A drug that slows the gut, given to people whose disease is measured partly by how often they go to the bathroom, produced a very large improvement in that measure. The first question is whether the measure moved or the disease did — the same question every endpoint choice forces.
What was found
Adults with ulcerative colitis at one academic health system who started liraglutide or semaglutide for metabolic reasons, and stayed on it at least 12 weeks, were matched one to one against patients with the condition who did not. [1] Symptomatic remission was 34.7% against 15.3% at four weeks, 54.7% against 18.0% at eight, and 66.7% against 25.3% at twelve. Mean partial Mayo scores fell from 5.8 at baseline to 2.1 against 4.6 in controls. The adjusted odds ratio for remission was 5.90, 95% CI 3.52 to 9.86.
Semaglutide did better than liraglutide, 72.5% against 60%, odds ratio 1.77, p=0.04. Weight loss was not associated with remission.
The obvious objection
Partial Mayo is a symptom score. Two of its components are stool frequency and rectal bleeding, and these drugs slow gastric emptying and intestinal transit as their central mechanism. A person going to the bathroom less often scores lower on a scale that counts how often they go — without anything having changed in their colon.
What still deserves suspicion
An adjusted odds ratio of 5.90 is very large. Effect sizes of that size in matched observational data usually shrink when somebody randomizes, and the reasons are familiar: people prescribed a new drug for metabolic reasons differ from people not prescribed one in ways that records do not capture, including how closely their gastroenterology is being managed. That is the same limit every matched-records study runs into, and it does not get smaller because the result is interesting.
One hundred and fifty pairs at one tertiary center over two years is also a small, particular population. Everyone in it was well enough and metabolically unwell enough to be started on a GLP-1, which is a specific kind of patient.
The finding that cuts the other way
Weight loss was not associated with remission. If the benefit were simply the downstream effect of losing weight — less visceral fat, less systemic inflammation — you would expect the people who lost most to improve most. They did not.
That points toward something more direct, which is biologically plausible: GLP-1 receptors are present in the gut and these drugs have measurable anti-inflammatory effects. It is also the kind of mechanistic inference that a study without randomization can support only weakly, however sensible it sounds.
What this means for a reader
Nothing to buy. No seller on this roster offers anything for ulcerative colitis, none claims it, and nobody should start a GLP-1 to treat inflammatory bowel disease on the strength of a single matched cohort.
What it is worth is a conversation. Someone who has ulcerative colitis and is considering one of these drugs for weight has a reason to raise it with their gastroenterologist rather than an intake form — and inflammatory bowel disease is exactly the kind of history a telehealth questionnaire may never ask about.