Abstract

Aims Patients with device-detected atrial fibrillation (DDAF) have a lower stroke risk than those with ECG-diagnosed AF, requiring careful evaluation of oral anticoagulation benefits vs. its inherent bleeding risk. Methods and results An unmatched win ratio analysis was performed of the NOAH-AFNET 6 trial dataset, using components of the primary efficacy and safety outcomes of the trial. The primary analysis used this hierarchical order: (1) all-cause death, (2) stroke, (3) systemic or pulmonary embolism/myocardial infarction, and (4) major bleeding. Two additional analyses replaced all-cause death with cardiovascular death or included patient-reported outcomes. Win odds were calculated to account for undecided comparisons. Among 2534 patients 77 ± 7 years old, 947 (37%) women, median CHA2DS2-VA score 3 [interquartile range (IQR), 3–4], median follow-up 21 months (IQR, 10–38) 1 605 280 win ratio pairs were analyzed. The win ratio comparing edoxaban to no anticoagulation was 0.87 (95% CI: 0.68–1.10; P = 0.23). Most comparisons resulted in no clear winner (undecided pairs 84.9%). In the remaining comparisons, edoxaban won in 46% of the cases, placebo in 54%. Death and major bleeding were the most common events. The win odds was 0.98 (95% CI: 0.94–1.01; P = 0.23). Conclusions This hypothesis-generating win ratio analysis, integrating death, thrombotic events, and major bleeds with and without quality of life, did not find an advantage of anticoagulation with edoxaban over no anticoagulation in patients with DDAF. The most common events were death and major bleeding.