Antidepressant Remission and Manic Switch in Bipolar Depression: A Propensity Score Analysis

Objective

Antidepressants remain among the most widely used class of drugs in treating bipolar disorder, despite their minimal efficacy in randomized clinical trials and concern for association with manic episodes. This study sought to evaluate the outcomes of antidepressant treatment in bipolar depression in a large naturalistic cohort study, STEP-BD, in terms of symptomatic remission as well as emergence of mania, using a propensity score (PS) analysis to reduce indication bias.

Methods

Propensity scores were developed to estimate the probability of antidepressant exposure using multivariate logistic regression models; these scores were then used to match antidepressant-exposed and non-exposed individuals. Cox regression models were used to estimate hazard ratios for manic switch and time to remission, adjusted for these scores in the matched population.

Results

Total sample included 2166 individuals, of whom 1085 were exposed to AD and 1081 were unexposed to AD; mean follow-up duration was 182.5 (SD: 44.6) days (median = 126, ICR: 87.4). Cox regression models for manic switch with antidepressant exposure versus non-exposure yielded an unadjusted hazard ratio (HR) of 0.93 (95% CI 0.67–1.14) and PS-adjusted HR of 0.77 (95% CI 0.51–1.08), neither of which was statistically significantly different from 1. Probability of symptomatic remission was also not significantly associated with antidepressant exposure, with unadjusted and PS-adjusted HR of 1.15 (95% CI 0.97–1.37) and 1.02 (95% CI 0.87–1.23), respectively.

Conclusion

With PS adjustment, there was no evidence of increased likelihood of manic switch or achievement of symptomatic remission associated with antidepressant use in bipolar depression. Our results underscore the ongoing need to identify alternative strategies for effective treatment of bipolar depression.