GLP-1 receptor agonist use tracked with more repeat colonoscopy after index procedures
A large real-world analysis links GLP-1 receptor agonist use with a higher likelihood of repeat colonoscopy within 12 months, adding new clinical pressure to a question that has moved quickly from pharmacology into endoscopy practice. The cohort came from the TriNetX Global Collaborative Network and included adults aged 18 to 90 years with an index procedure between January 2021 and January 2025. After 1:1 propensity-score matching across 44 covariates, 20,490 long-term GLP-1 receptor agonist users were compared with 20,490 non-users.
The primary endpoint was repeat colonoscopy within 12 months, and the association was not subtle. Relative risk reached 1.50 with a 95% confidence interval of 1.39 to 1.61. The signal appeared earlier as well, with similar associations at 6 months (RR 1.61, 95% CI 1.48–1.75), 3 months (RR 1.75, 95% CI 1.59–1.94), and 2 months (RR 1.86, 95% CI 1.66–2.08). In active-comparator analyses, the association remained present versus dipeptidyl peptidase-4 inhibitors and sodium-glucose co-transporter 2 inhibitors, with relative risks of 1.41 and 1.26, respectively.
A repeat colonoscopy is not a diagnosis. It is a utilization endpoint, and that makes interpretation inherently cautious.
Discontinuation analyses did not identify a clean pre-procedural cutoff
The most clinically tempting part of this paper is the exploratory discontinuation analysis, because it speaks directly to the practical question endoscopists ask before a procedure: how long, if at all, should a GLP-1 receptor agonist be held? Investigators examined discontinuation intervals from 0 to 8 weeks. The estimates trended downward with longer intervals, but overlapping confidence intervals and shrinking stratum sizes prevented any definitive threshold from emerging.
That result matters because it prevents a tidy translational leap from association to protocol. A patient who stops a drug for a few days, or even longer, may not necessarily move into a lower-risk category for repeat colonoscopy within this dataset. The observations are compatible with several biologic and operational explanations: delayed gastric emptying and slowed gastrointestinal transit may contribute to suboptimal preparation in some patients, but the endpoint here also captures rescheduling patterns, procedural deferrals, incomplete examinations, and follow-up behavior that may have little to do with bowel cleansing alone. In other words, the mechanism cannot be read directly from the endpoint.
The study’s own framing stays appropriately restrained. Repeat colonoscopy is a non-specific healthcare-utilization outcome, and these discontinuation analyses were observational and exploratory. No specific pre-procedural discontinuation interval can be justified from these data alone, even though the association is consistent enough to keep the issue on the procedural radar.
The biological plausibility is easy to understand, but the endpoint remains noisy.
Why delayed motility keeps drawing attention in endoscopy workflows
Interest in GLP-1 receptor agonists extends well beyond glucose control and weight reduction. Their known effects on gastrointestinal motility have raised concern that some patients may arrive for colonoscopy with retained stool burden or otherwise inadequate preparation, leading to repeat procedures. That concern is reasonable at the bedside because colonoscopy quality depends on a clean mucosal surface, and poor prep can reduce lesion detection, prolong procedure time, and force a second bowel purge and second sedation exposure. Still, the leap from slowed motility to repeat colonoscopy is only partial, and this study does not directly measure bowel-preparation quality.
Propensity matching across 44 covariates is a strength because it helps reduce the most obvious confounding in a retrospective dataset, particularly in a population where GLP-1 receptor agonist users may differ systematically from non-users in diabetes burden, obesity treatment patterns, and health-care engagement. The active-comparator analyses add another layer of reassurance, because comparisons against DPP-4 inhibitor and SGLT2 inhibitor users make it less likely that the observed association is simply a reflection of diabetes care intensity. Even so, residual confounding remains possible in any claims-based or network-based observational study, and the outcome still sits downstream of several decision points that are not visible in the dataset.
The practical question for gastroenterology is not whether GLP-1 receptor agonists matter at all; they probably do in some patients. The harder issue is identifying which patients, which procedures, and which medication schedules create enough risk to change management. This analysis supports heightened awareness, but it does not provide a universal stop rule. For now, the safest reading is narrower: clinicians should recognize an association with repeat colonoscopy, avoid treating that association as proof of inadequate bowel preparation, and resist using the exploratory discontinuation curves as a stand-alone guide for procedural timing.
Future work needs a different design. Prospective studies with directly measured bowel-preparation quality, standardized discontinuation schedules, and clear procedural endpoints would separate delayed transit from logistical and behavioral causes of repeat colonoscopy. Until then, the current evidence supports caution, not certainty, and it leaves the key bedside question unresolved: how much of the excess repeat-procedure risk reflects drug effect, and how much reflects everything surrounding the procedure itself.