Predictive Value of Early Improvement in Bipolar Depression Trials: A Post-hoc Pooled Analysis of Two 8-week Aripiprazole Studies

Objective: To evaluate the value of early improvement to predict treatment outcome in patients with bipolar depression. Methods: Data were pooled from two aripiprazole, 8-week, randomized, double-blind, placebo-controlled trials in patients with bipolar depression without psychotic features to determine whether early improvement (>= 20% reduction in Montgomery-Asberg Depression Rating Scale (MADRS) Total score at Week 2 or 3) predicts later response (>= 50% MADRS Total score reduction at Week 8) or remission (MADRS Total <= 10 at Week 8). Sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) were calculated (LOCF). Univariate and multivariate logistic regression models were used to evaluate early improvement and baseline demographic/clinical characteristics as predictors of response/remission. Results: In total, 311 patients were randomized to placebo and 306 to aripiprazole. Predictive values of early improvement (>= 20% MADRS Total score reduction) for remission with aripiprazole at Week 2/3, respectively, were: sensitivity 83%/94%; specificity 41%/33%; PPV 44%/45%; NPV 81%/91%. The corresponding values with placebo were as follows: sensitivity 70%/84%; specificity 60%/51%; PPV 50%/51%; NPV 77%/84%. Univariate linear regression showed that early improvement (>= 15%, >= 20%, >= 25%, >= 30% at Week 3) was a significant potential predictor of remission. Conclusion: Absence of early improvement after 3 weeks of treatment reliably predicted non-response/non-remission at study endpoint with high sensitivity and NPV. In patients with <20% improvement after 21 days of aripiprazole monotherapy, treatment should be modified, as continued use is unlikely to result in response/remission. Clinical decision-making to optimize treatment course in bipolar I depression may be appropriate after as little as 2 weeks and certainly within the first 3 weeks of treatment. Psychopharmacology Bulletin. 2010; 43(2): 5-27.