GLP-1 Drugs and Quitting Smoking: What the Trials Actually Tested

A GLP-1 does not appear to help you stop smoking. The largest randomized test of the idea gave 255 smokers either dulaglutide or placebo on top of varenicline and counseling, and after twelve weeks 63% of the drug group and 65% of the placebo group had quit — a difference of −1.9%, which is nothing.^[1] What the drug did do was hold the weight: down 1.0 kg while the placebo group went up 1.9 kg. That is a real effect, and it is the one worth understanding, because a year later it had gone too.^[2]

What the trial actually tested

The trial ran at the University Hospital Basel between 2017 and 2020. Everyone in it was a smoker with at least moderate cigarette dependence who wanted to quit, and everyone got the standard of care: varenicline at 2 mg a day plus behavioral counseling. The randomization decided only whether a weekly injection of dulaglutide 1.5 mg or a weekly injection of placebo was added on top for twelve weeks.^[1]

That design matters when you read the result. This was not a test of whether a GLP-1 beats nothing. It was a test of whether a GLP-1 adds anything to treatment that already works, which is the question a person who is already being helped to quit would actually ask. The answer was no.

Craving was measured directly, and it fell during treatment in both groups with no difference between them.^[1] That is worth saying plainly, because the reason anyone expected this to work was a craving mechanism — the idea that GLP-1 signaling dampens reward in a way that would reach nicotine as well as food. In this trial, it did not show up in what the participants reported.

The weight benefit did not survive the year

The same team followed the same participants to 24 and 52 weeks, and that follow-up is the part almost nobody quotes. Abstinence at one year was 32% in both groups — identical.^[2] The weight advantage narrowed to −1.0 kg by week 24 and to −0.35 kg by week 52, with a confidence interval spanning zero. Both groups had gained weight from baseline by then.

If you are weighing this up, that is the trade-off in one line: the drug holds post-cessation weight for as long as you are taking it, and does not help you quit at any point. Whether that is worth an injection is a different question from the one the headlines answered, and it is yours and your prescriber’s.

The smaller trial that pointed the other way

There is a second randomized trial, and it is more encouraging. Eighty-four prediabetic or overweight smokers were randomized to weekly exenatide 2 mg or placebo, all of them on a 21 mg nicotine patch with brief counseling, for six weeks. Abstinence was 46.3% on exenatide against 26.8% on placebo — a risk ratio of 1.70. Post-cessation body weight was 5.6 pounds lower in the exenatide group.^[3]

Read the interval before the ratio. The 95% credible interval on that 1.70 ran from 0.96 to 3.27, which includes the possibility of no benefit at all. The paper calls itself a pilot, it enrolled a quarter as many people as the Basel trial, it ran for half as long, and it selected for smokers who were already prediabetic or overweight. Those are the conditions under which a promising number most often fails to repeat.

The study that made the headlines is not a trial

In 2024 a paper in the Annals of Internal Medicine reported that semaglutide was associated with lower rates of tobacco-use-disorder care. It compared 5,967 new semaglutide users against people starting seven other diabetes drugs, out of 222,942 patients with both type 2 diabetes and a tobacco use disorder diagnosis. Against insulin the hazard ratio was 0.68; against other GLP-1 drugs it was 0.88.^[4]

What it counted is the important part. The outcomes were a medical encounter recording a tobacco use disorder diagnosis, a prescription for a smoking cessation medication, and a counseling session. Those are health care events. The paper’s own limitations section records missing data on current smoking behavior.^[4] It did not observe whether anyone smoked fewer cigarettes, and it does not claim to have.

Fewer appointments about smoking is not the same finding as less smoking. It can also be what happens when someone is already busy managing a new injectable.

There is one more tell. Most of the between-group differences appeared within 30 days of the first prescription.^[4] A real effect on an addiction that people take months to break would be an odd thing to see almost entirely inside the first month. The authors are careful about this: their stated conclusion is that the findings suggest a need for clinical trials, not that the question is settled.

Where that leaves you

• No GLP-1 is approved for smoking cessation, and none of this evidence asks for one to be.
• The best randomized evidence says a GLP-1 added to treatment that already works does not improve your odds of quitting, at 12 weeks or at a year.
• It does hold post-cessation weight while you take it. Three months of treatment did not hold it at twelve.
• Gastrointestinal side effects were very common in the trial — 90% of the dulaglutide group, though 81% of the placebo group reported them too.
• Varenicline and nicotine replacement are the therapies these trials were built on top of, because they are the ones with the evidence.

If you are already taking a GLP-1 and thinking about quitting smoking, none of this is a reason to stop either one. If you are considering starting a GLP-1 in order to quit, the trials do not support that, and the conversation to have is with the person who would prescribe it. We keep a dated price for every seller we can find on the live price tracker, and the methodology sets out how we check them.

This question sits inside a bigger one. For what the same drugs look like across every substance class measured — and for the randomized behavioral trial that missed its endpoints — see GLP-1s and substance use. The nearest sibling with a dedicated trial of its own is alcohol: semaglutide and alcohol use disorder and alcohol and GLP-1s.