Predictors of relapse in a study of duloxetine treatment in patients with major depressive disorder

Background: Using data from a relapse prevention study of duloxetine treatment for adults with major depressive disorder (MDD), we examined demographic- and illness-related variables to identify factors that may predict relapse of MDD. Methods: Post-hoc analyses, using the Cox proportional hazards model, were performed on data from a study designed to compare the time to relapse of MDD in duloxetine- and placebo-treated patients. Patients received open-label duloxetine 60 mg/day during a 12-week acute phase, and those who met response criteria were randomly assigned to duloxetine 60 mg/day (N= 136) or placebo (N= 142) during a 26-week double-blind continuation phase. Results: Significant predictors of relapse were VAS back pain score at entry >30, HAMD(17) total score at randomization >7, and geography (Europe vs. US). Four significant treatment-by-predictor interactions were identified: the SQ-SS pain subscale score at entry>median of 4, VAS overall pain score at entry >30, VAS overall pain score at entry>median of 26, and VAS overall pain score at randomization> median of 7. In the "greater severity" category, the risk of relapse was significantly lower for duloxetine-treated patients compared with placebo-treated patients. Limitations: These were post-hoc analyses. Conclusions: Higher levels of pain severity and depressive symptoms and a US geographical location were significant predictors of relapse in patients with MDD. (C) 2008 Published by Elsevier B.V.