A trial stopped early backs holding a GLP-1 dose before upper endoscopy

Drugs like semaglutide and tirzepatide slow the stomach down. That is part of how they curb appetite — and it is also why they have become a headache for the doctors who sedate patients for procedures. If food is still sitting in the stomach when a patient is sedated, it can be inhaled into the lungs. The practical question has been simple and unanswered: before an endoscopy, should a patient keep taking their GLP-1 drug or skip a dose? Until now, high-quality data to guide that call were lacking [s1].

OCULUS, published in JAMA Internal Medicine, is a randomised trial built to answer exactly that — and it was stopped early because the answer arrived faster than planned [s1].

What the trial did

OCULUS was a randomised, single-masked clinical trial run at two large tertiary referral centres in the United States [s1]. It enrolled adults undergoing elective upper endoscopy (EGD), with or without colonoscopy, under moderate sedation or monitored anaesthesia care, who had been taking a stable dose of a GLP-1 or GLP-1/GIP agonist for at least a month [s1]. Patients with prior foregut surgery, achalasia, documented gastroparesis, retained contents on a previous endoscopy, gastric outlet obstruction, planned general anaesthesia, or recent opioid use were excluded [s1].

Participants were randomly assigned either to continue their medication or to hold one dose before the procedure [s1]. The main outcome was clinically significant residual gastric volume (RGV) — a composite of retained stomach contents that stop the examination, force early termination or intubation, or cause an aspiration event needing extra monitoring, unplanned treatment or admission [s1]. The trial is registered as NCT06533527 [s2].

What happened

The trial did not run to its planned size. At the preplanned interim analysis there were 60 patients — 32 holding one dose and 28 continuing their medication — with a median age of 62.5 years and 50.0% women [s1].

Clinically significant RGV occurred in 3.1% of the hold group versus 25.0% of the continue group — an absolute difference of 21.9 percentage points (90% CI 7.0% to 36.7%, P = .003) [s1]. That crossed the prespecified O'Brien-Fleming stopping boundary, and the trial was terminated early because the risk of continuing exceeded what the safety rule allowed [s1].

The effect was starkest among patients having endoscopy alone. In that EGD-only subgroup of 35 patients, clinically significant RGV occurred in 46.7% who continued the drug versus 5.0% who held it — a difference of 41.7 percentage points (90% CI 17.9% to 65.4%, P = .001) [s1].

The part that complicates the headline

There is a second finding that matters as much as the first. In the subgroup of 25 patients having endoscopy plus colonoscopy — who were already on clear liquids the day before their procedure — no patients had clinically significant RGV, in either group [s1].

In other words, the standard bowel-prep diet appears to have emptied the stomach enough that the GLP-1 question became moot [s1]. The trial's own conclusion draws both threads together: continuing the drug increased retained contents but did not increase other adverse events, and clear liquids the day before "may mitigate the risk of clinically significant RGV regardless of GLP-1/GIP use" [s1].

What the trial does and does not establish

The direction of the result is clear and clinically plausible, and it lines up with the drugs' known effect on gastric emptying [s1]. But the evidence base here is thin by design. This was a single-masked trial at two centres, halted at an interim analysis of just 60 patients, with subgroup findings resting on a few dozen people each [s1]. Early-stopped trials tend to overstate the size of an effect, even when its existence is real.

It is worth noting what OCULUS did not report as different: continuing the medication did not raise the rate of other adverse events in this small sample [s1]. And the study was investigator-led — the lead sponsor on the registry is the Cleveland Clinic, not a drug manufacturer [s2].

What to watch

Professional societies have issued competing guidance on periprocedure GLP-1 management, some advising a pause and some a more individualised approach. OCULUS is among the first randomised data to enter that debate, and its most durable message may be the least dramatic one: that a clear-liquid day before a procedure did the job on its own in the patients who had it [s1].

Larger trials, and data on whether holding a dose changes actual aspiration outcomes rather than a surrogate volume measure, are what would move this from a stopping-boundary signal to settled practice [s1].

This article describes trial results. It is not medical advice, and nothing here should be used to start, stop or change any medication or to plan a procedure.

Sources

• Holding vs Continuing GLP-1/GIP Agonists Before Upper Endoscopy: The OCULUS Randomized Clinical Trial, JAMA Internal Medicine, 16 March 2026
• OCULUS trial registration (NCT06533527), ClinicalTrials.gov, first posted 1 August 2024