Nonvitamin K Antagonist Oral Anticoagulants Versus Warfarin in Atrial Fibrillation Patients and Risk of Dementia: A Nationwide Propensity‐Weighted Cohort Study

BACKGROUND: It is unclear whether nonvitamin K antagonist oral anticoagulants (NOACs) can mitigate dementia development in atrial fibrillation. We compared dementia development among users of NOACs or warfarin in patients with atrial fibrillation with no prior neurological diagnoses. METHODS AND RESULTS: We conducted a Danish nationwide cohort study including 33 617 new oral anticoagulant users with nonvalvular atrial fibrillation, of which 11 052 were aged 60 to 69 years, 13 237 were aged 70 to 79 years, and 9238 were aged 80 years and older. To exclude prevalent non‐oral anticoagulants–associated dementia, we considered the at‐risk population of patients alive and free of dementia at 180 days following inclusion. We compared rates of new‐onset dementia by age and treatment regimen using inverse probability of treatment weighting to account for confounding. Approximately 60% of patients were NOAC users and 40% were warfarin users. Mean follow‐up was 3.4 years. Dementia occurred in 41 patients aged 60 to 69 years, 276 patients aged 70 to 79 years, and 441 patients aged 80 years and older. Relative to warfarin users, dementia rates were nonsignificantly lower among NOAC users aged 60 to 69 years (0.11 events/100 person‐years versus 0.12 events/100 person‐years; weighted hazard ratio, 0.92 [95% CI, 0.48–1.72]) and NOAC users aged 70 to 79 years (0.64 events/100 person‐years versus 0.78 events/100 person‐years; weighted hazard ratio, 0.86 [95% CI, 0.68–1.09]), whereas NOACs were associated with significantly higher dementia rates (2.16 events/100 person‐years versus 1.70 events/100 person‐years; weighted hazard ratio, 1.31 [95% CI, 1.07–1.59]) in patients 80 years and older. CONCLUSIONS: This nationwide cohort of patients with atrial fibrillation revealed no clinically meaningful difference in dementia development between users of NOACs or warfarin apart from a higher risk in NOAC users 80 years and older, which may relate to residual confounding from selective prescribing and unobserved comorbidities.