Magnitude of the Placebo Response Across Treatment Modalities Used for Treatment-Resistant Depression in Adults: A Systematic Review and Meta-analysis

The placebo effect in depression clinical trials is a substantial factor associated with failure to establish efficacy of novel and repurposed treatments. However, the magnitude of the placebo effect and whether it differs across treatment modalities in treatment-resistant depression (TRD) is unclear. To examine the magnitude of the placebo effect in patients with TRD across different treatment modalities and its possible moderators. Searches were conducted on MEDLINE, Web of Science, and PsychInfo from inception to June 21, 2021. Randomized clinical trials (RCTs) were included if they recruited patients with TRD and randomized them to a placebo or sham arm and a pharmacotherapy, brain stimulation, or psychotherapy arm. Independent reviewers used standard forms for data extraction and quality assessment. Random-effects analyses and standard pairwise meta-analyses were performed. The primary outcome was the Hedges g value for the reported depression scales. Secondary outcomes included moderators assessed via meta-regression and response and remission rates. Heterogeneity was assessed with the I2 test, and publication bias was evaluated using the Egger test and a funnel plot. Cochrane Risk of Bias Tool was used to estimate risks. Fifty RCTs were included involving various types of placebo or sham interventions with a total of 3228 participants (mean [SD] age, 45.8 [6.0] years; 1769 [54.8%] female). The pooled placebo effect size for all modalities was large (g = 1.05; 95% CI, 0.91-1.1); the placebo effect size in RCTs of specific treatment modalities did not significantly differ. Similarly, response and remission rates associated with placebo were comparable across modalities. Heterogeneity was large. Three variables were associated with a larger placebo effect size: open-label prospective treatment before double-blind placebo randomization (β = 0.35; 95% CI, 0.11 to 0.59; P = .004), later year of publication (β = 0.03; 95% CI, 0.003 to 0.05; P = .03), and industry-sponsored trials (β = 0.34; 95% CI, 0.09 to 0.58; P = .007). The number of failed interventions was associated with the probability a smaller placebo effect size (β = -0.12; 95% CI, -0.23 to -0.01, P = .03). The Egger test result was not significant for small studies' effects. This analysis may provide a benchmark for past and future clinical RCTs that recruit patients with TRD standardizing an expected placebo effect size.