Predictors of relapse in a study of duloxetine treatment for patients with generalized anxiety disorder

Objective Data from a relapse prevention study of duloxetine treatment for adults with generalized anxiety disorder (GAD) were examined to identify predictors of relapse. Methods Patients responding to 6 months of open‐label duloxetine treatment were randomized to continuation with duloxetine or withdrawal to placebo for a 6‐month double‐blind continuation phase (duloxetine, N = 216; placebo, N = 213). Post hoc analyses compared time to GAD relapse during continuation phase by using predictor variables that included patient demographics, symptom severity measures (Hamilton Anxiety Scale Scores [HAMA], Hospital Anxiety and Depression Scale), functional outcomes, and visual analogue scale (VAS) pain measures. Univariate and multivariate analyses were performed using predictor variables from time of randomization into the continuation, withdrawal phase. Results Severity of anxiety symptoms, degree of functional impairment, and severity of pain at time of randomization were significantly predictive of likelihood of relapse during the continuation phase. Multivariate backwards elimination analysis of significant univariate predictors identified HAMA item one (anxious mood) ≥1 and severity of pain while awake (≥30 on VAS) as the strongest predictors of GAD relapse. Conclusions For patients with GAD responding to open‐label treatment with duloxetine, residual symptoms related to anxious mood, pain severity, and psychosocial function were associated with increased relapse risk, although the greatest risk was associated with anxious mood and increased severity of pain while awake. Copyright © 2011 John Wiley & Sons, Ltd.