Adults with obesity who had weight-loss surgery were roughly 60 percent less likely to suffer a serious heart problem over the following years than adults who took semaglutide, the drug sold as Ozempic and Wegovy. The gap held whether or not the patients had diabetes, and it was widest for heart failure.
That is the finding of a retrospective comparison published August 11 in Endocrine Practice ↗, the journal of the American Association of Clinical Endocrinology. Researchers drew on TriNetX, a network of United States hospital records, and pulled adults with obesity who either started semaglutide ↗ or underwent one of the two most common operations, sleeve gastrectomy or gastric bypass, between January 2018 and January 2025. They matched the two groups separately by diabetes status so the arms looked alike on age, weight, and prior illness, leaving 11,466 pairs without diabetes and 7,327 pairs with it.
What the numbers said
The primary outcome bundled coronary events, stroke and other cerebrovascular events, heart failure, and cardiac arrest over as long as five years. Among patients without diabetes it occurred in 4.4 percent of the surgery group and 6.6 percent of the semaglutide group. That is a hazard ratio of 0.40, meaning the surgery patients ran about 40 percent of the drug patients' risk. Among patients with diabetes the same composite hit 9.0 percent after surgery and 14.6 percent on the drug (hazard ratio 0.42). Surgery came out ahead on coronary events, cerebrovascular events, and heart failure separately in both strata. Heart failure showed the largest separation, with the surgery groups carrying roughly a third of the drug groups' risk.
On its face the result is not a shock. Surgery removes far more weight than semaglutide does, often two to three times as much, and it does so durably. Less weight means less strain on the heart, better blood sugar, and lower blood pressure. If two treatments shed different amounts of fat, they should not be expected to move heart risk by the same amount.
Who chose surgery
The problem is that a records study cannot randomize who gets a scalpel and who gets a pen. Surgery is not handed out at random. A patient has to be referred, cleared for anesthesia, and committed enough to finish the workup. That filtering tends to leave the surgery arm healthier and more motivated in ways that matching on age and weight never fully captures. This is confounding by indication, and here it runs in the drug's disfavor.
The semaglutide arm carries the opposite baggage. Everyone who filled a prescription counts, including the large share of real-world patients who never reach the higher doses or quit within months. A Lilly claims analysis last week ↗ put persistence on this class of drug in the range of two thirds at six months. The people who stay on and escalate are a select slice of that. Averaged across everyone who ever picked up the drug, the real-world dose is lower and the exposure shorter than a trial would deliver.
So the honest reading is narrower than "surgery is 60 percent better for your heart." What the data support is that, as semaglutide is actually used in American clinics, it is not a like-for-like substitute for surgery on hard cardiac outcomes. The drug arm's real-world doses and dropout are baked into that hazard ratio. Whether a patient who took a full dose and stayed on it for five years would close the gap is a question this design cannot answer.
The comparison keeps landing the same way
Semaglutide works by activating the GLP-1 receptor ↗, the gut-hormone switch that curbs appetite and slows the stomach. It is a genuine advance over the drugs that came before it, and randomized trials have shown it cuts heart events against placebo. This study is a different question: drug against surgery, in the messy real world, with no coin flip deciding who got which.
The same ledger tension shows up on cost. A comparison of the two treatments over twelve years ↗ found the drug ran about 109 dollars more per month per patient. Both analyses capture something real, and neither can escape the fact that the people who choose surgery are not the people who choose a shot.
The authors frame their result as evidence to inform individual decisions, which is the right register. It gives a patient weighing an operation against an injection a reason to know that surgery has the stronger real-world heart record so far. It is not a reason to read a hazard ratio out of a claims database as if it came from a trial.