Semaglutide and Liraglutide Linked to 36% and 28% Lower Alcohol Use Disorder Hospitalizations in Major Swedish Study

This was a nationwide Swedish register-based observational cohort study spanning January 2006 to December 2023. The population included all residents aged 16-64 with an AUD diagnosis, identified from inpatient, outpatient, sickness absence, and disability pension registers. The primary analysis used a Cox regression within-individual model, which compares each person's outcomes during periods of GLP-1 agonist use versus non-use — eliminating all time-invariant confounders (genetics, demographics, baseline health). Primary outcome was AUD hospitalization; secondary outcomes included any SUD hospitalization, somatic hospitalization, and suicide attempts.

Alcohol use disorder is a leading cause of preventable death, and current approved medications (naltrexone, acamprosate, disulfiram) have limited effectiveness and are underused. If GLP-1 drugs can substantially reduce alcohol-related harm — as this study suggests — it would represent a transformative addition to AUD treatment. The effect sizes seen here (36% reduction with semaglutide) far exceed those of approved AUD medications in this same population. Published in JAMA Psychiatry, this is among the strongest real-world evidence to date for GLP-1 drugs in addiction.

GLP-1 receptor agonists are increasingly linked to reduced addictive behaviors across multiple substances — alcohol, tobacco, and other drugs. The GLP-1 receptor is expressed in brain reward circuits, and animal studies have shown these drugs reduce the rewarding effects of alcohol. This study provides the largest human evidence to date supporting a real clinical effect. If confirmed in randomized trials, GLP-1 drugs could become the first effective pharmacotherapy for a condition that kills 3 million people per year globally. The findings also support the broader concept that metabolic peptide pathways intersect fundamentally with addiction neurobiology.

This is an observational study — despite the strong within-individual design, it cannot definitively prove causation. GLP-1 drug users had comorbid obesity/diabetes, and periods of GLP-1 use may coincide with periods of better overall health engagement (healthy user bias). The number of GLP-1 users (4,321 semaglutide, 2,509 liraglutide) is a fraction of the total AUD cohort. The study cannot determine whether GLP-1 drugs directly reduce alcohol craving/consumption or indirectly improve outcomes through weight loss, better diabetes control, or overall health improvement. Results may not generalize to AUD patients without metabolic comorbidities.