A living systematic review, meta-analysis, and open data resource of trials of MDMA-assisted therapy for PTSD

Post-traumatic stress disorder (PTSD) remains a substantial clinical and public health burden, with associated suicide risk, reduced quality of life, mortality, and large economic costs. Although trauma-focused psychotherapies are typically first-line and can reduce symptoms, they often have notable dropout rates, incomplete remission, and limited efficacy for some patients. Pharmacological options are limited, with only paroxetine and sertraline approved by the US Food and Drug Administration, and their benefits are modest. MDMA-assisted therapy has therefore attracted interest as a potential way to facilitate trauma processing, but the paper notes important uncertainties in the evidence base, including small trial sizes, limited external validity, and the risk of functional unblinding in psychedelic trials. Sevchik and colleagues set out to provide a living systematic review, meta-analysis, and open-data resource on MDMA-assisted therapy for PTSD so that the evidence can be updated as new trials appear. Their aim was to synthesise randomised controlled trial data on PTSD symptoms and related outcomes, while also making the underlying data and analytic choices transparent through a public database and dashboard. The paper positions this as a response to the rapid growth of studies and the problem that earlier meta-analyses quickly become outdated. The authors present this as an initial release within the SYPRES project, with plans for regular updates and public access to data, code, and results through an online resource.

This was a pre-registered living systematic review and meta-analysis under the SYPRES project, with a protocol registered on PROSPERO. The review focused on randomised controlled clinical trials published in English in peer-reviewed journals that compared MDMA for PTSD with a comparator in adults with elevated PTSD symptoms. Studies in healthy participants were excluded. Eligible interventions included any dose or formulation of MDMA, with or without therapy, provided the goal was to alter subjective experience; microdosing studies were excluded. Eligible comparators included placebo, lower doses intended to improve blinding, other psychotropics without known PTSD efficacy, waitlist, and usual care. For crossover studies, only pre-crossover data were used to avoid carry-over effects. The researchers searched PubMed, Embase, PsycInfo, Web of Science, Scopus, and reference lists of systematic reviews, with the final search conducted on 25 November 2025. Two reviewers independently screened studies, extracted data, and assessed risk of bias using Cochrane’s Risk of Bias 2.0 tool. Certainty of evidence was rated using GRADE. The authors state that the review and associated resources will be updated at least annually for five years. For continuous PTSD outcomes, the primary effect size was Hedges’ g, a small-sample corrected standardised mean difference, calculated from endpoint means, standard deviations, and sample sizes. Dichotomous outcomes were summarised with risk ratios. The primary analysis used inverse-variance random-effects meta-analysis of PTSD symptom scores at the primary endpoint specified by each study. Between-study heterogeneity was estimated with REML, and confidence intervals were calculated using the Knapp-Hartung adjustment. Funnel plots and Egger’s test were used to assess small-study bias, though the authors noted these methods have limited precision with few studies. To examine effects across all available timepoints and dosing conditions, the authors also ran a three-level correlated and hierarchical effects model on 15 effect sizes from the six studies. This model accounted for multiple effect sizes within studies using an assumed within-study correlation of 0.6, and cluster-robust variance estimation was used. Two meta-regressions tested whether the number of dosing sessions and cumulative MDMA dose predicted effect size. Additional analyses examined response, remission, and depression symptoms. Sensitivity analyses included fixed-effect models, substitution of medium-dose for high-dose arms in three-arm trials, a sweep of within-study correlations from 0 to 1, and a Bayesian meta-analysis with weakly informative priors. Analyses were conducted in R using Metapsy-related packages, and the living database is publicly accessible via Metapsy and the project website.

The searches yielded 1,586 reports, of which 26 full texts were assessed and six randomised controlled trials met criteria for the primary meta-analysis. The final database included 61 effect sizes from these six studies, covering continuous and dichotomous PTSD outcomes from four days after first dose to 12 months after baseline. All included studies combined MDMA with psychotherapy or psychological support before, during, and after dosing. Most studies were judged to have low risk of bias overall, while one study had some concerns because it appeared to use a per-protocol design in which dropouts were replaced with new enrollees. On the primary continuous outcome, MDMA-assisted treatment reduced PTSD symptoms more than control conditions, with Hedges’ g = -0.71, 95% CI -0.95 to -0.47, p < 0.001, based on six studies and 242 participants. Between-study heterogeneity was reported as low (tau² = 0.00; I² = 0.0%, although the confidence interval for I² was wide). Funnel plot inspection suggested little asymmetry, and Egger’s test did not indicate small-study effects, though the authors cautioned that the test was underpowered. In the three-level model using 15 effect sizes from the same six studies, the pooled effect remained significant and similar in direction, Hedges’ g = -0.60, 95% CI -0.86 to -0.34, p < 0.001, based on 286 participants. Meta-regression suggested that more dosing sessions were associated with a larger between-group difference in PTSD scores (β = -0.20 Hedges’ g per additional session, 95% CI -0.34 to -0.06, p = 0.01). Higher cumulative MDMA dose was also associated with a larger effect (β = -0.002 Hedges’ g/mg, 95% CI -0.003 to -0.0005, p = 0.01). For dichotomous outcomes, MDMA improved treatment response in five studies and 222 participants (RR = 1.35, 95% CI 1.10 to 1.66, p = 0.016) and remission in four studies and 210 participants (RR = 2.25, 95% CI 1.04 to 4.87, p = 0.044). The review found no statistically significant effect on co-morbid depression symptoms in three studies and 118 participants (Hedges’ g = -0.66, p = 0.30), with substantial heterogeneity. Sensitivity analyses were generally consistent with the primary results, including fixed-effect models, alternative dose selection in three-arm trials, the within-study correlation sweep, and the Bayesian analysis, which produced a posterior estimate centred at Hedges’ g = -0.70, 95% credible interval -1.05 to -0.36. The authors downgraded the overall certainty of evidence to low under GRADE, mainly because of indirectness. They argued that functional unblinding, especially in the larger Phase III trials using inert placebo, and the high proportion of participants with prior MDMA use could make the observed effects less generalisable to routine practice. They did not downgrade for risk of bias, inconsistency, or imprecision.

The authors interpret the findings as evidence that MDMA-assisted psychotherapy may be a promising treatment for PTSD. They emphasise that, compared with placebo or other controls, participants receiving MDMA had lower PTSD symptom scores and higher response and remission rates. They also view the meta-regression results as suggestive that more dosing sessions and greater cumulative MDMA exposure may be associated with larger symptom reductions, although they note that dosing frequency and cumulative dose are closely linked and cannot be separated cleanly with the available data. Sevchik and colleagues place their findings in the context of earlier research by noting that the overall direction of effect is consistent with previous meta-analyses, but that those reviews quickly became outdated in a fast-moving field. They present the open database, code, and dashboard as a contribution to transparency and reproducibility, and say that the living-review format is intended to keep pace with new trials. They also frame their results as potentially relevant to mechanistic work, citing preclinical and human studies suggesting MDMA may reduce fear responses, support therapeutic alliance, and aid memory reconsolidation and extinction. The authors stress several important limitations. The evidence base remains small, with only six eligible trials, and many studies were conducted under the oversight of the same organisation, which may have contributed to the low heterogeneity. They caution that standard risk-of-bias tools do not fully capture functional unblinding or expectancy effects, which are particularly relevant in psychedelic studies and may inflate apparent treatment effects. They also note variability across studies in dosing schedules, outcome timing, and control conditions, as well as limited long-term follow-up data, making durability of benefit uncertain beyond about two months in the available evidence. The paper further notes that the trial samples were carefully selected and may not reflect a typical treatment-seeking population with substantial psychiatric, medical, and substance use comorbidity. The discussion also references concerns raised in the FDA review of the Phase III programme, including inadequate safety data on positive or favourable adverse events, uncertain durability, and possible selection bias linked to prior MDMA use. The authors state that they did not analyse safety in this paper and plan to do so separately. They conclude that larger, better-controlled studies with broader inclusion criteria and longer follow-up are needed, particularly to assess generalisability, dose-response questions, and how benefits evolve over time.

The authors conclude that the available trial evidence suggests MDMA-assisted psychotherapy reduces PTSD symptoms and increases response and remission rates, but that the certainty of this evidence is low. They call for additional large controlled trials with rigorous methods, longer follow-up, and more representative samples. They also state that the living review and public dashboard will be updated regularly as new randomised trials emerge.