Skip to content

Bschor T, et al. Differential Outcomes of Placebo Treatment Across 9 Psychiatric Disorders: A Systematic Review and Meta-Analysis. JAMA Psychiatry. 2024;81:757-768. PMID 38809560

One-paragraph summary

Placebo is the only intervention systematically evaluated across every common psychiatric diagnosis, but had never been compared across them. MEDLINE and the Cochrane Database of Systematic Reviews were searched in March 2022 for the most recent systematic reviews meeting predetermined quality criteria for nine major psychiatric diagnoses; from those reviews, the ten highest-quality and most recent placebo-controlled RCTs per diagnosis were selected, giving 90 RCTs and 9,985 placebo-treated participants. Cross-diagnosis comparison used standardised pre-post effect sizes for each placebo group. Symptom severity improved with placebo in every diagnosis, and the pooled effects differed significantly across them (Q=88.5, df=8, p<0.001). Major depressive disorder had the largest placebo effect (d_av 1.40, 95% CI 1.24–1.56) and generalized anxiety disorder the second largest (d_av 1.23, 1.06–1.41). Panic disorder, ADHD, PTSD, social phobia and mania fell between 0.68 and 0.92; OCD was 0.65 (0.51–0.78) and schizophrenia 0.59 (0.41–0.76).

Key findings

  • GAD's placebo arms improve by more than one standard deviation, second only to depression.
  • The ordering is not random: the disorders with the largest placebo responses are the ones defined by subjective distress; those with the smallest are defined by observable behaviour or psychosis.
  • The comparison is restricted to high-quality, recent trials, so the effect is not an artefact of old or biased studies.

Limitations

  • Pre-post change in a placebo arm is not "the placebo effect": it contains natural course, regression to the mean, expectancy, and measurement reactivity, and this study cannot separate them.
  • Ten trials per diagnosis; diagnoses differ in trial duration, rating instrument and severity thresholds.
  • The GAD estimate derives from drug trials using clinician-rated scales, chiefly the HAM-A, whose properties differ from the self-report instruments used in psychotherapy trials.

Why it matters

This paper provides essential context for GAD efficacy estimates. The Cochrane responder analysis (RR 1.41, NNTB 7; Kopcalic 2025, PMID 39880377) and the network ranking (best drug −3.60 HAM-A points; Slee 2019, PMID 30712879) are measured against placebo arms in a condition with large pre-post placebo change. That finding may contribute to failed separation but does not by itself explain individual trial failures. It also reinforces why drug trials using pill placebo and psychotherapy trials often using waitlist controls cannot be compared naively (Carl 2020, PMID 30760112).

Cited by wiki pages

  • clinical-trials-landscape.md
  • overview.md