Abstract
English-language literature cited in MEDLINE from January, 1980 to October 30, 2014 was searched by using terms of antipsychotic, generic and brand names of atypical antipsychotics, "bipolar depression/bipolar disorder", "placebo", and "trial". The parameters of response (≥50% improvement on MADRS, Montgomery-Asberg Depression Rating Scale total score), remission (either ≤12 or 8 on MADRS total score at endpoint), discontinuation due to adverse events (DAEs), somnolence, ≥7% weight gain, overall extrapyramidal side-effects (EPSs), and akathisia, were extracted from originally published primary outcome papers. The number needed to treat to benefit (NNT) for response and remission or harm (NNH) for DAEs or other side effects relative to placebo were estimated and presented with the estimate and 95% confidence interval. Olanzapine monotherapy, olanzapine-fluoxetine combination (OFC), quetiapine-IR monotherapy, quetiapine-XR monotherapy, lurasidone monotherapy, and lurasidone adjunctive therapy were superior to placebo with NNTs for responses of 11-12, 4, 7-8, 4, 4-5, and 7, and NNTs for remission of 11-12, 4, 5-11, 7, 6-7, and 6, respectively. There was no significant difference between OFC and lamotrigine, and between aripiprazole or ziprasidone and placebo in response and remission. Olanzapine monotherapy, quetiapine-IR, quetiapine-XR, aripiprazole, and ziprasidone 120-160 mg/day had significantly increased risk for DAEs with NNHs of 24, 8-14, 9, 12, and 10, respectively. For somnolence, quetiapine-XR had the smallest NNH of 4. For ≥7% weight gain, olanzapine monotherapy and OFC had the smallest NNHs with both of 5. For akathisia, aripiprazole had the smallest NNH of 5. These findings suggest that among the FDA-approved agents including OFC, quetiapine-IR and -XR, lurasidone monotherapy and adjunctive therapy to a mood stabilizer, the differences in the NNTs for response and remission are small, but the differences in NNHs for DAEs and common side-effects are large. Therefore, the selection of an FDA-approved atypical antipsychotic for bipolar depression should be based upon safety and tolerability.