A new meta-analysis pooled 28 studies and 212,082 patients to test whether holding GLP-1 receptor agonists before an upper endoscopy ↗ actually changes the procedure. The odds of arriving with food still in the stomach were nearly four times higher in patients on the drugs (odds ratio 3.95, with 95 percent confidence between 2.69 and 5.82). Procedure cancellations rose by about the same multiple (OR 3.96, CI 2.87 to 5.56). Aspiration, the outcome the entire hold-the-drug guidance debate has been about, did not change (OR 1.09, CI 0.46 to 2.60).
GLP-1 drugs slow gastric emptying. That is one of the mechanisms by which they reduce appetite and blunt postprandial glucose. It is also why the American Society of Anesthesiologists issued an advisory in 2023 and the American Gastroenterological Association issued a clinical practice update the same year, both telling proceduralists to hold the drugs before sedation. The fear was that food retained past the usual fasting window would track up the esophagus during anesthesia and end up in the lungs. Three years and a hundred thousand patient-encounters of follow-up data later, the meta-analysis says the fear is real on imaging and real on the schedule, and not yet visible in the actual harm event.
The outcomes split cleanly. Retained gastric contents and procedure cancellations were the consistent signals, both at p less than 0.0001. Retained contents had substantial heterogeneity across studies (I-squared of 82.7 percent), which is what one expects when each center applies its own definition of "retained" and its own fasting window. The cancellation finding was tight (I-squared of 4.2 percent), meaning every center is taking the scheduling hit. Aspiration was null and the bound on its odds ratio crossed 1, which is also what one expects when the event is rare and centers vary in how aggressively they intervene.
That null on aspiration is the part worth thinking about. It does not necessarily mean the drugs are safe to scope through. It can equally mean the cancellations are doing the prevention work. A procedure that ends with the patient told to come back next week is not a procedure where anyone breathed stomach contents. The headline reading, "GLP-1 RAs do not raise aspiration risk," is true on the data. It also assumes the current cancellation rate is the price of safety, which is exactly what an unconstrained analysis on real-world cohorts cannot disentangle. The authors say so in the conclusion: further randomized trials are necessary to separate the cause from its workaround.
The clinical pattern matches what last week's tirzepatide cutaneous allodynia case series ↗ hinted at from a different angle. As GLP-1 prescription volumes widen past clinical trial populations, the side-effect surface widens with them, and most of what surfaces first is logistical before it is biological. The endoscopy paper documents one of the logistical edges in unusual detail because procedures generate paperwork the way clinic visits do not.
For the receptor pipeline, the timing of the next test is mechanical. Semaglutide ↗ and tirzepatide ↗ are weekly drugs with multi-day pharmacokinetic tails; a Wednesday procedure on a Sunday-dosed patient is still a drug-on-board procedure. Daily oral agents, the orforglipron and oral semaglutide class, will be the natural test of whether shorter pharmacokinetic windows cut the cancellation tax without reopening the aspiration question. The meta-analysis runs 2015 to 2023, so it captured almost no oral GLP-1 dosing; the next read on that question lies in cohorts the data here did not see.
The piece on the prescriber side has a smaller question. Hold the drug for the standard window, accept the cancellations as the safety valve, and wait for the trials the authors asked for. Three years after the ASA advisory, the safety event the advisory was about has not moved. That is news, even if it is not yet a license to scope through a full stomach.