The published evidence says a hold matters and does not say how long yours should be. Continuing the drug to the day of an upper endoscopy left retained stomach contents in 5.83% of patients against 0.4% of non-users, an odds ratio of 10.68 (95% CI 3.74–30.55) [1]. Whoever performs the procedure sets the interval, and the numbers below are what they are weighing when they do.
Anyone booked for a scope or an operation gets asked how long to stop their GLP-1 first, and until recently the honest answer was that nobody had measured it. A retrospective study at one center compared three groups having elective upper endoscopy: people not on these drugs, people who kept taking them, and people who held for two weeks. The overnight fasting question underneath it is covered in still food in there after twelve hours, and the answer there is worse than the fasting instructions assume.
What continuing the drug did
The continued-use finding is unambiguous and large. Retained stomach contents — solid residue or more than 100 mL of fluid — appeared in 5.83% of those who kept taking the drug, against 0.4% of non-users. Six patients in 103, against 63 in 15,902. The motility effect behind it is the same one that makes these drugs work at all, which is why it cannot simply be engineered away. See the break-even against surgery for what that mechanism is worth paying for.
The two-week hold group came in at 1.61%, which the authors describe as statistically comparable to not using the drug at all. That is where care is needed.
The overnight fast does not empty the stomach
Fasting instructions before surgery assume the stomach empties, and a prospective cohort scanned 67 adolescents with gastric ultrasound to check [2]. Twenty of them were on a GLP-1, and solids were still present in 16 of those 20 after a median fast of 13 hours. Healthy controls came in at 1 of 20.
Twenty patients is a small denominator and nobody in that study was an adult, so the direction is what travels rather than the figure. It does establish that the fasting rule a scheduler reads off a form was written for stomachs that behave differently from these ones. That is the practical reason the anesthesiologist has to be told the drug and the date of the last dose.
How long a hold, and before what
The interval is not one number, because the procedures are not one procedure. Before aesthetic abdominoplasty, 30-day complications fell in a straight line [3]. They ran 45% on the drug to the day of surgery, 30% at a two-week stop and 10% at four weeks, against 10% in patients who had never taken it. Each of those groups held 20 people, so 45% is nine patients and 10% is two, and the gradient is built from single-digit counts.
A cardiac surgery review shows how thin this literature is elsewhere. Four randomized trials of perioperative liraglutide across 446 patients found no difference in 30-day mortality, complications or hypoglycemia [4]. The intervals are wide enough to answer nothing: the one for nausea and vomiting runs from 0.26 to 35.27. That trial set was testing whether to give the drug during surgery rather than when to stop it, and it could not settle either.
The finding that cuts against the headline
The result that should carry most weight for anyone reading this desk is the one that cuts the other way. Obesity, defined here as a BMI of 28 or above, independently raised the risk, and did so even after the full two-week hold — roughly tenfold against non-users. The population that buys these drugs is the population in whom holding them worked least well, which turns the headline on its head for exactly the readers most likely to act on it.
One incidental result is worth knowing if a procedure is being scheduled. Having a colonoscopy at the same sitting was strongly protective, and no retained-contents events occurred at all among those who had held for two weeks and were having both procedures together. The bowel preparation for a colonoscopy empties the stomach as well, which is a practical lever a gastroenterologist can pull.
What to do with all of this
None of it is a guideline, and a single retrospective series at one center should not be used as one. The decision about how long to hold belongs to whoever is doing the procedure, who needs to know the drug, the dose and how recently it was taken. That is a reason to keep a record of what was actually dispensed. It matters more when the seller and the doctor are not the same organization, as discussed in sellers publishing two prices and the dose people stay on.
A pause also costs money when it lands inside a plan somebody has already paid for, and that is a question a surgeon will not ask. The term and prepay check works out what a four-week gap does to a prepaid term, which is the part of this decision that belongs to the buyer rather than the clinic.