Only one in ten adults with alcohol use disorder in the United States receives any form of evidence-based pharmacological treatment. That number has been stuck for years, resistant to guideline updates, payer reforms, and public health campaigns alike. So when a medication class already in widespread use for diabetes and obesity starts showing meaningful signals for addiction-related outcomes, the field has an obligation to look carefully rather than cautiously.
A study published this month in Lancet Psychiatry did exactly that. Using Sweden’s national patient registers and a within-individual design, researchers examined whether GLP-1 receptor agonist treatment was associated with lower rates of hospitalization in individuals with alcohol use disorder and substance use disorder. The within-individual approach is important: it compares each person’s hospitalization rate during treatment periods against their own rate during non-treatment periods, which strips away the confounding-by-indication problem that haunts most observational addiction research. The finding was significant. GLP-1 treatment correlated with reduced hospitalizations across both AUD and SUD populations. More striking, for the AUD group, the protective association persisted for up to 182 days after discontinuation. For the broader SUD group, it did not.
Why the Reward System Explanation Is No Longer Theoretical
The mechanistic story has been building for several years. GLP-1 receptors are expressed in dopaminergic regions of the brain, including the ventral tegmental area and nucleus accumbens. Animal models have shown that GLP-1 agonism attenuates alcohol-seeking behavior, reduces binge-drinking episodes, and blunts the dopaminergic response to reward cues. Until recently, the human clinical data were fragmentary, drawn from incidental observations in diabetes trials. That picture changed substantially in March 2026, when Washington University School of Medicine published a study in The BMJ analyzing 606,434 U.S. veterans with type 2 diabetes, finding that GLP-1 medications reduced risk of incident substance use disorders and adverse SUD-related outcomes across multiple substance classes. Now the Lancet Psychiatry study adds a hospitalization endpoint — a harder, more operationally meaningful signal — from a national register dataset that covers an entire country’s patient population.
The divergence between AUD and broader SUD after discontinuation deserves attention. Alcohol’s interaction with GLP-1 signaling may be mechanistically distinct from, say, opioid or stimulant-driven reward pathways. Alcohol is a GABAergic and glutamatergic molecule with secondary effects on endocannabinoid and opioid systems — the GLP-1 receptor’s modulatory influence on dopamine release may produce a more durable resetting of alcohol-specific reward salience than it can for substances that act more directly on opioid receptors. In my practice running ketamine therapy for patients with concurrent mood disorder and AUD, I see the parallel: rapid-acting interventions that modulate reward circuitry can produce extended benefit windows that outlast the acute pharmacological effect. The mechanism here may be related.
An earlier large cohort study published in The BMJ also found that initiation of GLP-1 receptor agonists was associated with reduced risk of developing incident substance use disorders, adding further convergent evidence that this signal is replicable across study designs and populations.
The Discontinuation Problem Nobody Has a Protocol For
Here is where the clinical implications become urgent and uncomfortable. GLP-1 agonists are not prescribed for AUD. They are prescribed for diabetes and obesity, and patients discontinue them for the reasons those patients always do: side effects, cost, insurance coverage changes, weight loss plateaus. Nobody in that prescribing chain is asking about the patient’s drinking history or monitoring for AUD relapse when the medication stops.
The Lancet Psychiatry authors explicitly call for clinical monitoring of substance use when GLP-1 treatment is stopped. That is a reasonable recommendation. It is also, at this moment, almost entirely unoperationalized. The StatPearls review on addiction relapse prevention notes directly that “there have been no standard relapse prevention programs established” — and that was before GLP-1 agonists entered the clinical picture as an unintentional protective factor. We have no screening protocol to capture AUD history before GLP-1 initiation, no structured monitoring window at discontinuation, and no handoff pathway to addiction medicine when a primary care physician stops semaglutide because a patient’s A1C normalized.
In my experience running SUD trials and managing dual diagnosis patients across telehealth and in-person settings, the discontinuation transition is where the field consistently underinvests. A medication that provides 182 days of post-treatment protection in AUD creates a clinically meaningful runway — but only if someone is watching during those 182 days. Right now, in the real-world prescribing environment, nobody has been assigned to watch.
For trial designers, the design implication is direct. Any prospective study of GLP-1 agonists in AUD or SUD populations needs pre-specified discontinuation monitoring arms, with relapse events and hospitalization rates tracked as primary or secondary endpoints through at least six months post-cessation. The Swedish register methodology here is useful precisely because it captures what happens in the gaps — but a prospective design could also capture why patients discontinue, which observational data cannot. Spanish-speaking and underserved populations, who carry disproportionate AUD burden and access GLP-1 medications inconsistently due to cost and insurance barriers, are the populations most likely to experience abrupt, unplanned discontinuation. They are also the populations systematically under-enrolled in trials that could answer this question. That enrollment gap represents both a scientific and an ethical failure the field should address before the next readout.
The BMJ Open analysis of five Swedish national registers confirms that single-register studies underestimate true AUD and SUD prevalence — which means even the Lancet Psychiatry findings may be understating the effect. I am watching for the first prospective, adequately powered, dual-diagnosis-inclusive trial of a GLP-1 agonist with AUD as a primary indication. When that protocol surfaces, the endpoint and discontinuation design choices will tell us everything about whether the field learned what this Swedish cohort was trying to show.
References
- Lancet Psychiatry — “Association of GLP-1 receptor agonists with alcohol use disorder-related and substance use disorder-related hospital admissions during treatment and after discontinuation: a Swedish register-based within-individual observational study”
- Washington University School of Medicine — “GLP-1 medications get at the heart of addiction, study finds” (The BMJ, March 4, 2026)
- The BMJ — Large cohort study: GLP-1 receptor agonist initiation associated with reduced risk of incident substance use disorders
- BMJ Open — “Cross-sectional population-based study on indications of alcohol and substance use disorders across five Swedish national registers”
- StatPearls — “Addiction Relapse Prevention” (updated July 21, 2023)
Dr. Leonardo Vando, MD is Principal Investigator of Psychiatry at Vando Medical Services.
