Predictors of relapse during fluoxetine continuation or maintenance treatment of major depression

Background: The goal was to examine predictors of relapse during continuati on/maintenance treatment of major depression that had remitted following 12 to 14 weeks of fluoxetine therapy. Method: The study utilizes data collected in a collaborative clinical trial including patients with DSM-III-R major depression at 5 university-affilia ted outpatient psychiatry clinics. Three hundred ninety-five patients who r emitted with fluoxetine therapy were randomly assigned to 1 of 4 treatments : fluoxetine for 14 weeks followed by placebo for 36 weeks, fluoxetine for 38 weeks followed by placebo for 12 weeks, fluoxetine for 50 weeks, or plac ebo for 50 weeks. Cox proportional hazard models were used to identify pred ictors of time to relapse. Results: In addition to the previously reported longitudinal pattern of res ponse during acute treatment, neurovegetative symptom pattern was a predict or of fluoxetine benefit compared with placebo. Greater chronicity predicte d poorer survival, which was not differential by treatment. The most robust advantage of fluoxetine was seen for patients with endogenous vegetative s ymptoms, chronic depression, and acute treatment response characterized by onset in the third week or later and persistence of response once attained. Conclusion: Both nonspecific pattern of response and neurovegetative sympto ms characteristic of atypical depression were predictive of lack of fluoxer ine efficacy in continuation/maintenance treatment. These findings have imp ortance for both clinical management and analyses of future maintenance tri als.