Semaglutide and tirzepatide cut alcohol-related hospitalizations by up to 65%

Evidence linking GLP-1 receptor agonists to reduced substance use has been accumulating across multiple substance categories this year, following a large veterans cohort study on substance use disorders broadly published earlier this year. A study published this month in BMJ Open narrows in specifically on alcohol, using a design that runs four separate comparisons to stress-test the finding from different angles [s1].

The design

Researchers used electronic health record data from a collective of US healthcare systems to conduct a retrospective cohort study using target trial emulation — a method for approximating a randomized trial's design using observational data [s1]. Adults with alcohol use disorder and either type 2 diabetes or obesity, who started a newer GLP-1 receptor agonist (semaglutide or tirzepatide) or a relevant active comparator between 1 January 2018 and 31 December 2024, were included across four separate target trials: an anti-diabetic medication comparison, an anti-obesity medication comparison, and two comparisons specifically against medications approved for alcohol use disorder — one in patients with diabetes, one in patients with obesity [s1]. The primary outcome was time to first alcohol-related emergency department visit or hospitalization within one year of treatment initiation; non-alcohol-related hospitalization was tracked as a negative control outcome, a check meant to help rule out that any finding simply reflects healthier people generally being prescribed GLP-1 drugs [s1].

What it found

A total of 40,703 adults met study criteria across the four trials [s1]. Starting a newer GLP-1 receptor agonist was associated with a lower hazard of alcohol-related hospitalization in every comparison run: against sulfonylureas in the anti-diabetic medication trial (hazard ratio 0.74, 95% CI 0.62–0.89) and against other anti-diabetic medications (hazard ratio 0.78, 95% CI 0.65–0.92); against other anti-obesity medications (hazard ratio 0.68, 95% CI 0.54–0.85); and, with the largest effects of all, against medications specifically approved for alcohol use disorder, both in the diabetes-comorbid group (hazard ratio 0.37, 95% CI 0.29–0.46) and the obesity-comorbid group (hazard ratio 0.35, 95% CI 0.26–0.47) [s1].

Why the comparator drug matters as much as the headline number

The most useful thing about this study's design is that it doesn't rely on a single comparison — it runs the same underlying question against four different comparator groups, and the effect size varies meaningfully depending on what GLP-1 drugs are being compared against. Against other diabetes or obesity medications, the reduction in alcohol-related hospitalization was real but moderate — roughly 22–32% lower hazard. Against medications specifically approved for treating alcohol use disorder itself, the reduction was much larger — roughly 63–65% lower hazard. That's a striking comparison: it suggests GLP-1 drugs performed considerably better in this observational data than the medications that exist specifically to treat alcohol use disorder, among people who had an alcohol use disorder diagnosis and were being compared on that specific outcome.

What the negative control outcome adds to the study's credibility

Including non-alcohol-related hospitalization as a negative control — a comparison where no effect would be expected if the association were driven purely by GLP-1 users being generally healthier or more health-engaged people — is a specific methodological safeguard against exactly the kind of confounding-by-indication concern that limits most observational drug studies. The study's overall framing implies this negative control behaved as expected, though the specific negative-control result numbers aren't detailed in the available summary.

What this doesn't establish

This remains observational data, even with the target trial emulation framework and negative control designed to strengthen causal inference — it is not a randomized trial, and residual confounding (from factors like disease severity, healthcare engagement, or unmeasured psychosocial factors that could influence both GLP-1 prescribing and alcohol-related outcomes) can't be fully ruled out. The population studied specifically had alcohol use disorder alongside diabetes or obesity — how this generalizes to people with alcohol use disorder without either comorbidity, who wouldn't typically be prescribed a GLP-1 drug in the first place, isn't addressed. The one-year follow-up window is relatively short for a chronic, relapsing condition like alcohol use disorder.

Why this adds to a broader, converging pattern

This study joins a growing list of similarly designed target-trial-emulation analyses this year finding reduced risk across different substances — the earlier, larger veterans cohort study covering multiple substance categories, and separate preclinical and early human research specifically on opioids — collectively building a more consistent observational picture, even as none of these studies individually proves causation on their own.

What to watch

Whether randomized controlled trials specifically testing GLP-1 drugs for alcohol use disorder — several are reportedly underway across the broader research landscape — confirm the magnitude of benefit suggested by this and other observational analyses. GLP-1 receptor agonists are not approved for alcohol use disorder; this article describes observational research and is not medical advice.

Sources

1. Association between GLP-1 receptor agonists and alcohol-related hospitalisations among adults with alcohol use disorder: multi-target trial emulation study — BMJ Open, 21 July 2026