GLP‑1 Agents and Associations with Reduced Substance Use in a 600,000‑Patient Cohort

An observational analysis in U.S. Department of Veterans Affairs patients with type 2 diabetes found that starting a GLP-1 receptor agonist was associated with lower rates of multiple substance-use–related outcomes over a three-year period compared with starting a sodium-glucose cotransporter-2 (SGLT-2) inhibitor.

The article frames the findings as associations rather than evidence of causality, placing them alongside discussion of biologic plausibility and uncertainty. It also notes that the reported patterns include both relapse-related events and incident diagnoses across several substance categories. Overall, the piece presents these results as signals to interpret within the limits of observational data, while noting that aspects of the study design—such as a new-user design, an active comparator, and extensive confounding control—were intended to strengthen causal interpretation.

Investigators used Veterans Affairs electronic health record data to emulate eight parallel target trials comparing initiation of a GLP-1 receptor agonist with initiation of an SGLT-2 inhibitor, a drug class with similar cardiometabolic indications. The analysis used statistical adjustment to account for measured baseline differences between groups. This is presented as an effort to make the groups more comparable on recorded characteristics while reflecting real-world prescribing patterns. The piece then distinguishes results for patients with pre-existing substance problems from those without a prior substance use disorder.

Among patients described as already living with a substance use disorder, the analysis is reported to have found lower rates of several severe outcomes among GLP-1 users compared with SGLT-2 inhibitor users: about 50% lower mortality related to substance use, 39% fewer overdoses, 26% fewer drug-related hospital admissions, and a 25% lower risk of suicidal ideation or attempt. The study also reported fewer substance-use–related emergency department visits. In absolute terms, these findings corresponded to roughly one to ten fewer events per 1,000 people over three years. These results are presented as adjusted associations rather than proof of benefit attributable to the medication exposure.

For patients without a prior substance use disorder, the article reports associations with lower incidence of new diagnoses across multiple substances. It describes about an 18% lower risk for alcohol use disorder, a 14% lower risk for cannabis use disorder, a 20% lower risk for cocaine use disorder, a 20% lower risk for nicotine dependence, and a 25% lower risk for opioid use disorder among GLP-1 users compared with SGLT-2 inhibitor users. In absolute terms over three years, the piece states this corresponds to roughly one to six fewer cases per 1,000 people. By presenting results across several categories together, the article emphasizes the breadth of the observed pattern rather than an association limited to a single substance class.

In discussing biologic plausibility, the article points to GLP-1 activity in brain regions involved in reward, motivation, and stress and describes preclinical findings in which GLP-1 receptor activation reduced reward-driven substance-seeking behaviors in animal models. It also references early randomized-trial signals, including a study of semaglutide in alcohol use disorder that reported reductions in alcohol craving and some drinking outcomes over short follow-up.

Alongside this context, the article lists caveats: the observational design leaves room for residual confounding; substance use disorders and overdoses may be underdiagnosed; differences in healthcare contact could influence outcome detection; and the cohort consisted predominantly of older men, which limits generalizability to women, younger people, and non-US populations. It also notes that the estimated effect reflects starting a GLP-1 receptor agonist rather than starting an SGLT-2 inhibitor, and therefore does not establish benefit compared with no treatment or with evidence-based addiction pharmacotherapies. The overall framing is that the reported associations are hypothesis-generating and accompanied by unresolved questions.

Key Takeaways:

• The article reports an association between GLP-1 receptor agonist initiation and lower rates of multiple substance-use–related outcomes over three years compared with initiation of an SGLT-2 inhibitor in a large VA electronic health record–based analysis.
• Reported patterns differ by baseline history, with fewer severe events described among patients with existing substance use disorders and fewer incident diagnoses across several substances among those without prior SUD.
• Proposed reward-circuit mechanisms and supportive preclinical and early clinical evidence are presented alongside stated uncertainties, including residual confounding, potential underdiagnosis of outcomes, and limited generalizability.