Exploring Reciprocal Associations Between Self-Reported Anxiety and Eating Disorder Symptoms Longitudinally: A Bivariate Latent Change Score Approach

Consistent data support an association between anxiety and eating disorders (EDs), and theoretical models of EDs suggest that anxiety may be involved in the etiology and maintenance of ED symptoms over time. However, the directionality of relations between these variables remains under-characterized, particularly within treatment settings. We used bivariate latent change score models to explore longitudinal associations between anxiety and ED symptoms in a sample of ED patients (N = 548, 93.2% female, Mage = 21.16, 78.8% White, 79.6% Non-Hispanic/Latinx) throughout intensive treatment and at 6- and 12-month follow-up. Best-fitting models exploring change in each variable independently (i.e., univariate models) suggested that changes in ED and anxiety symptoms decreased over time, but that change was influenced by an individual's symptom severity at the previous timepoint. Models exploring associations between changes in both variables over time (i.e., bivariate latent change score models) suggested the best fit for a model where both anxiety and ED symptoms at one timepoint were associated with later change in the other. Specifically, parameters within these models suggested that higher levels of anxiety were associated with increased subsequent reductions in ED symptoms, whereas elevations in ED symptoms were associated with decreased later reductions in anxiety. Our findings suggest that anxiety and ED symptoms are intricately related both within and outside of intensive treatment. Future multi-modal research exploring real-time links between anxiety and ED symptoms throughout treatment is critical to extend this work and inform improvements in targeted, mechanistic interventions for this population.