Can GLP-1 Drugs Reduce Drinking? Inside the New Alcohol-Use Trials

In one semaglutide trial, participants consumed less alcohol during a laboratory drinking session, yet their average drinks per calendar day did not significantly change.^[1] That apparent contradiction is a useful starting point. A medication can affect the intensity of drinking on some occasions without changing every measure of alcohol use. Three randomized trials now offer encouraging results, negative findings and a clearer picture of what still needs testing.

Start with the question each trial was built to answer

These studies tested semaglutide, a GLP-1 receptor agonist, in people with alcohol use disorder, or AUD.^[1]^[2]^[3] They did not test whether the entire drug class prevents problematic drinking in healthy consumers. Their populations, formulations and measurements also differed.

A primary outcome is a measurement designated in advance as central to a trial's main question. Secondary and exploratory outcomes add information, but a positive result among them does not erase a negative primary finding. The distinction matters especially when a small study measures several aspects of behavior.

Trial	Participants	Duration and formulation	Primary question and result
2025 trial ^[1]	48 adults with AUD, not seeking treatment	9 weeks; injection	Alcohol consumed in a laboratory task decreased
2026 obesity trial ^[2]	108 treatment-seeking adults with AUD and obesity	26 weeks; injection	Heavy drinking days decreased more than with placebo
2026 oral trial ^[3]	50 treatment-seeking adults with moderate-to-severe AUD	8 weeks; oral preparation	Laboratory cue-provoked craving did not significantly decrease versus placebo

All three original published abstracts were reviewed for this article; their full papers were not. That permits comparison of the reported findings, but limits assessment of detailed methods, sensitivity analyses and safety tables. An omitted detail cannot be treated as a method the researchers never used.

The first trial separated drinking intensity from drinking frequency

The 2025 investigators found less alcohol consumption in the posttreatment laboratory task, with reductions in both the amount consumed and peak breath alcohol concentration.^[1] A controlled session gives researchers a defined opportunity to observe behavior. It does not represent every drinking occasion across someone's life.

Outside that task, semaglutide did not significantly alter average drinks per calendar day or the number of drinking days, but reduced drinks per drinking day and reported craving; the analysis also indicated greater reductions in heavy drinking over time than placebo.^[1] Those findings are compatible. Drinks per drinking day concerns occasions when alcohol was consumed; drinks per calendar day also includes days without drinking. A change in one measure need not produce a statistically detectable change in the other.

The authors described the findings as initial prospective evidence supporting larger trials.^[1] That conclusion recognizes the positive signal while leaving room to learn whether it becomes a durable clinical benefit.

The longer trial: why the placebo improvement matters

The 26-week study enrolled treatment-seeking adults with both AUD and obesity, and everyone received cognitive behavioral therapy in addition to semaglutide or placebo.^[2] Both groups improved. The investigators reported an estimated between-group treatment difference of 13.7 percentage points favoring semaglutide, with a 95% confidence interval of 5.4–22.0 percentage points in that direction.^[2]

That last figure is not “13.7% less alcohol.” It describes a difference in the change in heavy-drinking-day frequency, not the volume consumed. Nor can the entire decline in the treated group be attributed to the medication. The comparison estimates an additional effect against a background in which both groups received care and both improved. It does not isolate how much of the placebo group's improvement came from therapy, study participation or other influences.

Of 108 randomized participants, 88 completed the intervention; the analysis included all 108 under an intention-to-treat approach and used multiple imputation to address missing outcomes.^[2] That approach preserves the randomized comparison, but completion and missing-data assumptions still matter when interpreting an estimated effect. A single-center study in people with obesity cannot establish the same benefit for every person with AUD.

The oral trial gave a mixed answer

In the eight-week oral semaglutide trial, the primary outcome was craving elicited by alcohol cues in the laboratory, assessed at week six.^[3] Semaglutide did not significantly improve that outcome or drinks per day compared with placebo, but it did significantly reduce heavy drinking days, one of the preregistered secondary outcomes.^[3]

The investigators also reported favorable exploratory findings for drinks per drinking day, craving outside the laboratory, alcohol-related consequences and movement to lower drinking-risk categories.^[3] These results make continued study reasonable. They do not transform a negative primary result into a positive one, or prove that an oral preparation is interchangeable with the injected preparations studied elsewhere.

Laboratory cue-provoked craving, everyday craving and actual consumption are related concepts, but they are different measurements. Reading them separately avoids dismissing useful signals or promoting secondary findings beyond what the study established. These trials address alcohol use disorder; they do not establish a treatment for every addiction.

Encouraging results still need a safety record

The 26-week investigators described adverse events as temporary, mainly gastrointestinal, usually mild or moderate, and more frequent with semaglutide than placebo.^[2] That is reassuring about the typical reported pattern, but does not establish an absence of serious events. The abstracts do not provide a basis for numerical claims about every adverse event or about rare and long-term harms.

The approved Wegovy semaglutide label adds relevant context: it warns about acute pancreatitis observed in treated patients, gallbladder disease and sometimes severe gastrointestinal reactions.^[4] Those warnings concern that approved product and its evidence base. They are not measured event rates from these AUD trials, and an observed event does not by itself prove medication causation in every case.

Funding also belongs in the account. The 26-week abstract lists Danish mental-health and charitable research funders, including the Novo Nordisk Foundation.^[2] Disclosing support helps readers assess the evidence without treating funding alone as either validation or disqualification.

A promising treatment question is not a self-treatment license

The Wegovy prescribing information reviewed here contains no alcohol-use-disorder indication.^[4] Evidence of benefit in an experimental setting and approval for a particular condition are separate milestones. The findings justify serious clinical research without establishing a universal prevention or treatment claim.

An approved medicine, a clinical-trial preparation, a compounded product and a laboratory reagent are also distinct products and contexts. Sharing the name semaglutide does not establish equivalence of formulation, quality or suitability for a particular person.

Research materials should not be consumed or self-administered. That boundary does not presume every vial is contaminated or untested; it recognizes that chemical testing alone cannot reproduce the human-use evidence and oversight behind a clinical preparation. AUD warrants individualized clinical care. These trials help refine a research question, not provide a protocol for self-treatment or alcohol withdrawal.