Gut and Liver published an analysis on 2 September 2026 asking whether the reflux signal around GLP-1 receptor agonists holds up outside one country’s reporting system.
The authors ran the same disproportionality analysis in the United States, Japan and Canada. All three produced a signal in the same direction.
One compound in the class went the other way, and that detail is the most informative part of the paper.
A framing note. GLP-1 receptor agonists are approved medicines and this is an analysis of spontaneous adverse event reports. Nothing below is guidance about human use.
What a GLP-1 disproportionality analysis measures
Spontaneous reporting databases collect adverse events that somebody chose to report. They do not have a denominator, so they cannot produce an incidence rate.
What they can do is compare proportions. If reflux makes up a larger share of the reports filed about one drug than of the reports filed about a comparator, that shows up as a raised reporting odds ratio. It is a signal that something may be worth investigating, not a measurement of how often the event happens.
The comparator choice is what separates a careful analysis from a careless one, and the authors used dipeptidyl peptidase-4 inhibitors as an active comparator rather than comparing against every other drug in the database. That matters, because DPP-4 inhibitors are given for overlapping reasons to a broadly similar population.
Events were identified using MedDRA preferred terms, and the authors report calculating reporting odds ratios with 95 percent confidence intervals, with subgroup analyses by drug, age and sex.
The GLP-1 numbers across three systems
Among 260,417 GLP-1 reports, the authors report significant signals in every database.
| Reporting system | Reporting odds ratio | Counts |
|---|---|---|
| FAERS, United States | 2.83, CI 2.39 to 3.35 | 9,245 versus 139 |
| JADER, Japan | 4.76, CI 2.45 to 9.27 | 19 versus 16 |
| Canada Vigilance | 2.91, CI 1.97 to 4.31 | 206 versus 29 |
All signals met both the reporting odds ratio and the proportional reporting ratio detection criteria, which are two separate thresholds rather than one applied twice.
The Japanese counts deserve a note. Nineteen events against sixteen is a very small base, and the confidence interval reflects that honestly, running from 2.45 all the way to 9.27. A ratio calculated from nineteen events is compatible with a modest effect and with a very large one. The point estimate being the highest of the three is not evidence that the signal is strongest in Japan.
Which GLP-1 compounds carried the signal
The drug-level breakdown is where the GLP-1 class stops behaving like a class.
| Compound | Range across databases |
|---|---|
| Semaglutide | 3.44 to 6.93 |
| Liraglutide | 2.36 to 5.90 |
| Tirzepatide | 2.76 to 5.77 |
| Dulaglutide | 2.35 to 3.32 |
| Exenatide | 0.60 in FAERS, CI 0.46 to 0.77 |
Exenatide showed an inverse association. Its interval sits entirely below 1, which in this framework means reflux made up a smaller share of exenatide reports than of the comparator’s.
An inverse finding inside the same GLP-1 receptor class is the sort of result that should slow anyone down before writing that GLP-1 receptor agonists cause reflux. Either the mechanism is not shared across the class, or something about how exenatide is used, reported or prescribed differs enough to reverse the arithmetic. A reporting database cannot distinguish those two, and the authors do not claim to.
Age-stratified analysis showed the signal strengthening with age, with a reporting odds ratio of 3.01 in those aged 65 and over and a trend p value of 0.028.
What the GLP-1 result adds and what it does not
The authors describe this as the first multi-database pharmacovigilance analysis of the question, confirming consistent signals across Western and Asian populations with approximately two to five-fold higher reporting.
Cross-national replication is genuinely worth something. A signal that appears only in FAERS could reflect American reporting culture, litigation environment or prescribing patterns. A signal that appears in three systems with different reporting cultures is harder to explain that way.
What it still cannot do is establish incidence or causation. Every constraint of spontaneous reporting applies: reporting is voluntary, publicity drives reporting volume, and a widely discussed medicine accumulates reports for reasons unrelated to biology. Slowed gastric emptying provides a plausible mechanism for reflux, which makes the finding easier to believe and does not make it measured.
The authors’ closing recommendation is directed at clinicians monitoring patients, which is outside what we cover and outside what a research supplier should comment on.
Where it sits among the other GLP-1 signals
This is the third pharmacovigilance study we have covered in five weeks, and the three of them together make a better point than any one does alone.
A FAERS analysis of cardiac reporting found a disproportionate signal and the authors attributed it to demographics and indication rather than toxicity. Six years of European reports on eye and psychiatric disorders produced a reporting odds ratio of 2.67 for eye disorders against liraglutide. Now three national systems agree on reflux, with one class member pointing the other way.
Same method, three organ systems, three different verdicts about what the signal means. That is what a method looks like when it detects rather than decides. It also connects to ten meta-analyses resting on one shared set of trials we covered last week: agreement between sources is only informative when the sources are independent, and here they genuinely are.
What this has to do with the catalogue
Nothing in this GLP-1 analysis involves a compound supplied for laboratory research, and the useful part is structural.
Tirzepatide appears here with its own range, 2.76 to 5.77, because enough of it has been prescribed for reports to accumulate. Retatrutide does not appear at all, and cannot, because a compound with no market generates no spontaneous reports.
That asymmetry is worth naming every time it comes up. The absence of a pharmacovigilance signal for a newer compound is an absence of exposure, not a finding about safety, and the two look identical in a database.
For laboratories, our European research buyer guide covers procurement and documentation, everything is supplied on a research use only basis, and our ranking of the most-studied research peptides counts primary studies rather than reports.
References
- Choi JG, Gong EJ, Bang CS, Lee JJ. Multi-Database Pharmacovigilance Analysis of Gastroesophageal Reflux Disease Associated with GLP-1 Receptor Agonists: A Cross-National Signal Validation. Gut and Liver, 2 September 2026. doi 10.5009/gnl260189
- Singla A, Kaur G, Singh G, et al. Comparative Cardiovascular Pharmacovigilance of Semaglutide and Tirzepatide: A FAERS Database Analysis (2022-2026). Cureus, 25 July 2026
- Ammendolia I, Mondello C, Esposito E, et al. Psychiatric and eye disorders associated with use of glucagon-like peptide 1 receptor agonists (GLP-1 RAs). Naunyn-Schmiedeberg’s Archives of Pharmacology, 22 August 2026
- Ramzy D, Isshak R, Patel D, et al. Glucagon-Like Peptide-1 Receptor Agonists, Semaglutide, and Atrial Fibrillation: An Umbrella Review. Cureus, 30 July 2026, 18(7):e113641
The products referenced on this site are supplied for laboratory research use only. They are not medicines and are not intended for human or veterinary use. This article summarises published research for informational purposes and is not medical advice. Statements about third-party studies belong to their authors.
