REVIEW OF: Hu Y et al, Psychiatry Res 2022;311:114468

STUDY TYPE: Meta-analysis

Antidepressants are thought to cause mania, mixed states, and rapid cycling in bipolar disorder, but they remain the most commonly prescribed medication in this illness. The last meta-analysis on this controversial subject found a small benefit in depressive symptoms (effect size 0.17) but no change in remission or response rates with their short-term use (McGirr A et al, Lancet Psychiatry 2016;3(12):1138–1146). The current paper updates that work with 13 additional trials, adding newer studies and including some in Chinese and French.

The analysis included 19 randomized placebo-controlled trials that tested antidepressants as add-ons to mood stabilizers or antipsychotics in acute bipolar depression (not manic or mixed). The studies included 2,587 patients, only 11% (280) with bipolar II. Most studies followed them for six to 10 weeks. Investigators treated real-world patients, both inpatient and outpatient, some with substance use, psychotic symptoms, and rapid cycling. The primary outcome was response, as measured by >50% improvement in depressive rating scales or a Clinical Global Impression-Improvement score of 1–2. Secondary outcomes included remission, time to remission, adverse reactions, mood switching, and dropouts. They also compared secondary outcomes: remission rates and time to remission, adverse reactions, mood switching, and dropouts.

Response rates were available for 16 studies. In those, 59% of the subjects on antidepressants achieved response, compared with 51% on placebo (RR=1.10). Eleven studies measured remission. In those, 49% of subjects on antidepressants achieved remission, compared with 42% on placebo (RR=1.09). Neither the response nor remission rates were statistically significant. Mood switching occurred in about 5% of patients in treatment and placebo arms, and neither arm proved better at preventing adverse reactions or dropouts.

Researchers then looked at whether antidepressants performed differently when paired with a traditional mood stabilizer (lithium, valproate, or carbamazepine) vs an antipsychotic. The results for antipsychotics looked better, but close analysis showed that two of the four antipsychotic trials were industry sponsored and larger than the others, weighing the statistical outcomes in their favor.

The authors identified quality concerns in about half of the studies. Chinese researchers did not describe randomization methods and presented open-label data in eight out of 10 articles. Nonetheless, their outcomes were not significantly different from other trials.

CARLAT TAKE

Antidepressants neither help nor harm bipolar depression in the short term, at least in bipolar I disorder, which accounted for 89% of the subjects in these trials. However, longer-term studies have linked antidepressants to rapid cycling in patients with bipolar disorder.