Ozempic curbs hunger… but not just for food
A new study shows the diet drug might also help people quit smoking.
If you’ve only been paying attention to the headlines, when you think of “Ozempic” you’ll think of a few things — a blockbuster weight loss drug, sure. The tip of the spear of an entire new industry — why not? A drug so popular the people it was invented for, those with diabetes, can’t even get it. Drugs like Ozempic, the GLP-1 receptor agonists are undeniably game changers. Insofar as obesity is the number one public health risk in the United States, anti-obesity drugs hold immense promise even if all they do is reduce obesity.
But if you’ve been looking a bit deeper than the headline-grabbing stories, if you’ve been reading some of the case reports or, you know, listening to your patients, you’ll start to wonder if Ozempic isn’t doing something more. In 2023, an article in Scientific Reports presented data suggesting people on Ozempic might be reducing their alcohol intake — not just their total calories.
A 2024 article in Molecular Psychiatry found that the drug might positively impact cannabis use disorder.
Here’s an article from Brain Sciences that suggests the drug reduces compulsive shopping.
A picture is starting to form. A picture that suggests these drugs curb hunger both literally and figuratively. That GLP-1 receptor agonists like Ozempic and Mounjaro are fundamentally anti-consumption drugs. And in a society that, some would argue, is plagued by overconsumption — well these drugs might be just what the doctor ordered.
If only they could stop people from smoking.
Oh wait. They can.
At least — it seems like they can, based on a new study appearing in the Annals of Internal Medicine.
Before we get too excited, no, this is not a randomized trial. There actually has been a small randomized trial of exenatide — Byetta — which is in the same class as Ozempic but probably a bit less potent — with promising results for smoking cessation.
But Byetta is the weaker drug in this class — the market leader is Ozempic. So how can you figure out if Ozempic can reduce smoking without doing a huge and expensive randomized trial? You can do what Nora Volkow and colleagues from the National Institute on Drug Abuse did — you can do a target trial emulation study.
A target trial emulation study is more-or-less what it sounds like. First you decide what your dream randomized controlled trial would be — you plan it all out in great detail. You define the population you would recruit, with all the relevant inclusion and exclusion criteria. You define the intervention and the control, and you define the outcome.
But you don’t actually do the trial. I mean, you could if someone would lend you 10–50 million dollars but assuming you don’t have that lying around, you do the next best thing, which is to dig into a medical record database to find all the people who would be eligible for your imaginary trial. And you analyze them.
In the case of the Annals paper, the authors define seven target trials. Ozempic vs. insulin, Ozempic vs. metformin, Ozempic vs. DPP-4 inhibitors, Ozempic vs. SGLT2 inhibitors, Ozempic vs. sulfonylureas, Ozempic vs. thiazolenediones and finally Ozempic vs. other GLP-1 receptor agonists. None of these trials are happening in the short term.
The authors wanted to study the effect of Ozempic on smoking among people with diabetes — that’s why all those comparator agents are anti-diabetic drugs. They figured out the folks were smoking based on having a medical record diagnosis of tobacco use disorder before they started one of the drugs of interest. This code is fairly specific — if a patient has it you can be pretty sure they are probably smoking — but not very sensitive — not every smoker has this code. This is an age-old limitation of using electronic health record data instead of, you know, asking patients — but it’s part of the trade off for not having to spend 50 million dollars.
After applying all those inclusion and exclusion criteria, they have a defined population of people who could be in their dream trial. And, as luck would have it, some of those people really did get treated with Ozempic, and some really did get treated with those other agents. Although the decisions about what to treat them with were not randomized, the authors account for this confounding-by-indication using propensity score matching. You can find a little explainer on propensity-score matching from an earlier column here.
OK it’s easy enough to figure out who has diabetes and who got what drug from the electronic health record. But how do you tell who quit smoking? Remember — everyone had a diagnosis code for tobacco use disorder prior to starting Ozempic or a comparator drug. The authors decided that if the patient had a medical visit where someone again coded tobacco-use disorder, they were still smoking. If someone prescribed smoking cessation meds like a nicotine patch or varenicline — they were obviously still smoking. If someone billed for tobacco-cessation counseling, the patient was still smoking. We’ll get back to the implications of this outcome definition in a minute.
But let’s talk about the results, which are pretty intriguing.
When Ozempic is compared to insulin among diabetic smokers, those on Ozempic are about 30% more likely to quit smoking. They are about 18% more likely to quit smoking than those who took metformin. They’re even slightly more likely to quit smoking than those on other GLP-1 receptor antagonists — though I should note that Mounajro — which is probably the more potent GLP1 drug in terms of weight loss, was not among the comparators.
This is pretty impressive for a drug that, you know, was not designed to be a smoking cessation drug. And it speaks to this emerging idea that these drugs do more than curb appetite by slowing down gastric emptying or something. They work in the brain — modulating some of the reward circuitry that keeps us locked into our bad habits.
There are, of course, some caveats. As I pointed out, this study captured the idea of “still smoking” through the use of administrative codes in the electronic health record and things like prescription of smoking cessation aids. You could see similar results to what was shown here if taking Ozempic just makes you less likely to address your smoking at all — maybe you shut down the doctor before they even talk about it with you, or you have too much to discuss during these visits to even get onto the subject of smoking. You could also see results like this if people taking Ozempic had fewer visits overall — but the authors showed that that, at least, was not the case.
And I’m inclined to believe this effect is real, simply because we keep seeing signals like this from multiple sources. If that turns out to be the case, these new “weight loss” drugs may prove to be much more than that — they may turn out to be the drugs that can finally save us from ourselves.
A version of this commentary first appeared on Medscape.com.