Seasonal and subtype differences in body mass index at admission in inpatients with anorexia nervosa

Action Editor: Ruth Weissman Abstract Objective: In the general population, body weight is—on average—higher in the winter than in the summer. In patients with anorexia nervosa (AN), however, the opposite pattern has been reported. Yet, only a handful of studies exist to date that suffer from small sample sizes and inconsistent results. Therefore, the current study examined seasonal effects on body weight in a large sample of patients with AN to dissolve previous inconsistencies. Method: Clinical records of N = 606 inpatients (95.4% female) who received AN treatment at the Schoen Clinic Roseneck (Prien am Chiemsee, Germany) between 2014 and 2019 were analyzed. Results: Patients with restrictive type AN had lower body mass index at admission in the winter than in the summer. This difference was not found for patients with binge/purge type AN and patients with atypical AN. Discussion: Individuals with restrictive type AN show seasonal variations in body weight that are opposite to seasonal variations in body weight in individuals without AN. These seasonal effects are specific to the restrictive subtype and cannot be found for the binge/purge or atypical subtypes. Future studies that replicate this effect in other cultures or latitudes and that examine the mediating mechanisms are needed.


| INTRODUCTION
Seasonal variations in body weight have been consistently reported.
In European and North American countries, body mass index (BMI) is on average higher in the winter than in the summer (Mehrang, Helander, Chieh, & Korhonen, 2016). This pattern has been found both in the general population (Visscher & Seidell, 2004) and in overweight adults participating in a behavioral weight loss intervention (Fahey, Klesges, Kocak, Talcott, & Krukowski, in press). Possible mechanisms that explain increases in body weight during the winter include a reduction in physical activity and increased consumption of high-calorie foods compared to the summer period (Lloyd & Miller, 2013;Ma et al., 2006;Sabba g, 2012;Sturm, Patel, Alexander, & Paramanund, 2016;Westerterp, in press).
In a small sample of 37 adolescents with anorexia nervosa (AN; 68% restrictive type) from the Netherlands, Carrera et al. (2012) reported the peculiar finding that participants during the cold season (October to April) had lower BMI than participants in the warm season, suggesting that seasonal changes in body weight in individuals with AN may be opposite to the seasonal changes found in persons without AN. Two later studies extended this finding, suggesting that seasonal variations in BMI in individuals with AN depend on AN subtype. Specifically, in a study from Spain, Fraga et al. (2015) found that BMI at admission in the cold season (November to April) was lower in adolescent inpatients with restrictive type AN than in binge/purge type AN, whereas subtypes did not differ in BMI in the warm season.
Similarly, in a study from Germany, Born et al. (2015) found that BMI at admission during autumn and winter was lower in adult inpatients with restrictive AN than in binge/purge type AN, whereas subtypes did not differ in BMI during spring and summer. Both studies, however, were based on small sample sizes (n = 86 and n = 68) with particularly small groups of patients with binge/purge type AN.
To overcome these limitations, Kolar et al. (2018) examined clinical records of 304 adolescent inpatients from a multi-centric database in Germany. Although they indeed found an interactive effect between season and AN subtype on BMI at admission, the nature of this interaction indicated higher body weight during the cold than the warm season in patients with restrictive type AN. Hence, the current state of affairs is that the study with the largest sample size to date found seasonal differences in BMI in patients with restrictive type AN that are opposite to the seasonal differences found in three small-scale studies and, therefore, no clear conclusions can be drawn about the existence and the direction of seasonal BMI variations in AN. Thus, we examined clinical records of more than 600 patients with AN with the aim of dissolving previous inconsistent findings.
Previous studies applied different categorizations of warm versus cold season (e.g., differing in the assignment of October and April; Carrera et al., 2012;Fraga et al., 2015;Kolar et al., 2018). Thus, we   (Carrera et al., 2012). In line with this, preliminary evidence based on case reports suggests that heat treatment results in a reduction of hyperactivity, anxiety, and depression in patients with AN (Gutierrez & Vazquez, 2001), although results from randomized controlled trials have been mixed (Carrera & Gutiérrez, 2018). Another explanation posits that individuals with AN have deficient insulation due to reduced subcutaneous fat, and thus, the body expends more energy for thermoregulation at cold ambient temperatures (Fraga et al., 2015). Hypothetically, this process may be mediated by brown adipose tissue, which is activated by hypothermia, dissipates energy, and generates heat (Freemark & Collins, 2018). Finally, it has also been suggested that reduced exposure to sunlight in the winter and lower vitamin D concentrations may also influence body weight, for example, through increased depressiveness (Kolar et al., 2018).
At least two methodological aspects need to be considered when interpreting findings from the current and from the previous studies. One aspect refers to the definition of AN subtypes. In the current study, this differentiation was based on the ICD-10-GM, whereas other studies referred to the criteria in the fourth (Carrera et al., 2012;Fraga et al., 2015) or fifth (Kolar et al., 2018)

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available in the supplementary material of this article. F I G U R E 1 Mean body mass index at admission as a function of subtype and season. Error bars represent the standard error of the mean