Adults with type 2 diabetes who started an SGLT2 inhibitor had a lower five-year risk of atrial fibrillation than adults who started a GLP-1 drug, and adding a GLP-1 on top of an SGLT2 inhibitor did not lower that risk any further. That is the finding of a study of 216,293 people published September 6, 2026 in the European Journal of Preventive Cardiology, led by a team at Soroka University Medical Center and Ben Gurion University of the Negev in Israel.
Key takeaways
- Over five years, atrial fibrillation or flutter occurred in 4.3% of people who started an SGLT2 inhibitor and 4.8% of people who started a GLP-1 drug, a gap of about half a percentage point.
- Starting an SGLT2 inhibitor was linked to about 12% lower risk of atrial fibrillation than starting a GLP-1 drug (relative risk 0.88).
- Among people already on an SGLT2 inhibitor, adding a GLP-1 drug early was linked to no change in atrial fibrillation risk (relative risk 1.00).
What the study found
The researchers used population health data from 2015 to 2025 to build a target trial emulation, a design that takes real prescribing records and analyzes them as if people had been randomly assigned. They sorted 216,293 adults with type 2 diabetes into three groups by what they were dispensed in a 30-day window: 114,572 got an SGLT2 inhibitor without a GLP-1, 91,524 got a GLP-1 without an SGLT2 inhibitor, and 10,197 got both early.
At five years, cumulative atrial fibrillation or atrial flutter was 4.3% in the SGLT2 group, 4.8% in the GLP-1 group, and 4.2% in the combined group, according to the study.
The SGLT2 group had about 12% lower relative risk than the GLP-1 group (relative risk 0.88, 95% confidence interval 0.83 to 0.94). The true reduction is very likely between 6% and 17%, and unlikely to be zero. In absolute terms, the difference between those two five-year rates is about 0.5 percentage points, or roughly 5 fewer cases per 1,000 people over five years. Put another way, about 200 people would need to start an SGLT2 inhibitor rather than a GLP-1 for one of them to avoid an episode of atrial fibrillation over five years.
Adding a GLP-1 early to an SGLT2 inhibitor tracked with no difference at all (relative risk 1.00, 95% confidence interval 0.84 to 1.19). The range there is wide, running from 16% lower risk to 19% higher risk, so this study cannot detect any rhythm benefit from stacking the two. The combined group was also by far the smallest.
Dr. Kumar’s take
GLP-1 drugs are being discussed as a single fix for everything cardiometabolic. This is one of the largest head-to-head comparisons available, and it points the other way: the two classes are not interchangeable organ by organ. Rhythm protection looks like an SGLT2 property that a GLP-1 does not add to.
That fits what is already known about how SGLT2 inhibitors behave in heart failure and fluid handling. Less pressure and stretch on the left atrium is a plausible route to fewer arrhythmias, though this study was not designed to prove that route.
The restraint worth stating plainly is the size of the gap. Half a percentage point over five years, in observational data, is an argument about which drug to reach for first in someone with diabetes and rhythm concerns. It is not a reason for anyone to stop a GLP-1 that was prescribed for weight, kidney, or heart failure reasons. Those are different questions this study does not touch.
The design has limits. People are not randomly assigned to drugs in real life. The researchers balanced the groups statistically and accounted for deaths that compete with an atrial fibrillation diagnosis, which is the right approach, but the groups likely still differed in ways prescribing records do not capture. Atrial fibrillation also only gets diagnosed when someone looks for it.
Which drug goes first keeps turning out to matter more than how many drugs get stacked, which echoes what was seen when starting tirzepatide early beat stacking other diabetes drugs.
What it means for you
If you have type 2 diabetes and are choosing a first drug with your doctor, atrial fibrillation risk is one more thing worth putting on the table, particularly if you have a family history of it, sleep apnea, or existing heart disease.
If you are already on a GLP-1, this study is not a signal to stop. It compared starting choices, not stopping ones.
None of this replaces the basics that lower atrial fibrillation risk on their own: blood pressure control, alcohol moderation, treating sleep apnea, and staying active. Lifting weights lowers your risk of type 2 diabetes in the first place, which is the version of this problem nobody has to medicate.