Background: Depression in bipolar disorder is a major therapeutic challenge associated with disability and excess mortality.
Methods: We reviewed findings from randomized placebo-controlled trials concerning efficacy and adverse effects of treatments for acute bipolar depression, including anticonvulsants, antidepressants, lithium, and modern antipsychotics, to compare numbers-needed-to-treat (NNT) versus -to-harm (NNH).
Results: Included were data from 22 reports involving 33 drug-placebo pairs. Antidepressants (especially modern drugs) had the most favorable (highest) risk/benefit ratio (pooled NNH/NNT=18.1). Anticonvulsants were effective agents (pooled NNT=5.06), but carbamazepine and valproate were not as well tolerated (NNH<10) as lamotrigine, and they had an unfavorable pooled NNH/NNT (3.75). Some antipsychotics (lurasidone, olanzapine+fluoxetine, and quetiapine (NNT all < 10) were effective though aripiprazole and ziprasidone were not (NNT≥45); olanzapine alone was weakly effective (NNT=11.3), and all but lurasidone (NNH=20.2) were not well tolerated (NNH≤4.18). Lithium appeared to be poorly effective but well tolerated in only one trial.
Conclusions: Some anticonvulsants and antipsychotics seemed effective for acute bipolar depression, but most antipsychotics were not well tolerated. Antidepressants were effective and well-tolerated; lithium remains inadequately tested.
Limitations: There are remarkably few short-term treatment trials (2.75/12 treatments), and fewer long-term trials for bipolar depression, possibly arising from exaggerated concerns about inducing mania.
Keywords: Bipolar depression; Number-needed-to-harm (NNH); Number-needed-to-treat (NNT); Treatments.
Copyright © 2015 Elsevier B.V. All rights reserved.