A Single Ketamine Infusion Combined With Motivational Enhancement Therapy for Alcohol Use Disorder: A Randomized Midazolam-Controlled Pilot Trial

Pathological alcohol use causes substantial global mortality and economic burden, yet most affected individuals are not in treatment and many who enter care do not respond to existing pharmacotherapies or behavioural interventions. The investigators frame problematic drinking as maintained both by preexisting vulnerabilities and by neural adaptations that undermine motivation, increase stress sensitivity, and reduce interest in nonalcohol pursuits. Earlier work in cocaine dependence by Dakwar and colleagues suggested that a single subanesthetic ketamine infusion can produce rapid effects on such vulnerabilities and, when paired with a behavioural intervention, increase rates of abstinence compared with an active control. This pilot trial set out to test whether a single intravenous ketamine infusion, administered on a therapist-designated quit day during the second week of a 5-week outpatient motivational enhancement therapy (MET) programme, would improve drinking outcomes relative to an active control (midazolam) in adults with alcohol dependence. Primary and secondary aims were to evaluate short-term abstinence, heavy drinking days, and time to relapse or study dropout, and to assess tolerability and safety in this clinical sample.

This was a randomized, midazolam-controlled outpatient pilot trial conducted from September 2014 to September 2017. Treatment-seeking adults meeting criteria for alcohol dependence were recruited; 95 were screened, 50 enrolled, and 40 were randomly assigned in an intent-to-treat sample. Participants were required to abstain from alcohol for at least 24 hours before the infusion and to fast overnight. The infusion was delivered on a predesignated quit day during week 2 of a 5-week MET programme. Participants were randomised by a statistician in randomly sized blocks to receive either ketamine (total nominal dose 0.71 mg/kg given as a 2-minute 0.11 mg/kg bolus followed by a 50-minute 0.6 mg/kg slow drip; N=17) or an active control of midazolam (2-minute saline bolus then 50-minute slow-drip midazolam at 0.025 mg/kg; N=23). Medication preparation and assignment were handled by the study pharmacy. Blinding included a minor deception: participants were told they might receive a variety of psychoactive compounds to reduce expectancy effects. Vital signs were continuously monitored, medical coverage was provided for up to 3 hours postinfusion, and relaxation/mindfulness exercises were used before and during infusions. All participants received six MET sessions over 5 weeks (initial session in week 1 to set goals, weekly sessions in weeks 2–5, plus an additional MET session 24 hours after infusion to capitalise on hypothesised short-term motivational effects). Alcohol use was measured at each visit using the timeline followback method with urine ethyl glucuronide testing to confirm abstinence; a positive urine test overrode self-reported abstinence. Secondary measures assessed craving, arousal, withdrawal, self-efficacy, perceived stress, mindfulness, and impulsivity, and urine toxicology was collected repeatedly. A six‑month telephone follow-up was attempted. The prespecified primary outcome was daily abstinence (0 drinks) over the 21 days following infusion, dichotomised for each calendar day. Days with missing data were treated as nonabstinent in the main analysis; sensitivity analyses treated missing days as missing. A longitudinal logistic mixed-effects model (logit link, random intercept) tested the effects of time (including quadratic terms), treatment, and time-by-treatment interaction, adjusted for baseline total drinks. Heavy drinking days and other secondary outcomes were analysed using longitudinal mixed-effects models with appropriate link functions (logit for binary outcomes, log link for right-skewed outcomes) and an autoregressive correlation structure. Time-to-event outcomes (time to relapse, first use, first heavy drinking day) were analysed with Kaplan-Meier curves and log-rank tests. Analyses were conducted in SAS v9.4 with two-sided tests at the 5% significance level.

The intent-to-treat sample comprised 40 participants, who were mostly middle-aged (mean age 53 years, SD=9.8), predominantly white (70.3%), and largely employed (71.8%), reporting an average of five drinks per day before study entry. Across groups, infusions were generally well tolerated. The most common adverse effects were sedation (midazolam: N=12; ketamine: N=8) and headache (midazolam: N=4; ketamine: N=6), typically lasting about 12 hours postinfusion. Dissociative symptoms measured by the Clinician-Administered Dissociative States Scale were significantly higher immediately after ketamine (median=19, IQR 9–30.75) than after midazolam (median=2, IQR 0.25–9.25; χ2=7.87, p=0.005). Two ketamine-treated participants had brief mild agitation; there were no reports of persistent psychoactive effects or initiation of benzodiazepine, opioid, or ketamine misuse. On the primary outcome of daily abstinence over the 21 days after infusion, the longitudinal logistic mixed-effects model (with time treated as a quadratic term) indicated that ketamine significantly increased the likelihood of abstinence compared with midazolam. Sensitivity analyses treating missing days as missing, and analyses without adjustment for baseline drinks, yielded similar results, suggesting robustness to these assumptions. Across the 21 days, 47.1% (8/17) of ketamine participants used alcohol and 17.6% (3/17) had a heavy drinking day; corresponding figures in the midazolam arm were 59.1% (13/22) using alcohol, 40.9% (9/22) with a heavy drinking day, and 52.2% (12/23) who met the study definition of relapse at some point. Six participants dropped out of the midazolam group (one immediately after infusion, five after the first week); no participants dropped out of the ketamine group. For the secondary outcome of heavy drinking days, there was a significant study week-by-treatment interaction (F=12.34, df=1,798, p<0.001). In the midazolam group the odds of a heavy drinking day increased with each postinfusion day (odds ratio=1.19, 95% CI=1.14–1.25, p<0.001), whereas the odds did not change significantly over time in the ketamine group (odds ratio=0.98, 95% CI=0.89–1.08, p=0.74). The two-way interaction of study week-by-treatment was not significant for the other secondary measures (craving, withdrawal, mindfulness, impulsivity, stress sensitivity, self-efficacy). Time-to-event analyses showed a longer time to relapse (defined as first heavy drinking day or dropout) in the ketamine group by log-rank test (χ2=4.2, p=0.04), but no significant differences in time to first alcohol use or time to first heavy drinking day. At six-month telephone follow-up 19 of 40 participants were reached (47.5%); among respondents, 75% (6/8) in the ketamine group reported abstinence versus 27% (3/11) in the midazolam group.

Dakwar and colleagues interpret these preliminary data as evidence that a single subanesthetic ketamine infusion, given on a planned quit day and combined with motivational enhancement therapy, reduced short-term alcohol use and heavy drinking and prolonged time to relapse relative to an active control. They situate these findings within prior work showing transient laboratory effects of ketamine and longer-lasting benefits when an infusion is embedded within a psychosocial framework, proposing that an integrated pharmacotherapy–behavioural model may produce synergy. The investigators suggest psychological mechanisms—such as perspectival shifts that increase motivation to change—may interact with neurobiological effects (neurotrophic and modulatory changes, potential mTOR involvement, altered prefrontal connectivity) to support sustained behaviour change beyond the pharmacokinetic clearance of ketamine and its metabolites. The authors acknowledge several important limitations. The sample was small and relatively homogeneous, limiting statistical power and generalisability; treatment duration and follow-up were brief, with only 21 days of detailed alcohol monitoring after infusion and incomplete six-month follow-up. More than a quarter of control participants dropped out, and the standard assumption of treating dropout as relapse could bias results despite sensitivity checks that produced similar findings. Blinding remains a concern given ketamine's distinctive dissociative effects; although an active control and limited deception were used, the integrity of the blind was not formally assessed. The comparator (low-dose midazolam) may have had some transient benefit for withdrawal or craving, which could have reduced the apparent effect size of ketamine compared with an inert placebo. Safety considerations for repeated administration—neurotoxicity and bladder complications—are highlighted as areas requiring further study, particularly outside specialised research settings. Finally, the investigators emphasise that these results are a first step. They recommend replication in larger, more diverse samples with longer follow-up, more frequent assessments, and design features (for example, factorial designs and measurement of ketamine metabolites) that can clarify whether ketamine's clinical benefits depend on concurrent behavioural therapy and elucidate underlying mechanisms.

In this pilot randomised trial, a single ketamine infusion administered on a designated quit day and combined with motivational enhancement therapy was associated with greater short-term odds of alcohol abstinence, fewer heavy drinking days, and longer time to relapse compared with an active midazolam control. The findings support further research to determine durability of effect, safety with repeated use, and whether there is true synergy between ketamine and behavioural treatments, ideally tested in larger samples with longer follow-up and mechanistic assessments.