Percentage of Heavy Drinking Days Following Psilocybin-Assisted Psychotherapy vs Placebo in the Treatment of Adult Patients With Alcohol Use Disorder

Interest in psilocybin and other classic psychedelics for neuropsychiatric disorders, including substance use disorders, has grown over the past two decades. The authors note that these drugs act at serotonin 2A receptors and produce downstream changes that may enhance plasticity at multiple levels (neuronal structure, networks, cognition, affect, and behaviour), while acknowledging that some therapeutic effects may not be solely mediated by serotonin 2A activation. Prior controlled research from the 1960s suggested benefit of LSD for alcohol dependence, and a 2015 open‑label study reported large reductions in drinking after moderately high doses of psilocybin, but controlled evidence for psilocybin in alcohol use disorder (AUD) remained lacking. This multisite randomized clinical trial therefore set out to evaluate whether two administrations of high‑dose psilocybin, given within a manualised psychotherapy programme, would reduce the percentage of heavy drinking days (PHDD) in adults with AUD compared with an active placebo (diphenhydramine) plus the same psychotherapy. The report covers drinking outcomes during the 32‑week double‑blind phase following the first medication session, presenting efficacy and safety data and situating results relative to earlier findings in psychedelic research.

This was a double‑blind, parallel‑group randomized clinical trial conducted at two academic centres (University of New Mexico and New York University). Adults aged 25 to 65 years meeting DSM‑IV criteria for alcohol dependence and reporting at least 4 heavy drinking days in the 30 days before screening were eligible; exclusions included major psychiatric or drug use disorders, recent hallucinogen use, contraindicating medical conditions or medications, and current treatment for AUD. Recruitment occurred from March 2014 to March 2020. Race and ethnicity were self‑reported to describe sample representativeness. The protocol and statistical analysis plan are referenced as available in Supplement 1. Participants were randomised 1:1, stratified by site, to receive psilocybin or diphenhydramine in two eight‑hour medication sessions scheduled at weeks 4 and 8. Study medication was encapsulated to be visually identical; first‑session doses were psilocybin 25 mg/70 kg versus diphenhydramine 50 mg. Second‑session psilocybin dose was increased to 30 mg/70 kg or 40 mg/70 kg based on the participant's Mystical Experience Questionnaire (MEQ) score from the first session; diphenhydramine was increased to 100 mg for the second session regardless of subjective response. Sessions began about 9 AM, lasted at least 8 hours, and included standardised music, eyeshades and two attending therapists. Medications for acute hypertension, severe anxiety, or psychotic symptoms were available per protocol. All participants were offered 12 manualised psychotherapy sessions (motivational interviewing and cognitive behavioural therapy for AUD with preparation/support for psychedelic experiences): four sessions before the first medication, four between the two medication sessions, and four after the second medication session. Blinding was maintained for participants, therapists and most staff; randomisation sequences were held by site pharmacists. Primary outcome was percentage of heavy drinking days (PHDD) during weeks 5 to 32, assessed using the timeline followback (a calendar‑based interview). Secondary outcomes included percentage of drinking days (PDD), mean drinks per day (DPD), dichotomous outcomes (abstinence, no heavy drinking days, WHO risk level reductions), and drinking‑related problems measured by the Short Index of Problems (SIP‑2R). Hair or fingernail ethylglucuronide (EtG) assays were collected at week 24 to corroborate self‑reported abstinence. Safety monitoring included frequent blood pressure and heart rate checks during sessions and systematic adverse event elicitation at follow‑up visits. Subjective drug effects were assessed with the 43‑item MEQ immediately after each medication session. The planned statistical approach used multivariate repeated‑measures analysis of variance for continuous drinking outcomes with baseline covariates, site and assessment effects, and treatment interactions; missing monthly drinking values were imputed using multivariate imputation by chained equations (MICE). T tests compared MEQ scores, χ2 tests were used for dichotomous outcomes, Hedges g reported continuous effect sizes, and odds ratios were reported for binary outcomes. The trial initially planned up to 180 participants with an interim re‑estimation, but enrollment stopped at 95 randomised participants because of a COVID‑19 recruitment suspension.

Ninety‑five participants were randomised (49 to psilocybin, 46 to diphenhydramine); 93 received at least one dose and were included in the primary analyses. The mean (SD) age was 46 (12) years and 44.2% were female. Valid drinking‑outcome data were obtained for 717 of 744 follow‑up months (96.4%) across the 93 treated participants. Attendance at psychotherapy sessions was high and similar between groups (mean [SD] completed sessions 11.75 [0.76] for psilocybin vs 11.47 [1.20] for diphenhydramine). EtG confirmation at week 24 was available for 50 of 93 participants (53.8%). The prespecified primary outcome, PHDD during the 32‑week double‑blind period (weeks 5 to 36), was lower in the psilocybin group: 9.7% versus 23.6% in the diphenhydramine group, a mean difference of 13.9% (95% CI, 3.0%–24.7%; F1,86 = 6.43; P = .01). The authors also report that mean daily alcohol consumption was lower in the psilocybin group. In relative terms, PHDD in the psilocybin arm was 41% of that observed in the diphenhydramine arm over the observation period. Exploratory analyses of secondary continuous outcomes (PDD and DPD) similarly favoured psilocybin in multivariate tests. For dichotomous outcomes, participants receiving psilocybin were more likely to have no heavy drinking days and to achieve a 2‑level reduction in WHO risk level during weeks 5 to 36; these differences persisted in the final month of follow‑up (weeks 33 to 36), when abstinence and 1‑ and 3‑level WHO reductions were also more common in the psilocybin group. Numbers needed to treat ranged from 4.0 to 8.2 and odds ratios from 2.03 to 4.74 for these outcomes. Drinking‑related problems were reduced to a greater extent in the psilocybin group: mean (SD) total problems score at week 36 was 6.59 (8.80) for psilocybin versus 13.00 (10.48) for diphenhydramine (mean difference 6.4; 95% CI, 2.22–10.60; Hedges g = 0.67; P = .003). Regarding subjective drug effects and blinding, MEQ scores differed by group (psilocybin produced higher MEQ scores), and both participants and therapists overwhelmingly guessed treatment assignment correctly (participants guessed correctly in 93.6% of first sessions and 94.7% of second sessions; therapists guessed correctly in 92.4% and 97.4%, respectively), with high reported certainty. Safety data showed no serious adverse events among participants who received psilocybin. Three serious adverse events occurred, all in the diphenhydramine group: two psychiatric hospital admissions for suicidal ideation during binge drinking and one hospitalisation for Mallory‑Weiss tear from severe vomiting. Treatment‑emergent events within 48 hours of drug administration included headache (reported in 21 of 48 psilocybin recipients [43.8%] vs 2 of 45 diphenhydramine recipients [4.4%]), and higher rates of anxiety and nausea with psilocybin. Two psilocybin‑assigned participants received diazepam 10 mg during a session for anxiety; one participant experienced brief passive suicidal ideation during a session that resolved without sequelae. No persistent psychosis or hallucinogen persisting perception disorder was observed. Among the 14 participants who reported total abstinence at week 24 and had EtG results, all had negative EtG, providing some objective corroboration of self‑report.

Bogenschutz and colleagues interpret their findings as evidence that psilocybin, administered with manualised psychotherapy, was associated with substantially improved drinking outcomes over 32 weeks of double‑blind follow‑up compared with active placebo plus the same psychotherapy. The primary outcome showed a clinically meaningful reduction in percentage of heavy drinking days, and exploratory analyses found consistent between‑group effects across several secondary drinking measures. The authors place these results in the context of older LSD trials and recent open‑label psilocybin work, noting consistency with prior signals of efficacy. Safety observations are described as mostly mild and self‑limited, aligning with reports from other recent psilocybin trials; however, the authors emphasise that these safety findings arise from a tightly controlled research environment with careful screening, two trained therapists including a psychiatrist present during sessions, and provision for acute medical and psychiatric management, and therefore cannot be generalised to other settings. The trial's strengths include what the authors describe as the largest sample of any published psilocybin trial to date, rigorous assessment, high retention over 32 weeks, and use of manualised psychotherapeutic procedures drawn from empirically supported addiction treatments. Several limitations are acknowledged. Blinding was not maintained—participants and therapists usually identified treatment assignment—which could have introduced expectancy bias. Objective EtG corroboration was available for only 53.8% of participants. The enrolment stopped early because of the COVID‑19 suspension, leaving a smaller sample than planned and limiting power for subgroup analyses (for example by sex, race/ethnicity or psychiatric comorbidity) and for investigation of mechanisms, dosing optimisation, or predictors of response. The sample had lower drinking intensity at screening than many AUD pharmacotherapy trials, limiting generalisability to more severely affected populations. The two‑group design does not permit separation of medication effects from psychotherapy effects or their interaction, and the study provides no data on durability beyond the 32‑week double‑blind observation period. The authors conclude that further studies are needed to address these open questions and to expand understanding of psilocybin‑assisted treatment for AUD.

In this randomised clinical trial, psilocybin administered alongside manualised psychotherapy was associated with robust and sustained reductions in drinking over a 32‑week double‑blind period, exceeding the reductions observed with active placebo plus the same psychotherapy. The authors state these results support further research into psilocybin‑assisted treatment for adults with alcohol use disorder.