Meta-analysis of placebo group dropout in adult antidepressant trials

Minimizing dropouts across antidepressant, placebo-controlled trials remains a major opportunity to improve the efficiency of trials. This meta-analysis investigated placebo dropout rate and its predictors in second generation antidepressant (SGA) for anxiety, depression and obsessive-compulsive disorder (OCD).

A random-effects meta-analysis was performed to examine placebo group dropout rate in SGA trials for depression, anxiety and OCD using Freeman - Tukey transformation. Stratified subgroup analysis by diagnostic indication was performed to examine the dropout rate across disorders. Meta-regression was performed to identify correlates between placebo dropout rate and trial and subject characteristics.

Meta-analysis included 148 trials with 18,016 participants receiving placebo. Across antidepressant trials the overall placebo dropout rate was 25% (dropout rate ± standard error (SE) = 0.25 ± 0.01, 95% CI: 0.23-0.27, z = 23.95, p < .001) and was similar across disorders (χ2 = 1.09, df = 2, p = .58). The placebo group dropout rate was 26% in depressive disorders, 25% in anxiety disorders and 22% in OCD. Across all diagnostic indications, earlier publication year, placebo lead-in, studies conducted in a single country (instead of internationally), longer trial duration, fewer study sites, more study visits and less baseline illness severity were associated with higher placebo dropout rate. Significant predictors of placebo dropout did not replicate across disorders.

No significant difference was found in placebo dropout rate between internalizing disorders with overall dropout rate for placebo groups in antidepressant trials being around 25%. Placebo dropouts in trials can be minimized by reducing subject burden in trials, enrolling more severely affected subjects and foregoing placebo lead-in periods.