A panel of 45 clinicians and patients has set the first structured rules for prescribing GLP-1 weight-loss drugs to people who have, or are at risk of, an eating disorder. The short version: screen everyone before the first dose, and generally do not give these drugs to someone in the grip of anorexia or bulimia.
The recommendations, published in World Psychiatry ↗, try to hold two facts at once. GLP-1 receptor agonists, the class that includes semaglutide ↗ (sold as Ozempic and Wegovy) and tirzepatide ↗ (Mounjaro and Zepbound), blunt appetite so effectively that they cut binge-eating episodes in early studies. That same appetite suppression is a hazard for someone prone to eating too little. A drug that quiets the urge to binge can, in the wrong person, feed the urge to restrict.
An eating disorder is not one thing. Binge-eating disorder, where people lose control and eat large amounts, is the one these drugs might actually help. Anorexia nervosa and bulimia nervosa, marked by severe restriction, purging, or a terror of gaining weight, are the ones the panel worries the drugs could worsen or unmask. Weight loss is the goal for an obesity patient and a symptom for an anorexia patient, and the same prescription reads very differently across that line.
How the lines were drawn
The guidance came out of a modified three-round Delphi process, a structured way of turning scattered expert opinion into agreed statements by voting, feeding the results back to the group, and voting again. The panel, led by Aaron Keshen at Dalhousie University in Halifax, mixed diabetes and obesity specialists, eating-disorder researchers including Cynthia Bulik of the University of North Carolina and Karolinska Institutet and B. Timothy Walsh of Columbia University, and people with lived experience of an eating disorder or of taking the drugs. A statement needed at least 70 percent agreement to survive. Average agreement across the final recommendations was 91.7 percent, so most of what shipped was close to unanimous.
Four clinical recommendations made the cut. Screen every patient for a current or past eating disorder before starting a GLP-1 drug, using a short validated questionnaire rather than a hunch. Monitor for eating-disorder symptoms during treatment, more closely for higher-risk patients. Build standardized education for both patients and prescribers. And offer real psychotherapy alongside the drug to anyone who already has an eating disorder, rather than treating the weight in isolation.
The firmest line is around active illness. These drugs should generally be avoided in people currently living with anorexia nervosa, atypical anorexia, or bulimia with heavy dietary restraint or significant weight suppression, and used cautiously in anyone with that history. For the patient in front of a prescriber, that is a concrete instruction: ask first, and in the clearest cases, do not prescribe.
What the document is, and is not
What it is not is evidence. A Delphi consensus is expert judgment, formalized. The panel is blunt that the underlying data "remains limited and inconsistent," which is exactly why it reached for structured opinion in the first place. High agreement among 45 experts is not the same as a trial showing that screening prevents harm, and none of these recommendations has been tested against a real-world outcome. They are a floor to stand on while the studies get done.
The same paper sketches what those studies should be: long-term tracking of eating-disorder rates in GLP-1 users versus non-users, regulatory work on safer prescribing, and trials of the drugs specifically in binge-eating disorder, where the early signal of benefit lives. Until those exist, prescribing sits on judgment, which is why the panel wanted the judgment written down and shared.
The tension it is trying to manage showed up plainly in the trial record. A meta-analysis of 25 trials ↗ found GLP-1 drugs cut binge eating, loss-of-control eating, and emotional eating, but pushed deliberate dietary restraint up, and no one could say whether that added restraint was healthy discipline or the early shape of a problem. This consensus is the clinical answer to that open question. Assume it could be a problem, and check before you prescribe.