Beyond control in psychiatric research: systematic review and meta-analysis of placebo treatment responses across major psychiatric conditions in citalopram and escitalopram RCTs.
Level 1 - systematic review of randomized trials
Systematic review and meta-analysis of randomized controlled trials
PubMed 42639721 · doi:10.1192/bjp.2026.10665
What was done
A PRISMA-guided, PROSPERO-registered systematic review and meta-analysis evaluated randomized controlled trials (RCTs) containing at least one placebo arm and one active medication arm (citalopram or escitalopram). Trials spanned major psychiatric conditions: depressive disorders, anxiety disorders, obsessive-compulsive disorder (OCD), and post-traumatic stress disorder (PTSD). Random-effects models were used to calculate Cohen's d effect sizes for placebo and drug responses and to identify factors explaining heterogeneity across trial designs, participant demographics, and outcome metrics.
What was found
Across 80 RCTs (312 outcome measures, 19,776 participants aged 7–91 years): - Placebo effect size was large overall (d = 1.01; 95% CI: 0.93–1.10) with extreme heterogeneity (I² = 99.61%). - Medication effect size was d = 1.43 (I² = 99.76%) and did not differ significantly across psychiatric conditions (P = 0.61). - Placebo responses differed significantly across conditions (P < 0.01): highest in PTSD (d = 1.14) and depressive disorders (d = 1.02); moderate in anxiety disorders overall (d = 0.86, but lowest in specific phobia at d = 0.23, P < 0.01); and lowest in OCD (d = 0.62, P < 0.01). - Significant moderators of larger placebo effects included: more study arms (b = 0.181, P < 0.01), younger mean age (b = -0.010, P < 0.01), clinician-rated vs. self-reported outcomes (d = 1.07 vs. 0.76, P < 0.01), multicentre status (P < 0.01), and intention-to-treat datasets (d = 1.08 vs. 0.86, P = 0.01).
Why it matters
This analysis shows that placebo responses in psychiatric trials are condition-specific and heavily influenced by trial methodology, such as rater type and arm count, rather than being uniform background noise. These findings directly inform the sample sizing, design, and outcome selection of future psychiatric clinical trials.
Limits
The dataset was restricted to trials evaluating citalopram or escitalopram, which may limit generalizability to other pharmacological classes or non-drug interventions. Heterogeneity was exceptionally high across all analyses (I² > 99%). The abstract does not report individual study risk of bias or publication bias assessments.