Among 216,293 adults with type 2 diabetes, starting an SGLT2 inhibitor was associated with a lower five-year risk of atrial fibrillation than starting a GLP-1 receptor agonist: 4.3% against 4.8%. Starting both classes together came in at 4.2%, statistically indistinguishable from SGLT2 alone. For anyone weighing the two classes with an eye on heart rhythm specifically, this is a point in favour of SGLT2 inhibitors — and only on that one endpoint. It was published on 6 September in the European Journal of Preventive Cardiology.
How the comparison was built
The authors emulated a three-arm target trial using population-based healthcare data covering 2015 to 2025. Participants were assigned to one of three initial strategies — an SGLT2 inhibitor without a GLP-1 drug, a GLP-1 drug without an SGLT2 inhibitor, or early combined therapy — according to what they were dispensed during a 30-day assignment window. Follow-up began at the end of that window under a landmark design, groups were balanced with inverse probability of treatment weighting, and cumulative incidence was estimated with weighted Aalen-Johansen methods that account for death as a competing risk.
The arms were 114,572 on an SGLT2 inhibitor alone, 91,524 on a GLP-1 drug alone, and 10,197 on early combined therapy. At five years, cumulative incidence of atrial fibrillation or atrial flutter was 4.3%, 4.8% and 4.2% respectively. SGLT2 alone carried a lower risk than a GLP-1 drug alone, with a risk ratio of 0.88 (95% CI 0.83–0.94). Half a percentage point over five years is about one additional atrial fibrillation diagnosis for every 200 people — a real difference, and a small one.
The combined arm was statistically identical to SGLT2 alone, RR 1.00 (95% CI 0.84–1.19). That interval is wide because the combined arm holds only 10,197 people, under a twentieth of the cohort. It is consistent with a 16% reduction and with a 19% increase, so it establishes no advantage to adding a GLP-1 drug on this endpoint, and rules out very little.
What it does and does not say about GLP-1 drugs
Atrial fibrillation is one endpoint, not the cardiovascular picture. GLP-1 drugs carry approved cardiovascular indications of their own, and nobody chooses between these two classes on rhythm alone — heart failure, kidney function, weight and cost all sit on the same decision. Read this as one more input, weighted about as heavily as a half-point absolute difference deserves.
What is not yet known. This is observational. Weighting is not randomisation, and confounding by indication is the central difficulty: SGLT2 inhibitors are preferentially prescribed to people with heart failure and chronic kidney disease, GLP-1 drugs more often for weight, and those populations start with different atrial fibrillation risk and receive different amounts of cardiac monitoring — which itself governs how much atrial fibrillation gets found rather than how much exists. The abstract does not name the country or the database; the authors are based at Soroka University Medical Center and Ben-Gurion University of the Negev in Israel. It does not report which specific GLP-1 agents were dispensed, and the 2015–2025 window spans a decade in which GLP-1 prescribing changed enormously, so the average GLP-1 patient in 2016 and in 2024 are not the same person. No funding was declared.
Sources: Ostrovsky D, Alnsarsra H, Shamia D, Westreich R, Star A, Tsaban G. “Sodium-Glucose Cotransporter-2 Inhibitors, Glucagon-Like Peptide-1 Receptor Agonists, and Incident Atrial Fibrillation in Type 2 Diabetes: A Population-Based Target Trial Emulation.” European Journal of Preventive Cardiology, published online 6 September 2026. PubMed 42700963
