Adults who had both obesity and an autoimmune disease and were taking a GLP-1 drug had about 31 percent fewer pulmonary embolisms than near-identical patients who were not on one. A pulmonary embolism is a clot that lodges in the lungs, and it is exactly the kind of event this group is already primed for. The same patients had fewer strokes, fewer clots in the legs and veins, fewer emergency room visits, and were far less likely to die during the study window. What did not move was the heart attack.
That split is the story. The benefit landed on the clotting and stroke side of the ledger, not the coronary side, in a population that the large cardiovascular trials of these drugs mostly left out.
Who these patients are
People with autoimmune diseases like lupus, rheumatoid arthritis, and inflammatory bowel disease carry an elevated risk of clots and cardiovascular events on top of whatever their joints or gut are doing. Chronic inflammation makes blood more likely to clot, and some of these conditions, antiphospholipid syndrome above all, are defined by it. Add obesity, which independently raises the same risks, and you have a group sitting at the high end of the clot-and-stroke curve.
The landmark trials that earned GLP-1 drugs their cardiovascular labels, the studies in diabetes and in plain obesity, were not built around these patients. Whether the heart and clot benefits carried over to someone with active autoimmune disease was an open question. This is the first large dataset to take a direct look.
What the analysis found
The numbers come from a retrospective cohort published online June 6 in the Journal of the American Heart Association ↗. The authors used a target trial emulation, a method that arranges messy real-world records to imitate the structure of a randomized trial, on electronic health records from the OneFlorida+ network spanning 2014 to 2024. They matched 13,204 adults with obesity and an autoimmune disease who were on a GLP-1 drug against 13,204 who were not, one to one, balancing them on the variables that usually distort these comparisons. The group was middle-aged (mean 54.7 years), mostly women (73.4 percent), with an average body mass index of 37.
On the clotting axis the signal was clean. Pulmonary embolism dropped by roughly a third (hazard ratio 0.69, 95 percent confidence interval 0.56 to 0.86). Venous thromboembolism, the broader category of clots in the deep veins, fell about 17 percent (HR 0.83, 0.72 to 0.95). Stroke or transient ischemic attack came down about 13 percent, though the confidence interval nearly touched the no-effect line (HR 0.87, 0.76 to 0.99). Emergency room visits dropped about 21 percent (HR 0.79, 0.75 to 0.83), and all-cause mortality was cut nearly in half (HR 0.56, 0.47 to 0.66), meaning the GLP-1 group was about 44 percent less likely to die over the follow-up.
The two arterial outcomes told a different story. Myocardial infarction and coronary revascularization, the heart-attack-and-stent end of the cardiovascular panel, were prespecified primary outcomes, and the analysis did not report a significant drop in either. The protection clustered where these patients clot, not where they have heart attacks.
What it does and does not show
This is an association pulled from health records, not a randomized result, and the authors are explicit about that. Target trial emulation and time-dependent propensity matching narrow the gap between the two groups, but they cannot rule out that the people who got prescribed and stayed on a GLP-1 were healthier, more engaged, or better followed than those who did not. The mortality number in particular, a near halving, is the kind of effect size that observational cardiovascular data routinely overstate. Read this as a strong hypothesis, not a verdict.
The direction, though, matches what the platform has already seen elsewhere. When GLP-1 drugs were blamed for blood clots and the FDA's adverse-event database was checked ↗, the reports came in lower than the background rate would predict, not higher. Two very different methods, a disproportionality scan of a spontaneous-reporting system and a matched cohort, now point the same way on the clot question, and this one does it in a population built to clot. The class is best known through semaglutide ↗, sold as Ozempic and Wegovy, which acts on the GLP-1 receptor ↗.
For a rheumatologist deciding whether to add an antiobesity drug for a patient with lupus or rheumatoid arthritis, there has been almost no direct evidence to weigh. This does not settle it. It does say that the clot-prone group has the most to gain on the outcome they fear most, and that the next move is a trial designed around them rather than one that excluded them.