Obese patients who took a GLP-1 shot and an SGLT2 pill together had fewer irregular heartbeats over five years than patients who took the GLP-1 shot alone. The drop was about 18 percent, and almost all of it was atrial fibrillation, the most common rhythm disorder and a leading cause of stroke.
That is the headline from a real-world analysis of 52,212 non-diabetic obese adults, published online July 14 in the International Journal of Cardiology Cardiovascular Risk Prevention ↗. A team led by George Blankson at Case Western Reserve University and University Hospitals Cleveland Medical Center pulled the records from TriNetX, a network that aggregates de-identified electronic health data from dozens of health systems. They matched 26,106 patients on the combination against 26,106 on GLP-1 alone, balancing the two groups on age, weight, blood pressure, and other risk factors so the comparison was closer to like-for-like.
What the two drugs are
A GLP-1 receptor agonist is the injected class that includes semaglutide ↗ and tirzepatide ↗, the drugs behind the current weight-loss wave. They work on the GLP-1 receptor ↗ to blunt appetite and slow the stomach. An SGLT2 inhibitor is a different animal: a daily pill, a small molecule rather than a peptide, that makes the kidney dump glucose into the urine. It was built for diabetes, then turned out to protect the heart and kidney on its own, so cardiologists now reach for it in people who never had high blood sugar.
Stacking the two is increasingly common, including in patients without diabetes who are taking both for weight and cardiovascular risk. Until now there was almost no data on what that combination does to heart rhythm specifically, as opposed to heart attacks and heart failure.
The numbers, and the catch
At the five-year mark, the combination group had lower odds of any arrhythmia (odds ratio 0.825, meaning about 17.5 percent lower). Atrial fibrillation and flutter drove the whole effect (odds ratio 0.821, about 17.9 percent lower). Ventricular fibrillation, the most dangerous rhythm, was no different between the groups at any point.
The catch sits in the ventricular tachycardia numbers. Early in follow-up, the combination group showed a higher rate of that rhythm, an odds ratio of 1.186. The confidence interval ran from 0.943 to 1.491, which crosses 1, so the signal was not statistically firm, and it faded as the years passed until the two groups looked the same by 60 months. The authors read this as an early, transient risk that resolves rather than a lasting hazard, but a wide interval that includes no effect is exactly the kind of result a randomized trial exists to settle.
Why the design matters
This is a retrospective look at records, not a trial. Nobody was randomly assigned to one regimen or the other. Doctors chose who got the second drug, and the reasons they chose are not fully captured in claims data, so some of the apparent benefit could be the healthier patients getting the combination rather than the combination making patients healthier. Propensity matching narrows that gap but never closes it. TriNetX also identifies both the drugs and the arrhythmias through diagnostic and prescription codes, which miss the paroxysmal atrial fibrillation that never got recorded.
What the study does well is scale and duration. Fifty thousand patients tracked out to eight years is a sample no trial will run for a rhythm question in this population, and the direction is consistent with the mechanism. SGLT2 inhibitors have a track record of cutting atrial fibrillation in heart-failure trials, and the leading theory is that draining fluid and easing pressure on the left atrium leaves it less prone to misfire. This analysis extends that signal to obese people without diabetes, on top of a GLP-1 drug, which is a group those earlier trials did not study.
It also lands next to a related finding from this section: a separate cohort where adding a GLP-1 drug to empagliflozin beat adding an older pill ↗ on broad cardiovascular risk. That piece was about major events like heart attacks. This one is about the electrical system, and the two together sketch a case that the pairing does more than either drug's label promises. Neither replaces the trial that would prove it.