IMPORTANCEThe 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.OBJECTIVE To examine the magnitude of the placebo effect in patients with TRD across different treatment modalities and its possible moderators. DATA SOURCESSearches were conducted on MEDLINE, Web of Science, and PsychInfo from inception to June 21, 2021. STUDY SELECTION 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. DATA EXTRACTION AND SYNTHESIS Independent reviewers used standard forms for data extraction and quality assessment. Random-effects analyses and standard pairwise meta-analyses were performed. MAIN OUTCOMES AND MEASURES 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 I 2 test, and publication bias was evaluated using the Egger test and a funnel plot. Cochrane Risk of Bias Tool was used to estimate risks. RESULTS 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 pooledplacebo 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. (continued) Key Points Question What is the placebo effect magnitude in different treatment modalities used for management of patients with treatment-resistant depression? Findings In this systematic review and meta-analysis of 3228 patients with treatment-resistant depression in 50 randomized clinical trials, the placebo effect size was large and consistent across treatment modalities. Response and remission rates associated with placebo effect were comparable across modalities.Meaning The findings of this study suggest a placebo effect size benchmark may be used to interpret the findings of past and future clinical trials.

AimsThe placebo response in depression clinical trials is a major contributing factor for failure to establish the efficacy of novel and repurposed treatments. However, it is not clear as to what the placebo response in treatment-resistant depression (TRD) patients is or whether it differs across treatment modalities. Our objective was to conduct a systematic review and meta-analysis of the magnitude of the placebo response in TRD patients across different treatment modalities and its possible moderators.MethodSearches were conducted on MEDLINE and PsychInfo from inception to January 24, 2020. Only studies that recruited TRD patients and randomization to a placebo (or sham) arm in a pharmacotherapy, brain stimulation, or psychotherapy study were included (PROSPERO 2020 CRD42020190465). The primary outcome was the Hedges’ g for the reported depression scale using a random-effects model. Secondary outcomes included moderators assessed via meta-regression and response and remission rate. Heterogeneity was evaluated using the Egger's Test and a funnel plot. Cochrane Risk of Bias Tool was used to estimate risks.Result46 studies met our inclusion criteria involving a total of 3083 participants (mean (SD) age: 45.7 (6.2); female: 52.4%). The pooled placebo effect for all modalities was large (N = 3083, g = 1.08 ,95% CI [0.95-1.20)I 2 = 0.1). The placebo effect in studies of specific treatment modalities did not significantly differ: oral medications g = 1.14 (95%CI:0.99-1.29); parenteral medications g = 1.32 (95%CI:0.59-2.04); ayahuasca g = 0.47 (95%CI:-0.28-1.17); rTMS g = 0.93 (95%CI:0.63-1.23); tDCS g = 1.32 (95%CI:0.52-2.11); invasive brain stimulation g = 1.06 (95%CI:0.64-1.47). There were no psychotherapy trials that met our eligibility criteria. Similarly, response and remission rates were comparable across modalities. Heterogeneity was large. Two variables predicted a lager placebo effect: open-label prospective design (B:0.32, 95%CI: 0.05-0.58; p:0.02) and sponsoring by a pharmaceutical or medical device company (B:0.39, 95%CI:0.13-0.65, p:0.004)). No risk of publication bias was found.ConclusionThe overall placebo effect in TRD studies was large (g = 1.08) and did not differ among treatment modalities. A better understanding of the placebo response in TRD will require: standardizing the definition of TRD, head-to-head comparisons of treatment modalities, an assessment of patient expectations and experiences, and standardized reporting of outcomes.