Blinding Integrity in Psychedelic Randomized Clinical Trials A Systematic Review

Psychedelic randomised clinical trials for psychiatric disorders have expanded rapidly, but these studies face a methodological problem: the acute subjective effects of psychedelics can make it difficult to keep participants and raters blinded to treatment assignment. The paper explains that when people can infer what they received from their experiences, this functional unblinding may introduce expectancy effects, such as placebo or nocebo responses, and thereby bias efficacy estimates. The authors also note that blinding is especially important for psychiatric outcomes, where measures are often not fully objective, and they point to regulatory concerns about trial validity as a reason this issue matters. Orsini and colleagues set out to systematically assess how often psychedelic RCTs report blinding integrity and how severe functional unblinding appears to be across studies. Their aim was to quantify the extent of blinding assessment in trials of psychedelics used for psychiatric indications, and to summarise participant and rater guessing of treatment allocation. They frame this as the first systematic evaluation of blinding integrity in psychedelic RCTs, intended to identify methodological gaps and inform future trial design and regulatory evaluation.

Orsini and colleagues conducted a systematic review in line with PRISMA and registered the review in PROSPERO. They searched OVID databases, including MEDLINE, Embase, and APA PsycINFO, for studies published from January 2020 through July 2025, and then extended the search through December 11, 2025 using updated search strings from three earlier systematic reviews to capture older studies. They also manually screened reference lists from relevant reviews to identify studies published before January 2020. Eligible studies were randomised clinical trials of psychedelics for psychiatric disorders. The authors used a broad definition of psychedelics based on the presence of acute psychological effects, so both classic serotonergic agents such as psilocybin and LSD and non-classic agents such as MDMA and ketamine were included. Trials without a comparator group or without blinding methodology were excluded, because blinding success could not be assessed. There were no age restrictions or diagnosis-based exclusions, and trials of peri-procedural anxiolytic or antidepressant effects in surgical settings were excluded because the indication was not a primary psychiatric disorder. Three reviewers independently screened studies and extracted data on the intervention, comparator, route, dosing regimen, diagnosis, sample size, study design, and whether participant and rater blinding had been assessed. They also recorded whether authors described blinding as a limitation. Risk of bias was assessed with the Cochrane RoB 2.0 tool by two reviewers. The review did not report a formal quantitative meta-analysis, and the extracted text indicates that heterogeneity in blinding methods made quantitative synthesis infeasible.

The review identified 112 RCTs in total: 11 psilocybin studies, 7 LSD studies, 78 ketamine studies, 11 MDMA studies, and 5 studies of other psychedelics (2 ayahuasca, 2 DMT, and 1 noribogaine). Only 33 trials, or 29.5%, reported any assessment of blinding integrity, although 64 trials, or 57.1%, cited blinding as a limitation. The authors emphasise that this shows blinding is often discussed as a concern even when it is not formally measured. Functional unblinding was common across compounds. In psilocybin trials, 6 of 11 reported blinding integrity, but only 4 assessed participant blinding directly; several studies found more than 90% correct guesses of allocation, and one placebo-controlled study reported 80% correct guesses. Rater unblinding was also substantial, with more than 87% correct guesses in four studies. In LSD studies, all trials that assessed blinding found that more than 80% of participants and raters guessed allocation correctly, and in one study all participants guessed correctly. Most psilocybin and LSD trials that reported blinding data described it as a limitation. Among ketamine trials, blinding assessment was less common: 14 of 78 studies assessed participant blinding and 6 assessed rater blinding. One inert placebo-controlled study found 100% correct guessing among ketamine recipients, while another using remifentanil as a control reported 88% blinding integrity. Studies using midazolam as an active control generally showed better blinding preservation, with reported integrity values ranging from 55% to 75%, although some still showed unblinding at higher ketamine doses. Anesthesia-based blinding in one study reduced correct guessing to 36%. For MDMA, 8 studies assessed participant blinding and 6 assessed rater blinding; low-dose MDMA controls were associated with lower correct-guess rates of 50% to 65%, whereas inert placebo controls produced more than 86% correct guesses. In the smaller group of other psychedelics, both ayahuasca studies reported blinding assessments, and one placebo-mixture control study found 100% correct guessing by both participants and raters. Across the review, no control strategy consistently achieved ideal blinding. The authors also note that intranasal esketamine trials did not report blinding data in the included set. They report that midazolam generally performed better than saline as a ketamine comparator, whereas low-dose LSD, diphenhydramine, and ayahuasca-like active placebos were largely ineffective. The review also states that 66.7% of included trials had blinding integrity values above 70%, though the paper does not present a pooled quantitative estimate because of methodological heterogeneity.

The authors interpret their findings as evidence that functional unblinding is pervasive in psychedelic RCTs and may threaten the validity of efficacy estimates. They argue that, because participants and raters can often infer treatment allocation from subjective drug effects, expectancy may be unevenly distributed between arms and could inflate apparent benefit or contribute to nocebo responses. They state that the problem is especially pronounced for psilocybin, LSD, ayahuasca, and MDMA trials, where correct allocation guesses were often above 90%, but that ketamine studies showed more variable results and appeared better protected when midazolam rather than saline was used as a comparator. The review is positioned as extending earlier concerns about blinding in psychiatric research by systematically showing how often psychedelic trials fail to assess blinding at all. The authors compare the 29.5% rate of blinding assessment in psychedelic RCTs with much lower rates reported in general psychiatric trials, which they suggest reflects growing awareness of the issue in this field. They also note that no universally effective control strategy emerged, despite attempts such as low-dose active comparators, incomplete disclosure, fixed-order dosing, and anesthesia-based masking. The main limitations identified by the authors are that the review excluded healthy-volunteer trials, which means many psychedelic studies were not examined. They also highlight substantial inconsistency in how blinding was measured, including differences in timing, methods such as Bang’s Blinding Index versus simple guess proportions, and whether both participants and raters were assessed. Trial designs and active-control doses were heterogeneous, preventing quantitative meta-analysis, and the review did not analyse blinding by psychiatric diagnosis or compare compounds directly in a formal subgroup framework. In terms of implications, the authors call for standardised, validated blinding and expectancy measures, including broader use of tools such as Bang’s Blinding Index and expectancy scales. They also suggest that future psychedelic studies may need alternative designs when placebo control is not feasible, such as comparison with standard treatment, multi-dose paradigms, incomplete disclosure, or factorial designs that separately manipulate what participants are told and what they actually receive. They caution that even if blinding is imperfect, regulatory decisions may still need to consider the full package of efficacy, reproducibility, durability, and safety.

The authors conclude that psychedelic RCTs show consistently high rates of functional unblinding in both participants and raters, while formal assessment of blinding remains uncommon. They argue that the field needs more rigorous evaluation of expectancy and blinding integrity to separate true pharmacological effects from expectancy-driven responses.