Introduction
Responses in eating disorder symptoms following behavioural weight management interventions vary between individuals, and some may be at greater risk of poorer outcomes. We aimed to identify the risk factors in global eating disorder and binge eating symptoms among adolescents and adults receiving behavioural weight management interventions.
Methods
We systematically identified randomised trials of behavioural weight management interventions that assessed eating disorder outcomes and invited investigators to join the Eating Disorders In weight-related Therapy Collaboration and share individual participant data. Thirty-six trials contributed data from 4,455 participants. Associations between baseline demographic, mental health and psychosocial factors and global eating disorder symptoms (primary outcome) and binge eating symptoms (secondary outcome) at the latest follow-up were examined using two-stage random-effects individual participant data meta-analysis, adjusted for covariates. Multivariable prediction models were also developed using LASSO regression and multilevel modelling. Apparent performance, optimism and internal-external validity were assessed.
Results
Global eating disorder and binge eating symptoms decreased from baseline to the latest follow-up. Female sex and lower baseline self-esteem, quality of life and body image predicted smaller reductions in global eating disorder symptoms. Racial or ethnic majority status and higher baseline emotional eating, stress and disinhibition predicted smaller reductions in binge eating symptoms. No baseline factor predicted symptom increases relative to baseline. Prediction models included 13 trials (n=1,322) for global eating disorder symptoms and 11 trials (n=1,381) for binge eating symptoms. Baseline eating disorder symptoms, sex and depressive symptoms were the strongest predictors. Models explained 34% and 46% of outcome variance, respectively, with minimal evidence of overfitting.
Conclusions
Higher baseline eating disorder symptoms, female sex and depressive symptoms predicted smaller symptom reductions following behavioural weight management interventions. No individual factor predicted a clinical increase from baseline. Screening may help identify individuals who could benefit from monitoring and support.