Onset of antidepressant effect of olanzapine and olanzapine/fluoxetine combination in bipolar depression

Objectives: The current analysis investigated the onset of antidepressant effect of olanzapine/fluoxetine combination. Methods:Data for these post hoc analyses were obtained from a clinical trial comparing olanzapine, placebo, and olanzapine/fluoxetine combination in bipolar depression (BD). Subjects were 833 patients with a DSM-IV diagnosis of bipolar I disorder, depressed. The Montgomery-angstrom sberg Depression Rating Scale measured depressive symptoms. Multiple analytic methods were applied, including traditional (mean differences) analysis, pattern analysis, survival analysis of sustained response, mixed-effects regression, and area-under-the-curve analysis. Results:Traditional analysis showed significantly greater improvement in depression scores at week 1 for olanzapine/fluoxetine combination versus placebo (-9.55 versus -5.08, p < 0.001) and for olanzapine versus placebo (-8.31 versus -5.08, p < 0.001). Pattern analysis revealed olanzapine/fluoxetine combination had a significantly greater percentage of early persistent responders than placebo or olanzapine (32.4% versus 12.7%, p < 0.001; and 18.3%, p < 0.05, respectively). Survival analysis showed a significantly shorter time to sustained response for the combination versus placebo (p < 0.001), for olanzapine versus placebo (p = 0.04), and for the combination versus olanzapine (p = 0.03). Mixed-effects regression analysis revealed a significant therapy-by-time interaction (p < 0.001). Early area-under-the-curve analysis revealed a significantly greater percentage of improvement for the combination versus placebo (26.7% versus 13.9%, p < 0.001) and for olanzapine versus placebo (22.0% versus 13.9%, p < 0.001). Conclusions::Based on consistent results from related methods of measuring onset, olanzapine/fluoxetine combination demonstrated rapid onset of antidepressant effect (within 7 days) compared to placebo that was sustained over 8 weeks of treatment in a sample of BD patients. Using multiple statistical techniques may help profile a drug's onset of effect.