GLP-1 drugs, alcohol and addiction: what is known so far

The reports started with patients. People taking semaglutide or tirzepatide for weight or diabetes described losing interest in a second glass of wine, or in alcohol altogether. Animal studies had already shown that GLP-1 receptor agonists reduce alcohol intake in rodents and non-human primates, and that GLP-1 receptors sit in the brain's reward circuits. Researchers have since tested the idea in people in two ways: small randomised trials, and large studies of health records.

No GLP-1 drug is approved for alcohol use disorder or any other addiction. What follows is research, not a treatment option.

The randomised trials

The semaglutide alcohol trial

The most cited trial is a phase 2 study led by Christian Hendershot, published in JAMA Psychiatry in 2025. It randomised 48 adults with alcohol use disorder who were not seeking treatment, 71 percent of them women, to semaglutide or placebo for nine weeks. The dose was low by weight-loss standards: 0.25 mg for four weeks, 0.5 mg for four weeks, then 1 mg for one week.

The primary outcome was a laboratory session in which participants could choose to drink. After treatment, the semaglutide group drank less in that session, both in grams of alcohol and in peak breath alcohol, with medium to large effect sizes. In daily life, semaglutide reduced drinks per drinking day and weekly craving, and predicted greater reductions in heavy drinking over time. It did not change the number of drinking days or average drinks per day. In a subgroup who smoked, cigarettes per day fell more on semaglutide.

The authors called it initial evidence that justifies larger trials. That is the correct weight to give it: 48 people, nine weeks, a lab end point, and a population that was not trying to stop.

The exenatide trial

An earlier Danish trial, led by Mette Kruse Klausen, gave 127 treatment-seeking people with alcohol use disorder exenatide 2 mg weekly or placebo for 26 weeks, on top of cognitive behavioural therapy. Exenatide did not reduce heavy drinking days, the primary end point. Brain imaging showed reduced reactivity to alcohol cues in reward areas. In an exploratory analysis, people with a BMI above 30 did drink less. Exploratory subgroup results are hypotheses, not findings, but this one has shaped how later trials are designed.

The record studies

Several large studies have looked back at health records to compare people prescribed semaglutide with people prescribed other drugs. They are much bigger than the trials, and much weaker for proving cause.

• Alcohol, US records. Wang and colleagues, Nature Communications 2024, used records of 83,825 people with obesity in the TriNetX network. Over 12 months, semaglutide compared with other anti-obesity medications was associated with a 50 to 56 percent lower risk of both a first and a recurrent alcohol use disorder diagnosis. The finding was repeated in 598,803 people with type 2 diabetes.
• Alcohol, Swedish registers. Lähteenvuo and colleagues, JAMA Psychiatry 2025, followed 227,866 people with alcohol use disorder in Sweden, comparing each person's periods on and off a drug. Semaglutide use was associated with a lower risk of hospital admission for alcohol use disorder (adjusted HR 0.64, 95% CI 0.50 to 0.83), and liraglutide with HR 0.72. Approved alcohol medications were associated with a much smaller reduction (HR 0.98). Only 4,321 people in the cohort used semaglutide.
• Cannabis. Wang and colleagues, Molecular Psychiatry 2024, found semaglutide associated with a lower risk of a new cannabis use disorder diagnosis in 85,223 people with obesity (HR 0.56, 95% CI 0.42 to 0.75) and of a recurrent one (HR 0.62).
• Opioids. In 33,006 people with type 2 diabetes and opioid use disorder, semaglutide was associated with a lower risk of opioid overdose than each of eight other diabetes drug classes, with hazard ratios from 0.32 to 0.58 (JAMA Network Open, 2024).

Why the record studies are not proof

They show association, not cause. The people prescribed semaglutide differ from those prescribed other drugs in ways statistics can only partly correct: access to care, insurance, motivation, how often they see a doctor. A diagnosis code for alcohol use disorder depends on someone asking and recording it. The studies used careful matching and negative control outcomes, and the consistency across countries and substances is notable, but only randomised trials can settle whether the drug causes the change.

What is and is not shown

• Shown, in small trials: semaglutide can reduce craving and some drinking measures over weeks, at low doses, in people not seeking treatment.
• Not shown: that a GLP-1 drug helps people with alcohol use disorder stop or stay stopped, that the effect lasts, or what dose is best.
• Not shown: any benefit for opioid, cannabis or nicotine dependence in a randomised trial. The 2026 smoking trial missed its primary end points.
• Not approved: no regulator has approved a GLP-1 drug for any substance use disorder. Larger phase 2 and 3 trials are under way.

If you drink and take a GLP-1

Alcohol on a slowed stomach can hit differently, and it adds to nausea and dehydration. With insulin or a sulfonylurea it raises the risk of low blood sugar. If drinking is a concern in itself, the established treatments, counselling and the approved medications, work and are available now. The side effects page covers alcohol and nausea, the fatty liver page covers the liver, and how peptide research works explains why phase 2 results so often shrink in phase 3.

If you are curious whether your own drinking changes, a plain count is enough: drinks per day, logged against the dose and the day of the week since the shot, for a few weeks. It is your data, not a trial, but it is honest. The free GLP-1 tracker keeps the dose, the weight, side effects and, on Premium, lab results on one timeline, so the numbers your prescriber asks about are in one place rather than spread across portals and notes. It is free to download.