
[Adobe stock]
“The problem when you take people off a diabetes medicine a week before surgery is that they can have an elevated glucose level at the time of surgery,” said Klonoff in an interview. “We always tell people with diabetes not to change medicines or anything they’re doing just before surgery.”
Rethinking the GLP-1 guidelines for patients undergoing surgery
In a study involving 13,361 adults with diabetes, 16.5% of whom were treated with GLP1-RAs, researchers found that GLP1-RA users had a lower risk of peri- and post-operative complications related to decelerated gastric emptying and antiemetic use compared to non-users. The study also revealed no significant differences in the risk of ileus, aspiration/pneumonitis, hypoglycemia and 30-day mortality between GLP1-RA users and non-users. A sensitivity analysis conducted on patients with both diabetes and obesity yielded similar findings.
In its guidance, ASA noted that “GLP-1 agonists are associated with adverse gastrointestinal effects such as nausea, vomiting and delayed gastric emptying.” The guidance also suggested that “given the concerns of GLP-1 agonists-induced delayed gastric emptying and associated high risk of regurgitation and aspiration of gastric contents,” patients on daily dosing should consider holding the medication on the day of surgery, while those on weekly dosing should stop the drug one week prior, regardless of the indication, dose, or procedure type. Yet the ASA acknowledged that “the evidence to provide guidance for preoperative management of these drugs to prevent regurgitation and pulmonary aspiration of gastric contents is sparse.” The guidelines noted that its conclusions were “limited only to several case reports.”

Dr. David Klonoff
“Usually, when a recommendation is made, there’s a reason for it,” Klonoff noted. “In this case, they weren’t using real-world evidence. It wasn’t evidence-based, it was strictly expert opinion.”
Grounding guidelines in evidence
To illustrate his concerns, Klonoff offered an analogy, suggesting that the ASA’s recommendation was akin to advising the entire country to stop flying based on the opinion of one expert who believed that flying would lead to severe complications. “That would be absolutely crazy. But that’s what we’re dealing with, something analogous,” he explained, highlighting the absurdity of making broad recommendations without evidence.
The real-world impact of the ASA’s guidelines has been significant, leading to increased surgical cancellations and economic consequences for health systems. As Klonoff explained, “A person comes in for surgery and it turns out that they were taking a GLP-1 drug and the hospital says, ‘Well, you can’t have your surgery today. Go home and reschedule.'” This highlights the impact of the ASA’s guidelines leading to last-minute surgical cancellations, causing significant inconvenience for patients and logistical issues for healthcare providers.
Moreover, these cancellations have ripple effects on hospital operations and finances. “Meanwhile, the hospital wants to keep their operating rooms running all day long,” Klonoff noted. “So all of a sudden, there’s a hole in the calendar.”
The ASA did not respond to a request for comment regarding its guidelines.
Tapping AI for real-world evidence
In Klonoff’s retrospective observational cohort analysis, which was published by Diabetes, Obesity and Metabolism, the researchers started with a dataset including more than 130 million deidentified U.S. adults with diabetes and found no increased risk of perioperative complications among those taking GLP-1 receptor agonists compared to non-users.
To tap such a broad swath of patients, Klonoff collaborated with Atropos Health, a healthcare technology company specializing in generating real-world evidence using AI-based analytics. The company’s platform enabled the researchers to analyze a dataset of more than 130 million anonymous patient records. By applying rigorous inclusion criteria and techniques such as propensity score matching, Atropos helped whittle down the dataset to focus on approximately 13,000 patient records that fit the study criteria. In particular, the research compared the odds of six peri- and post-operative complications in GLP-1 receptor agonist users and non-users undergoing nine common surgical procedures requiring general endotracheal anesthesia. The two cohorts were matched using high-dimensionality propensity scoring to account for potential confounding factors.
Challenging the need to discontinue GLP-1 agonists
The analysis suggested that for patients with diabetes, discontinuing GLP-1 agonists before surgery may be unnecessary and potentially harmful. As the authors note, “The potential risks of withholding GLP1-RAs for 1-3 weeks prior to a procedure/surgery are unclear but could lead to deterioration of glucose tolerance.” The paper also noted that “GLP1-RAs can have glucose-lowering effects within the first day after injection, and withholding this important therapeutic agent could lead to perioperative hyperglycemia, which is known to increase risk for surgical site infections and urinary tract infections.”
“In fact, when we looked at six outcomes, for two of them, we actually found that the people on GLP-1s had statistically significantly fewer complications,” Klonoff said in an interview. “I wasn’t going to say GLP-1s prevent complications…but the evidence did not show that there’s an increased risk of complications.”
Toward precision medicine
The study’s findings not only challenge the guidelines but also highlight the potential of real-world evidence to inform personalized — and more precise —treatment decisions, Klonoff said. “I look at real-world evidence now as a new tool for reaching conclusions,” he explained. “Precision medicine means you’re using information about each person based on the electronic health record and as much information you can find.” He continued: “Evidence is more important than opinion these days.”
When projecting the impact of the analysis, Klonoff said: “I think that some hospitals are going to rescind it, others might not.”
Filed Under: clinical trials, Drug Discovery, Endocrinology, machine learning and AI, Metabolic disease/endicrinology, Regulatory affairs, Uncategorized