Risk for stroke after newly diagnosed atrial fibrillation during hospitalization for other primary diagnoses: a retrospective cohort study

Background — Atrial fibrillation (AF) that is first diagnosed during hospitalization for other causes can subside with resolution of the inciting stressor.

Objective — To describe the risk for stroke after newly diagnosed AF during hospitalization for other causes.

Design — Population-based retrospective cohort study.

Setting — Ontario, Canada.

Participants — Patients aged 66 years or older discharged alive from the hospital between April 2013 and March 2023 with a first diagnosis of AF.

Intervention — Newly diagnosed AF during hospitalization for other causes, categorized into cardiac medical, noncardiac medical, cardiac surgical, and noncardiac surgical.

Measurements — The primary outcome was hospitalization for stroke. The cumulative incidence function was used to estimate crude incidence, censoring on anticoagulant dispensation. Inverse probability of censoring weights were used to account for informative censoring.

Results — Atrial fibrillation was diagnosed in 20 639 patients (mean age, 77.1 years; 58.1% male) while hospitalized for other causes: 8340 (40.4%) for noncardiac medical, 7097 (34.4%) for cardiac surgical, 3553 (17.2%) for noncardiac surgical, and 1649 (8.0%) for cardiac medical diagnoses. At 1 year, anticoagulants were being dispensed to 26.4% of patients with CHA2DS2-VA scores of 1 to 4 and 35.2% of those with CHA2DS2-VA scores of 5 to 8. The 1-year risk for stroke without anticoagulation was 1.3% (95% CI, 0.7% to 2.3%) for cardiac medical, 1.2% (CI, 0.9% to 1.5%) for noncardiac medical, 1.1% (CI, 0.8% to 1.7%) for noncardiac surgical, and 1.0% (CI, 0.7% to 1.3%) for cardiac surgical patients. Patients with CHA2DS2-VA scores of 1 to 4 had a 1-year stroke risk of 0.7% (CI, 0.6% to 1.0%) without anticoagulation, compared with 1.8% (CI, 1.4% to 2.2%) at CHA2DS2-VA scores of 5 to 8.

Limitation — Long-standing AF may have been misclassified as newly diagnosed, leading to overestimation of stroke risk.

Conclusion — Among patients with newly diagnosed AF during hospitalization for other causes, a substantial proportion with low CHA2DS2-VA scores receive anticoagulation, with modest increases in this proportion at higher scores. The stroke risk in patients with CHA2DS2-VA scores greater than 4 approximated the 2% threshold commonly used to initiate anticoagulation in AF.