Antipsychotic effects on anthropometric outcomes in anorexia nervosa: a retrospective chart review of hospitalized children and adolescents

BACKGROUND: Off-label antipsychotic use is not uncommon in youth with anorexia nervosa (AN), aiming to enhance suboptimal weight restoration, yet its efficacy remains debated, especially in youth. METHODS: Retrospective chart review of consecutively admitted inpatients (ages 8–18 years) with restricting/binge-purge AN, comparing youth with versus without antipsychotic treatment regarding baseline factors, treatment, and anthropometric outcome characteristics including all patients and matched subgroups. Matched subsamples were also compared regarding faster versus slower weight change (median split). Furthermore, within-subject analyses compared weight gain trajectories before versus after antipsychotic use in antipsychotic-treated youth. These results were then compared in a pre-/post design with the matched control group without antipsychotic treatment, using the mean duration until antipsychotic use in the antipsychotic-treated group as the dividing timeline, controlling for a potential order effect, in that later rather than earlier antipsychotic treatment for AN may be more successful. RESULTS: Of 294 youth with AN (median age = 15.2 (interquartile range = 14.0, 16.6) years, females = 96.6%, restricting subtype = 81.0%, hospitalization duration = 98.2 ± 43.2 days), 44 (15.0%) underwent 52 antipsychotic trials (olanzapine = 63.5%). In multivariable analyses, antipsychotic use was independently associated with younger age, childhood physical abuse history, comorbid borderline personality traits, and lower pre-antipsychotic weight gain (p < 0.0001). In unmatched groups, antipsychotic-treated versus non-treated youth had significantly lower discharge anthropometric parameters, longer inpatient treatment, and lower weight change/week (p < 0.001–p = 0.005), without significant differences between olanzapine and non-olanzapine antipsychotics (p = 0.27–0.44). Non-significant antipsychotic effects on weight outcomes were confirmed in (1) matched subgroups of antipsychotic-treated versus non-treated youth (n = 43 each), (2) youth with faster versus slower weight gain (n = 21 vs. n = 22), and (3) antipsychotic-treated youth when comparing weight change before versus after antipsychotic use (n = 31). Moreover, in antipsychotic-treated youth, weight change/week remained significantly lower versus matched non-antipsychotic-treated youth (n = 31) both before (p = 0.053) and after (p = 0.006) the median time (5 weeks) until antipsychotic use. CONCLUSIONS: In this naturalistic study, clinician's antipsychotic choice, given to a more severely ill subgroup of adolescents with AN, did not significantly improve overall worse weight change trajectories during inpatient treatment, even in matched subgroups. Randomized trials in individuals reflecting real-world samples are needed to evaluate the utility of antipsychotic treatment in youth with AN. SUPPLEMENTARY INFORMATION: The online version contains supplementary material available at 10.1186/s40337-023-00862-4.