Abstract
OBJECTIVE: Recent preliminary studies indicated a seasonal association of BMI at admission to inpatient treatment for anorexia nervosa (AN), indicating lower BMI in the cold season for restrictive AN. An impaired thermoregulation was proposed as the causal factor, based on findings in animal models of AN. However, findings regarding seasonality of BMI and physical activity levels in the general population indicate lower BMI and higher physical activity in summer than in winter. Therefore, we aimed to thoroughly replicate the findings regarding seasonality of BMI at admission in patients with AN in this study. METHOD: AN subtype, age- and gender-standardized BMI scores (BMI-SDS) at admission, mean daily sunshine duration and ambient temperature at the residency of 304 adolescent inpatients with AN of the multi-center German AN registry were analyzed. RESULTS: A main effect of DSM-5 AN subtype was found (F(2,298) = 6.630, p = .002), indicating differences in BMI-SDS at admission between restrictive, binge/purge and subclinical AN. No main effect of season on BMI-SDS at admission was found (F(1,298) = 4.723, p = .025), but an interaction effect of DSM-5 subtype and season was obtained (F(2,298) = 6.625, p = .001). Post-hoc group analyses revealed a lower BMI-SDS in the warm season for restrictive AN with a non-significant small effect size (t(203.16) = 2.140, p = .033; Hedges'g = 0.28). Small correlations of mean ambient temperature (r = -.16) and daily sunshine duration (r = -.22) with BMI-SDS in restrictive AN were found. However, the data were widely scattered. CONCLUSIONS: Our findings are contrary to previous studies and question the thermoregulatory hypothesis, indicating that seasonality in AN is more complex and might be subject to other biological or psychological factors, for example physical activity or body dissatisfaction. Our results indicate only a small clinical relevance of seasonal associations of BMI-SDS merely at admission. Longitudinal studies investigating within-subject seasonal changes might be more promising to assess seasonality in AN and of higher clinical relevance.