Diabetes patients taking GLP-1 drugs showed up for their colonoscopy with a dirtier colon. In a Japanese study, about a third of them (32.2 percent) had bowel preparation rated inadequate, versus one in six (16.7 percent) of patients not on the drugs. That is close to double the risk (relative risk 1.93, 95 percent confidence interval 1.12 to 3.35), and it matters because a colon that is not clean is a colon the endoscope cannot fully read.
The study appeared online August 24 in the Journal of Diabetes Investigation ↗. It is a single-center retrospective look at Japanese patients with diabetes who had a colonoscopy at Kameda Medical Center between October 2021 and September 2024. Out of 997 eligible patients, the authors built 90 matched pairs. They lined up drug users and non-users on age, sex, kidney disease, long-term blood sugar (HbA1c), and insulin use, so the comparison was not just sicker versus healthier. The bowel-prep grade came from the Aronchick scale, a standard four-step rating where anything below "good" counts as inadequate.
Why a clean colon is the whole point
A screening colonoscopy exists to find and remove polyps before they turn into cancer. The scope can only catch what the doctor can see, and a film of retained stool hides small or flat lesions. Inadequate prep is not a cosmetic complaint. It means missed adenomas, shorter intervals before the next scope, and sometimes a repeat procedure with a fresh round of laxatives.
GLP-1 receptor agonists slow the gut. That is part of how they curb appetite and blunt blood-sugar spikes, and it is the same mechanism behind the well-documented risk of food left in the stomach before an upper endoscopy. The colonoscopy problem is the downstream version of the same story. Bowel prep works by flushing a large volume of laxative solution through the intestine fast enough to sweep it clean. If the gut is moving in slow motion, the sweep stalls and the far end of the colon stays coated. The most common weekly GLP-1 drugs are semaglutide (Ozempic and Wegovy ↗) and dulaglutide (Trulicity ↗). The dual GIP and GLP-1 agonist tirzepatide (Mounjaro and Zepbound ↗) shares the same gut-slowing effect. So does the daily injectable liraglutide (Victoza and Saxenda ↗), on a shorter clock.
The direction is solid. The details are not.
One 90-pair study from a single hospital would not be worth much on its own. It is worth reading because it lands on the same answer as a much larger pool. A 2026 systematic review and meta-analysis in Clinical Endoscopy ↗ combined 12 studies and 123,858 patients, of whom 57,699 were on GLP-1 drugs. It found the same signal. After adjustment, the odds of inadequate prep were more than doubled (odds ratio 2.35, 95 percent confidence interval 2.02 to 2.74). The Japanese cohort adds an Asian population, where the authors note the evidence had been thin.
Where the new study reaches past what it can support is dosing frequency. It reports that weekly formulations fouled the prep more often than daily ones (44.2 versus 22.9 percent, relative risk 1.93). That is a tidy narrative, longer-acting drug means longer-lasting gut slowdown, but the ground under it is soft. The authors are candid that their additional adjusted analyses lost statistical significance, the matched sample is only 90 pairs, and it is one center. The larger meta-analysis, working from far more patients, did not find a clear difference between short-acting and long-acting agents at all. Treat the weekly-versus-daily claim as a hypothesis, not a finding.
The question every gastroenterologist actually wants answered is how long to hold the drug before the scope. Neither paper can say. The meta-analysis authors write plainly that skipping a single dose may not be enough to restore normal gut motility, and no study has pinned down the right washout interval. That is the practical gap: the risk is real and now replicated at scale, but the fix is still guesswork.
There is a small reassurance buried in the pooled data. The meta-analysis found no significant rise in repeat colonoscopies (odds ratio 1.5, but the confidence interval crossed 1). Inadequate prep on the day is common; a canceled-and-redone scope is not yet proven to follow at scale.
This is the mirror image of an earlier finding on the other end of the gut, where GLP-1 users left food in the stomach before an upper endoscopy and a liquid-diet prep erased the risk ↗. Upstream, better prep solved the slowdown. Downstream, the prep itself is what the slowdown defeats. The same slow gut did not raise bowel-obstruction risk in 298,000 diabetics ↗. It turns out to carry a quieter, more common cost, one that shows up not in the emergency room but on the prep-quality line of a screening report.