Eating disorders more common in schools with more girls or more educated parents


Eating disorders are extremely prevalent in adolescents, occurring in 2.8% of the adolescent population (Swanson et al., 2011). Eating disorders have the highest mortality rate of psychiatric disorders in adolescents (Arcelus et al., 2011). However, the risk factors for eating disorders are not well understood.

Previous studies have shown that parental characteristics have a great impact on adolescent eating disorders, such as parent eating disorder prevalence. Other factors, such as the school environment, may also play an important role in how eating disorders manifest.

A new cohort study out today, led by fellow Mental Elf blogger Dr Helen Bould, aimed to investigate how school characteristics impact eating disorder prevalence. The present research examined whether female student populations and the higher levels of parental education is associated with changes in eating disorders prevalence.

Eating disorders have the highest mortality rate of any psychiatric disorder.

Eating disorders have the highest mortality rate of any psychiatric disorder.


As with their previous study, the researchers used data from the Sweden Youth Cohort to collect information from all 0-17 year olds in Stockholm County. Parental information was obtained via The Multi-Generation Register.

They had a strict inclusion criteria:

  • Eating disorders were defined as anorexia nervosa, bulimia nervosa or any other unspecified eating disorder. Attendance at a specialist eating disorder clinic was also included (see limitations)
  • The sample were born after 1982 and left high-school (or Gymnasiums as they’re called in Sweden) between 2002-2010
  • Only individuals who were born in Sweden were used in the sample, as it would be more difficult to get information about their parents at their birth
  • Only females were studied, due to different rates of eating disorders between the sexes
  • Only females who first presented eating disorders between the ages of 16-20 were included. This excluded any individuals who presented an eating disorder before the age of 16, and any individuals who did not have full data entry.

Information about 55,059 females from 409 schools was collected.

The researchers conducted mixed effects logistic regression analyses to determine changes in odds ratios (OR) for eating disorder occurrence when studying both individual and school variables.

The main analyses measured the change in odds ratios across schools with different proportions of female students and highly educated parents.


2.4% of adolescents between the ages of 16-18 in the sample had an eating disorder.

When controlling for individual risk factors, rates of eating disorders were higher in schools with:

  • A higher female student population (p = .018)
  • A higher proportion of children with at least one parent with higher education (p < .001).

These variables remained significant even when adjusting the regression model for school-related risk factors and each other.

This can be shown in the simplified table below.

% Parents with education beyond high school % Girls Predicted probability of Eating Disorders (%)
25 25 1.3
25 75 1.8
75 25 2.5
75 75 3.3
Schools with 75% girls have higher rates of eating disorders.

Schools with 75% girls have higher rates of eating disorders.


This research suggests that eating disorders are more common in schools with a greater proportion of a) female students and b) pupils with highly educated parents.

The authors propose multiple different reasons for the results:

  • Highly educated parents may encourage greater ‘perfectionism’ in their children, an attribute highly correlated with eating disorders (Holland et al., 2013)
  • Girl-only schools tend to view academic success as more important than co-ed schools (Tiggerman et al., 2001), and so this may result in an environment that encourages ‘perfecting oneself’.
Schools with more highly educated parents also have increased eating disorders rates.

Schools with more highly educated parents also have increased eating disorders rates.


  • As the authors note, anorexia nervosa and bulimia nervosa were not separated in the analyses due to limited statistical power. However, this is an important distinction, as bulimia nervosa proportions have shown to be more common in lower socioeconomic backgrounds compared to anorexia nervosa (Gard and Freeman, 1996). It may be the case that certain changes in school characteristics apply differently to anorexia nervosa over bulimia nervosa.
  • 64% of the cases of eating disorders were ‘inferred’ based on visits to eating disorders clinics. The remaining 36% were clinical diagnoses of anorexia nervosa and bulimia nervosa (22% and 14% respectively). This may be problematic, as some schools may have greater access to clinics whilst having similar rates of diagnosed eating disorders.
  • The researchers only counted eating disorder incidence if eating disorders were first diagnosed between the ages of 16-18. However, it is difficult to know for how much longer before this age adolescents may have been presenting symptoms of the disorder, and whether the onset occurred before the new school environment could have an effect. Considering the average age of onset of eating disorders is age 12 (Swanson et al., 2011), it would be important to study school factors on children at an earlier age.


This study has a vast and well-detailed sample. Only with such a great number of participants can a reliable logistic regression model be calculated. Register-based studies are very useful in population cohort research, and more research should be conducted in countries where this information is collected.

These findings are extremely important if we wish to make changes to school designs in the hope of reducing mental illness in adolescents. The mechanisms in single-sex girl schools need to be studied to better understand how eating disorder rates increase. Pupils need to be taught from an earlier age that eating disorders do not equal positive, better or perfect outcomes.

Interventions are needed in schools where students are at high risk of eating disorders.

Interventions are needed in schools where students are at high risk of eating disorders.


Primary paper

Bould H, De Stavola BL, Magnusson C, Micali N, Dal H, Evans J, Dalman C, Lewis G. (2016) The influence of school on whether girls develop eating disorders. International Journal of Epidemiology.

Other references

Ahrén JC, Chiesa F, Koupil I, Magnusson C, Dalman C, et al. (2013) We are family – parents, siblings and eating disorders in a prospective total-population study of 250,000 Swedish males and females. International Journal of Eating Disorders 46, 693–700. [PubMed abstract]

Arcelus J, Mitchell AJ, Wales J, Nielsen S. (2011) Mortality rates in patients with anorexia nervosa and other eating disorders. A meta-analysis of 36 studies. Arch Gen Psychiatry, 68, 724–731. [PubMed abstract] [Mental Elf blog of this paper]

Gard MC, Freeman CP. (1996) The dismantling of a myth: a review of eating disorders and socioeconomic status. International Journal of Eating Disoders, 20(1), 1-12. [PubMed abstract]

Holland LA, Bodell LP, Keel PK. (2013). Psychological factors predict eating disorder onset and maintenance at 10-year follow-up. European eating disorders review: the journal of the Eating Disorders Association, 21(5), 405-10.

Swanson S, Crow S, Le Grange D, Swendsen J, Merikangas K. (2011) Prevalence and correlates of eating disorders in adolescents. Archives of General Psychiatry, 68, 714–723. [PubMed abstract]

Tiggemann M. (2001) Effect of gender composition of school on body concerns in adolescent women. International Journal of Eating Disorders, 29(2), 239-43. [PubMed abstract]

Photo credits


Share on Facebook Tweet this on Twitter Share on LinkedIn Share on Google+