- Design
- phase 2, double-blind, randomised, parallel-arm, placebo-controlled trial
- Population
- 50 treatment-seeking adults with moderate to severe alcohol use disorder
- Primary outcome
- laboratory-based alcohol cue-elicited craving at week 6
- Effect
- no significant effect on the primary outcome; heavy drinking days reduced (b=−0.580, 95% CI −1.012 to −0.148)
Fifty treatment-seeking adults with moderate to severe alcohol use disorder were randomised double-blind to oral semaglutide, 3 mg daily for four weeks then 7 mg daily for four, or placebo for eight weeks. The primary outcome was laboratory-based, cue-elicited craving at week six. Preregistered secondary outcomes were heavy drinking days and drinks per day over the final four weeks.
The primary outcome did not move, and neither did drinks per day. Heavy drinking days fell (b=−0.580, 95% CI −1.012 to −0.148), as did drinks per drinking day (b=−1.177, 95% CI −2.307 to −0.047), naturalistic craving reported outside the laboratory (b=−2.195, 95% CI −4.174 to −0.216), alcohol-related negative consequences (b=−4.618, 95% CI −8.651 to −0.585) and cannabis use days (b=−1.434, 95% CI −2.568 to −0.301). More participants on semaglutide dropped at least one World Health Organization risk drinking level than on placebo.
The interesting part is where the signal sits. Cue-elicited craving measured in a laboratory is the mechanistic endpoint; what changed was how much people drank when they drank, and what happened to them as a result. That dissociation either means the laboratory paradigm is the wrong probe for this drug, or that the consumption findings are chance across many comparisons. Both readings are live on 50 participants.
Nothing here supports prescribing a glucagon-like peptide-1 receptor agonist for alcohol use disorder. What it does support is asking about alcohol in patients already on one for diabetes or weight, and documenting what happens — many psychiatric patients in India are now on these drugs for metabolic reasons, and that is where the observations will accumulate first.
- Do not start semaglutide for alcohol use disorder outside a trial — this is phase 2 with a negative primary endpoint
- Do ask about alcohol in patients already taking a glucagon-like peptide-1 receptor agonist for another indication, and record the pattern
- Note the cannabis signal — worth asking about, not worth claiming
- Keep naltrexone, acamprosate and psychosocial treatment as the actual offer for alcohol use disorder
- Measure drinks per drinking day, not just abstinence — that is where this drug's effect appeared
Why it matters
A drug patients are already taking in large numbers may change their drinking, and the trial evidence is still too thin to act on.
Don't overread it
Phase 2, 50 participants, primary outcome not met — the consumption findings are hypothesis-generating, not grounds for an off-label prescription.
The statistics, in plain English
The confidence interval for drinks per drinking day, −2.307 to −0.047, only just excludes zero, and the same is true of naturalistic craving and alcohol-related consequences. With one negative primary outcome and six or more secondary comparisons, at least one boundary-significant result is expected by chance. That is why the authors describe continued development as warranted rather than describing the drug as effective — a distinction worth preserving when a patient arrives having read the headline.
Read the rest in the app
You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.
Scan to keep reading on your phone. No account needed to start.
Tomorrow morning, before your first patient
One edition a day for psychiatry, written by the desk, every claim tied to its paper. Six minutes.
Get the app — free