Response to acute monotherapy for major depressive disorder in randomized, placebo controlled trials submitted to the US Food and Drug Administration: individual participant data analysis.
Level 1 - systematic review of randomized trials
Individual participant data meta-analysis of randomized controlled trials
PubMed 35918097 · doi:10.1136/bmj-2021-067606
What was done
This was an individual participant data analysis of 232 randomized, double-blind, placebo-controlled trials of acute drug monotherapy for major depressive disorder submitted to the US FDA between 1979 and 2016. The dataset included 73,388 adult and pediatric participants. Efficacy outcomes were converted to 17-item Hamilton Rating Scale for Depression (HAMD17) equivalent scores. Multivariable models and finite mixture models evaluated the effects of age, sex, baseline severity, and study year, as well as the underlying distributions of treatment response.
What was found
The random-effects mean difference between drug and placebo favored drug treatment by 1.75 points on the HAMD17 (95% CI, 1.63 to 1.86). Drug-placebo differences increased significantly with higher baseline severity (P < 0.001), with no significant trends in treatment effect or placebo response over calendar time. Finite mixture modeling identified three distinct response distributions with mean improvements of 16.0 points (Large), 8.9 points (Non-specific), and 1.7 points (Minimal). Participants receiving active drug were more likely to show a Large response (24.5% vs. 9.6%) and less likely to show a Minimal response (12.2% vs. 21.5%), indicating an absolute difference of approximately 15% achieving a substantial response beyond placebo.
Why it matters
While the mean population-level advantage of antidepressants over placebo is small (1.75 HAMD17 points), response distribution modeling indicates that approximately 15% of patients experience a substantial antidepressant effect beyond the placebo response.
Limits
The study is limited to clinical trials submitted to the FDA, which have strict eligibility criteria that may not reflect real-world clinical populations. Efficacy measures that were not originally HAMD17 required conversion to equivalent scores, which may introduce conversion error. The abstract does not detail individual drug classes, long-term outcomes, or adverse effects.