The dataset was 606,434 U.S. veterans with type 2 diabetes. The question was whether GLP-1 medications — semaglutide, liraglutide, dulaglutide, the drugs that became household names as weight-loss tools — were doing something unusual to the people who took them. Something beyond blood sugar and appetite.
Published June 3, 2026 in The BMJ, the study from Ziyad Al-Aly and colleagues at Washington University School of Medicine in St. Louis found that yes, they were. Among participants without a baseline substance use disorder, GLP-1 medications were associated with a 14% lower risk of developing any SUD compared to SGLT2 inhibitors — the control comparator. The risk reductions by substance: opioids down 25%, cocaine down 20%, nicotine down 20%, alcohol down 18%, cannabis down 14%.
Among participants who already had a diagnosed substance use disorder, the numbers were more striking. GLP-1 users showed a 40% reduction in overdose events, a 50% reduction in drug-related deaths, a 30% reduction in emergency department visits, and a 25% reduction in hospitalizations.
These are the findings. What they mean requires more care.
The Mechanism: Quieting the “Drug Noise”
Al-Aly’s framing of the mechanism is useful. GLP-1 receptor agonists, he told reporters, appear to act “against the craving itself” — not against a specific substance, but against the underlying dopaminergic signal that addictive substances hijack. The analogy he drew was to “food noise” — the persistent, intrusive mental chatter around eating that GLP-1 drugs demonstrably quiet in people with obesity. In that framing, compulsive substance use is “drug noise,” and the drugs appear to turn down that volume across substance classes.
The biology is consistent with this. GLP-1 receptors are expressed in the ventral tegmental area and nucleus accumbens — the brain structures at the center of the reward system. When GLP-1 agonists activate those receptors, they appear to dampen the dopamine response to rewarding stimuli, including substances. Earlier research from Washington University published in March 2026 showed that GLP-1 medications reduced both craving and consumption in a range of addictive behaviors in preclinical models, including alcohol, cocaine, and nicotine, by targeting what the researchers described as a “common biological pathway underlying addiction.”
A separately published randomized clinical trial in JAMA Psychiatry found that once-weekly semaglutide reduced alcohol craving and consumption in adults with alcohol use disorder, though the effects were significant on craving more than on consumption measures at 9 weeks. The BMJ observational study, drawing on a population three orders of magnitude larger, suggests the craving effect may translate to real-world risk reduction at scale.
What the Study Can and Cannot Say
Al-Aly’s team was careful about the design’s limits, and so should anyone reading the findings. This is an observational study, not a randomized controlled trial. The veteran population is predominantly male, older, and has a high prevalence of comorbidities that may not generalize to the broader SUD population. SGLT2 inhibitors are a reasonable comparator for cardiovascular outcomes research, but they are not an inert placebo — they have their own pharmacological effects.
The study also can’t answer the question that matters most clinically: whether GLP-1 drugs work as an addiction treatment when prescribed for that purpose, in people whose primary diagnosis is SUD rather than diabetes or obesity. The veterans in this cohort were taking semaglutide for metabolic indications. The SUD effects were observed, not intended. Whether a clinical trial specifically targeting SUD populations and dosing GLP-1 medications for addiction — rather than for blood sugar — would replicate these findings is unknown. More than a dozen clinical trials are underway.
What It Means for the Treatment Field Right Now
The FDA has not approved GLP-1 drugs for any addiction indication. Prescribing them for SUD is currently off-label, with all the insurance and reimbursement complications that implies. No treatment guideline currently recommends semaglutide as part of an SUD treatment protocol.
No treatment guideline currently recommends semaglutide as part of an SUD treatment protocol.
That will change if the RCT data continues to develop this way. The observational signal across 600,000 people — across six substance classes, across both prevention and harm reduction contexts — is not noise. It is a direction worth following urgently. Given that the existing pharmacotherapy toolkit for most substance use disorders is thin (effective medications exist primarily for opioid and alcohol use disorder; for stimulants and cannabis, there is essentially nothing with comparable evidence), a drug class that appears to reduce craving cross-substance and at scale represents one of the more significant treatment signals in years.
The implication for technology platforms like Rize is also worth noting: as GLP-1 prescribing for SUD moves from off-label exception toward potential standard of care, the insurance navigation complexity will grow. Which plans cover it. Which diagnoses justify it. Which prior authorization pathways exist. These are exactly the frictions that an AI-guided treatment navigator is built to reduce.
For now, the most honest framing is this: a very large study found a very striking result using a drug class that wasn’t designed to treat addiction. The result is consistent with a plausible biological mechanism. The result has not been replicated in a controlled addiction-treatment trial. It should be taken seriously and acted on in research prioritization. It should not yet be taken as a basis for changing clinical practice.
But if you work in SUD treatment and you’re not watching the GLP-1 literature right now, you will be behind when it resolves.
Filed Under
sciencebiologytreatment