Objective The risk-benefit profile of antidepressant medications in bipolar disorder is controversial. When conclusive evidence is lacking, expert consensus can guide treatment decisions. The International Society for Bipolar Disorders (ISBD) convened a task force to seek consensus recommendations on the use of antidepressants in bipolar disorders. Method An expert task force iteratively developed consensus through serial consensus-based revisions using the Delphi method. Initial survey items were based on systematic review of the literature. Subsequent surveys included new or reworded items and items that needed to be rerated. This process resulted in the final ISBD Task Force clinical recommendations on antidepressant use in bipolar disorder. Results There is striking incongruity between the wide use of and the weak evidence base for the efficacy and safety of antidepressant drugs in bipolar disorder. Few well-designed, long-term trials of prophylactic benefits have been conducted, and there is insufficient evidence for treatment benefits with antidepressants combined with mood stabilizers. A major concern is the risk for mood switch to hypomania, mania, and mixed states. Integrating the evidence and the experience of the task force members, a consensus was reached on 12 statements on the use of antidepressants in bipolar disorder. Conclusions Because of limited data, the task force could not make broad statements endorsing antidepressant use but acknowledged that individual bipolar patients may benefit from antidepressants. Regarding safety, serotonin reuptake inhibitors and bupropion may have lower rates of manic switch than tricyclic and tetracyclic antidepressants and norepinephrine-serotonin reuptake inhibitors. The frequency and severity of antidepressant-associated mood elevations appear to be greater in bipolar I than bipolar II disorder. Hence, in bipolar I patients antidepressants should be prescribed only as an adjunct to mood-stabilizing medications.

We conducted meta-analyses of findings from randomized, placebo-controlled, short-term trials for acute mania in manic or mixed states of DSM (III-IV) bipolar I disorder in 56 drug-placebo comparisons of 17 agents from 38 studies involving 10 800 patients. Of drugs tested, 13 (76%) were more effective than placebo: aripiprazole, asenapine, carbamazepine, cariprazine, haloperidol, lithium, olanzapine, paliperdone, quetiapine, risperidone, tamoxifen, valproate, and ziprasidone. Their pooled effect size for mania improvement (Hedges' g in 48 trials) was 0.42 (confidence interval (CI): 0.36-0.48); pooled responder risk ratio (46 trials) was 1.52 (CI: 1.42-1.62); responder rate difference (RD) was 17% (drug: 48%, placebo: 31%), yielding an estimated number-needed-to-treat of 6 (all po0.0001). In several direct comparisons, responses to various antipsychotics were somewhat greater or more rapid than lithium, valproate, or carbamazepine; lithium did not differ from valproate, nor did second generation antipsychotics differ from haloperidol. Meta-regression associated higher study site counts, as well as subject number with greater placebo (not drug) response; and higher baseline mania score with greater drug (not placebo) response. Most effective agents had moderate effect-sizes (Hedges' g ¼ 0.26-0.46); limited data indicated large effect sizes (Hedges' g ¼ 0.51-2.32) for: carbamazepine, cariprazine, haloperidol, risperidone, and tamoxifen. The findings support the efficacy of most clinically used antimanic treatments, but encourage more head-to-head studies and development of agents with even greater efficacy.

No abstract