A study published in BMJ on March 4 analyzed more than 600,000 American veterans prescribed GLP-1 receptor agonists for diabetes and found that they were substantially less likely to develop substance use disorders than a matched comparison group. The reductions were not marginal. GLP-1 users had an 18 percent lower risk of new alcohol use disorder, a 14 percent lower risk of cannabis use disorder, a 20 percent lower risk of cocaine use disorder, a 26 percent lower risk of nicotine use disorder, and reduced risk across opioid and stimulant categories. [1] [2]

The study is observational, which means it can identify association but not causality. But the signal runs across every major substance category tested — alcohol, tobacco, cannabis, opioids, cocaine — in a cohort of more than 600,000 people matched against a control group taking a different diabetes medication. The breadth of the effect across substances is what distinguishes this finding from prior smaller studies and makes the dopamine hypothesis harder to dismiss. [2] [3]

The dopamine hypothesis holds that GLP-1 drugs work, in part, by modulating reward circuitry in the brain. The drugs act on GLP-1 receptors that are expressed not only in the pancreas and gut — where their primary metabolic effects occur — but also in the ventral tegmental area and nucleus accumbens, the structures at the core of addiction's neurobiological architecture. In animal studies, semaglutide has been shown to reduce alcohol consumption in rats, cocaine self-administration, and nicotine seeking. The BMJ study extends that signal to the largest human dataset yet assembled. [1] [4]

NPR reported this as "addiction next" — the story of a drug class that began as a diabetes treatment, became a weight loss intervention, revealed cardiovascular benefits, and is now showing evidence of addiction protection. Each expansion has followed the same pattern: unexpected benefit discovered in large outcome studies, mechanism partially understood, clinical application years behind the evidence. [4]

The clinical implications are constrained by the evidentiary base. The BMJ study was not designed to test GLP-1 drugs as addiction treatments. Its subjects were diabetic veterans, not addiction patients. Prescribing semaglutide off-label for alcohol or opioid use disorder requires a different evidence standard than observational association — it requires randomized trials with addiction-specific endpoints. Several such trials are underway; none has reported results. [3]

What the study does is change the terms of the conversation about why GLP-1 drugs work. The weight loss effect is explained by reduced appetite. The cardiovascular effect is partially explained by metabolic improvement and partially by direct cardiac GLP-1 receptor activity. The addiction effect, if it proves causal, would require a neurobiological explanation that repositions these drugs from metabolic agents to something closer to neuromodulators. [2]

The field is not ready to say that. But 600,000 people are saying something, and the pattern across every substance category tested is not the shape of coincidence.