The influence of school in the development of eating disorders: a record-linkage study

Abstract

BackgroundClinical impression is that rates of eating disorders vary between schools; we are not aware of any previous research on this topic. We aimed to investigate whether rates of eating disorders in 16–20-year-old girls vary between upper secondary schools, and to test the hypothesis that school characteristics are associated with rates of eating disorders, even after accounting for characteristics of individual students. MethodsThis multilevel longitudinal study made use of record-linkage data from Stockholm County, Sweden. Participants were 55 824 Swedish-born girls completing secondary education in 2001–10 at 409 schools. Outcome was any diagnosed eating disorder at 16–20 years, as defined by an ICD (9 or 10) or DSM-IV code, or inferred from an appointment at a specialist eating disorder clinic. Multilevel modelling was used to separate individual and school level effects. The Stockholm Regional Ethical Review Board approved the study. FindingsA 4·4% variation in incidence of eating disorders between schools was seen; after taking individual risk factors into account variation between schools was 2·9% (95% CI 1·5–5·0). Schools with a higher proportions of girls than boys had an increased incidence of eating disorders: for each 10% increase in the proportion of girls at a school, the odds ratio for eating disorders was 1·07 (95% CI 1·01–1·13, p=0·017). For each 10% increase in the proportion of parents with post-secondary education, the odds ratio for eating disorders was 1·14 (1·09–1·19, p<0·0001). InterpretationOur findings show that the contextual aspects of a school environment are associated with increased incidence of eating disorders. Incidence rates of eating disorders are higher in schools characterised by a high proportion of female students and of students with highly educated parents. To our knowledge, this is the first study to investigate whether rates of eating disorders vary between schools; however, use of registry data means that individuals who did not seek treatment would not have been studied. FundingHB was supported by a Wellcome Trust Institutional Strategic Support Fund (via the Elizabeth Blackwell Institute).