Cohort study associates GLP‑1 use with fewer new atrial-fibrillation diagnoses

What happened

A retrospective cohort study examined new atrial fibrillation among 12,812 patients who started a GLP-1 receptor agonist at one academic center from 2020 through 2023. Researchers matched them one to one with controls using known atrial-fibrillation risk factors. People with earlier atrial fibrillation were not the population of interest.

The team reviewed charts for new atrial-fibrillation diagnoses and recorded weight direction after six months of treatment. Its time-to-event analyses included methods that treated death as a competing event and methods intended to account for changes in treatment over time.

Compared with the matched controls, GLP-1 use was associated with a hazard ratio of 0.67 for atrial fibrillation and 0.34 for death. Both associations met the study's statistical threshold. A hazard ratio below 1 indicates a lower estimated event rate over follow-up, not the number of events prevented in an individual.

The atrial-fibrillation association appeared in each weight-change category, including among patients whose weight rose. In analyses of individual medicines, only semaglutide had a statistically clear association. The reported reduction became apparent after 24 months of use.

What it means

The study asks whether the rhythm association might involve more than weight reduction. Its results did not disappear when patients were grouped by the direction of their weight change, and the researchers used several adjusted time-to-event models. That makes the question suitable for further study in randomized trials and in other health systems.

For patients, the result is a research signal rather than a new treatment use. Atrial fibrillation can raise stroke and heart-failure risk, so prevention research matters, but care still relies on established evaluation, stroke-risk assessment, medicines, procedures, and risk-factor management.

What it does not mean

Researchers did not randomly assign treatment. People who receive a GLP-1 medicine can differ from controls in diabetes care, cardiovascular health, access, adherence, monitoring, and other measured or unmeasured ways. Propensity matching and advanced models reduce some bias but cannot prove that the medicine caused the lower rates.

The weight analysis used direction of weight change recorded after six months. It cannot rule out effects from the amount or timing of weight change, changes later in follow-up, or metabolic changes not captured by body weight. The single-center design also limits how confidently the result applies elsewhere.

The study does not establish a GLP-1 medicine as an antiarrhythmic drug, show that semaglutide is superior to every other agent, or replace anticoagulation and rhythm or rate treatment. No patient should start, continue, or stop a medicine solely to act on this observational result.