Coded symptoms fell, which is not the same thing as bowels settling. In a matched analysis of 6,665 people per arm who already had irritable bowel syndrome, abdominal pain was coded in 31.6% against 35.8%, constipation in 19.8% against 22.0% and bloating in 8.3% against 10.9% [1]. Every one of those outcomes is a billing code entered at a visit, chronic diarrhea included at 8.9% against 10.6%, and the gaps run two to four percentage points. Nothing here was a symptom diary, and no randomized trial has tested a GLP-1 in irritable bowel syndrome.
Nausea, diarrhea, constipation and bloating are the defining side effects of this drug class, described from the mechanism side in nausea and appetite on one circuit. So a study finding fewer of those symptoms recorded in people who already have irritable bowel syndrome is worth reading carefully [1].
The direction is consistent across every outcome measured. In the 90-day cohort, chronic diarrhea was coded in 8.9% of the treated group against 10.6% of controls, constipation in 19.8% against 22.0%, abdominal pain in 31.6% against 35.8%, and bloating in 8.3% against 10.9%. Similar patterns appeared in the diarrhea-predominant and constipation-predominant subtypes.
The authors provide a clue that points the same way. They note that differences in outcome recurrence — how many separate episodes people had — were smaller than differences in incidence. If the drug were genuinely settling irritable bowels, you would expect ongoing symptom burden to fall at least as much as first occurrences. A gap concentrated in first coding is more consistent with a change in how symptoms are attributed than in how often they happen.
Two other guts, measured the same year
The same slowing mechanism has been read in two other settings and it lands differently in each. At the severe end, a systematic review searching four databases to October 2025 found twelve case reports covering thirteen patients with drug-associated gastroparesis worldwide [2]. At the other end, a chart review of 117 patients having elective posterior lumbar fusion found bowel function returning sooner in GLP-1 users, at 1.35 days against 1.90 (P = 0.013) [3]. A drug that slows gut transit producing an earlier return of bowel function is the opposite of the expected direction, and the reading of it is in the post-surgical cohort. None of the three findings contradicts the others; they describe one mechanism meeting three different guts.
The absolute differences are also modest: two to four percentage points across the outcomes. With roughly 6,665 people in each arm, small differences reach significance easily, and the P values say nothing about whether a patient would notice — the distinction drawn in the FLOW numbers needed to treat applies here too.
A real effect is nonetheless plausible. Slower gastric emptying could genuinely help diarrhea-predominant IBS, and weight loss reduces the systemic inflammation that some IBS phenotypes involve. What this design cannot do is separate that from a coding artifact, and a prospective study with symptom diaries rather than billing codes would settle it — the same gap running through the Crohn's disease cohort and the gastroparesis evidence.