). Moreover, R. Edward Freeman through his'stakeholder theory' has explained that the only way for a declining business to revive is by managers and entrepreneurs accepting and promoting the interconnections and interdependence between various stakeholders, instead of letting them thrive in an isolated ecosystem (Freeman, 2009). The study emphasises on the application of this theory in the tourism industry as well. --- Who are travel influencers? Influencers are themselves the consumers but with the fervent following, who hold the tendency to shape the opinion of their followers as they have greater expertise, knowledge, or experience about the product (Gretzel, 2018). We have observed a paradigm shift from traditional influencers being just celebrities (Glover, 2009) to the recent trend of influencers on social media. In today's era of the tech-savvy world, social media influencers have become an ingrained part of the social fabric of any society. With the Finnish government calling social media influencers as 'critical actors' to help flatten the curve during the pandemic, alongside 'doctors, bus drivers, and grocery store workers', they play a pivotal role in providing digital information to its followers in real-time, before any other form of media, or where government communication is not accessible (Heikila, 2020). According to (Abidin et al., 2020), influencers perform a rainbow of activities like promotion of digital economy, dissemination of information, and maintaining public relations. A subset of influencers is the travel influencers on social media whose niche lies in promoting the tourism industry through their content which may take the form of social media posts, blogs, or vlogs. These digital-first personalities (Hutchinson, 2019) are a major player in today's information ecosystem. They usually post their live travel stories and pictures on Instagram, Facebook, Youtube, etc. and are not hesitant to show their back-stage behaviour, a term coined by Erving Goffman (Cole, 2019), which in turn creates an imprint of legitimacy in their followers (MacCannell, 1973). A study by Lalangan (2020) reveals that it is not just the influencers who shape the content, but, the content also, in turn, motivates travel influencers to earn a living through their hobby of travelling, in addition to having an impact on the audience. These'social media celebrities' may further collaborate with a few brands or companies, and promote them in accordance to their experiences with it (Baruah, 2017). Their importance to tourism can be drawn from considering the example of Indonesian government allocating US$ 5.2 million to influencers to boost and safeguard tourism from the web of COVID-19 (Gorbiano, 2020). --- Operational Definitions For the purpose of the present study, Travel influencers and Coping strategies are defined as following. 1. Travel influencers: Here, travel influencers refers to those Indian influencers who promote travelling on Instagram through content creation in the form of posts, blogs, vlogs, or Youtube videos (shared through Instagram). It includes the scope of both who have taken this as a profession as well as as a hobby. Further, it covers tourism business managers and owners who promote their travel businesses through Instagram. Conclusively, it refers to all the respondents who had responded to the social survey. --- Coping strategies: In the present study, coping strategies include all the behavioural and psychological efforts put by the travel influencers to minimise the impact of COVID-19 on them. It was measured through the self-developed google form. --- Methodology Social survey, a type of descriptive survey study was conducted to obtain data from the respondents. The theoretical framework finds its place through the works of Peter Burn and R. Edward Freeman. The population for the present study included all the Instagram travel influencers in India. The sample was selected by employing the two types of non-probability sampling techniques, i.e., snowball and purposive sampling. It included the 55 Instagram travel influencers in India who had responded to the google form on Instagram and email. The google form sent to the respondents was in the form of a questionnaire, including a total of 14 items. It included both open and closed-ended questions. The responses to the open-ended questions were tabulated and organised in various categories using tally marks, which were further converted into frequencies and percentages. Similarly, the responses to closed-ended questions were also converted into percentages. Suitable diagrammatic and graphical representations were used to further analyse and draw conclusions from the information. The review conducted was used to support the various claims. Further, I added all the 55 respondents to my Instagram account and followed their posts for more than three weeks to get a clear picture. Hence, a mixed-method research was employed, i.e., QUAL-quan (Creswell & Clark, 2017). The 55 respondents were from different states of India and had a varied following, as shown in figure 2. Figure 2 reveals that nearly 87% of respondents in the present study had less than ten thousand followers. The profile of most of the respondents was public. While the majority of the respondents shared only travel content, a few respondents shared content other than their travel saga also, as shown in figure 3. 65.5% of respondents had their blogs/vlogs as well, though only 21.8% of respondents were making money through being an influencer. Here, it becomes important to mention a trend that was observed, wherein 91.67% of respondents who were earning through being an influencer had their blogs/vlogs. Further, only 10.9% of respondents had made this their profession, while the remaining 89.09% of respondents took it as a hobby. Interestingly, the study revealed that the number of followers did not have much effect on social media travel influencing to be taken as a profession, in contrast to the expected belief. It was also observed through the posts of the selected travel influencers that while some respondents were amateur, others were professional. They were monetising not just through their blogs/vlogs/websites, but also through collaborations, promotion of brands and products, paid campaigns, freelancing assignments, giveaways, and travel businesses. 69.09% of respondents incurred a financial loss due to the pandemic. As speculated, my intensive following of the respondents on Instagram confirmed that it was mainly due to the cancellation of their travel bookings and freelance assignments. Also, 66.67% of respondents from the ones who took it professionally were the ones to face the economic blow. The change in the followers' frequency during the pandemic is given in figure 4. Surprisingly, nearly 36% of respondents witnessed an increase in their followers in contrast to 6% of the respondents who lost them, as seen in figure 4. It indicates that maximum respondents were able to find out ways to engage their audience. According to @travelburpwithsurya, one of the respondents, "Creating content has become very difficult in this pandemic, but somehow we are managing it." The various coping strategies used by the respondents to get through the pandemic are given in figure 5. --- Research findings, analysis and interpretation 1. Participant demographics --- Ramifications of COVID-19 on the respondents --- Coping strategies The coping strategies given in figure 5 Coping strategies a. Responsible tourism: 60% of the respondents were still travelling despite the pandemic. Moreover, 83.33% of the respondents who had taken travel influencing as their profession were the ones to continue their travel. While @Mannu_Sambyal said that he was following all the Standard Operating Procedures (SOP's), @mussafir_hun_yaro and @visualsofmylife said that they were carrying the "safety gears" and "covid essentials" while travelling. It somehow helped in the adaptation of social distancing norms by their followers as well, thereby making it a new trend. Moreover, it was work from the home regime, which enabled a few of the respondents, like @visualsofmylife and @akhilesh1125, to travel more. b. Increasing content consumption: The respondents were posting the content, as shown in figure 6, to keep the audience involved. Figure 6 shows that majority of the respondents were sharing travel throwbacks, while nearly 36% of the respondents did share their present travelling content as well. Nearly 9% of the respondents changed their content from travelling to something else, in order to stay buoyant. Further, 12.7% and 1.81% of the respondents were posting advisory content and covid related information, respectively. With over 12% of the respondents sharing content related to lifestyle, and a few others sharing diverse content, like motivational quotes, reposts, photography, own pictures, and old unpublished travel stuff, respondents were doing all they could to increase the content consumption among the followers. It becomes vital to mention that 3.6% of the respondents were not updating anything at all. c. Going local: Nearly 27% of the respondents started bringing into limelight the local travelling places, with @dhritikapur rightly saying that it was "time to support local travel businesses." While @karavanlife started "exploring new places which are not crowded", @akhilesh1125 was "able to explore the overrated places during COVID-19 in a much better way." d. Saving money: 3.6% of the respondents saw COVID-19 as an opportunity to save money for future travels. @dhritikapur said that she was coping through the pandemic by "booking in advance to capture good deals for future travel." e. Positive outlook: Since every person has a unique defence mechanism, coping strategies, for some were to stay in the competition and balance out the monetary losses, but, for the others were to safeguard their mental health. 10.90% of the respondents looked at the pandemic optimistically and saw it as an opportunity to wind up their pending works. While @rajeshwar.s.jamwal saw this time perfect for spending it with his family and pets, @gobanjare said that for him, COVID meant "taking time off." @the_iffy_explorer did her pending "room renovation" and @meghana_goes_places got herself indulged in "reading about travel followers, sustainable options, and why it is important to travel slow." f. #StayAtHome: COVID-19 had drowned 32.72% of the respondents in the limbo of an apparent endless pause, all of whom reiterated and advised their followers to stay safe at their home. Hopelessness amidst the respondents was reflected with @indian__travellers saying that "nothing much" was left to do now for travel influencers. According to @traveldiaries_nabanitaroy, "travel will not be possible in the near future and will have to wait for a long time." So, like most of us, influencers like @breathe_dream_travel_repeat were also coping up by "building immunity, praying, and hoping." g. Getting future ready: 7.27% of the respondents took out this time to get ready for the future. For this, a few respondents like @m_i_s_i_x, @my_wandering_list, and @bagsonwheels utilised this time to prepare the bucket list of the places they would like to go, post-COVID. Also, @davidangami said that his travel company was taking this time for "advertising places for people to come and visit once travel starts." In parallel to this, his company was also "setting and polishing the travel packages to be COVID SOP compliant." h. Learning digital skills: For 14.54% of the respondents, coping up through the pandemic was facilitated by learning the essential digital skills. While @creating_destinations learnt the various editing skills to help them with their future content, @jammu360 emphasised on learning those digital skills which would help them "monetise their travelling career." Further, @hkmurali worked on refining and polishing both his profile and collaborations. i. Promoting engagement: One of the respondents, @trek_a_break, had found a unique way of coping through the pandemic, by working on their social relations with their existing followers and thereby promoting engagement on their Instagram page. j. Unpublished content: @travelure, one of the respondents, said that "I have a plethora of unpublished content. So, I can go on, with or without pandemic, and keep bringing out stories to my audience." This could be a great insight for somebody who loves travelling and has already travelled a lot, but is sceptical about turning into a travel influencer presently, given the COVID-19 situation. k. Webinars: @india.trail, a travel company, had found an innovative way of managing stress during the pandemic, by "giving and attending webinars, and other social online meetups." --- Way forward In the post-COVID time, the travel industry needs a good kick-start. However, I argue that social media travel influencers can not save themselves or the tourism industry in isolation. The need of the hour is a nexus of the various stakeholders in tourism, given by Peter Burns, to work collaboratively and cooperatively. Trends such as increasing ICT readiness and destination brand management through a multi-stakeholder approach can go a long way in reviving tourism from the aftermath of the pandemic. Thus, it would have a high probability of reviving the almost collapsed travel industry, in congruence with the'stakeholder theory' by R. Edward Freeman. Building on the existing works, I suggest the following coping strategies for the travel influencers: 1. Increasing awareness: Taking examples from the Finnish (Heikila, 2020) and the UK's government (Pritchard, 2020), India can also mobilise various social media influencers to not only spread awareness but keep a check on fake news. Travel influencers can help to a great extent in reiterating the social distancing norms and other safety protocols to be followed while travelling during the pandemic. In the long run, they can prove to be a vital agent to curb environmental pollution created by travellers. Thus, a collaboration with the government will help the influencers in coping up not just economically, but also by becoming mainstreamed in the tourism industry. 2. Promoting local tourism: Going a step ahead of travelling locally, and taking the idea from Mariellen Ward, influencers can consider launching a "digital marketing company" (Kaushik, 2020) or a startup to promote local tourism professionally. It can pave a way to preserve the local culture and environment as well, along with increased tourism, and effective dissemination and knowledge about the local practices to the tourists. --- Government funding: The Indonesian government has already set the benchmark by funding the travel influencers to boost its tourism (Gorbiano, 2020). I suggest a similar type of arrangement for Indian travel influencers as well. 4. Virtual tourism: The social media travel influencers may cope up through the cascading impact of the pandemic by being the digital vehicles to promote virtual tourism. A perfect example to explain this is the 'Experience Mecca' app created by a travel company, BSocial, which provides a virtual tour to all the Hajj pilgrimages (Cheng, 2017). Further, companies like 'andBeyond, Singita, and Lewa Wildlife' have found a temporary fix to rescue kneeling tourism by providing virtual safaris (Virtual Safari Tours Pick up as Global Travel Hit by Restrictions, 2020). It can provide a great insight to social media travel influencers to work on some innovative ways to thrive and flourish despite pandemic. 5. Learning from others: In conjunction with the'stakeholder theory', it is suggested that even the travel influencers must learn from each other, and must-see this uncertain time as an opportunity to do so. As an example, @the_nomad_couple, one of the travel influencers, has been reading other travel blogs for "pleasure and inspiration" during the pandemic (David, 2020). To further reinforce the idea of collaboration between the various stakeholders, it is suggested to organise frequent webinars. It can not only offer a platform to discuss their common problems and ponder over its solutions, but also a commonplace to help mitigate their mental health problems. 6. Weighing the pros and cons: Finally, it is suggested that the travel influencers can utilise this time for spiritual growth and personal journey. Introspection and self-appraisal, along with "revaluating choices" (Joseph, 2020) of the brands they are associated with could be a good way to cope up presently, thereby making them future-ready. Further, they may work on building new business tieups and collaborations. --- Suggestions for future research The present study has some limitations that could pave the way for future research, written as follows. First, the scope of the present study is limited to travel influencers only. Further studies could be initiated covering influencers from other fields as well, along with the various stakeholders of the tourism industry. Second, only the travel influencers from Instagram were included in the present study. Further studies may cover travel influencers from other social networking sites, like Facebook, Twitter, etc. Third, the study can be improved by communicating directly with the respondents, through a blended interview, depending on feasibility. The present study is limited to collecting responses from travel influencers through a google form. In addition to this, the some unexpected trends were noticed while analysing the data. These include: (i) the study did not observe any association between the number of followers and motivation in the travel influencers to take it as their profession, and (ii) despite the majority of respondents having taken travel influencing as their hobby, they did not opt it as their career. It calls for future research to unlock if any correlation exists between the number of followers and career choice for a travel influencer. Further, the factors that motivate a travel influencer to take it his profession, apart from just hobby, may also be looked into intensively.

In the wake of the COVID-19 pandemic, the global business landscape has experienced unparalleled disruptions, particularly impacting the tourism industry. This article delves into the repercussions of the pandemic on Indian Instagram travel influencers and its subsequent effects on the tourism sector. Grounded in a social survey encompassing 55 Indian Instagram travel influencers, this study sheds light on the challenges they confronted and the strategies they employed to navigate the crisis. The scholarly work by R. Edward Freeman ( 2009) and Peter Burns (2013) provides a theoretical framework for understanding the responses of Instagram influencers. The study's findings reveal that a substantial majority of the respondents faced financial losses due to the pandemic. In response, these influencers diversified their approaches, engaging in responsible and local travel, adapting their content to enhance audience engagement, acquiring digital skills to prepare for the future, and hosting webinars, among other measures. In conclusion, this research presents innovative coping strategies, tailored not only to Indian Instagram travel influencers but also applicable to influencers worldwide who faced similar challenges.

INTRODUCTION The new respiratory disease COVID-19 is highly contagious disease. The rate of its spread is much higher than the previous biological pandemics (Varshney et al., 2020). In order to prevent its outspread, ruling authorities across the world have announced mandatory home quarantine as a preventive measure. To minimize its spread a temporary shutdown has been imposed nationally in different ways, i.e., closing down of academic institutions, ban in public meetings, restrictions in social gatherings and social distancing. On contrary to the basic instinct of human kind, people were obliged to stay physically disconnected. While following these preventive measures, reports indicate the use of digital apps in connecting people worldwide and providing knowledge to be healthy (Iqbal, 2020). Being in home quarantine and isolation can take a toll on both the mental and physical health. Various negative feelings such as anxiety, depression have been noticed among people during the lockdown period due to the pandemic (World Health Organization, 2020a). Along with this, numerous literatures have emphasized on the psychological effects experienced by individuals during previous pandemic eruptions like Ebola, MERS-Cov, SARS, and H1N1 (Sun et al., 2020). Besides, studies have statistically provided evidence of older patients' co-morbidity of Ebola with anxiety, depression, and post-traumatic stress disorder to be 24.9, 47.2, and 21%, respectively (Rojek et al., 2017). Notably, even the caregivers of these patients had been diagnosed with depression and anxiety; in fact, about 45% of them opted for psychological counseling albeit at a later date (Smith et al., 2017). On the brighter side, few studies reported that acts of social wellness, such as community support, positive interactions with patients and empathy have led to positive psychological wellbeing of patients in such crisis (Kama et al., 2017). Studies like this have illustrated the role of social connectedness to ease psychological impacts. During COVID-19, due to imposed restrictions, modern technology could be the only way to maintain social relations. Thus, this study tries to examine the role of social connectedness in social cure. --- Social Connectedness The feeling of connectedness is the basic need of human beings. During this pandemic, it has been reinstated since the time people were advised to follow "social distancing" norms even within the family. Recent researchers have argued to change the terminology of "social distancing" to "physical distancing, " which would effectively reflect the actual meaning of the preventive measures to reduce the feeling of social disconnection (Sanderson et al., 2020). Nevertheless, the foremost drawback through the norm of "social distancing" that has been observed is that people at large began to develop an apathy to "social wellness behavior, " whereby they began to discriminate COVID-19 patients, their family members, and at times even the "caregivers/care providers, " i.e., doctors, medical personnel, police officers among others (Graupensperger et al., 2020). However, if we were to question ourselves on what is the phenomenon of "social wellnesses!" In order to understand the same, one needs to understand the experiences of patients and the attribution of the term "social connectedness." Rossi et al. (2012) defined it as "a person's subjective awareness of being in close relationship with the social world." Notably, this attribution of self reflects cognitions associated with an individual's on-going interpersonal closeness with the social world. Therefore, "social wellness" may refer to the way a person develops social health, while expanding connection with others in the dwelling/society. It alludes to a person interacting in a positive way, while creating and maintaining healthy relationships, which in essence serve a meaningful purpose in life. Thus, both social health and social wellness become an essential dimension for emotional, psychological as well as physical health (Kingsep, 2019). Research in the past have shown the positive correlation of "social connectedness" with emotional resilience, conflict management, life satisfaction, and self-esteem (Fraser and Pakenham, 2009;Stavrova and Luhmann, 2016). Richards (2016) stated that despite an increase in income, the participants of his study were found to be happier when their social connections grew positively, which effectively meant that people did value social connections more as opposed to "income." On the contrary, chronically lonely people showed negative health outcomes, including the aspect of "addiction, " be it to drugs, alcohol and others. Importantly, people with low social connections proved to be more toxic than obesity, hypertension and high blood pressure (American Psychological Association, 2017;Tate, 2018). Interestingly, it was found that people who maintain healthy social connections could actually produce more antibodies through their immune systems, which eventually help them to fight diseases and provide longevity to the body (Marchant, 2013;Pappas, 2013). Studies have also highlighted the role of mirror neurons that help in producing empathy, sympathy, compassion and a feeling of belongingness within or others, often by inducing pain in selves and/or others. This goes on to prove that the aspect of "empathy" does make a person take proper moral or humane decisions eventually increasing the feeling of belongingness, while triggering an intent to help others (Lamm and Majdandzic, 2015). Psychological researchers have explained this phenomenon as reinforcement and vicarious learning (Kazemi and Rostamian, 2016;Pfefferbaum and North, 2020). Furthermore, people having good social connections were found to have better eating and exercise habits, which could be the result of reinforcing healthy habits by friends. --- Social Connectedness and Digital Technology The "small-world problem" hypothesis showed social connections to be so dense that every person on this planet could be connected to just a few intermediaries. The experiment "Six Degrees of separation" conducted by Stanley Milgram (Milgram and Travers, 1969;Maier, 2019) found that to help a person in receiving a letter from one end of the world to the other, there would be in a need of only six hands exchange. In other words, on an average, only the involvement from six people could create nodes in establishing connections. Researchers have also shown that the "small-world network model" could be used to explain social networks, computer networks, neural networks and wireless networks (Watts and Strogatz, 1998). Sohn's (2017) "small-world network model" highlighted the difference in the number of people that could be connected indirectly through social network and a wireless network connection (Uzzi et al., 2007). Although social connectedness and belongingness has been identified as the third key need postulated by Maslow in his "hierarchy of needs, " extensive research has shown and even reiterated the importance of "connectedness" to maintain a good physical as well as mental health (Baumeister and Leary, 1995). Bronfenbrenner (1994) also discussed the interpersonal interactions of human at different levels that include micro, meso, exo-, and macro systems. With modern technology, interactions between these spheres have become more compatible (Ashiabi and O'Neal, 2015). The different palette of communication tools such as email, text messaging, WhatsApp, Facebook, etc. have all provided a platform to interact across the globe, albeit in a virtual mode. A few researchers divide social connectedness into two prominent divisions: (a) one to one connectedness (through text messaging) and (b) large basis of connectedness (through Facebook or Instagram) (Bel et al., 2009). Both one to one and large basis connectedness do help individuals to "connect" with family, friends, and the world at large. Since the outbreak, Social media has been over-embellished with its persisting and exclusive coverage on the pandemic. For instance, the News18's (2020) strange news of two women doctors being threatened and abused, and asked to leave the grocery shop and empty their flat on an immediate basis by their neighbors had gone viral in social media. On contrary, India TV on 23rd April 2020 reported certain pleasant incidents of the warm welcome of recovered patients by the neighbors (India Tv, 2020). Interestingly, the Washington Post, on 8th April (2020) in one of its news bulletins, opined that such contrary behaviors have been observed across the world (The Washington Post, 2020). Social media has been a disseminating tool for both information and news, including many fake ones, ranging from the recovery rate, death rate, and vaccine development. World Health Organization team has stated, "The 2019-nCoV outbreak and response has been accompanied by a massive 'infodemic,'some accurate and some not, that makes it hard for people to find trustworthy sources and reliable guidance when they need it" (World Health Organization, 2020b). The term "infodemic" was meant to allude to an over-abundance of news. During this pandemic, World Health Organization (2020b) prompt response about infodemic calls for one's immediate attention to cross check the reliability of incidents that happen in the community. --- COVID-19 and Digital Technology Scientists and researchers have used technology, ranging from high-tech robots to low-tech masks, to the best, to fight this pandemic. Robots have been used massively for cleaning, sterilizing, emitting Ultra Violet (UV) rays and delivering food to patients or the needy during quarantine in order to limit humanto-human contact. Digital technology has also been harnessed to provide innovative ideas; e.g., when patients were short of ventilators, researchers and engineers have formed several online groups on platforms such as Telegram, Facebook, Instagram etc. to share knowledge on manufacturing ventilators. This apart, Artificial Intelligence (AI) was also used; in fact, Blue-Dot software had triggered an alert even before WHO on this pandemic by analyzing data provided by the different agencies (Steig, 2020). The usage of AI for image-based analysis in Computerized Tomography-scan (CT-scan), understanding the working of coronavirus to produce drugs, telehealth has been an immediate relief provided by the technology. Although largely technology has played an uplifting role, it has its own hazards; for instance, the aspect of individual privacy. When the world was facing global lockdowns, various countries using technologies such as AI tried to track down its people. Even though this was intended for a "noble" cause, it is also disadvantageous, as it leads to intrusion upon an individual's privacy (Tétrault, 2020). Media coverage, especially through television has also worked bi-directionally. While this pandemic continues to cause harm to society, scientists and researchers are no longer bound by geographies due to technology-based support. The technological advancements has helped immensely to provide different COVID-19 testing aids and treatment facilities across the globe within a short span of time. --- Social Connectedness and COVID-19 Following preventive measures and use of digital apps help to connect people worldwide and providing knowledge to be healthy (Iqbal, 2020). Being in home quarantine and isolation can take a toll on both the mental and physical health (Altschul, 2020). The role of social connectedness cannot be ignored during the pandemic. Thus this study aims to record insights from the patients and caregivers' perception and experiences toward the COVID-19 as well as the usage of technological aids to maintain social connections in the unforeseen and disastrous situations such as this pandemic. In addition, this study explores attitudinal differences of people toward COVID-19 patients. The salience of this study is to explore and attempt to understand how social media network and social connectedness aided through digital technology play an important role in maintaining relationships as well as to see if they act as a social cure during the pandemic. --- METHODS --- Design and Participants This study used an exploratory research design to understand the proposed objective. The participants of this study were chosen, using a purposive sampling method. Before conducting the interviews, informed consent was taken, and confidentiality was assured. This study included 38 adults (Male = 21, Females = 17) across India, between the age group of 21 to 40 years (Mean = 28.5, and SD= 4.5) having at least graduate-level qualification. Out of 38 participants, 20 were COVID-19 recovered patients, while 18 were caregivers. --- Procedure This qualitative study aims at exploring the first-hand experiences of the recovered COVID-19 patients, along with their caregivers during the pandemic. Telephonic interviews with the participants were recorded after getting their due consent. They were also assured of total confidentiality, and were told that the data would exclusively be used for research. On an average, each interview lasted about 40-45 min. A set of semi-structured questions were prepared to conduct the interviews, in consultation with the experts. Semi-structured interview questions were framed after several scrutiny of experts. To ensure the inclusion and exclusion criteria few questions were asked initially during telephonic interview. For instance, a few questions like following were asked: (1) During the crisis of COVID-19, what was your experience of social distancing? (2) Has it affected your emotional well-being? and (3) What were your strategies to cope up with the same? Few questions asked to COVID-19 patients: (1) What was your first reaction the moment you got positive report for COVID-19? (2) Who all has helped you to fight this situation? How did they help you? and (3) What was the perception of others toward you once you recovered from such illness? Few questions asked to caregivers: (1) What was your experience and approach to help the person who was diagnosed positive for COVID-19? (2) Except being COVID-19 as a contagious disease, what all could be the reason for not helping needy? --- Data Analysis The data of this study was analyzed using Braun and Clarke's ( 2006) Thematic Analysis research method. Thematic analysis is a method for identifying, analyzing, organizing, describing and reporting themes within a data set. The approach of thematic analysis suits to explore and identify prominent themes of the data collected here. After each telephonic interview, recorded content was carefully converted into verbatim and thoroughly reviewed by the researchers. This was done to assure the accuracy of the information provided by the participant transferred to the paper. During the analysis of the data, researchers have gone through verbatim again and again to refine the understanding of data and identifying the words, sentences which were providing similar meaning. Further, responses provided by the participants were noted under different columns. Each column represented the responses which provided similar meaning and through literature review suitable sub-themes and themes were assigned. The criteria provided by Guba and Lincoln (2011) for evaluation of qualitative research was taken under knowledge during analysis. Through data analysis each criterion, firstly credibility of information was assured through repeated reviews of data gathered and transferred in verbatim. Secondly, transferability of knowledge via mentioning actual responses of participants, their characteristics was assured, which could help other researchers to generalize their findings in respective findings. The proper division of step-by-step administration by researchers assures the non-biasing and removes dependability on limited researchers. Lastly, the fourth criterion of conformability was assured when the final result of the data analyzed and the accompanying process involved was reviewed by researchers. --- RESULTS AND DISCUSSION The present situation of COVID-19 has brought diverse experiences to the human kind. This outbreak has led people to follow various preventive measures such as home quarantine them physically distanced from each other. At the same time, technology is on the peak which smoothen the route to connect with each other which may have positive as well as negative impacts. In this research, data was collected from 38 participants, in which 20 were COVID-19 recovered patients, while 18 were caregivers. Upon employing thematic analysis, four major themes, along with 10 sub-themes emerged. They include Psychological Experiences, Attitude toward the patients. The description of these four themes and overall, 10 sub-themes are as follows. --- Psychological Experiences of People COVID-19 has affected people both physically and psychologically. Participants have reported various psychological struggles that they experienced on the basis of which three sub-themes evolved such as: perceived experiences, concern toward patients and the negligence of psychological counseling. The significant findings from the experiences that have been observed in the obtained data of this study are mentioned below. --- Perceived Experiences Participants, who recovered from COVID-19, as well as their caregivers had gone through the distressing psychological experiences during the critical time of their illness. This primarily included depression and anxiety, while some were stressed, and a few others traumatized. However, the causal factors for the disturbances between these two groups were found to be different. Herein, it is vital to note that these experiences were also the result of social distancing, home quarantine and unexpected financial crisis faced by the participants. However, the overall responses under this sub-theme of "perceived experiences" were majorly negative because people were afraid of this uncertain situation. This has been also observed in previous epidemics in history (Smith et al., 2017;Varshney et al., 2020). --- Unpleasant Experiences of Patients and Caregivers Among several psychological distress faced by the participants, anxiety was the most dominant, followed by depression, stress and insomnia. Rehman et al. (2020) reported a wide range of psychological experiences, such as the perception of risk, fear, anxiety, stress, and depression level among the different populous, caused by COVID-19. A male participant AK for instance, encountered a lot of anxiety during his self-isolation. He stated "one day... I heard from the cabin man that... my neighbors patient suffering from breathlessness and sent to a different ward. And it created tremendous anxiety in me. I had chosen a single cabin room because I don't want to see the others' packed body." While a few reported their experiences of anxiety, others shared their reactions to various traumatic incidents. For example, a female participant DK stated, "I was so much traumatized that everyone came to know about my reports. I felt like I can't do anything. It was so depressing." The anxiety level was higher among healthcare professionals and women during pandemic events (Hacimusalar et al., 2020). However, in the present study, such gender-based differences could not be found. Few participants, irrespective of them being patients or caregivers, experienced higher stress levels. For example, a male participant NJ stated, "Me, my wife and my youngest daughter were tested positive for Covid-19. The day before the report came; I had cremated my mother.... Due to this lockdown... caused me a lot of financial crisis. And the report of my daughter has shaken me inside and made the situation more stressful." In certain studies, the anxiety level was high among the caregiver community (Cici and Yilmazel, 2020). Caregivers in specific have reported insomnia symptoms. A caregiver SD stated "I'll get up at two o'clock/three o'clock.... she was in a separate room.... I used to peek from the window itself and come back. Like the Attention, apprehension what we'll happen? will she recover or not?" Some of the caregivers claimed that the disturbances were related to the uncertainty of future and fear of loss of their family members. These psychological disturbances led the participants to develop insomniac conditions too. Caregivers in specific have reported insomnia symptoms. A caregiver SD stated "I'll get up at two o'clock/three o'clock.... she was in a separate room.... I used to peek from the window itself and come back. Like the Attention, apprehension what we'll happen? will she recover or not?" The caregivers claimed that the disturbances were related to the uncertainty of future and fear of loss of their family members. This category therefore, revealed that both patients and caregivers experienced anxiety, depression, stress, and insomnia due to various issues. However, the causes of experiences differed in both groups. While patients' psychological disturbances were mostly concerned with isolation and tensed panic environment, caregivers' anxiety was more associated with an uncertain future, and the well-being of their loved ones due to lack of proper treatment. --- Pleasant Experiences of Patients and Caregivers Despite negative psychological experiences, a few people found home quarantine and lockdown to be pleasurable in the initial phase, as they were having quality time. This sub-theme emphasizes on the positive experiences of the participants. A female participant NJ stated, "my mom cooked a lot so I was so happy. I am a very homey type of person.... stay with my parents. So, it was very fortunate.... I enjoyed talking and spending good time with family. Also, I wasn't living in my house for few years and I felt that I need to spend time." Concisely, the participants enjoyed being at home and spending quality time with their families. Hacimusalar et al. (2020) noted that the anxiety level was higher among healthcare professionals and women. However, in the present study, such gender-based differences could not be found. Furthermore, findings similar to this study were also noted by Cici and Yilmazel (2020) which stated that the anxiety level was high among caregivers. However, the overall responses under this sub-theme of "perceived experiences" were majorly negative because people were afraid of this uncertain situation. This has been also observed in previous epidemics in history (Smith et al., 2017;Varshney et al., 2020). --- Concern Experienced by Patients (Causing Others) While interviewing patients, most of them were concerned more about their family members toward COVID-19. Fear for their significant partners and children were more prominent. For instance, a female participant NY stated "my husband did not have any fear of getting corona we had some fear for children... because if they got... then how to take care of them we can't go near them." These findings reflect the phenomenon of higher risk perception (Cowling et al., 2010) that is during the pandemic, people perceive more risk of spreading infection. This increases the motivation to protect oneself and others to avoid infection. Notably, Simione and Gnagnarella (2020) also found that higher the risk perception in a given group of respondents, the higher is their concern toward health. Perceived severity and self-efficacy have been found to be positively correlated with self-isolation in order to prevent spreading of COVID-19 (Chen et al., 2020). --- Negligence of Psychological Counseling This sub-theme focuses mainly on the lack of psychological counseling services that participants thought was necessary. The Indian government-initiated state-specific intervention strategies, tele-psychiatry consultations, toll-free numbers specific for addressing the psychological and behavioral issues faced by the COVID-19 victims (Roy et al., 2020). However, efficacy of these psychological counseling on patients and caregivers has still been unreported and is thereby ambiguous. A male participant PG stated, "Apart from taking medicines there should be a programs like psychological counseling, which the govt or anyone can arrange so that they can help covid patients from psychological issues they are facing." Another caregiver participant, AB stated, "As I say that it is more on the mental side, not on the physical side." In fact, some of the caregivers went on to highlight that although they did not contract COVID-19, their mental health was as affected as of the patients. While patients were being treated by the hospitals, there was no psychological support provided from the government for the caregivers. Das (2020) emphasized the need for psychological interventions at a large scale to counter the after-effects of post-traumatic stress, frustration, stigmatization etc. Thus, there is indeed a need to have a strong policy for such provisions. --- Attitude of Others Toward the Patients and Caregivers This main theme, which originated based on shared experiences, sheds light on attitudes of family members and support from neighbors as well as peers. This theme has been sub-divided into two, i.e., during recovery and post-recovery, which in essence, shows the differences in people's attitudes toward participants both during and after recovery. This theme also involves some pleasant as well as unpleasant experiences, which are stated as below. --- During Recovery This was a time when patients were asked to isolate themselves, and avoid making contact with family members and others. The differences in the experience of social connectedness have been due to their varied situations, such as when the participant was infected by COVID-19 vis a vis after him/her getting cured. Although family members were concerned about patients, another dominant social support system, i.e., the "neighborhood" was found to be failing in establishing and maintaining social connections with the patients. --- Family Care In the initial stages, parents have had difficulty in accepting their son and/or daughter diagnosed with COVID-19. However, family members reported that they gradually provided the required support to the patients with proper care. A male participant AK stated "they all were just motivating us, saying that no problem, it happened just happened, be positive." Though family members and near ones initially displayed denial and non-acceptance, eventually they accepted and found the courage to support their loved ones, which increased their feeling of togetherness. --- Societal Attitude This sub-theme highlights the discriminating, stigmatizing and traumatic treatments by the neighbors toward the patient and caregivers. World Health Organization (2020) stated that having more ambiguity due to a new disease may lead to unconscious discrimination, which reflects through stigma and stereotype. A male participant PG stated "They ignored me.....my name was given to the newspaper and people announced it in the whole colony. People used to run away from our house as if there is a ghost inside. Everyone got to know." A few statements highlighted the discrimination and stigmatization of the disease, resulting in name-calling such as "virus home." Contrary to these negative attitudes, a few interviewees indicated a positive attitude by their neighbors also. A female participant AA noted, "One of our neighbors is so good. They used to come home only in night, till then they took care of children. We didn't feel the need for friends; our neighbors helped us thoroughly." Since caregivers were directly or indirectly in touch with the patients, the stigma toward patents might have been generalized toward caregivers as well. Overall, the neighbors' attitudes toward the patients as well as the caregivers have been negative. The behavior shown could also be understood through negative dominance theory (Covello et al., 2001;Glik, 2007), which states that when people are fearful, they give more weight to negative information and respond accordingly. --- Peer Support Support provided by the friends and colleagues has worked as a healing system to many survivors. It has been noticed that friends of the patients have helped their family members too and assured their presence in an emergency. A female participant MB noted "They were being too supportive....actually my mum was Home alone. my friends... used to bring all the fruits and required materials by ordering it online and it would be delivered at home. So that was a good thing." A positive social connectedness and concern by the friends stayed prevalent during as well as after the recovery period. --- After Recovery This second sub-theme focuses on the attitude and perception of patients after their recovery. In addition, caregivers reported that a few neighbors initially avoided them due to risk perception but, such behaviors subsided eventually. A male participant AK responded, "Since, no one in my area had it before me, I had become like the godfather for everyone. Whoever got it later called me for suggestions." Similar differences in attitudes were observed by caregivers, as others sought help to know the ways to maintain hygiene and prevent contracting COVID-19 from patients. This showed that people sought help from both caregivers and patients but for different reasons. This could be explained with the "mental noise theory" (Baron et al., 2000), which explains that people during a pandemic, get stressed, and they attend to a great deal of internal "mental noise, " albeit unconsciously, which in turn results in a lesser ability to attend to externally generated information. Notably, this indicates the change of behavior in people in which, negativity toward someone is not intentional or fabricated, but rather a representation of chaos and fear. --- Digital Technology This main theme highlights the dual role played by the technology. The digital technology has worked as a carrier to spread fear as well as knowledge about the disease condition and at the same time extent support to the community. In its sub theme, at one place social media is working as a carrier to spread fear causing anxiety, depression. On other hand through ease in communication through digital technology has worked wonders to provide support and strength to the society. --- Influence of Digital Media This main theme emphasizes the influence media has on the individual's thoughts, perceptions and behaviors.. Gao et al. (2020) found that 80% of the participants who were exposed to social media had a prevalence of depression and anxiety. Das (2020) provided evidence that social media news has been the leading cause of over-reactive behavior, especially of the Indian people. With technological advancement, there are various platforms, which provide information to the viewer. Some platforms are very recent and some have been there for a long time. The experience of listening to unpleasant news related to COVID-19 significantly influenced the people. For instance, a male participant LD stated "In the initial times.... hearing all such news that is many people died and there is no cure there is no treatment for this disease hearing all those stuffs I was getting disturbed that if I will get what will happen. So later, I didn't follow them at all to avoid panicking." This sub-theme showed that most of the people relied on newspapers than on news channels or any social media platform. On the contrary, the contemporary news feed from digital media platforms such as WhatsApp, Facebook, and Twitter form to be posing as a major problem than the news from television. A female participant ST shared "I remember someone posted a list with my name. I got mad. I said okay, you take precautions and all... but don't disclose the names of people. And this resulted in back to back comments and quarrel and I just left that group." Apejoye (2015) stated that information and facts shared through the means of both newspaper and television are more credible than social media platforms. The report provided by World Health Organization (2020) regarding infodemic seems to be true in this study also. However, results of this study show more of the negative aspects of the digital media, which is an indication for the government, and respective organizations to look upon this issue seriously, as news from these sources could be channelized to help create positive influence, which would bring the community closer. --- Technology Aids: Perception and Attitude During home quarantine and isolation, the only convenient method to stay connected was through mobile phones and social media platforms. This advancement has provided an opportunity to embrace social connectedness in a distinct way. Chen et al. (2020) mentioned about the ease provided by technological advancements during COVID-19 in order to connect with acquaintances. The perceived support provided by technological advancement worked as a mental component in recovery of patients. The advancement commencing from cell phones to the development of various tools to diagnose COVID-19, as well as to restrain the spread, has created an enormous difference in stalling the pandemic. This theme emerged because of shared experiences by the participants about these technological advancements by further focussing on social connection and social media attitude toward treatment and related aids. More or less the technology evolved as a companion who provided relief by even connecting the stranded people with their close ones. --- Opinion Toward Treatment and Related Aids With the advancement in technology, treatment to certain diseases has been made possible to diagnose and be cured. Researchers worked day and night to provide credible diagnostic systems to COVID-19. The purpose of this cannot meet until the common man does not trust and perceive it as a cure to the disease. This sub-theme is further divided into two divisions that include the perception of people toward COVID-19 testing, and applications, which promote a healthy lifestyle. --- Opinion Toward COVID-19 Tests Two prominent testing, i.e., Antigen Test and RTPCR test were widely used to diagnose the presence COVID-19 virus in human body. Antigen test also known as Rapid Antigen Test-(RAT), provides results within 4-5 h, while the RTPCR test provides results after a span of 2-3 days. It was an attempt to know the patients/ caregiver's perception related to the credibility and reliability of both the tests for diagnosis. A male participant KJ mentioned, "what I know is if rapid test there as false negatives... as far as I have seen if it comes positive, RTPCR also comes positive around 99% cases.... So, if it is negative with symptom better to go with RTPCR...." On contrary to this, a few participants mentioned the use of antigen test as the need of the hour. A few reasons mentioned by participants included the point of India's humongous populous, coupled with the virus spread being so high. They further stated that for the government, it would be more convenient to do testing through rapid tests. On other hand, experimented medicines and treatment provided by the government found to be not more reliable than the Ayurvedic Kada, which is a drink prepared with the use of herbs to boost immunity.. A female ST mentioned, "I trusted more on ayurvedic treatments." --- Opinion Toward Apps This subtheme emerged in accordance with the claim done by various health-based applications to provide a good knowledge base to stay healthy (Warman, 2015). In contrast, the majority of participants denied the utility of such apps and any exercisebased application. Most of them favored ancient Indian practices to stay fit, such as practicing "Yoga, "and "Pranayama, " and consuming "Kada or maintain a balanced diet" to stay healthy and boosting immunity. A male participant PG responded, "I didn't prefer it that much. I used to do yoga and kappalbharti and I have continued that in lockdown also and when I was in quarantine that time also." The responses provided by the participants were not in the favor of using any of the health-based applications. Husain and Spence (2015) provided evidence that a few health applications may accurately show some aspects of health, but it would not be wise to solely rely on such apps. This shows the trust and beliefs of people on traditional knowledge. Although, in this study the advancement of technology has been found to be beneficial in some domains, the importance of traditional practices could not be overlooked. In support of this, the traditional knowledge for "yoga, " exercises and to prepare medicines and "Khada" to cure the illness has also been preferred and given the importance of traditional and indigenous knowledge. --- Social Connectedness In order to follow social distancing and home quarantine, the only way to connect acquaintances was through social media platforms and mobile phones. In this theme, experiences of patients and caregivers are described which highlights the usage of digital media platform. --- Social connection and Social Media This theme emerged in the viewpoint of an increase in the use of social media as a helping aid for people to connect with others. It has been observed that the order of staying at home became easier due to technological aids. A male participant for instance, reported: "Thank god! We have this internet thing otherwise it could be difficult to stay home for so long." Moore and March (2020) found that social media helps to connect people with their loved ones and bring down the feeling of being in isolation during this pandemic. Most of the recovered patients mentioned that the frequent calls by the acquaintances through social media platforms they received during the lockdowns provided a lot of strength and mental support. When interviewees were further probed, they revealed that through social media connection and in-person meet, they got to know

Although COVID-19 pandemic has re-orientated humans to be more physically healthy and hygienic, it has also persuaded humans to create affiliations and experience a sense of belongingness through social networks and digital technologies. However, amidst these changes, experiences of COVID-19 patients and their perception of the outside world's attitudes toward them appears to be less attended in literature which formed the basis for the current study's objectives. Using qualitative methodology, the present study explored the experiences, perceptions and attitudes of patients and their care-givers' toward COVID-19. The thematic analysis emerged with four major themes. Psychological Experiences of People was generated prominently with sub-themes indicating the perceived experiences like fear of spreading diseases to others, and the need for psychological counseling. Attitude of others toward patients and caregivers revealed that family members and relatives played a major positive role on the patient's mental health, however, the neighbor's stigmatized attitude led to several undesired behaviors. Social Connectedness was another major theme derived from the study. Altruistic volunteers, a sub-theme of Social connectedness have indicated that amidst these negative factors, one can spread social harmony by motivating and supporting the victims with basic needs, financial support, hope and social empathy. Opinions of participants for digital technology through technological aids and preventive measures emphasized an overall positive attitude as it helped the society, in general to maintain social connections as well to curb the rate of COVID-19 cases.

hada" to cure the illness has also been preferred and given the importance of traditional and indigenous knowledge. --- Social Connectedness In order to follow social distancing and home quarantine, the only way to connect acquaintances was through social media platforms and mobile phones. In this theme, experiences of patients and caregivers are described which highlights the usage of digital media platform. --- Social connection and Social Media This theme emerged in the viewpoint of an increase in the use of social media as a helping aid for people to connect with others. It has been observed that the order of staying at home became easier due to technological aids. A male participant for instance, reported: "Thank god! We have this internet thing otherwise it could be difficult to stay home for so long." Moore and March (2020) found that social media helps to connect people with their loved ones and bring down the feeling of being in isolation during this pandemic. Most of the recovered patients mentioned that the frequent calls by the acquaintances through social media platforms they received during the lockdowns provided a lot of strength and mental support. When interviewees were further probed, they revealed that through social media connection and in-person meet, they got to know certain new trends being carried around them. A female participant AK for instance, stated, "It was a new experience that we were exploring and we were doing something.... Every Saturday and Sunday, there were some Tam bola Night, some other means or some other puzzle game. So, it was very interactive." None of the participants preferred in-person meeting against social media connection, and added that it was the need of the hour. Thus, interactions in social media platforms helped people to cultivate a sense of social connectedness. --- Preventive Measures Most of the participants talked about their respective concerns, and followed preventive measures, such as staying at home in order to restrain the spread of the virus either to themselves or to others. A male participant noted, "Home quarantine was mandatory for oneself and others." Other preventive measures, such as wearing masks, sanitization, and "social distancing" were acknowledged and approved by the participants. Some of them even talked about educating others to strictly follow these preventive measures. A male participant, PG stated, "We ourselves need to sanitize our society." Thus, people took care of themselves as well as others in order to remain safe from the virus' onslaught. --- Altruistic Volunteers This sub-theme has emerged as an additional finding to the primary objectives that were framed. Although COVID-19 has been a contagious disease, most participants have voluntarily helped others in need. The life instincts have been believed to be the cause of such self-preserving behaviors. It is believed that people due to life instinct would self-preserve themselves. Mawson (2005) explained such contrary behavior by linking it to a massive panic among people during such pandemic. On this contrary belief, literature shows that pro-social and altruistic behaviors are predominant. A male participant, PG stated, "Yes me and my social club members used to cook food for 200 people who were migrants. It went for 2 months. We use to prepare food for two meals. It was just a blessing." Moreover, separation and guilt of not helping others is an unbearable psychological stressor than the physical danger (Glik, 2007). This may be attributed to the altruistic behavior shown by most of the participants during this pandemic. This kind of helping attitude builds the community as a whole, and provides a mental cure that strengthens the feeling of togetherness, which possibly plays the lead role in the healing process. Thus, it may be affirmed that social connectedness does enhance the sense of belongingness to the society that can sustain even during a pandemic. This study reported that both patients and caregivers have experienced discrimination, stigmatization, as well as psychological disturbances during the battle against COVID-19 and find certain meaning to their lives. Conspicuously, negative experiences faced by both the caregivers and recovered patients have felt the need for psychological counseling. The changes in the attitude of neighbors, both before and after a patient's recovery has been seen clearly, indicates the necessity of socio-psychological theories and their applications. Although many caregivers did not contract the disease, they emphasized that their mental health was ignored drastically, while they were attending to the patients. Patients, caregivers and Doctors could interact more feasibly than ever in this pandemic, which alters the way to provide support and guidance to the patient and caregiver which impact psychologically (Sanderson et al., 2020). Importantly, this study also highlights the need for counseling and therapeutic services in all the government and private health sectors. Therefore, proper policies, strategies must be implemented through top-down approach. The study also provides insights regarding the role of self-introspection to be healthy, while enabling others to remain healthy and safe. On the contrary, negative influence of digital media, which highlights the need for the government to implement policies to regulate the information that the media delivers, especially social media. The conceptual framework derived from the findings of the study could be found in Figure 1. The study also provokes researchers, employees and health officials to design guidebooks and prepare oneself as well as others to create immediate awareness and knowledge among the common people for any such similar catastrophes in future. --- CONCLUSION The major findings of this study highlighted the positive role of the strength provided by the social connections to the patients and caregivers. In the cases where support was not provided, although the reasons varied, both caregivers and patients did experience psychological disturbances. While patients were mostly concerned with isolation and tensed surrounding, the caregivers were anxious about their uncertain futures, financial crisis and well-being of their loved ones. Although the caregivers were not infected with the virus, they reported that they could also not be free from fear and agony due to multiple situational causes. Further, patients opined that adequate counseling services should be mandatory for the patients as well for themselves; whereas caregivers shared that, their mental health also needs greater attention. In addition, although family members and friends supported the patients, as well as caregivers, the support of the neighbors was compromised most of the time due to the fear of contacting the disease. However, after recovery, the neighbors welcomed them to the extent that they sought help regarding preventive measures. This study also highlighted the ease to connect with loved ones through digital technology which was found to be a blessing in disguise. The voluntary help provided by the participants highlighted the concern for social welfare and shows the positive use of digital technology propounding social connectedness. The study also revealed the way the attitude toward the COVID-19 patients is considerably different among the individuals due to certain existing stigma, varied information through social media as well as uncertainty about the prognosis of the disease and their treatment. Further, this states a dire need of proper awareness, counseling, and several other researches, which can bridge the gap between the existing practices, varied perceptions and attitude which may further help us in treatment of COVID-19 as well to enrich the theoretical, cultural as well as policy-based contributions. --- LIMITATIONS AND FURTHER RECOMMENDATION Owing to the tensed situation created by the pandemic and reduced interaction among individuals, the researchers experienced difficulty in obtaining a varied representation of sample, e.g., socio-demographics that in essence play a major role in attitudes of individuals toward study variables. Moreover, several important variables, such as personality, religion, caste, age, comorbidities, possible financial constraints, and traditional practices to avoid the illness, variations of administration to tackle such issue and available resource on or before COVID-19 could be incorporated in future studies which may provide a cure paradigm for any such calamities in future. Integration of multiple disciplines, i.e., social science subjects with medical science and technology could be taken together to understand, as well as map the understanding of acts toward such situations in the future. --- DATA AVAILABILITY STATEMENT The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation. --- ETHICS STATEMENT The studies involving human participants were reviewed and approved by Departmental Review Board, Central University of Karnataka. The patients/participants provided their written informed consent to participate in this study. --- AUTHOR CONTRIBUTIONS VP, AA, NM, and RG conceived and performed the study design, data collection, and mastered the data. SJ, AA, RG, and GP ran the data analysis. ER, GL, and GP discussed the results. VP and AA wrote the manuscript with the support of RG, GP, and ER. SJ, ER, and GL supervised the project and manuscript preparation. All authors contributed to the article and approved the submitted version. --- SUPPLEMENTARY MATERIAL The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpsyg. --- 2021.634621/full#supplementary-material Conflict of Interest: The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. Copyright <unk> 2021 Pandey, Astha, Mishra, Greeshma, Lakshmana, Jeyavel, Rajkumar and Prabhu. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

Although COVID-19 pandemic has re-orientated humans to be more physically healthy and hygienic, it has also persuaded humans to create affiliations and experience a sense of belongingness through social networks and digital technologies. However, amidst these changes, experiences of COVID-19 patients and their perception of the outside world's attitudes toward them appears to be less attended in literature which formed the basis for the current study's objectives. Using qualitative methodology, the present study explored the experiences, perceptions and attitudes of patients and their care-givers' toward COVID-19. The thematic analysis emerged with four major themes. Psychological Experiences of People was generated prominently with sub-themes indicating the perceived experiences like fear of spreading diseases to others, and the need for psychological counseling. Attitude of others toward patients and caregivers revealed that family members and relatives played a major positive role on the patient's mental health, however, the neighbor's stigmatized attitude led to several undesired behaviors. Social Connectedness was another major theme derived from the study. Altruistic volunteers, a sub-theme of Social connectedness have indicated that amidst these negative factors, one can spread social harmony by motivating and supporting the victims with basic needs, financial support, hope and social empathy. Opinions of participants for digital technology through technological aids and preventive measures emphasized an overall positive attitude as it helped the society, in general to maintain social connections as well to curb the rate of COVID-19 cases.

Introduction Isolation has been defined in various ways in the scientific literature and is often confused with loneliness and exclusion. It has been defined as the objective measure of having minimal interactions with others (social isolation) and as the subjective feeling of dissatisfaction with the low number of social contacts maintained (emotional isolation or loneliness) [1]. Social isolation has also been defined as living without companionship, social support or social connection [2]; as a state in which a person or group feels the need or desire to become more involved with others but is unable to establish that contact [3]; and as a state in which the individual lacks a sense of social belonging and bonding or commitment to others and has few social contacts that do not constitute quality relationships [4]. Along the same lines, loneliness or emotional isolation has been defined as the (subjective) feeling of lack or loss of companionship, while they consider that social isolation is the (objective) absence of contacts and interactions between an older person and their social network [5]. The parallelism between "objective loneliness-social loneliness-social isolation" and "subjective loneliness-emotional loneliness-emotional isolation" is obvious and, for many authors, shapes the meaning of both terms. In general, social isolation refers both to the absence of satisfactory social relationships and to a low level of participation in community life. Several studies analysed the relationship between social isolation and loneliness on the health of the population. The review conducted by Mushtaq et al. [6] found that loneliness and social isolation can lead to various psychiatric disorders such as depression, alcohol use disorder, child abuse, sleep problems, personality disorders and Alzheimer's disease. Other authors have found that having social support is related to a lower perception of pain [7] and experiencing a greater life satisfaction [8,9]. It should be added that the health and well-being consequences of social isolation and loneliness in old age are increasingly being recognized. Courtin and Knapp [10] searched nine databases for empirical papers investigating the impact of social isolation and/or loneliness on a range of health outcomes in old age (128 items from 15 countries were included in the scoping review). These authors found that depression and cardiovascular health are the most often-researched outcomes, followed by well-being. Almost all studies found a detrimental effect of isolation or loneliness on health. However, causal links and mechanisms are difficult to demonstrate, and these authors suggest that further investigation is warranted. They found a paucity of research focusing on at-risk sub-groups and in the area of interventions. This also leads to various physical disorders such as diabetes; autoimmune disorders such as rheumatoid arthritis and lupus; and cardiovascular diseases such as coronary heart disease, hypertension, obesity, physiological ageing, cancer, poor hearing and poor health. In addition, the review by Crewdson [11] indicates that loneliness and social isolation in the elderly population is shown to be a significant independent risk factor for several conditions such as poor health behaviours (e.g., smoking and alcohol consumption), physical health problems (e.g., hypertension and motor decline), and psychiatric conditions (e.g., depression and cognitive impairment). In addition to this, a systematic review and meta-analysis of longitudinal studies [12] report that loneliness was positively associated with an increased risk of dementia; while older people's sense of identity helped them to compensate and adapt to the negative effects of the well-being of memory loss [13]. In the last 20 years, several interventions were developed to reduce the impact of social isolation and loneliness on the health of older people. The effectiveness of these interventions has been analysed in various systematic reviews [5,[14][15][16][17][18][19][20][21][22]. Among the conclusions of these reviews are that only a small proportion of these interventions are effective. It is observed that the interventions that demonstrated some effectiveness are those that include some educational component, social or group activities, well-trained staff facilitating social activity or support and those that are aimed at a specific population group (e.g., widowers, women, etc.). In addition, when thinking about interventions with people in social isolation, who are unlikely to attend a senior centre voluntarily, it is important to highlight the more individualized interventions, with a home approach by a professional, to assess the health and needs of the elderly person in social isolation and to serve as a link between the elderly person and the network of normalized social and health resources. In addition, interventions seem to be more effective if they include more than one methodology. In this same sense, outreach processes are not sufficient as the only element of the intervention but can be very useful if they are combined with engagement and intervention strategies. In addition, the high-quality selection, training and support for intervention facilitators and coordinators is one of the most important factors underpinning successful interventions. On the contrary, two factors are noted that make interventions more likely to be successful. On the one hand, interventions are more effective if the older person is involved in all steps: the planning, implementation and evaluation of the intervention. On the other hand, interventions are more likely to be successful if they use community resources and provide the community with tools and the capacity to help. Fakoya [21] points out that the individuality of the experience of loneliness and isolation may cause difficulty in the delivery of standardized interventions. In this regard, there is no one-size-fits-all approach to addressing loneliness or social isolation, hence the need to tailor interventions to suit the needs of individuals, specific groups or the degree of loneliness experienced. Therefore, future research should aim to discern which intervention works for whom, in what particular context and how. Furthermore, the quality of the evidence base is weak and further research is required to provide more robust data on the effectiveness of interventions [20]. Based on the above evidence, it is necessary to design and implement interventions at the community level to reduce the loneliness and social isolation of older people, to order to minimize the negative effects they can have on the health of this population. This study aims to describe the components of the Psychological Support Service for Socially Isolated Elderly People (PSIE), as well as analysing the effectiveness of a community intervention based on an outreach strategy to combat situations of social isolation in the elderly population of the city of Madrid (Spain). --- Materials and Methods --- Design and Procedure for the Application of the Psychological Support Service for Socially Isolated Elderly People (PSIE) To design this service, on the one hand, an extensive search of bibliographic information was carried out with different sources, trying to locate research projects or services in operation at a public or private level whose object of intervention was elderly people in social isolation; on the contrary, a qualitative study was carried out through groups of experts in the area of social exclusion, the elderly and other vulnerable groups (for example, homeless people and severe mental illness). The result of this work was the proposal of a psychological home care service for elderly people in social isolation. This research aims to present the effectiveness of the data after their implementation. This service is based on 5 basic principles: quality, outreach, case management, personal assistance and continuity of care, and it is composed of three protocols: contact and engagement, evaluation and intervention. In this case, it is a psychological care service integrated within the municipal care network of Madrid City Council. The general objective of the PSIE is to achieve an approach to frail elderly people with a clear risk of social isolation that guides their situations towards inclusion and normalization, assigning the appropriate social resources in each case, or as a last resort, supporting involuntary institutionalization and/or legal incapacitation. The specific objectives are the following: 1. To carry out an engagement work (through the outreach strategy) on people in social isolation, both with the PSIE and with the normalized social-sanitary network. 2. To carry out an assessment that is as exhaustive as possible depending on the case, including data on health assessment, psychosocial functioning, level of disability and unmet needs. 3. To carry out a psychosocial intervention to cover the person's needs through a process of inclusion and normalization or, as a last resort, to support the process of nonvoluntary institutionalization and/or legal incapacitation. --- 4. Strengthen the social network of the elderly person in isolation, working with potential informal support agents in the process of contact/engagement, assessment and intervention. --- 5. Destigmatize the person in isolation, both in their community environment and in their family context. The three PSIE action protocols are described below: (1) Contact and engagement protocol The strategy for approaching the elderly person is through home visits, carried out by PSIE psychologists alone or accompanied by facilitating agents to establish the therapeutic relationship (professional facilitating agents or from the social and family network). The system of appointments in a centre is not used, but instead, the professionals of the street team contact the person in the community. Contact is always made in a person's own environment, preferably in their home, and if it is impossible to contact them in their home, attempts are made in the places in the community that they frequent, with total respect and care for data protection and the person's privacy. Non-face-to-face contact can be carried out via emails, phone calls or by sending letters. The frequency of contacts depends on the response of the elderly person to the presence of the team; however, there will always be a minimum of one contact per month, either personal or non-face-to-face. The objective is to contact the person regularly to help him or she get used to the team's presence and gradually build trust. The presence of other professionals during PSIE visits is assessed according to the intervention needs and the degree of trust and engagement with the elderly person. (2) Assessment Protocol Once an initial level of engagement has been achieved, the psychological and social assessment of the new case begins in parallel. A clinical (physical health); functional (global functioning and disability); psychological, cognitive and social functioning; and needs assessment are performed. For this purpose, the following assessment instruments are used and will be administered in all cases: -Home Health Outcome Scales for People Over 65 (HoNOS65+) [23] in its Spanish adaptation [24]. -Global Assessment of Functioning (GAF) [25] in its Spanish adaptation [26]. -WHO Brief Disability Assessment Scale (WHO-DAS-S) [27] in its Spanish adaptation [28]. -Camberwell Needs Assessment Questionnaire for the Elderly (CANE) [29] in its Spanish adaptation [30]. -Other assessment scales can be added to this assessment as considered appropriate according to the specific case. Given the characteristics of the sample regarding the rejection of previous interventions and reluctance to the presence of professionals, none of the selected instruments will be completed by the person being assessed; instead, they will be assessed by the PSIE professionals based on the information gathered through direct observation and interview, both from the user and from other sources of information (professional facilitators or from the social and family networks of the person being assessed). In this way, the process of assessment of this population is simplified. (3) Intervention Protocol During the intervention, the main objective of the PSIE is not to directly treat the social or mental health problems that the person may present; instead, the priority approach is to normalize the care and connect or "reconnect" the person with normalized social and health networks, where a solution can be given to those social or health problems that the person may present. In other words, the PSIE facilitates contact between the elderly person and social and health services (both primary and specialized care) in their area, so that they can receive the care they need on each occasion. In these cases, the work of PSIE focuses on the detection of problems, is a social and psychological accompaniment for the beginning and continuity of the necessary treatments and interventions in each case, and helps to change the attitude of the elderly person in social isolation, encouraging them to accept the help they need, which they previously rejected. This research of awareness, contact with elderly people's realities and the acceptance of resources is conducted globally through psychological intervention strategies: -Intervention in the clinical area (physical health) and psychological and cognitive functioning: The objective is the acceptance and engagement of the person (or in some cases, maintenance) with normalized health services, both at primary level (primary health) and specialized level (especially neurology, geriatrics and mental health: Psychiatry/psychology). -Intervention in global functioning and disability: The objective is the acceptance and engagement of the person in those resources that assist in the activities of daily living, both basic and instrumental (home help service, home meal service, laundry service, etc.). These services are usually provided by social services, but may also come from other organizations, such as religious or voluntary organizations, etc. -Intervention in social functioning: The objective is to create, improve and/or maintain the person's social support network, both at the informal and formal level, favouring the person's approach to family and friends or creating new social networks through volunteering, municipal senior centres or community associations; working on neighbourhood support and the stigma of the person in social isolation in the community; and promoting leisure activities, etc. -Intervention in the area of unmet needs: The objective is to cover the unmet needs of the person. This area is transversal to the areas mentioned above. For the intervention, support will be provided by professional facilitators and the social and family network of the person in social isolation whenever possible. PSIE psychologists participate in training sessions on social isolation in the elderly for professionals of the social-health network, as well as specific training for professionals involved in specific cases, as well as family members, friends or other members of the support network of the elderly person in social isolation. In those cases, in which the level of deterioration is irrecoverable or there is an imminent risk to the health of the person or that of others, the opening of a file in the Office of the Public Prosecutor for the Elderly will be supported to assess the advisability of nonvoluntary institutionalization in a hospital or residential centre and/or legal incapacitation as a protective measure in a situation of clear lack of protection and risk. Once the intervention has been completed, the PSIE follows up on all cases closed within three to six months after closure. The follow-up consists of retrieving information on the persons served through contact with the user, the professionals involved (mental health, social services, residential centre, etc.) and/or the person's informal network (family members, caregivers, etc.). To elaborate on this follow-up, the same measures are used as in the assessment protocol. If the person has returned to a situation of social isolation, the case is reopened by the social services. On the other hand, if the situation remains non-isolated, the PSIE closes the case definitively, leaving the follow-up to the municipal social services. --- Study of the Effectiveness of the Psychological Support Service for Socially Isolated Elderly People (PSIE) 2.2.1. Participants and Procedure Cases are detected and referred to PSIE through the General Directorate for the Elderly and the Social Services of the Madrid City Council (Spain). For the PSIE to begin its intervention, the case must meet certain inclusion criteria: be 65 years of age or older; live alone or live with other people over 65; have uncovered social and/or health needs; have little or no social support network; refuse the assistance offered to cover their needs from the normalized social and/or health services; municipal social services have intervened, without having resolved the situation of social isolation, and consider that all avenues of intervention have been exhausted. The intervention was carried out with all the people who were detected and referred to the PSIE, so the sample is incidental. The sample of the effectiveness study consists of 68 people over 65 in social isolation in the city of Madrid (Spain), in which the PSIE carried out the complete intervention, until its closure, which allowed for the collection of pre-and post-treatment data. All participants in whom at least one post-treatment measurement could be taken were considered for analysis. Regarding the procedure carried out, the pre-measures were taken at the initial stages of the intervention, reflecting the situation of the elderly person at the time of referral to the PSIE; while the post-measures were taken 3 to 6 months after the end of the PSIE intervention. Given that the conditions of the sample are of great deterioration and chronicity in their state of isolation, the post-treatment measures are delayed until a few months after the end of the intervention to assess not only the achievement of the intervention objectives but also their consolidation in the medium term. For ethical reasons, it was not possible to have a control group, since, given the seriousness of the situation of the cases of elderly people in social isolation detected and referred to the PSIE, the intervention was started for all of them. Informed consent was requested from the people in the sample. The study was conducted following the Declaration of Helsinki, and the protocol was approved by the Department of Programming, Evaluation and Development of the General Directorate for the Elderly of the Madrid City Council, code 001/085/233.01/489.01. --- Measures The following instruments were included in the assessment: -Health and Psychosocial Functioning, through the use of the Spanish adaptation of the Health Outcome Scales for People Over 65 (HoNOS65+) [24]: An outcome assessment instrument for the elderly population consisting of 12 scales that assess various physical and psychosocial problems, which are scored in a range of severity from 0 (no problem), 1 (minor problem), 2 (mild problem), 3 (moderate problem) to 4 (severe problem), with a total sum score added. Scores range from 0-48 points, with higher scores indicating poorer health and functioning. The Spanish adaptation of the scale [24] has adequate psychometric criteria (statistically significant concurrent validity values ranging from 0.23 to 0.89; and inter-rater reliability values close to 1). -Global functioning, using the Spanish version of the Global Assessment of Functioning (GAF) [26]. This scale, included in Axis V of the DSM-IV-TR diagnostic manual (American Psychiatric Association, 2000), measures the psychosocial functioning of the individual, through the assessment of the severity of symptoms and activity impairment. It is designed for the general population. It is a 100-item scale, in which scores close to 0 indicate poorer functioning. It has good psychometric properties with an intraclass correlation coefficient of 0.57. -Disability, using the Spanish adaptation of the WHO Disability Assessment Short Scale (WHO-DAS-S) [28]. The short version of the semi-structured interview WHO-Psychiatric Disability Assessment Schedule (WHO-DAS). It consists of 4 areas to be assessed by the clinician during the interview. Based on the information obtained, the clinician must score the person's impairment on a visual analogue scale from 0 ("no disability") to 5 ("maximum disability") according to defined criteria. It has good psychometric properties, with an intraclass correlation coefficient ranging from 0.40 (for the family/housing category) to 0.74 (personal care), so that half of the specified disability categories have kappa values above 0.50, and the other half between 0.40 and 0.50. -Social health needs, using Section 1 of the Spanish adaptation of the Camberwell Needs Assessment Questionnaire for the Elderly (CANE) [30]: This instrument is specifically designed to measure the needs of people with a possible mental disorder over 65 years of age. Section 1 (whether there is a current problem in any of the areas assessed) was used, which consists of 24 areas of assessment relating to the user/patient, plus another 2 (A and B) that refer to the main caregiver but have not been included in the present study given the special characteristics of isolation of the sample and the lack of a main caregiver. Each of these 24 areas is scored: 0 (no serious problem); 1 (there is a problem, but it is covered due to the help provided); 2 (a serious problem is present and not covered, despite any help being provided). The Spanish version of the scale has a good concurrent validity, high test-retest reliability (kappa between 0.65 and 1) and high inter-rater reliability (kappa between 0.60 and 1) [30]. --- Data Analysis Descriptive statistics were calculated for socio-demographic and psychological variables (mental health, cognitive impairment, health and psychosocial functioning, global assessment of functioning, disability and social health needs). An analysis was carried out to study the possible influence of gender in the initial sample on the different variables assessed, using Chi-squared and Student's t-tests for independent samples, with measures of effect size in each case. A study of the effectiveness of PSIE was carried out with an analysis of pre-and post-treatment measures. Student's t statistic for related samples was used, as well as the effect size of Cohen's d statistic. For the assessment of the possible influence of gender on the results of the intervention, a 2 <unk> 2 repeated measures ANOVA (pre-/post-measures <unk> gender) was carried out. All analyses were conducted using SPSS v. 22.0 (IBM, Armonk, NY, USA) [31]. --- Results --- Sociodemographic Data of the Sample The sample (N = 68) contained a majority of women (73.5%), and the mean age was 81.8 years (MIN = 66, MAX = 97, SD = 6.9). The majority of the sample (86.8%) did not have a partner, 72.1% lived alone and 79.4% did not have children (Table 1). Regarding mental health, 65.2% of the sample presented symptoms compatible with a severe mental disorder, the most frequent being psychotic disorder (22.7%), alcohol use disorder (16.7%), personality disorder (15.2%), anxiety disorders (10.4%) and mood disorders (10.4%). In 66.7% of the cases, this mental health disorder had not been detected before the PSIE intervention and only 4.8% were receiving some type of mental health treatment. Forty-seven per cent of the sample presented moderate (37.9%)-to-severe (9.1%) cognitive impairment. This level of serious deterioration concerning the mental health of the sample is reflected in the fact that 20.6% of the persons attended by the PSIE already had an open judicial file requesting protection and custody measures at the time of referral (Table 2). Concerning health and psychosocial functioning, (HoNOS65+ scales), it can be seen that the scales where the lowest scores are found are those related to depressive disorders, psychotic disorders and suicide risk. On the other hand, the sample presents moderateto-severe difficulties on scales 9, 10, 11 and 12, related to constructive leisure and social support, activities of daily living and living conditions. This result is supported by the mean score on the GAF instrument close to 50, indicating any severe impairment in social or occupational activity. Scale 5, related to physical health problems and disability, also shows a high mean, indicating relevant problems in this area. These data are corroborated in the disability scale WHO-DAS-S, where the difficulties in the four subscales reach a mean of serious disability (Table 2). Regarding the needs assessment, the sample shows an average unmet needs score of almost 11, which is a high level of need considering that the instrument includes such basic needs as food, medical follow-up or housing conditions (Table 2). --- Description and Analysis of Gender Differences in Psychological Variables As mentioned above, the percentage of men and women in the sample is very different, with a greater presence of women (73.5%). An analysis was carried out to study the possible gender differences and their relevance with respect to the initial situation of the sample. The results are shown in Table 3. There are statistically significant differences in the five measures. First, cognitive impairment, had a higher percentage of moderate-severe impairment in women (56.2%) than in men (22.2%); however, if we look at the effect size, the relationship of this measure with the gender variable is not very strong (Phi = 0.304). Medium effect sizes are found in the other four measures: problems associated with alcohol use (scale 3, HoNOS65+) showed a mild alcohol use problem in men and was close to 0 in women; problems related to hallucinations or delusions (scale 6, HoNOS65+) showed a mild problem in women and was close to 0 in men; other mental problems (scale 8, HoNOS65+) showed mild problems in both cases, but slightly more pronounced in men; and an average level of disability in terms of personal care and survival (scale A, WHO-DAS-S) was higher in men (level close to moderate disability) than in women (level close to serious disability). In the rest of the measures evaluated, there are no statistically significant differences according to gender. Regarding the reasons for the closure, 63.3% of the cases were resolved after taking legal protection actions to ensure the well-being of the elderly; and in the remaining 36.7%, the situation of social isolation was redirected towards normalization through the acceptance of formal aid and/or the strengthening of the social network of the elderly. --- Effectiveness Study --- Health and Psychosocial Functioning (HoNOS65+) The general trend is towards maintenance or improvement in all variables of the HoNOS65+ scales (Table 4). In all of them, the mean score decreases, indicating improvement (lower scores indicate better functioning). Significant differences are found between pre-and post-treatment in 7 of the 12 scales included in the instrument. The differences are found both in scales of psychopathological symptomatology and behavioural alteration, such as scales 1, 3, 6 and 8; scales of functioning more related to social aspects, such as scales 9 and 12; and scale 11, more related to housing conditions, both in terms of hygiene and habitability, decoration, stimuli present, etc. Concerning the total score of the HoNOS65+, significant differences were also observed between pre-(21.5) and post-treatment (13.72), showing a significant improvement of the sample in this variable. Medium-to-large effect sizes were found in all scales where there were significant differences, with the largest being found in scales related to social support, leisure activities and living conditions. Differences in the HoNOS65+ total score also indicate a large effect size. --- Global Assessment of Functioning (GAF) The general trend is towards improvement in global functioning, with a higher posttreatment mean (in this instrument, higher scores indicate better global functioning). These differences between pre-and post-treatment are statistically significant and have a medium effect size (Table 5). --- Disability (WHO-DAS-S) The pre-/post-data in the WHO-DAS-S instrument are for 46 persons. There are significant differences and large effect sizes in the four areas of disability included in the instrument, with the trend in all of them being towards improvement, that is, towards the more adequate functioning of the persons in the sample (Table 6). In all of them, the mean score decreases, which indicates an improvement, since, in this scale, lower scores indicate lower levels of disability. For the analyses carried out with this instrument, "obvious disability" (2) to "maximum" (5) was chosen as the criterion for the highest severity, since from a score of 2 on this scale (obvious disability), interference with a person's social adaptation begins to be considered. --- Social Health Needs (CANE) In the 24 areas of need, the score 0 indicates that there are no needs; 1 means that the need exists but is covered; and 2 indicates that the need is not covered. The general trend is towards fewer unmet needs, decreasing from an average of 10.68 unmet needs at pre-treatment to an average of 1.44 unmet needs at post-treatment (Table 7). --- Study of the Influence of Gender on Intervention Outcomes The gender variable does not seem to have an influence on any of the outcome measures studied, except for alcohol use. In this variable, the results indicate that a reduction in use was achieved in both men and women, but in women a total reduction in alcohol consumption was achieved. (Table 8). --- Discussion This study aimed to describe the components of the PSIE and to analyse the effectiveness of a community intervention. The sample shows a moderate-to-severe level of mental health deterioration, with high percentages of symptoms compatible with at least one severe mental disorder, including psychotic symptoms, alcohol use disorder, personality disorders, anxiety and mood disorders; the high presence of moderate and severe cognitive impairment; and even judicial records requesting protection measures before a referral to PSIE. Given the characteristics of the sample of isolation and rejection of social and health resources, in most cases, they are not receiving any mental health treatment. The data highlight the need for mental health care in this population. In addition, the people in the sample of this study had a very high mean score of unmet social and health needs before the intervention (10.68) compared to the average in the general population (7) [32]. Concerning interventions, the duration and number of contacts required in each case will depend on the individual characteristics of each person treated, as evidenced by the variability in the minimum and maximum ranges of both the duration and number of contacts. However, the data point to interventions of a medium and long duration with frequent contacts (around an average of 3 contacts per month). Men seem to show a tendency towards greater alcohol use and greater difficulties in self-care; while women in the sample present a worse cognitive status and a greater (although low) presence of problems with hallucinations or delusions. However, it should be noted that the results of the analysis by gender show that the differences in this variable in most of the measures are not significant and the effect is low or medium. The results of the effectiveness study indicate that the PSIE is an intervention programme that serves to improve the scores of the people in the sample in all the variables studied. Regarding the changes in the measure of health and psychosocial functioning (HoNOS65+), the general tendency is towards maintenance or improvement in all the variables of the HoNOS65+ scales, with significant differences in 7 of the 12 subscales. It should be noted that the intervention is effective for reducing behavioural disturbance, alcohol use, psychotic symptomatology, other emotional problems such as anxiety, and depressive symptomatology, as well as improving the frequency of social contacts, a person's living environment, and participation in leisure activities, the latter being where the intervention shows the clearest effects. These achievements were made possible because the PSIE has engaged with elderly people in a situation of social isolation and is able to put them in contact with regular health and social services. In this way, many people with mental disorders have received care from mental health services, which has led to a reduction in mental health symptoms (mainly delusions, anxiety, depression and sleep problems). On the other hand, the PSIE has facilitated the acceptance of different social resources by people in social isolation, such as the home help service for household chores, personal hygiene, shopping, medical appointments, the home meal service, the Tele assistance service, or access to a residence for the elderly. This could explain the improvement in the HoNOS65+ subscales of living conditions, occupational activities and social relations. In addition, these social relationships may also have been improved by the inclusion of volunteers in the cases of people in whom emotional loneliness was detected. On the other hand, some HoNOS65+ subscales have not experienced significant improvements, such as cognitive problems (which are difficult to improve given that in many cases there are processes of dementia), and problems of disability and functioning in activities of daily living. It seems reasonable that this last aspect would be improved by incorporating a person from the home help service. However, the home help service was accepted on many occasions for cleaning the home but not for personal hygiene. This could explain the limited improvement in this variable. As for the more general measures of functioning (GAF) and disability (WHO-DAS-S), statistically significant changes and medium and large effect sizes were obtained, so it can be affirmed that the PSIE, being an intervention that connects the person in social isolation with regular social and health services, helps to improve general functioning in the specific areas of personal care, occupational, family and leisure functioning. Finally, the PSIE has proven to be effective in meeting most of the 24 needs identified by the CANE, with a general trend towards fewer unmet needs, decreasing from an average of 10.68 unmet needs at the pre-intervention time to an average of 1.44 unmet needs at the post-intervention time. It should be noted that the PSIE has served as a link to cover these needs by encouraging the elderly person in a situation of social isolation to accept different services from social and health professionals, facilitating the social support of people in their neighbourhood (shopkeepers, neighbours) and resuming contact and relationships with family members. Regarding the effect of gender on the results of the intervention, we only found an effect on the alcohol use variable. PSIE has achieved positive results in both men and women, without gender appearing to be a relevant variable in these results. Studies included in systematic reviews and meta-analyses on the effectiveness and efficacy of interventions to reduce social isolation do not examine gender as a predictor of intervention outcomes [14][15][16][17][18][19][20][21][22]. This makes it difficult to compare our results on gender differences with previous studies. Previous studies are interested in the analysis of variables such as the intervention type (face-to-face, group or individual intervention, etc.) or the variable age, as possible predictors of the effectiveness of the interventions. Future research should focus on the study of the gender variable as a predictor of the effectiveness of interventions that aim to reduce emotional loneliness and social isolation in the elderly, in order to better plan interventions to prevent situations of loneliness and social isolation. The effectiveness of PSIE might be because it is a programme that includes many of the ingredients recommended by different authors in different meta-analyses and reviews regarding the effectiveness of interventions to reduce social isolation and loneliness [5,[14][15][16][17][18][19][20][21][22]. Some of the components of PSIE that could explain its effectiveness are individualized interventions, with a home-based approach by the professional, aiming to assess the health and needs of the older person in social isolation and serving as a link between the older person and the normalized social health network of resources. In addition, PSIE has carried out high-quality selection, training and support for the facilitators and coordinators of the interventions (home helpers, social and health workers, volunteers, among others). Another factor to be taken into account in the approach to older people in social isolation is the need to involve the elderly person in all steps, from the planning, implementation and assessment of the intervention. This factor is relevant in the PSIE, always respecting the times and rhythms of change established by the people in social isolation. Finally, a decisive factor in the effectiveness of the service is the use of community resources (volunteers, neighbours, senior centres, etc.) that have served to provide tools and support the capacity of the community itself. A strong aspect of the present study is that it has been applied to a large sample of elderly people in social isolation and that it is one of the few studies that has been applied in the environment of the elderly person and not in a social or health centre. In other words, it was possible to access people in a situation of extreme social isolation, who do not ask for help, and who refuse any service. One of the main limitations of the study that we found was related to the lack of a control group. For ethical reasons, it was not possible to have a control group, since, given the seriousness of the situation of the cases of elderly people in social isolation detected and referred to the PSIE, the intervention was initiated in all of them. Regarding future research directions, it is important to include efficacy studies with a control group, in addition to conducting studies with even larger samples that include sufficient numbers of men and women to perform gender-differential analyses. On the other hand, since the COVID-19 pandemic, it seems interesting to promote the study of the efficacy of remote interventions to mitigate emotional loneliness and social isolation in the elderly [17,18,[33][34][35][36]. Information and communication technologies and other traditional networks can foster social support to deal with social isolation and loneliness. Further research is required to provide more robust data on the effectiveness of interventions, including gender differences. It is interesting to further develop theoretical understandings of how successfully interventions mediate social isolation and loneliness. --- Conclusions Older people in social isolation in the city of Madrid are in most cases women living alone, without children, and with a high

Loneliness and social isolation in the elderly population can be shown to be a significant independent risk factor for several conditions, such as poor health behaviours, physical health problems and psychiatric conditions. Although, in the last 20 years, several interventions have been developed to reduce the impact of social isolation and loneliness on the health of older people. However, only a small proportion of these interventions are effective. This study aims to describe the components of the Psychological Support Service for Socially Isolated Elderly People (PSIE), in addition to analysing the effectiveness of a community intervention based on an outreach strategy to combat situations of social isolation in the elderly population. The sample consisted of 63 people over 65 years of age from the city of Madrid (Spain), detected by the socio-health services as people at risk of social isolation. Sociodemographic, mental health, health and psychosocial functioning, global functioning, disability and socio-sanitary needs were evaluated with observational scales. Descriptive statistics were calculated for sociodemographic and mental health variables. An analysis was carried out to study the possible influence of gender in the initial sample on the different variables assessed, using Chi-squared and Student's t-tests for independent samples, with measures of effect size in each case. A study of the effectiveness of PSIE was carried out with an analysis of pre-and post-treatment measures. A Student's t-test was used for related samples, as well as the effect size of Cohen's d statistic. For the assessment of the possible influence of gender on the results of the intervention, a 2 × 2 repeated-measures ANOVA (pre-/post-measures × gender) was conducted. Regarding mental health, 65.2% of the sample presented symptoms compatible with a severe mental disorder, the most frequent being psychotic disorder (22.7%), alcohol use disorder (16.7%), personality disorder (15.2%), anxiety disorders (10.4%) and mood disorders (10.4%). The gender variable does not seem to have an influence on any of the outcome measures studied. The results of the effectiveness study indicate that the PSIE is an intervention programme that serves to improve the scores of people in the sample in all variables that the programme studied. Some of the components of PSIE that could explain its effectiveness are individualized interventions, with a home-based approach by professionals, serving as a link between the older person and the normalized social-sanitary network. Further research is required to provide more robust data on the effectiveness of interventions.

related to the lack of a control group. For ethical reasons, it was not possible to have a control group, since, given the seriousness of the situation of the cases of elderly people in social isolation detected and referred to the PSIE, the intervention was initiated in all of them. Regarding future research directions, it is important to include efficacy studies with a control group, in addition to conducting studies with even larger samples that include sufficient numbers of men and women to perform gender-differential analyses. On the other hand, since the COVID-19 pandemic, it seems interesting to promote the study of the efficacy of remote interventions to mitigate emotional loneliness and social isolation in the elderly [17,18,[33][34][35][36]. Information and communication technologies and other traditional networks can foster social support to deal with social isolation and loneliness. Further research is required to provide more robust data on the effectiveness of interventions, including gender differences. It is interesting to further develop theoretical understandings of how successfully interventions mediate social isolation and loneliness. --- Conclusions Older people in social isolation in the city of Madrid are in most cases women living alone, without children, and with a high percentage of severe mental disorders. This study was able to demonstrate the effectiveness of a psychological intervention with elderly people in social isolation with some specific components, such as individualized attention, intervention in the person's own home and environment, the engagement of the elderly person with the normalized social and health care networks to meet their needs, and an additional intervention strategy focused on training and support for both the professionals involved in the case and the elderly person's informal social support network. --- Data Availability Statement: The datasets used during the current study are available from the corresponding author on reasonable request. --- Conflicts of Interest: The authors declare no conflict of interest.

Loneliness and social isolation in the elderly population can be shown to be a significant independent risk factor for several conditions, such as poor health behaviours, physical health problems and psychiatric conditions. Although, in the last 20 years, several interventions have been developed to reduce the impact of social isolation and loneliness on the health of older people. However, only a small proportion of these interventions are effective. This study aims to describe the components of the Psychological Support Service for Socially Isolated Elderly People (PSIE), in addition to analysing the effectiveness of a community intervention based on an outreach strategy to combat situations of social isolation in the elderly population. The sample consisted of 63 people over 65 years of age from the city of Madrid (Spain), detected by the socio-health services as people at risk of social isolation. Sociodemographic, mental health, health and psychosocial functioning, global functioning, disability and socio-sanitary needs were evaluated with observational scales. Descriptive statistics were calculated for sociodemographic and mental health variables. An analysis was carried out to study the possible influence of gender in the initial sample on the different variables assessed, using Chi-squared and Student's t-tests for independent samples, with measures of effect size in each case. A study of the effectiveness of PSIE was carried out with an analysis of pre-and post-treatment measures. A Student's t-test was used for related samples, as well as the effect size of Cohen's d statistic. For the assessment of the possible influence of gender on the results of the intervention, a 2 × 2 repeated-measures ANOVA (pre-/post-measures × gender) was conducted. Regarding mental health, 65.2% of the sample presented symptoms compatible with a severe mental disorder, the most frequent being psychotic disorder (22.7%), alcohol use disorder (16.7%), personality disorder (15.2%), anxiety disorders (10.4%) and mood disorders (10.4%). The gender variable does not seem to have an influence on any of the outcome measures studied. The results of the effectiveness study indicate that the PSIE is an intervention programme that serves to improve the scores of people in the sample in all variables that the programme studied. Some of the components of PSIE that could explain its effectiveness are individualized interventions, with a home-based approach by professionals, serving as a link between the older person and the normalized social-sanitary network. Further research is required to provide more robust data on the effectiveness of interventions.

Introduction As a standpoint for the development of modern society, a smart city is also an important part of modern society. With the development of the Internet in recent years, there has been a lot of room for development in community construction innovation, especially in the construction of smart communities, ecological construction for the elderly, and community grid construction [1,2]. e smart community mainly has the following factors. e first is to make full use of modern science and technology; the second is to achieve the integration of resources; the third is to improve the comprehensive level of social governance; the fourth is to enable the modernization and facilitation of residents' lives [3]. So, in this sense, smart communities are a relatively new type of new concept of social governance. Building an intelligent community can not only improve the overall construction level of the community, but also reduce the cost of social governance and maximize the improvement of residents' living services with minimal investment so as to better build a multilevel intelligent social service system [4,5]. Driven by new social governance development requirements and a new round of technological and industrial revolution, smart communities will become an important way to innovate community governance, improve community services, and meet the needs of people's good lives. Smart communities rely on advanced technology, information technology, and infrastructure to effectively integrate the needs of the community's property services, convenience services, government affairs, medical treatment, communication, and communication in an intelligent information system to provide community residents with more convenience and comfortable, efficient, and safe living environment, so as to realize people's daily life and services in an intelligent and convenient way. e development of a smart community is not only the result of the continuous maturity, development, innovation, and change of the community, but also the extension of intelligence in the field of social governance [6]. e smart community service system is jointly constructed by the government and the social forces under its leadership [7]. It relies on advanced information technology and infrastructure in the smart community to meet the diverse and multilevel needs of community residents [8]. e goal is to live and innovate the community governance model, to provide community residents with various community service systems and operating mechanisms. Niaros et al. [9] have proposed a multidimensional, multilevel study on the construction of smart communities. e research content includes support at the government level, socioeconomic needs, the dimension of the construction environment, and the level of interconnected transportation. Aiming at the dimensions of individual needs, intelligent community management, and family-level dimensions, American administrative scientists Li et al. [10] and Ted Kriv<unk> [11] proposed that public services are not necessarily fully provided or directly provided by the government. e government provides this function of public services. A variety of other more efficient ways of providing services can be sought by community workers within the community [12]. Raven et al. [13] pointed out that nonprofit organizations should provide community services as the main body, and the role of the government is mainly to advocate and support. Verrest and Pfeffer [14] believe that the government, market, community, and other entities should complement each other and establish a multicenter supply model. Community services should not be a single or one-to-one supply model, emphasizing the diversity of supply [15]. Barns et al. [16] believe that the smart community is an autonomous system that can be thoroughly perceived and interconnected by relying on advanced scientific and technological information technologies such as the Internet of ings and cloud computing. is system has many functions such as community home detection, emergency alarm, and nursing services [17]. Eckhoff and Wagner [18] have used the perspective of distance learning to build a model of smart community network learning, which includes the three dimensions of smart community development, knowledge innovation, and coaching and learning support, and explain the connotation of smart community from a content of smart community. Sandulli et al. [19] believe that a smart community is a kind of converged network built in the context of the Internet of ings. e structure of a smart community is a network architecture, which is malleable and inclusive [20]. Kumar et al. [21] believe that smart communities are different from traditional communities. ey are provided for the residents of the community. ey are a united front and are efficient, healthy, safe, low-carbon, energy-saving, and environmentally friendly living environment and methods. e construction of smart communities should adhere to sustainable development. Sookhak et al. [22] considered that smart communities are key construction subprojects of smart cities and analyzed the connotation of smart communities from the perspective of smart cities, as a component of smart city construction. Kitchin [23] believes that a smart community is a community security network integration system. e construction and development of the community should first consider the issue of community security [24]. Using wireless sensor technology, a sensor network can be built, and community security can be placed in it. e research on smart communities has covered many aspects including the basic concept connotation, information technology support and standards, construction status issues and countermeasures, and smart community evaluation. e theoretical framework of the smart community has been gradually improved, but the service system of the smart community is the core component of the smart community. Its research is still in the exploratory stage, and there is less research in this area. Concentrating on the research of a single part of the service content such as old-age care and medical care, the research is relatively scattered, there are many deficiencies, and there is a lack of systematic analysis and design. If smart communities rely solely on technology and do not cooperate with the appropriate service system, they will restrict the development of smart communities, hinder the intelligent development of social governance, and build a process of coconstruction and shared governance. e selection of this paper is mainly based on the intelligent management of the community as the research target. e theoretical concept of smart city construction under dynamic games is combined with the specific construction practice of the smart community to analyze the problems and shortcomings of the current status. e research in this paper has both a certain theoretical significance and corresponding practical value. On the one hand, it can enrich the united front related theories of social services and community governance, and on the other hand, it can provide some data theory and practical experience to the latecomers studying smart communities. --- Study on Community Governance of Smart City Construction in Dynamic Game --- Analysis of the Game Relationship between Smart City Construction and Community Governance. Smart city planning and construction groups, business groups, and social public groups are the most important stakeholders in urban planning and construction. e three groups' decision-making interaction directly affects the process of local urban planning and construction. Smart city planning and construction groups make decisions intending to maximize the public interest in the region. ey are the leaders, guides, and maintainers of urban planning and construction. ey regulate the production and living behavior of enterprise groups through effective policy regulations and incentives. Decision making with the goal of maximizing benefits is the main body and promoter of green production and determines its production strategy according to market preferences and government policies; social public groups make decisions with the goal of maximizing personal interests and are the supervisors and markets of society of consumers, 2 Complexity determining their decision-making behaviors based on the utility of the goods to themselves and the individual gains and losses in social construction. Due to different interest preferences among the three major groups, different strategies will be selected, which will affect the effectiveness of urban planning and construction. erefore, by establishing an interactive game model among the three main bodies, a united front is formed. Using evolutionary game method to analyze the strategy evolution process and trend of each group has positive significance for the optimal mechanism design and system construction of urban planning and construction. is draws an expanded game model for participation in smart city construction (Figure 1). e smart community service system is a system architecture and operating mechanism for smart communities to implement community governance and services. It aims to provide community residents with efficient, convenient, and intelligent services through information technology and facilities. e smart community service system is supported by technology and information such as the Internet of ings, information technology, and urban big data. It is the responsibility of the government. rough various methods such as government purchase, it integrates the multiresources of community neighbourhood committees, enterprises, social organizations, and community residents to jointly achieve and provide diverse and multilevel services such as government services, life services, elderly services, united fronts, medical services, volunteer services, and property services, integrate the realization of various services, and use information technology and infrastructure to create online intelligence. e service application platform and offline intelligent service management platform provide various types of quality services to community residents through two major platforms. Community residents can participate in community governance through online smart service application platforms and offline smart service management platforms, or by directing prosecutions and expressing opinions and satisfaction with service subjects. e operation mode of the community service system is shown in Figure 2. To meet the needs of residents is the fundamental purpose. Under the new public service theoretical framework that advocates "service instead of helm" and "pursuit of public interest," the construction of a smart community service system should take service as the king and pursue the public interests of community residents as the foundation. It is necessary to accurately grasp community residents demand for various services, such as public service government affairs, social security, employment services, family planning and other services, convenient water and electricity provision, communications, pensions, medical care, housekeeping, shopping, and other services. To understand the degree of community residents' demand for various services, supply methods, and efficiency of service delivery, keep up with the times, dynamically update, and continuously optimize the service supply methods to improve service levels. e construction of a smart community service system should focus on the needs of residents, build high-rise houses, formulate top-level designs, and arrange service methods and approaches. --- Experiment Apparatus Hypothesis 1. ere are two important groups in the process of urban planning and construction: (1) Group A is the department responsible for urban planning and construction; (2) Group B is a relevant functional department that belongs to Group A and involves urban planning and construction. ey each have two decision-making schemes. Group A can choose to actively implement the country's construction requirements for urban planning (Strategy 1) or passively implement this requirement (Strategy 2), while Group B can choose to support Group A work (Strategy 1) or not to support Group A work (strategy 2). Individuals in both groups have limited rationality and will adopt imitation strategies to adjust their game strategies based on the choices of others in the group. --- Hypothesis 2. e cost of Group A's active execution is C 1, which is expressed as the human and financial input for execution; at the same time, it will receive the opportunity benefit I 1, which is expressed by the superior government's reward and the government's reputation improvement. At this time, if Group B supports, they will get auxiliary gains, which shows that they have saved manpower and material resources. When Group A performs negatively, it will be punished by the higher government for future financial penalties, which will be recorded as a fine L 1. <unk> Hypothesis 3. e cost of Group B's support for construction is C 2, which is also manifested in human and financial input. At this time, when Group A is actively performing, it will also receive opportunity income I 2, which is shown as an increase in reputation and political performance. When it is not supported, if Group A insists on actively implementing the financial penalty L 2 for Group B, it will show the reduction of financial support for the next year. <unk> Hypothesis 4. In the process of comprehensive urban planning, the proportion of Group A which chooses to actively execute the strategy is represented by x (x <unk> [0, 1]), and the proportion of people who choose to execute the passive strategy is (1-x). e proportion choosing to support the construction strategy is represented by y (y <unk> [0, 1]), and the proportion choosing not to support the construction strategy is (1-y). Table 1 shows the model parameters. Based on the above assumptions, list the game matrix between Groups A and B as shown in Table 2. E x1 y I 1 -C 1 + A 1 <unk> +(1 -y) I 1 -C 1 + L 2 <unk>.(1) e expected return from negative construction can be expressed as E x2 y -L 1 <unk> +(1 -y) -L 1 <unk>.(2) e average expected return can be expressed as 4 Complexity E x xE x1 +(1 -x)E x2 x y I 1 -C 1 + A 1 <unk> +(1 -y) I 1 -C 1 + L 2 <unk> <unk> <unk> +(1 -x) y -L 1 <unk> +(1 -y) -L 1 <unk> <unk> <unk>.(3) Write the replicated dynamic equation of Group A: <unk> G(x) dx dt x E x1 -E x <unk> x(1 -x) I 1 -C 1 + L 2 + yA 1 -yL 2 + L 1 <unk>. (4 ) It is not difficult to find that the independent variables in this equation only exist in the part of x (1-x), so let the part that has nothing to do with x (I 1 -C 1 + L 2 + yA 1 -yA 2 + L 1 ) be expressed by a formula. erefore, the expected return of Group B's choice of transformation and upgrading strategy can be expressed as E y1 x I 2 -C 2 <unk> +(1 -x) -C 2 <unk>.(5) e expected benefits of choosing a nontransformation upgrade strategy can be expressed as E y2 x -L 2 <unk>.(6) e average expected return can be expressed as E y yE y1 +(1 -y)E y2 y x I 2 -C 2 <unk> +(1 -x) -C 2 <unk> <unk> <unk> +(1 -y) x -L 2 <unk> <unk> <unk>.(7) Write the replicated dynamic equation of Group B: G(y) dy dt y E y1 -E y <unk> <unk> y(1 -y) xI 2 -C 2 + xL 2 <unk> <unk>.(8) By analyzing the evolutionary system composed of dynamic equations G (x) and G (y), we analyze the evolutionary stability strategy (ESS) of Group A and Group B, and the equilibrium point stability of the system can be determined by referring to the local stability analysis method of the Jacobi matrix. e Jacobi matrix of the game is listed as Jacobi (1 -2x)<unk> x(1 -x) A 1 -L 2 <unk> y(1 -y) I 2 + L 2 <unk> (1 -2y)<unk> <unk> <unk>. (9 ) Determinant of J: Traces of J: Det J (1 -2x) I 1 -C 1 + L 2 + yR 1 -yL 2 + L 1 <unk>(1 -2y) xI 2 -C 2 + xL 2 <unk> <unk> -y(1 -y) I 2 + L 2 <unk>x(1 -x) R 1 -L 2 <unk>. (10 ) Tr J (1 -2x) I 1 -C 1 + L 2 + yR 1 -yL 2 + L 1 <unk> +(1 -2y) xI 2 -C 2 + xL 2 <unk> <unk>.(11) --- Result Analysis --- Analysis of Evolutionary Stability in Dynamic Games. To more intuitively express the evolution process of emergency decision information sharing behavior, this paper uses MATLAB simulation tools to simulate conclusions 1, 2, and 3 at different initial points under different circumstances and evolve to the aggressive strategy. In Figure 3, the horizontal axis represents time (t) and the vertical axis represents the proportion of active strategies selected in smart city planning and construction (x) or number to govern (y). at is, when the value of the parameters meets the above conditions, the evolutionary stabilization strategy of smart city planning and construction is active integration. For the convenience of analysis, under the conditions that the above evolutionary stability strategy is satisfied, it is assumed that the parameter value is E 3; <unk> 0.4; <unk> 1 0.8; <unk> 2 0.2; P 0.5; L 0.6; C 1 1.5. It can be seen from the evolution process in Figure 3 that no matter what the initial proportion of the active integration strategy for smart city planning and construction is, the final evolutionary stability strategy ratio converges to 1, that is, the active integration strategy. It can be seen that the smaller the initial proportion of smart city planning and construction, the faster the convergence rate, and all the curves will not cross before reaching stability. e cost of actively integrating information in smart city planning and construction is less than the benefit of information value brought about by the improvement of the ability of community governance to actively share information and transform it. Regardless of the initial proportion of community management's choice of active sharing strategy, the final evolutionary stable strategy ratio converges to 1, that is, the active sharing strategy. It can be seen that the smaller the initial proportion of community governance, the faster the convergence rate, and all the curves will not cross before reaching stability. At this time, the risk brought by enterprises to reduce the density of information is greater than the net income of active sharing and less than the sum of the net income of active sharing, the reward of active sharing, and the penalty. e evolutionary stability strategy for smart city planning and construction is 1, and the evolutionary stability strategy for smart city planning and construction is 0. e evolution diagram is shown in Figure 4. It can be seen from Figure 4 that when the initial proportion x0 of the smart city planning and construction cluster is greater than 0.72, that is, 0.8 and 0.9 in the figure, the final evolutionary stability strategy of the smart city planning and construction cluster is 1 (active integration). When the initial ratio x0 of the planning and construction cluster is less than 0.72, that is, 0.2, 0.4, and 0.6 in the figure, the final evolutionary stabilization strategy for the planning and construction cluster of the smart city is 0 (negative integration). In the same way, under the abovementioned prerequisites, the evolutionary stability strategy of community governance clusters under different initial proportion conditions. When the cost of integrating information in the smart city planning and construction department is low enough, the strategy of smart city planning and construction has nothing to do with the choice of community governance strategies. Active integration of information is an evolutionary and stable strategy for smart city planning and construction, that is, the difference between the benefits of actively integrating information and the cost of actively sharing information. When it is large enough, the strategy of community governance has nothing to do with the strategic choice of smart city planning and construction. Active sharing of information is an evolutionary and stable strategy for community governance. In the planning and construction process of smart city, we need to plan and construct information, integrate cost, share information of community governance, and improve the ability of information conversion, improving information hiding coefficients, and actively integrating information conversion capabilities and smart city planning and construction, government incentives both the level and the active sharing of profits by enterprises can promote the evolution of smart city planning and community governance to a dualpositive strategy (active integration, active sharing), forming a united front. --- Analysis of Dynamic Game Decision Methods. e effectiveness of the corresponding emergency plan for different levels of community governance E: the prior probability is P (i) of the four levels of community governance in the first stage, and when the actual level of the nuclear accident is i, but the joint meeting judges the level of accident, the conditional probability is P (ij). e above data in the actual nuclear emergency are discussed and decided by the emergency decision-making group, that is, the joint meeting on countermeasures for atomic disasters, relying on the judgment of experts in nuclear emergency decisionmaking. e analysis of this example mainly explores the operability and theoretical feasibility of the model. To facilitate the analysis, the above three types of fuzzy data will be assumed. It is assumed that the corresponding scheme corresponds to the utility E of different levels of community governance, as shown in Figure 5. It is also assumed that when the actual level of the nuclear accident is i, the conditional probability that the 6 Complexity joint meeting judges that the level of the accident is j is P (ij), and the value of the conditional probability is shown in Figure 6. According to the calculation steps, all posterior probabilities of the four levels of nuclear accidents can be calculated, as shown in Figure 7. --- Dynamic Game Optimization Analysis. e effect of the user's strategy level on the expected equilibrium benefits of community governance is mainly studied through numerical simulations. In order to control the impact of the data transmission rate ratio on the experiment, the data transmission rate ratio in the experiment is fixed at 0.7. e abscissa of Figure 8 represents the user's strategy level, and the ordinate represents the expected total benefits of community governance. e image visually shows that the expected total benefits of community governance in both communities decrease as the user's policy level increases. is is because a higher user policy level implies that more users will choose to delay access to the edge network. erefore, the user's strategy will reduce the expected total benefits of community governance in the competitive pricing model. Next, compare the differences in expected equilibrium returns of community governance in the face of "irrational" users and strategic users, as shown in Figure 9. e abscissa indicates the user's strategy level, and the ordinate indicates the percentage difference in the expected equilibrium benefits of community governance from the transition from 'irrational' users to strategic users. It can be found from the figure that the image function values are all below 0, which verifies the previous conclusions: the user's strategic behavior will reduce the equilibrium benefits of community governance. In addition, it can be found that the higher the user's policy level, the greater the loss of community governance benefits, because strategic users will choose to delay access to the edge network, making the price elasticity of demand greater. Changing the value from 0 to 1, we found that the simulation results are consistent with the above, so this simulation result can explain the problem. Because the distributed dynamic optimization scheme we proposed is cooperative, each independent subsystem has its own objective function, but in the cooperative dynamic game optimization calculation, all the subsystem goals are to make the entire overall optimization performance of the system the largest. So the objective function used by each subsystem is to improve the optimization performance of the entire system and should be consistent. For the selection of the objective function, the general approach is to linearly add the objective functions of each independent subsystem, and the coefficients are determined according to the actual production conditions, situation, and united front. erefore, we integrate the above three objective functions into the overall objective function of a large system, as shown in Figure 10. Under the RCGO dynamic optimization scheme, the iterative convergence and number of iterations of the optimization solution in each optimization cycle are shown in Figure 11. e Euclidean norm is that of the difference between the optimal controls obtained by the last two consecutive iterations in each optimization cycle. e numbers are all less than the set iteration termination tolerance <unk> 0.001. is means that, in each optimization cycle, the RCGO dynamic optimization solution can successfully converge when solving large-scale dynamic optimization problems of process networks. At the same time, the number of iterations of the optimization solution in each optimization cycle is less than the set maximum number of iterations. e results of these numerical simulations show that the RCGO dynamic optimization scheme can effectively solve large-scale dynamic optimization problems. Next, we compare the economic performance indicators of the process network, closed-loop control performance indicators, and real-time calculation performance indicators to the advantages and disadvantages of RCO and RCGO dynamic optimization solutions. When the RCO and RCGO dynamic optimization solutions are used to solve the largescale dynamic optimization problem of the process network, the economic benefit indicators of the process network are 1.3541 and 0.9160, respectively, and the closed-loop control performance indicators of the process network during the 100-s online optimization operation are as shown in Figure 12. From the results of the above numerical simulations, it can be seen that when the RCO scheme is used to solve the large-scale dynamic optimization problem of the process network, the economic benefit indicators and closed-loop control performance indicators of the process network obtained are optimal. is is because the RCO scheme simultaneously considers all the control variables in the process network when optimizing the solution, so the results of the optimization solution under this scheme often take When using the RCGO scheme, the scheme considers not only the overall optimization of the process network but also the optimality of each subsystem. is will cause the global optimality of the process network dynamic optimization problem to be often not guaranteed. In other words, the quality of the economic benefit index and the closed-loop control performance index of the process network is the result of weighing the optimality of the subsystem and the overall optimality of the process network. When using the RCO scheme, the average optimization calculation time in each sampling period is longer than that using the RCGO scheme. In each optimization cycle, the RCO scheme needs to solve a large-scale dynamic optimization problem. Due to the large scale and complexity, it often takes more time and cost. At the same time, the average optimization calculation time in each sampling period under this scheme will Complexity exceed the sampling period length, which means that the RCO scheme will be difficult to apply to the real-time solution of the process network dynamic optimization problem. When using the RCGO solution, the system model size of the optimization solution is smaller and the solution complexity is lower than that of the RCO solution. erefore, the solution can guarantee the real-time solution of the dynamic optimization problem of the process network. In summary, when solving complex large-scale dynamic optimization problems, the RCGO scheme has greater advantages than the traditional centralized optimization schemes in the comprehensive index consisting of system economic benefits, closed-loop control performance, united front achievement, and optimization real-time performance. --- Conclusion is article summarizes the theoretical research of smart communities, identifies and classifies the main interests of local smart cities based on previous studies, and finally determines that smart cities and community governance are the main objects of this research. Secondly, based on the assumption of bounded rationality, this study builds a dynamic game model between smart cities and community governance and finally draws the stable evolution conditions of the ideal equilibrium ESS. Guiding the behavior of smart cities mainly depends on establishing appropriate incentive mechanisms. Guiding smart cities mainly depends on setting up a reasonable reward and punishment system. e enthusiasm for stimulating public participation is to increase their recognition of smart city construction. Strengthening social supervision is beneficial for local governments to change their behaviors and fulfil their regulatory responsibilities. en, for local governments to carry out research on the game of interests among their internal departments, the attitude of community governance is crucial. Whether it is an internal department or an external department, the degree of cooperation has a positive impact on the attitude of community governance construction. e concentration of regional environmental functions is more conducive to advancing the united front of smart cities. --- Data Availability e data used to support the findings of this study are available from the corresponding author upon request. --- Conflicts of Interest e authors declare that they have no known conflicts of interest or personal relationships that could have appeared to influence the work reported in this paper.

We present a research on smart city planning and community governance using dynamic game methods, analyze the current status and problems of the current smart community service system, and put forward countermeasures and suggestions based on the global smart community development experience. rough smart city planning and construction, the game model of government governance and information sharing between communities and decision making is obtained, and the two-dimensional replication dynamic system equations of smart city planning and construction agencies and communities are obtained from smart city planning and construction. e evolutionary stability strategy and a united front are analyzed. e strategy achieves the ideal stable conditions. We define the relevant definition of a smart community and discuss the purpose and significance of the research, especially the importance and necessity of the construction of the smart community. Based on the current situation of the construction of smart communities, we conduct specific research, discuss the deficiencies and shortcomings in the current construction of smart communities, and find out the causes of these problems. Finally, based on learning from the relevant work experience, corresponding improvement measures are proposed for the existing problems from the united front, top-level design, social participation, platform construction, and talent training.