This random-effects model meta-analysis of double-blind, randomized placebo-controlled trials compared recurrence rates in bipolar disorder (BD) patients between antipsychotic/mood stabilizer discontinuation and maintenance groups.

We conducted systematic literature search of Embase, PubMed, and CENTRAL databases without language restriction from inception until 22 May 2020. Independent investigators assessed studies and extracted data. We calculated risk ratios (RRs) and numbers needed to benefit or harm (NNTB/NNTH). Primary outcome was the recurrence rate of any mood episode at 6 months. Secondary outcomes were recurrence rates of depressive episodes and manic/hypomanic/mixed episodes and all-cause discontinuation at 6 months. We also investigated these outcomes at 1, 3, 9, 12, 18, and 24 months.

We identified 22 studies (n = 5462) receiving aripiprazole, asenapine, divalproex, long-acting injectable (LAI)-aripiprazole, LAI-risperidone, lamotrigine, lithium, olanzapine, paliperidone, or quetiapine. Mean study duration was 64.50 ± 69.35 weeks. The maintenance group demonstrated lower recurrence rates of any mood episode, depressive episodes, and manic/hypomanic/mixed episodes as well as reduced all-cause discontinuation at every observational point. The RRs (95% confidence interval, NNTB/NNTH) of recurrence rate at 6 months were 0.61 (0.54–0.70, 5) for any mood episode, 0.72 (0.60–0.87, 13) for depressive episodes, and 0.45 (0.36–0.57, 6) for manic/hypomanic/mixed episodes. The RR for all-cause discontinuation at 6 months was 0.71 (0.61–0.82, 6).

Maintaining drug treatment during clinically stable BD prevented recurrence for up to 24 months. Discontinuation of medications for ⩾1 month significantly increased recurrence risk. However, 47.3% of patients who discontinued drugs for 6 months did not experience recurrence.