The impact of psychedelics on patients with alcohol use disorder: a systematic review with meta-analysis

Alcohol use disorder is highly prevalent and associated with substantial morbidity and mortality. Nehring and colleagues cite 2022 survey estimates that tens of millions of Americans meet criteria for alcohol use disorder, and they note elevated risks of accidental death, suicide, homicide, liver disease and dementia that together account for many thousands of deaths annually. Current guideline-recommended pharmacotherapies such as naltrexone and acamprosate are offered alongside psychosocial interventions, but alternative or adjunctive treatments are being explored, including psychedelic-assisted therapies. The authors identify weaknesses in the prior systematic review and meta-analysis of LSD for alcohol use disorder and set out to update and correct the evidence base. Their aim was to re-evaluate the impact of LSD and other psychedelics (including psilocybin and others) on abstinence or substantial reductions in drinking, using a systematic search and meta-analytic pooling of randomised trials while addressing methodological concerns in earlier work. This work therefore functions as an updated systematic review with meta-analysis focused on randomised controlled trials in predominantly alcohol use disorder populations.

The investigators conducted a PubMed search covering 1960 to 9 September 2023 using terms for a range of psychedelics and substance use disorders. Eligible studies were randomised controlled trials (RCTs) in which more than 90% of participants had alcohol use disorder and a psychedelic therapy (LSD, MDMA, psilocybin, mescaline, ayahuasca, dipropyltryptamine or dimethyltryptamine) was compared with any control. Non-randomised designs, reviews, letters, abstracts and trials conducted wholly in people with schizophrenia or psychosis were excluded. Only trials reported in English were considered. Data extracted included publication details, participant characteristics (age, gender, group sizes), intervention and control descriptions (including dose), outcomes (abstinence, drinking frequency, haemodynamics), follow-up timing, and risk of bias domains (sequence generation, allocation concealment, blinding, incomplete data and selective reporting). Where trials reported continuous outcomes or composite endpoints differently, the authors applied pre-specified decision rules to convert data to a dichotomous outcome representing abstinence or a substantial reduction in drinking (for example, Smart 1966 data were dichotomised using a >75% gain threshold). The review followed PRISMA 2020 guidance and risk of bias was assessed with the Cochrane Risk of Bias 2.0 tool. Primary analyses were restricted to randomised, double-blind, placebo-controlled trials; placebo was defined broadly to include inert placebo or active control agents not known to affect drinking (including very low dose psychedelics). Results were pooled with inverse-variance random-effects meta-analysis in R (Paul-Mandel estimator for tau2) and 95% confidence intervals adjusted with the Hartung–Knapp method. Effects on dichotomous outcomes are presented as odds ratios (OR) with 95% confidence intervals (CI). Heterogeneity was quantified with I2, with values above 60% considered high. Intention-to-treat methods were used when possible, and data read from figures were digitised when necessary.

The literature search yielded 174 citations; after screening and full-text review six trials met inclusion criteria and were analysed. Several potentially relevant controlled trials were excluded for reasons such as not enrolling a majority with alcohol use disorder (Nicholas 2022), restricting participants to people with schizophrenia (Tomsovic 1970b), reporting only composite endpoints that could not be separated (Bowen 1970), or presenting results as covariate-adjusted group means without usable dispersion measures (Rhead 1977). Of the six included trials, three randomised, double-blind, placebo-controlled studies assessed LSD (Hollister 1969, Pahnke 1970 plus Smart 1966’s qualifying arms), and one contemporary trial assessed psilocybin (Bogenschutz 2022). Additional LSD trials from the 1960s–1970s that were randomised but not placebo-controlled (Ludwig 1969, Tomsovic 1970a) were included in broader analyses. Most LSD trials were conducted in the United States in predominantly male samples; Bogenschutz 2022 enrolled a majority non-male proportion (45% female). Follow-up reporting varied: some trials reported outcomes at 2–3 months, most at 6 months, and only two reported 12-month data. Loss to follow-up ranged from <2% in some trials to 8–36% in others. Pooling only randomised, double-blind, placebo-controlled LSD trials at the first recorded follow-up produced an OR of 1.99 (95%CI: 1.10 to 3.61; n=3; I2=0%) for abstinence or substantial reduction in drinking versus placebo. At 2–3 months (n=1) the point estimate was OR 2.07 (95%CI: 0.79 to 5.41) and at 6 months (n=3) the estimate was OR 1.80 (95%CI: 0.98 to 3.30), neither reaching conventional statistical significance. When the pool was broadened to include randomised trials with any control (i.e. not all placebo-controlled), LSD effects at the first follow-up remained significant (n=5, OR 1.79; 95%CI: 1.36 to 2.34; I2=0%) and at 2–3 months (n=3, OR 1.83; 95%CI: 1.12 to 2.97; I2=0%), but effects attenuated by 6 months (n=5, OR 1.45; 95%CI: 0.81 to 2.60; I2=0%) and were negligible at 12 months (n=2, OR 1.05; 95%CI: 0.20 to 5.53) with high heterogeneity (I2=83%). Adding the single psilocybin trial to the double-blind placebo-controlled pool (LSD + psilocybin; n=4) yielded a first-follow-up OR of 2.16 (95%CI: 1.26 to 3.69; I2=0%) and at 6 months an OR of 2.00 (95%CI: 1.16 to 3.44; I2=0%). Including non-placebo-controlled randomised trials in the LSD/psilocybin pool (n=6) produced a first-follow-up OR of 1.89 (95%CI: 1.42 to 2.50; I2=0%) and a 2–3 month OR of 1.83 (95%CI: 1.12 to 2.97; I2=0%), but the effect was not statistically significant at 6 months (n=6, OR 1.58; 95%CI: 0.93 to 2.67) and was essentially absent at 12 months (n=2, OR 1.05; 95%CI: 0.20 to 5.53; I2=83%). Safety data were limited. Bogenschutz 2022 reported haemodynamic monitoring for 360 minutes post-dosing and documented significant transient increases in systolic and diastolic blood pressure and heart rate after psilocybin sessions (p<0.001). Estimated maximal differences versus control were approximately 20–21 mmHg systolic, 10–12 mmHg diastolic and 8–12 beats per minute for heart rate. Suicidality was sparsely reported: one brief 15-minute episode of suicidal ideation occurred during a psilocybin session, while two post-therapy severe suicidal-ideation episodes associated with binge drinking were reported in a placebo participant. Risk-of-bias judgements found Ludwig 1969 and Bogenschutz 2022 had some concerns, whereas the other trials were rated at high risk of bias. The authors highlight pervasive blinding challenges in psychedelic trials and variability in outcome definitions across studies.

Nehring and colleagues conclude that pooled data from randomised trials suggest LSD and psilocybin are promising interventions for reducing alcohol consumption or achieving abstinence in the short term. Meta-analyses restricted to double-blind, placebo-controlled trials produced statistically significant effects at the earliest follow-up, and pooled analyses that included non-placebo-controlled randomised trials showed similar early benefits. However, effects tended to attenuate by 6 months and were inconsistent or absent at 12 months, with substantial heterogeneity in the longer-term data. The authors position their findings as an update and methodological correction of an earlier meta-analysis and stress important limitations in the available evidence. Four of six included trials were judged to have high risk of bias, most LSD trials date from the 1960s–1970s and may not reflect contemporary practice, outcome definitions varied across studies, and the distinct subjective effects of psychedelics make adequate blinding difficult. They also raise concerns about selection bias when trials allow prior psychedelic users to enrol, since prior positive or negative experiences could influence participation and outcomes. Safety data were limited: haemodynamic effects (noted in the psilocybin trial) were transient but potentially important for patients with hypertension or autonomic vulnerability, and suicidal ideation was observed infrequently but requires prospective monitoring in future work. On the basis of these limitations, the authors refrain from definitive claims about clinical value and call for further contemporary, rigorous randomised trials. They recommend future studies recruit psychedelic-naïve, non-psychotic participants; use standardised primary endpoints and follow-up intervals; incorporate appropriate active control or masking strategies; prospectively assess suicidality and haemodynamics; and carefully consider exclusion of patients at cardiovascular risk. Until such evidence accumulates, the balance of benefits and harms for psychedelic therapy in alcohol use disorder remains uncertain.

Despite several trials showing relatively consistent short-term benefits of psychedelic therapy for alcohol use disorder, important limitations in study quality, generalisability and safety data prevent a conclusive determination of clinical value at this time. Nehring and colleagues therefore urge larger, methodologically rigorous contemporary trials to clarify effectiveness and the balance of benefits and harms.