Psychedelic therapy for depressive symptoms: A systematic review and meta-analysis

Earlier research has identified multiple putative psychological and biological mechanisms by which classic psychedelics might reduce depressive symptoms, including increased insight, improved social cognition and emotional processing, greater psychological flexibility, mystical-type experiences, enhanced neuroplasticity, and altered fronto-limbic activation. Psychedelic drugs are typically administered intermittently with psychological support; reported acute adverse effects tend to be mild-to-moderate (for example anxiety, nausea, transient hypertension and headache) and usually resolve with the acute drug effect. Interest in these agents has been driven in part by limitations of standard treatments for complex presentations such as treatment-resistant depression (TRD), where roughly 30% of people with Major Depressive Disorder fail to respond to at least two antidepressants of different classes according to commonly used operational criteria. Ko and colleagues designed this review to assess recent clinical trials of classic psychedelics for depressive symptoms, improving on prior reviews by including both open-label studies and randomised controlled trials and by incorporating the largest Phase 2b randomized trial published to date. The study aimed to synthesise outcome data quantitatively via meta-analysis where trials were sufficiently similar and to describe the broader clinical evidence base for psilocybin, ayahuasca (DMT-containing brew), and LSD administered with psychological support.

The review protocol was registered on PROSPERO (CRD42022318972) and followed PRISMA guidance. Electronic searches were performed in Embase, MEDLINE, PsycINFO, and CENTRAL for publications from January 1990 to March 2022 using a comprehensive search string for psychedelics and depression/distress. Two reviewers (KK and EK) independently screened records, resolved disagreements by discussion, and examined reference lists of included studies and relevant reviews. From 1,091 records identified, 13 studies met inclusion criteria and a further large trial published after the initial search was added, giving 14 studies for the systematic review; seven clinically similar randomised controlled trials (comparable doses and placebo rather than active psychiatric medication controls) were pooled in the meta-analysis. Eligible studies were clinical trials (randomised or open-label) in adults with depressive disorders and/or distress associated with life-threatening illness, using a classic psychedelic (psilocybin, LSD, mescaline, DMT/ayahuasca), reporting pre/post depression outcomes on validated instruments (for example BDI, MADRS, HAM-D), and published in English. Exclusions included studies of healthy volunteers and microdosing trials. Quality of uncontrolled trials was assessed with a modified Newcastle-Ottawa Scale and randomised trials were appraised with the Revised Cochrane Risk of Bias Tool; two reviewers independently rated risk-of-bias domains (blinding, selective reporting, randomisation, completeness of outcome data and other bias), resolving disagreements by discussion or third-party adjudication. Data extraction collected primary depression outcome measures, timing of follow-up, substance, dose, number of dosing sessions, psychological support model, and participant demographics and diagnoses. For the meta-analysis, means and standard deviations of depression scores for experimental and control arms were extracted. When authors could not provide raw data, graphical extraction (Plotdigitizer) and imputation procedures following the Cochrane Handbook were used; in some cases the formula reported in the text (sqrt(N) * (upper 95% CI − lower 95% CI) / 3.92) or approximations from other studies were applied to estimate SDs. The authors also describe heterogeneity in dosing regimens and washout procedures across trials: psilocybin doses varied (reported either per kg, e.g. 0.2–0.43 mg/kg, or fixed doses such as 10–25 mg), washout periods for concomitant psychotropic medications varied, and psychological support was provided nondirectively during dosing and variously before/after sessions.

Fourteen studies were included in the systematic review (11 clinical trials and 3 long-term follow-ups); seven randomised controlled trials contributed to the quantitative meta-analysis. Trials administered psilocybin, ayahuasca, or LSD together with psychological support. Overall, the qualitative synthesis indicated significant reductions in depressive symptoms across conditions and timepoints reported in the included studies. The meta-analysis of seven RCTs found statistically significant symptom reduction at three of four sampled timepoints: 1 day, 1 week, and 3–5 weeks after dosing. The 6–8 week follow-up point was less conclusive in the pooled analysis. The extracted text does not present pooled effect sizes, heterogeneity statistics, or p-values for each timepoint beyond the qualitative statement of significance at those three timepoints. Individual study findings reported in the extraction include: an LSD double-blind pilot (n≈12) in cancer-related anxiety where mean HADS-D decreased from 10.0 to 7.5 in the active group at 2 months versus 9.3 to 8.7 in placebo; significance testing for these secondary measures was not performed. Two ayahuasca studies examined recurrent MDD; one reported a HAM-D response rate of 64% in the ayahuasca arm versus 27% in placebo (p<.04), with an effect estimated (95% CI 0.67 to 2.32) in the excerpt. Psilocybin was the focus of eight clinical trials. Small open-label TRD cohorts (for example an n=12 feasibility study) reported large within-group reductions on QIDS-SR16 from baseline (mean 19.2, SD 2.0) to 1 week (mean 7.4, SD 4.9) and other follow-ups (all p<.005), with 8/12 in complete remission at 1 week and 5/12 remaining in remission at 3 months. A longer follow-up that included 19 participants reported large effect sizes (Cohen's d = 2.2 at 1 week; d = 2.3 at 5 weeks; d = 1.5 at 3 months; d = 1.4 at 6 months; all p<.001) and retention of response in a majority of responders at 6 months. The largest psilocybin trial to date randomised 233 participants across 22 sites to single-dose psilocybin arms of 25 mg, 10 mg and 1 mg (1:1:1), with follow-up to 12 weeks. Least-squares mean change in MADRS at 3 weeks was −12.0 (95% CI −14.6 to −9.3) for 25 mg, −7.9 (−10.6 to −5.2) for 10 mg, and −5.4 (−8.1 to −2.7) for 1 mg; the 25 mg versus 1 mg difference was significant (−6.6, 95% CI −10.2 to −2.9; p<.001). Response and remission rates at the main assessment were 36.7% and 29.1% for 25 mg, 18.7% and 9.3% for 10 mg, and 17.7% and 7.6% for 1 mg. Some benefit in the 25 mg arm was sustained to 12 weeks (response rate 20.3%; OR 2.2, 95% CI 0.9–5.4). A double-blind RCT by Carhart-Harris and colleagues compared two psilocybin doses plus placebo capsules to low-dose psilocybin plus daily escitalopram; on the primary outcome (QIDS-SR-16 at 6 weeks) between-group difference favoured psilocybin (mean change −8.0 vs −6.0) but did not reach statistical significance (p=.17). Secondary clinician-rated measures (HAM-D-17, MADRS, BDI) generally favoured psilocybin with between-group differences whose confidence intervals were unadjusted for multiplicity. Another randomized trial using a waitlist control (n=27) reported large between-group effects on GRID-HAMD and sustained large effect sizes through 12-month follow-up (Cohen's d in the 2.0–2.6 range), with response/remission rates of 71%/54% at 1 month and 75%/58% at 12 months. All MDD-focused trials reported positive outcomes, but many were small and several used open-label or cross-over designs.

Ko and colleagues interpret the body of evidence as indicating that psychedelic-assisted therapy is associated with robust short- and medium-term reductions in depressive symptoms across studies of psilocybin, ayahuasca and LSD when administered with psychological support. The meta-analysis supported significant symptom reductions at acute and early follow-up timepoints (1 day, 1 week, 3–5 weeks), whereas evidence at 6–8 weeks was less consistent. The investigators note that long-term efficacy has not been definitively demonstrated within double-blind randomised designs, although a small number of long-term follow-ups (including one with average follow-ups at about 3.2 and 4.5 years) report sustained benefit. The authors place these findings in the context of early-stage research: most trials remain small, blinding is difficult because of the obvious subjective effects of psychedelics, and many studies used crossover or open-label designs that raise bias concerns. Specific limitations highlighted include the necessity to perform graphical extraction and imputation for missing data in two studies, limited statistical power in all but the large recent RCT, the absence of long-term parallel-group RCT follow-up, and exclusion of people at high suicide risk which limits generalisability. The review also emphasises lack of participant diversity (ethnic, cultural, gender and socioeconomic), heterogeneity in washout regimens for concurrent psychotropic medications and variability in psychological support models, all of which complicate attribution of effects to the pharmacological agent alone. Practical and policy implications discussed by the authors include the intermittent dosing model and concentrated therapeutic supervision of psychedelic therapy compared with daily antidepressants, the generally tolerable acute safety profile observed in trials, and potential health-economic and access challenges given the resource intensity of supervised administration. In conclusion, the authors argue that while the accumulating evidence is encouraging—strengthened by inclusion of a large Phase 2b RCT and multi-timepoint analyses—definitive clinical efficacy requires further large-scale, well-blinded randomised trials with more diverse samples and standardised methods to address the methodological limitations identified.

This systematic review and meta-analysis suggests that classic psychedelic therapy administered with psychological support is associated with significant reductions in depressive symptoms at several timepoints, most clearly in the short and medium term, with some indications of longer-term benefit from limited follow-up data. However, the small number of studies, generally low sample sizes (except for one large recent trial), challenges with blinding, and limited diversity among participants mean the findings should be interpreted cautiously. The authors conclude that further randomised clinical trials with larger, more diverse samples and improved methodological rigour are needed.