In March 2026, The BMJ published the largest study yet on GLP-1 medications and substance use disorders. Analyzing medical records from more than 606,000 U.S. veterans with type 2 diabetes, the researchers found that people prescribed a GLP-1 receptor agonist were 14% less likely than those on a different class of diabetes drug to develop any new substance use disorder — a signal that held across alcohol, opioids, nicotine, cannabis, and cocaine.
But the study didn't stop at new diagnoses. Among veterans who already had a substance use disorder, GLP-1 users had significantly fewer overdoses, hospitalizations, emergency department visits, and substance-related deaths.
Why This Study Stands Out
Most previous GLP-1/addiction research examined one substance at a time — Wang/Volkow studied alcohol, then opioids, then cannabis, then tobacco in separate papers. This study looked at all substance use disorders simultaneously in a single cohort, using the same methodology and comparison group. That unified approach is important because it answers the broader question: is GLP-1's effect substance-specific, or does it operate across the board?
The answer: across the board. The reduced risk was present for every substance category examined.
The Veteran Population: Uniquely Informative
The VA healthcare system captures medical data with unusual completeness — prescription records, diagnoses, hospitalizations, lab work, and mortality data all flow through a unified system. This makes VA data particularly reliable for observational research, with less risk of the information gaps that plague commercial insurance databases. Additionally, veterans have elevated rates of substance use disorders compared to the general population, providing a large enough sample to detect effects across substance categories.
The Key Findings
New Substance Use Disorders (Prevention)
Among veterans with type 2 diabetes and no prior substance use disorder, those who started a GLP-1 agonist were compared to those who started an SGLT2 inhibitor (another newer diabetes drug class). GLP-1 users showed a 14% lower overall risk of being diagnosed with any new SUD. For individual substances, the risk reductions ranged from modest to substantial, with some categories showing up to 25% lower risk.
Severe Outcomes (Harm Reduction)
Among veterans who already had a substance use disorder diagnosis, GLP-1 medication use was associated with sharp reductions in the worst outcomes: overdoses, substance-related hospitalizations, emergency department visits, and substance-related deaths. This harm reduction finding is arguably more clinically significant than the prevention data, because it suggests GLP-1 medications might reduce the severity of existing addictions — not just prevent new ones.
Study Design Strengths
The study used an active comparator design — comparing GLP-1 users against SGLT2 inhibitor users rather than against the general population or against patients on no diabetes medication. This is important because both drug classes are prescribed to a similar patient profile (relatively well-managed type 2 diabetes), reducing confounding from differences in healthcare engagement, comorbidity burden, and socioeconomic factors.
Additional methodological strengths included propensity-score matching, intention-to-treat analysis, and multiple sensitivity analyses testing the robustness of the findings against various potential biases.
What Experts Said
Lorenzo Leggio, an addiction researcher at the National Institute on Drug Abuse who is leading a separate semaglutide-for-AUD trial, characterized the findings as important but insufficient on their own to justify prescribing GLP-1 medications for addiction. He emphasized that observational data, no matter how large, cannot substitute for randomized controlled trials — particularly when it comes to recommending these medications over established addiction treatments like naltrexone.
This measured response reflects the scientific community's broader stance: the observational evidence is now overwhelming in quantity and remarkably consistent in direction, but the field still needs more RCTs to confirm causation and guide clinical practice.
How This Fits the Evidence Pyramid
The BMJ veterans study occupies a specific position in the growing evidence base:
| Evidence Type | Study | What It Shows |
|---|---|---|
| Randomized Trial | Hendershot (JAMA Psych) | Semaglutide directly reduces drinking in AUD |
| Within-Individual Registry | Lähteenvuo (JAMA Psych) | Same people hospitalized less on GLP-1 vs. off |
| Multi-Substance Cohort | BMJ Veterans Study | GLP-1 signal spans all substances simultaneously |
| Single-Substance Cohort | Wang/Volkow (multiple) | Individual substance associations replicated across cohorts |
| Small RCT | Grigson/Penn State | Liraglutide reduces opioid craving 40% |
| Preclinical | Jerlhag, Merkel, others | Animal models show reduced drug-seeking |
The BMJ study is the widest lens in this evidence stack — not the sharpest (that's the Hendershot RCT), but the broadest. Its value is in demonstrating that the GLP-1/addiction signal isn't an artifact of studying one substance in one population with one methodology. It appears across every substance, in a different population (veterans vs. the general or Swedish populations in other studies), using different comparators.
Observational study. Cannot prove causation. Confounding by indication remains possible — veterans prescribed GLP-1 medications may differ from SGLT2 users in unmeasured ways. Healthy user bias (people who start and maintain GLP-1 therapy may be generally more health-engaged) is plausible. These findings justify further research, not clinical recommendations for off-label use.
Interested in GLP-1 Treatment?
GLP-1 medications are prescribed for weight management and diabetes. We compare 20+ telehealth providers on pricing, speed, and medication options.
Compare Verified Providers →Sources
- GLP-1 drugs linked to lower addiction rates in large study of veterans. Science/The BMJ. March 2026.
- Hendershot CS, et al. Once-weekly semaglutide in AUD. JAMA Psychiatry. 2025;82(4):395-405.
- Lähteenvuo M, et al. Repurposing semaglutide and liraglutide for AUD. JAMA Psychiatry. 2025;82(1):94-98.
- Volkow ND. GLP-1R agonist medications for addiction treatment. Addiction. 2025.