Skip to content
This GLP
← Research
Safety

Reflux reports rose in three countries, and fell for one drug

The same signal appeared in American, Japanese and Canadian databases. Exenatide, in the same class, pointed the opposite way.

Glenn Torres6 min read
Reflux reporting ratio, vs DPP-4 drugsFAERS (US)9,245 vs 1392.83Canada Vigilance206 vs 292.91JADER (Japan)19 vs 164.76The largest ratio rests on the fewest reports.

Stomach complaints are the best-known side effect of these drugs, and reflux specifically has been harder to pin down than nausea. A cross-national study ran the same disproportionality analysis over the American, Japanese and Canadian adverse event databases, using DPP-4 inhibitors as an active comparator rather than the whole database [1]. The motility question underneath it is covered in the gastroparesis evidence.

All three databases produced a signal. FAERS gave a reporting odds ratio of 2.83 (95% CI 2.39–3.35), Canada Vigilance 2.91 (95% CI 1.97–4.31), and Japan's JADER 4.76 (95% CI 2.45–9.27). Semaglutide carried the strongest drug-specific signals across all three, and the signal grew with age, reaching 3.01 in people aged 65 and over — a gradient that matters more than it looks, because older readers are also the group carrying the most other medication, as the frailty analysis sets out.

The finding worth the most attention is the one that inverts. Exenatide, an older drug in the same class, showed a reporting odds ratio of 0.60 (95% CI 0.46–0.77) in FAERS — fewer reflux reports than the comparator, not more. Within-class disagreement of that kind is hard to explain by the usual objection to this method, because notoriety bias should lift every drug in a famous class together rather than push one below the line.

That objection still applies to the rest. Semaglutide is the most discussed medicine in the world at the moment, attention drives reporting, and the drug with the most attention has the strongest signal. This design cannot separate those two facts, and no amount of consistency across countries fixes it, since the same drug is famous everywhere.

What a reader can take from this is direction rather than magnitude. Reflux belongs on the list of things these drugs plausibly cause, it is not obviously a class effect given exenatide, and nothing here tells anyone how likely it is to happen to them. The trial-based pooled figures, which do have denominators, are the place to look for that, alongside the dose-escalation pattern described in titration and nausea and the broader safety picture in the cancer risk evidence.

Frequently asked

How likely am I to get reflux on a GLP-1?
This study cannot say. A reporting odds ratio compares how often a term appears among reports, with no count of how many people took the drug, so it produces no incidence or personal risk.
Why did exenatide show the opposite result?
It is unexplained, and it is the most interesting result in the paper. An inverse signal within the same drug class is difficult to attribute to reporting bias, which would tend to lift the whole class together.
Is the Japanese figure the strongest evidence?
It is the largest ratio and the weakest evidence. It rests on 19 reports against 16, which is why its confidence interval runs from 2.45 to 9.27.

Sources

  1. [1] Choi JG, Gong EJ, Bang CS, Lee JJ (2026). Multi-Database Pharmacovigilance Analysis of Gastroesophageal Reflux Disease Associated with GLP-1 Receptor Agonists: A Cross-National Signal Validation Study Gut and Liver. PMID 42682267

More in Safety