Skip to content
This GLP
← Research
Evidence

Five subgroups and a pilot trial

A small trial found a GLP-1 drug helped smokers quit. A secondary analysis then found five different groups it helped most — which is what exploratory subgroup analysis reliably produces.

Ruth Alvarez5 min read
Posterior probability of benefit, by subgroupsmoked >20 a day81.7%no prediabetes76%no obesity94.4%no depression symptoms91.2%one genotype88.6%⛔ Posterior probabilities, from a secondary analysis of a pilot trial.

This page is not really about smoking. It is about what happens when a small positive trial is asked who it worked for, because the answer is almost always “several different groups” and that answer is almost always less informative than it looks — a cousin of the problem in what happens outside a trial.

What the original trial found

In a pilot trial, exenatide added to a nicotine patch improved smoking abstinence compared with the patch alone [1]. That is the primary result, and it is the one with a design built to support it.

What the secondary analysis found

Exploring baseline characteristics, the analysis reported stronger benefit among participants smoking more than twenty cigarettes a day (posterior probability 81.7%), among those without prediabetes (76.0%), among those without obesity (94.4%), among those with no or minimal depression symptoms (91.2%), and among carriers of one genotype at a nicotinic receptor gene (88.6%).

Read quickly, that is five findings. Read carefully, it is five ways of dividing one small sample, each producing a smaller group than the trial was sized for.

Why this pattern matters beyond one trial

Because subgroup findings are the raw material of marketing. “Works best for people who...” is a persuasive sentence, and it can be sourced to a real analysis of a real trial while resting on a group of a few dozen people identified after the fact.

The same caution applies to every claim about who these drugs suit. When a result is presented as applying especially to some readers, the question is whether that group was named before the data arrived or after — which is a version of the reading discipline in what a composite endpoint hides and what a comparator can and cannot tell you.

What is actually established

That a GLP-1 drug added to a nicotine patch improved abstinence in one pilot trial. That is a genuine and interesting result, and it is the only one here with a design behind it. Everything after it is a direction for the next study.

None of this is available to buy for smoking cessation, and no seller tracked on this site prescribes for it — the same distance between a trial result and a purchasable product that runs through what the semaglutide trials are worth.

Frequently asked

Does a GLP-1 drug help people stop smoking?
In one pilot trial, adding exenatide to a nicotine patch improved abstinence compared with the patch alone. That is the primary result and the only one with a design built to support it.
Who did it work best for?
The secondary analysis reported five different answers — heavier smokers, people without prediabetes, without obesity, without depression symptoms, and carriers of one genotype. Five ways of dividing one small sample is not five findings.
Are those percentages significance levels?
No. They are posterior probabilities — statements of belief under a model — and describing them as statistical significance would be a category error.
Can I buy this for smoking cessation?
No. It is not an approved use and no seller tracked on this site prescribes for it.

Sources

  1. [1] Yammine L, et al. (2025). Exploring Predictors of Treatment Response to GLP-1 Receptor Agonists for Smoking Cessation Nicotine & Tobacco Research. PMID 39780397

Where to get it

Best GLP-1 injections

Every injectable seller we can verify, with the price each one publishes and an honest read of what the trials measured.

Compare providers →

More in Evidence