Scientific deep-dive

GLP-1 Drugs and Addiction: A Huge Signal, and a Trial That Missed

A 524,817-person cohort found fewer new substance use disorders of every kind measured. A randomized trial in people who smoke then missed both its primary endpoints. Both landed in 2026.

By Nora Bissett · Pricing Editor
Editorially reviewed. Not medical advice · How we verify contentLast reviewed
7 min read·4 citations

Two studies landed in 2026 and they point different ways. A cohort of 524,817 US veterans found people starting a GLP-1 developed fewer new substance use disorders of every kind measured (alcohol, cannabis, cocaine, nicotine, opioids) than people starting a comparable drug.[1] A randomized trial of semaglutide in people who smoke daily then missed both its primary endpoints: craving fell and weight fell, but resistance to smoking and cigarettes per day did not move significantly.[2] That trial randomized 24 people. Both results need to be read together.

What the cohort found

The design is better than most observational work. Drawing on 606,434 US veterans with type 2 diabetes, the researchers emulated eight parallel target trials, comparing people who started a GLP-1 against people who started an SGLT-2 inhibitor. Comparing against an active drug rather than against nobody controls for the fact that people who get prescribed something differ from people who get prescribed nothing.[1]

New substance use disorders over up to three years, GLP-1 initiators against SGLT-2 inhibitor initiators. Hazard ratios below 1 favor the GLP-1.
DisorderHazard ratio (95% CI)
Opioid use0.75 (0.67–0.85)
Cocaine use0.80 (0.72–0.88)
Nicotine use0.80 (0.74–0.87)
Alcohol use0.82 (0.78–0.85)
Cannabis use0.86 (0.81–0.90)
All substance use disorders combined0.86 (0.83–0.88)

Among veterans who already had a substance use disorder, the outcomes that matter clinically moved too: fewer emergency department visits (0.69), fewer hospital admissions (0.74), fewer overdoses (0.61), and substance-related mortality at 0.50, or half.[1]

Read the absolute numbers alongside the ratios. The composite works out to about 6.6 fewer diagnoses per 1,000 people over three years. A hazard ratio of 0.86 is a real effect, but on its own it does not transform anyone’s life expectancy. Both numbers describe the same finding; the ratio is the one that gets quoted.

What the randomized trial found

Also in 2026, a phase 2a trial gave semaglutide or placebo for nine weeks to adults smoking at least five cigarettes a day, with laboratory sessions measuring how well they resisted smoking and how much they smoked when they could.[2]

Neither co-primary outcome reached significance. Smoking resistance came in at β 0.16 (95% CI −0.07 to 0.40, P = .16); number of cigarettes at β −0.08 (95% CI −0.25 to 0.08, P = .30).[2] What did move was craving for cigarettes, which fell significantly, along with body weight.

Wanting a cigarette less is not the same as smoking fewer of them, and this trial separated the two. (What nicotine itself does to weight is set out in does vaping or nicotine cause weight loss.)
It randomized 24 people. The authors did not hide this. It is a phase 2a laboratory study, designed to detect mechanism rather than to establish cessation, and their own conclusion calls for larger trials. But the null results are weakly powered, and a supplementary analysis that did reach significance should be read as a hypothesis rather than a finding. Small trials produce both kinds of error.

Why the two disagree, and which to trust for what

This is the same pattern our alcohol article documents, in a different substance and by the same research group: a large observational signal, and randomized tests that shrink or vanish when they measure behavior rather than desire.

The most likely explanation is not that one study is wrong. Observational data compare people who took a drug against people who did not, and no weighting fully removes the differences between those groups. Someone prescribed a GLP-1 is engaged with healthcare in ways that themselves predict fewer substance problems. Randomized trials remove that, and pay for it in size and duration. Nine weeks in 24 people is not a fair test of whether someone quits smoking.

The signal is real enough to be worth a properly powered trial, and there is not yet convincing evidence that these drugs help anyone quit anything. Those two statements are compatible.

One result that checks out against something else

Among veterans with a pre-existing substance use disorder, suicidal ideation or attempt came in at a hazard ratio of 0.75 (0.67–0.83).[1] That points the same way as the evidence behind the removal of the suicidality warning from these labels in February 2026, which we cover in the suicidality article. It comes from an entirely separate dataset and population.

Two independent bodies of evidence agreeing is worth more than either alone. It is still observational, and it is still in a population that is 90%-odd male veterans with type 2 diabetes.

Smoking: five findings from eighty-four people

A pilot trial randomized 84 smokers with prediabetes or excess weight to exenatide or placebo, everyone also receiving a nicotine patch and brief counseling, with biologically confirmed abstinence at six weeks as the outcome. The drug arm did better. A follow-up analysis then asked who it worked for, and reported five moderators.[4] A larger trial has since tested the question head-on and found nothing; see GLP-1s and quitting smoking for that evidence and its 12-month follow-up.

Reported moderators of exenatide’s effect. PP is posterior probability that an effect exists; the study’s threshold for reporting was 75%.[4]
Stronger effect in…Posterior probability
People smoking more than 20 cigarettes a day81.7%
People without prediabetes76.0%
People without obesity94.4%
People with no or minimal depression symptoms91.2%
Carriers of one CHRNA rs16969968 genotype88.6%
Read the threshold before the findings. These are Bayesian posterior probabilities, not P values, and the bar for reporting was 75%. A result could therefore be included while carrying close to a one-in-four chance that no effect exists at all. That is an exploratory standard, appropriate for generating hypotheses and much weaker than the significance testing most readers will assume.

Now look at rows two and three. The trial enrolled smokers with prediabetes and/or excess weight, yet reports the drug working best in participants who had neither. When five subgroups are carved out of 84 people and the treatment looks most effective in the one least like the eligibility criteria, suspect noise before biology.

The genetic moderator deserves the same caution as every other pharmacogenomic subgroup in this field. Real associations do exist, and their effects are small (see the genetics are real, and small). A genotype effect found in a fraction of 84 people is a long way short of that.

What survives is the parent trial’s own result: adding a GLP-1 to a nicotine patch improved short-term abstinence in a pilot. The authors ask for larger prospective studies, which is the right request.

A larger emulation, and the split within the class

A 2026 analysis went further, emulating four separate trials inside a records network covering more than 120 million patients. Adults with type 2 diabetes and no prior alcohol-use disorder starting tirzepatide, semaglutide, liraglutide or dulaglutide were each matched one-to-one against patients starting a DPP-4 inhibitor, and followed 18 months for a first diagnosis.[3]

Incident alcohol-use disorder against DPP-4 inhibitors, after propensity matching.
DrugMatched pairsHazard ratio (95% CI)
Tirzepatide7,1650.47 (0.29–0.75)
Semaglutide20,1980.68 (0.52–0.89)
Liraglutide6,565Not significant
Dulaglutide19,061Not significant

Head to head, tirzepatide beat liraglutide directly (0.47, 0.24–0.92). The split is the interesting part: the two newer, more potent agents show the effect and the two older ones do not, in the same drug class.

That pattern has at least two explanations and the study cannot separate them. It could be dose or potency (more receptor activation, more effect on the reward pathways implicated here). Or it could be who takes what: tirzepatide and semaglutide are prescribed in a different era, to different patients, for different reasons, than liraglutide and dulaglutide. Propensity matching handles what is recorded and not what is not.

One more limit: the outcome is a diagnosis code. Someone has to seek care, and a clinician has to record alcohol-use disorder. That measures diagnosis rather than drinking, and it under-counts in both arms by an unknown amount. The authors conclude that randomized evidence for this indication is now appropriate, and that is the right conclusion.

Frequently Asked Questions

Not on current evidence. The randomized trial of semaglutide in people who smoke daily missed both its co-primary endpoints: neither resistance to smoking nor cigarettes per day changed significantly. Craving did fall, and so did weight. That trial randomized 24 people, so its null results are weakly powered too. A pilot of 84 smokers found exenatide added to a nicotine patch improved abstinence at six weeks, but a larger trial, of dulaglutide added to varenicline, found no benefit.
Among 524,817 people with type 2 diabetes, those starting a GLP-1 developed fewer new substance use disorders than those starting an SGLT-2 inhibitor. Hazard ratios ran from 0.75 for opioid use to 0.86 for cannabis, with 0.86 for all combined. In absolute terms that is roughly 6.6 fewer diagnoses per 1,000 people over three years.
They are answering slightly different questions with different weaknesses. Observational data cannot fully separate the drug from the kind of person who gets prescribed it. Randomized trials can, but this one ran nine weeks in 24 people, which is not a fair test of whether someone quits. The signal justifies a properly powered trial; it is not itself evidence of benefit.
No GLP-1 is approved for any substance use disorder, and the randomized evidence on behavior is null or absent. Substance use disorders have treatments that are approved and tested in far larger numbers, and a clinician can set those out.
It points the same way. Among veterans with a pre-existing substance use disorder, suicidal ideation or attempt came in at a hazard ratio of 0.75. That is consistent with the evidence behind the removal of the suicidality warning from these labels in February 2026, and it comes from an entirely separate dataset.

References

  1. 1.Cai M, Choi T, Xie Y, Al-Aly Z. Glucagon-like peptide-1 receptor agonists and risk of substance use disorders among US veterans with type 2 diabetes: cohort study BMJ. 2026. PMID: 41781010.
  2. 2.Hendershot CS, Bremmer MP, Paladino MB, et al. Once-Weekly Semaglutide in Adults With Daily Cigarette Use: A Randomized Clinical Trial JAMA Network Open. 2026. PMID: 42189538.
  3. 3.Henney AE, Riley DR, Heague M, et al. Relative efficacy of GLP-1 and GLP-1/GIP receptor agonists in the prevention of alcohol-use disorders using a target trial emulation approach Diabetes, Obesity and Metabolism. 2026. PMID: 41058240.
  4. 4.Yammine L, de Dios C, Suchting R, et al. Exploring Predictors of Treatment Response to GLP-1 Receptor Agonists for Smoking Cessation Nicotine & Tobacco Research. 2025. PMID: 39780397.

MOTS-c: What the Evidence Shows, and What FDA Found

MOTS-c slowed weight gain and improved insulin sensitivity in mice. No published trial has given it to people, and there is no established human dose. The 503A bulks list leaves it out; under an interim HHS policy dated October 2026, HHS does not intend to take enforcement action against 503A compounding under strict conditions.

11 min read

GLP-1 Drugs and HIV-Associated Lipohypertrophy: The Randomized Evidence

A phase 2b randomized trial found once-weekly semaglutide cut abdominal visceral fat by about 30.6% over 32 weeks in people with HIV-associated lipohypertrophy, alongside a 9.3% fall in psoas muscle volume that did not cost measurable function.

6 min read

Jardiance or Farxiga With Ozempic or Mounjaro: Taking an SGLT2 and a GLP-1

SGLT2 inhibitors like Jardiance and Farxiga are routinely taken with Ozempic or Mounjaro, and no label forbids it. Sugar and weight effects add up in trials; heart and kidney gains rest on subgroups. The real overlap is dehydration and ketoacidosis.

12 min read

GLP-1s and Atrial Fibrillation: Reading a 42% Headline

A meta-analysis of ten trials in 12,651 patients found a 42% lower risk of new atrial fibrillation on semaglutide. None of those trials was designed to measure it, and the absolute numbers were never published.

6 min read

Dry Mouth on a GLP-1: Three Published Cases, and What to Tell Your Dentist

The most detailed published evidence that GLP-1 drugs cause dry mouth is three case reports, alongside an adverse-event reporting signal for semaglutide. Nobody has measured how often it happens, but reduced saliva raises decay risk regardless of cause, which is the part worth acting on.

5 min read

Exenatide (Byetta, Bydureon): The First GLP-1, Its Weight Loss and Where It Went

Exenatide, sold as Byetta and Bydureon BCise, was the first GLP-1 drug. It is approved only for type 2 diabetes, produces far less weight loss than semaglutide, and both brands are discontinued in the US. Its one generic is now listed as discontinued too.

10 min read

Where to get semaglutide online, safely: sellers our editors have checked

These are telehealth sellers our editors have checked. For each one we hold a price, the form the drug comes in, and the states it reaches.

No insurance needed · vetted by our editors

Some of the links on this page earn us money. If you sign up with a provider after following one, that provider may pay GLP Watchdog a commission. Learn more

6.6

HumeCare+

An oral route if you will not self-inject

9.1

Collective

From $59/mo + membership

Get started →
Read full review →

About this analysis

GLP Watchdog publishes independently. Each seller in our dataset is checked against that seller's own website, and the dataset is updated continuously. Read our full methodology →