Across three national drug-safety databases, acid reflux turned up in the reports about GLP-1 drugs two to five times more often than in the reports about an older diabetes pill. That is the headline from the first study to line up the reflux question across countries. The more interesting part is what the number can and cannot say.
A team at Hallym University College of Medicine in Chuncheon, South Korea pulled adverse-event reports from three systems that collect them: the United States FAERS, Japan's JADER, and Canada Vigilance. Published September 2 in Gut and Liver ↗, the analysis counted how often gastroesophageal reflux disease (GERD), the medical name for chronic heartburn, showed up in the pile of reports about GLP-1 receptor agonists compared with how often it showed up in reports about DPP-4 inhibitors, the older once-daily diabetes pills sold as Januvia and its cousins.
The reflux signal was there in all three. In FAERS it was 2.83 times higher than the comparator (9,245 reflux reports against 139 in the comparison group), in Japan's JADER 4.76 times, in Canada 2.91 times. Every one of those cleared the statistical bar the field uses to call something a signal. Broken out by drug, semaglutide ↗, the molecule in Ozempic and Wegovy, threw the strongest signal (roughly 3.4 to 6.9 times higher across the databases), followed by liraglutide ↗, tirzepatide ↗, and dulaglutide ↗. The signal also grew with age: in people 65 and older it ran about three times higher, with a statistically real trend across age bands.
Here is the catch, and it is the whole story. A reporting odds ratio is not a rate. Nobody in this study counted how many patients on these drugs got reflux. The number measures how often heartburn appears in the paperwork about one drug relative to another. Spontaneous-report databases are voluntary, uncontrolled, and shaped by who is paying attention. GLP-1 drugs are the most talked-about medicines on the planet right now, and prescriptions have surged. Drugs under a media spotlight generate more reports of everything, a well-documented effect called stimulated reporting. The comparator, a quieter older pill, sits under no such spotlight. So part of a two- to five-fold gap can be the attention, not the acid.
The mechanism, though, is not far-fetched. GLP-1 drugs slow the stomach down on purpose, which is part of how they blunt appetite. A stomach that empties more slowly holds its contents, including acid, longer, and that can push reflux the wrong way. So the signal is biologically plausible even if the size is inflated by reporting habits. Plausibility and proof are different things, and a disproportionality study delivers only the first.
One result cuts against a tidy class-wide story. Exenatide ↗, the oldest GLP-1 drug and the original twice-daily injection, ran the other way in FAERS, with reflux reported less often than the comparator, not more (0.60). If reflux were a clean, uniform property of the whole drug class, the founding member should not invert. The likelier explanation is mundane: exenatide is barely prescribed now, its reporting base is old and small, and small old piles of reports behave erratically. That is exactly the kind of noise that should make a reader hold the headline number loosely.
What this changes for a patient is modest and practical. It is not a reason to stop a drug that is doing its job. It is a reason for a doctor to ask about heartburn, especially in older patients, and to treat it if it shows up. And it is a flag for the study nobody has run yet. That would be a real cohort that measures how many patients actually develop reflux, at what rate, on which drug, against a proper comparison group. That study would settle what this one can only raise.
The drugs named here map to live entries on peptidemodel. Semaglutide, liraglutide, tirzepatide, dulaglutide, and exenatide all sit on the GLP-1 receptor ↗, the switch every one of them flips. The DPP-4 inhibitors used as the yardstick are small-molecule pills, not peptides, and have no card here. The signal belongs to the injections.