Abstract / Summary
Children and adolescents receiving inpatient psychiatric treatment may be at increased risk of short-term weight gain-related changes, particularly during antipsychotic exposure. In this retrospective cohort study we analyzed a patient-level cohort of 776 unique patients younger than 18 years treated between 2022 and 2024, after exclusion of eating disorder diagnoses and repeated hospitalizations. Clinical data were extracted from electronic medical records and included demographic and clinical characteristics, antipsychotic treatment, and anthropometric measurements. The median hospitalization duration was 3.0 days (IQR 2.0-7.0). Repeated BMI-SDS measurements permitting calculation of BMI-SDS change were available for 228 patients. Mean BMI-SDS change was +0.071 (SD = 0.242). BMI-SDS increased significantly among patients exposed to antipsychotics (n = 152; mean change +0.107, p < 0.001), whereas no significant change was observed among patients without antipsychotic exposure (n = 76). BMI-SDS change differed across antipsychotic treatment groups (Kruskal-Wallis p = 0.0006; ϵ² = 0.081). However, olanzapine monotherapy was not significantly different from no antipsychotic treatment after correction for multiple comparisons, although exploratory analysis showed a higher risk of BMI-SDS increase ≥0.5 with olanzapine exposure (NNH = 3.7, 95% CI 2.2–10.2). In multivariable linear regression analysis, longer hospitalization (B = 0.0035 per day, 95% CI 0.002-0.005, p < 0.001) and any antipsychotic exposure (B = 0.0668, 95% CI 0.003-0.131, p = 0.042) were associated with higher follow-up BMI-SDS after adjustment for baseline BMI-SDS. Sex, age and F20-F29 diagnosis were not significantly associated with follow-up BMI-SDS. These findings suggest that short-term BMI-SDS change during child and adolescent psychiatric hospitalization is associated with antipsychotic treatment and length of hospitalization. The results do not support a significant olanzapine dose-response relationship and should be interpreted as anthropometric associations rather than evidence of metabolic dysfunction or causality.