Background: Treatment of anorexia nervosa (AN) sometimes requires hospitalisation, which is often lengthy, with little ability to predict individual trajectory. Depicting specific profiles of treatment response and their clinical predictors could be beneficial to tailor inpatient management. The aim of this research was to identify clusters of weight recovery during inpatient treatment, and their clinical predictors.
Methods: A sample of 181 inpatients who completed a treatment programme for AN was included in a retrospective study. A latent class mixed model approach was used to identify distinct weight-gain trajectories. Clinical variables were introduced in a multinomial logistic regression model as predictors of the different classes.
Results: A four-class quadratic model was retained, able to correctly classify 63.7% of the cohort. It encompassed a late-rising, flattening, moderate trajectory of body mass index (BMI) increase (class 1), a late-rising, steady, high trajectory (class 2), an early-rising, flattening, high trajectory (class 3) and an early-rising, steady, high trajectory (class 4). Significant predictors of belonging to a class were baseline BMI (all classes), illness duration (class 2), and benzodiazepine prescription (class 3).
Conclusion: Predicting different kinetics of weight recovery based on routinely collected clinical indicators could improve clinician awareness and patient engagement by enabling shared expectations of treatment response.
Keywords: anorexia nervosa; body mass index; cluster analysis; eating disorders.
© 2024 Eating Disorders Association and John Wiley & Sons Ltd.