Abstract
BACKGROUND: Pharmacological efforts to treat anorexia nervosa (AN) have predominantly repurposed medications that treat conditions with overlapping symptoms and yielded generally disappointing results. Despite limited empirical support, SSRIs are often prescribed to patients with AN. Whether SSRIs are effective in a subgroup of individuals with AN, such as those with depression, is not known. METHODS: A secondary analysis of a randomized trial of fluoxetine versus placebo for relapse prevention in AN was conducted. Participants (n = 92) were weight-restored women with AN who completed the Beck Depression Inventory (BDI) at the time of randomization. BDI scores were dichotomized to reflect moderate/severe depression (BDI > 20, n = 26). A Cox Proportional Hazards model estimated the association of the level of depression, medication, and their interaction with time to relapse. Mixed effects models examined the effects of medication on symptom trajectories in high versus low depression groups and whether depression severity modified the effect of the drug on symptom trajectory. RESULTS: There was a significant interaction between medication and depression severity in time to relapse (hazard ratio = 0.46, 95% CI: [0.25, 0.85], p = .01). Depression severity modified the effect of fluoxetine on the time course of symptoms of depression (β = -0.27, 95% CI: [-0.42,-0.12], p = 0.001) and bulimia (β = -0.15, 95% CI: [-0.25,-0.05], p = 0.004) in the twelve month follow-up period. CONCLUSIONS: Fluoxetine was more effective than placebo in reducing relapse among more depressed, weight-restored individuals with AN. These results require replication but provide support for the use of antidepressant medication for patients with AN who remain depressed following weight restoration.