The Blinding Problem: What a Living Review of MDMA Trials Actually Found

In October 2026 a living systematic review, meta-analysis and open data resource on trials of MDMA-assisted therapy for PTSD was published in Neuropsychopharmacology. It is the closest thing the field has to an independent accounting, and its most important feature is not its conclusion — it is that the underlying data were made available.

What the review pooled

Meta-analysis requires studies that are comparable enough to combine, and the eligible set was small: six studies met the criteria. The review notes that MAPS and Resilient Pharmaceuticals/Lykos Therapeutics provided oversight of all of these studies — a fact that shapes how much independent weight the pooled estimate can carry, and which the authors state rather than bury.

The primary model used inverse-variance random-effects meta-analysis of standardized mean differences on PTSD outcomes, and MDMA-assisted therapy showed a greater reduction in PTSD symptoms than control conditions. In other words: across the trials that could be combined, the drug arm did better than the control arm.

Why six studies is not a settled answer

Six studies is enough for a pooled estimate and not enough for the questions that matter most. With a set that small, you cannot reliably test whether the effect holds across different populations, different trauma types, different numbers of dosing sessions, or different countries. You cannot detect whether one unusually large trial is driving the result. And you cannot say much at all about durability, because durability requires follow-up the trials were not designed to compare.

The open data resource is the review's answer to that limitation. If the raw data are available, other groups can run the analyses the sponsors did not, test the sensitivity of the pooled estimate to the largest trial, and check whether the effect survives the exclusion of studies with the most enthusiastic expectation of benefit. Independent reanalysis is how a small evidence base matures, and it is the most useful thing this publication produced.

The limitation that keeps recurring

Every serious account of this evidence lands on the same two problems: challenges with blinding, and a lack of active comparator conditions. Both were raised in the VA's own state-of-the-science summary, and both were central to the FDA's decision to ask for a new randomised, double-blind Phase 3 trial.

The reason is structural rather than sloppy. MDMA produces unmistakable subjective effects. A person who receives it knows, and so does the therapist watching them. In a trial where the outcome is a clinician-rated score and the clinician may suspect the assignment, and where the participant's own expectation of relief is raised by the experience itself, the difference between drug and placebo becomes very hard to isolate. A pooled estimate that averages across studies with this problem does not average it away.

What this means for how you should read the headline numbers

• Direction is consistent. Across six eligible trials, the MDMA arm improved more than the control arm. That is not in serious dispute.
• Magnitude is uncertain. The size of the drug's specific contribution — as opposed to the therapy both arms received — is exactly the thing blinding problems inflate.
• Independence is limited. Sponsor oversight of every eligible study is a real constraint on how strong the pooled claim can be, which is why the open data matter.
• Durability is unproven. Nothing in the pooled estimate tells you whether improvement holds at two years.

The useful conclusion

MDMA-assisted therapy probably does something for PTSD that placebo-plus-therapy does not. How much, for whom, for how long, and whether it justifies the screening and supervision costs are still open questions — and they are open in a way that better trials, not more advocacy, will settle. The living-review format, updated as new evidence arrives, is the right place to watch.