GLP-1 Drugs Linked to 55% Lower Alcohol Use Disorder Risk After Bariatric Surgery
In nearly 8,000 matched bariatric surgery patients, semaglutide, liraglutide and tirzepatide were tied to 55% fewer new alcohol use disorder diagnoses.
Summary
Weight-loss surgery has been linked to a higher risk of problem drinking, possibly amplified by faster alcohol absorption and changes in gut hormones afterward. Researchers used a large US health record network to compare 3,990 matched patients on each side after surgery. One group took GLP-1-type drugs (semaglutide, liraglutide or tirzepatide). The other took older weight-loss pills, mostly phentermine. Over two years, the GLP-1 group was 55% less likely to be newly diagnosed with alcohol use disorder and 41% less likely to start a medication for it. The results held across several stress tests. This was an observational study, so it cannot prove cause and effect. It does suggest that drug choice after bariatric surgery may matter for brain and behavior as well as weight.
Detailed Summary
Bariatric surgery is among the most effective treatments for obesity, but it carries a less familiar risk: higher rates of alcohol use disorder (AUD), sometimes called "addiction transfer." Surgery speeds alcohol absorption and raises peak blood levels, and altered gut hormone signaling may change reward pathways. Risk appears to climb roughly two years after the operation, which is often when patients with weight regain are offered additional antiobesity medications.
This retrospective cohort study used TriNetX, a federated network of more than 70 US health care organizations. It identified 15,382 adults who had Roux-en-Y gastric bypass or sleeve gastrectomy and then started an antiobesity medication between 2020 and 2024. Of these, 11,194 received incretin-based therapies (IBTs; semaglutide in 83.7%, plus liraglutide or tirzepatide), and 4,188 received non-IBT drugs, mostly phentermine (91.4%). Patients with preexisting AUD were excluded.
The groups differed at baseline. IBT patients were older (51.4 vs 45.1 years) and far more likely to have type 2 diabetes (55.1% vs 26.1%) and hypertension (73.0% vs 52.2%). One-to-one propensity matching produced 3,990 patients per group, about 85% women, with a mean age near 46 and good covariate balance. A small residual BMI difference remained (38.8 vs 37.7). Median time from surgery to medication start was about 3.5 years, and median follow-up was 2.0 years.
New AUD diagnoses occurred in 16 IBT patients and 35 non-IBT patients, or 2.4 vs 5.2 per 1,000 person-years. That is a 55% lower hazard (HR 0.45; 95% CI 0.25-0.81; P=.006). Medication for AUD (naltrexone at 50 mg or more, acamprosate or disulfiram) was started by 103 vs 171 patients, or 15.2 vs 25.6 per 1,000 person-years (HR 0.59; 95% CI 0.46-0.75; P<.001).
The signal held in every sensitivity analysis. Restricting to medication starts within 5 years of surgery gave an AUD HR of 0.43. A 6-month landmark analysis gave 0.36. Three-year follow-up gave 0.44 for AUD and 0.71 for medication initiation. Comparing only against phentermine gave 0.46 and 0.66, and requiring three or more prescriptions gave 0.38 and 0.65. The authors point to preclinical and early human data. GLP-1 receptors in the ventral tegmental area and nucleus accumbens appear to blunt alcohol-related dopamine reward. Small trials of exenatide, dulaglutide and semaglutide have shown reduced drinking.
The authors suggest clinicians consider prioritizing IBTs over older drugs for post-surgical patients with weight regain or AUD risk factors. Important caveats apply. The design is observational, so confounding by indication cannot be excluded. AUD events were few in absolute terms, which makes the estimates imprecise. Outcomes rely on diagnosis and prescription codes, with no data on actual drinking. Prospective trials and liver outcome data are still needed.
Key Findings
- IBT use after bariatric surgery was associated with a 55% lower hazard of new-onset AUD (HR 0.45; 95% CI 0.25-0.81; P=.006)
- AUD incidence was 2.4 vs 5.2 per 1,000 person-years (16 vs 35 events) for IBT vs non-IBT antiobesity medications
- Initiation of medications for AUD was 41% lower with IBT (HR 0.59; 95% CI 0.46-0.75; P<.001), at 15.2 vs 25.6 per 1,000 person-years
- A 6-month landmark analysis strengthened the AUD association (HR 0.36; 95% CI 0.18-0.72), and at 3 years it remained significant (HR 0.44)
- Compared with phentermine alone, IBT was still associated with lower AUD risk (HR 0.46; 95% CI 0.25-0.85) and lower medication initiation (HR 0.66)
- Among patients with 3 or more prescriptions, AUD risk was 62% lower (HR 0.38; P=.007)
- After 1:1 matching, 3,990 patients per group (about 85% women) were compared; before matching, 92.2% of IBT recipients were in obesity class II or III at medication start
Methodology
This retrospective cohort study used deidentified electronic health records from the TriNetX US Collaborative Network (more than 70 organizations) and followed STROBE guidelines. It included 15,382 adults with Roux-en-Y gastric bypass or sleeve gastrectomy who started an antiobesity medication in 2020-2023. After 1:1 greedy nearest-neighbor propensity matching (caliper 0.1), 3,990 patients per group remained. Outcomes were assessed from 1 month after the index prescription for up to 2 years, using Kaplan-Meier curves, log-rank tests and Cox models. Incidence per 1,000 person-years was estimated from mean follow-up time, and five sensitivity analyses were run.
Study Limitations
The study is observational and retrospective, so residual confounding and confounding by indication are possible. A small BMI imbalance remained after matching (SMD 0.15). AUD and medication outcomes depend on diagnosis and prescription codes, with no measures of actual alcohol intake, and AUD event counts were small. Person-years were estimated from mean follow-up, and long-term liver outcomes were not assessed. The provided text was cut off before the authors' full limitations and disclosures, so conflicts of interest could not be reviewed.
Enjoyed this summary?
Get the latest longevity research delivered to your inbox every week.
Enter your email to subscribe: