The association between study design and antidepressant effects in psychedelic-assisted therapy: A meta-analysis

Different study designs of psychedelic trials may impact the blinding and expectance, leading to biased treatment effects. This study aimed to examine the association between antidepressant efficacy and study designs in psychedelic trials. Six databases were systematically searched. Eligible trials were required to investigate the efficacy of psychedelics (psilocybin, lysergic acid diethylamide [LSD], 3,4-Methylenedioxymethamphetamine [MDMA], and ayahuasca) in adult patients with depressive symptoms. We only considered oral psychedelicassisted therapy without concomitant use of antidepressants. The primary outcome was the change in depressive symptoms. There were five study designs of psychedelic trials, including non-active-drug-as-placebo, activedrug-as-placebo, waitlist-as-control, fixed-order, and pre-post designs. In non-active-drug -as-placebo design, psilocybin (k = 4, Hedges' g [g] = 0.87, 95 % confidence intervals[CIs] = 0.58 to 1.16) and MDMA (k = 2, g = 0.65, 95%CIs = 0.26 to 1.05) were associated with large and medium effect sizes, respectively. In active-drug-asplacebo design, both psilocybin (k = 2, g = 0.71, 95%CIs = -0.01 to 1.43) and MDMA (k = 3, g = 0.53, 95%CIs = -0.23 to 1.28) were not statistically significant. In pre-post single-arm (k = 3, g = 2.51, 95%CIs = 1.00 to 4.02) and waitlist-as-control (k = 1, g = 2.88, 95%CIs = 1.75 to 4.00) designs, psilocybin showed a large effect size of antidepressant effect. Ayahuasca also showed a large effect size in both pre-post (k = 2, g = 1.88, 95%CIs = 1.18 to 2.57) and non-active-drug-as-placebo (k = 1, g = 1.60, 95%CIs = 0.84 to 2.36) designs. LSD was associated with a significant antidepressant effect only in non-active-drug-as-placebo design (k = 1, g = 1.49, 95%CIs = 0.80 to 2.17) but not in active-drug-as-placebo design (k = 1, g = 0.44, 95%CIs = -0.90 to 1.78). The antidepressant effects of psychedelics may be overestimated in studies with pre-post single-arm, non-activedrugs-as placebo, and waitlist-control designs. Restricted sample size, difficulty with establishing blinding for participants, and over expectancy limit the estimation of the antidepressant effect of psychedelic-assisted therapy.