Safety and Effectiveness of Direct Oral Anticoagulants vs Warfarin in People With Atrial Fibrillation and Dementia.

ABSTRACT

OBJECTIVE: To determine risks of embolic events, bleeding, and mortality with direct oral anticoagulants (DOACs) vs warfarin in people with atrial fibrillation (AF) and dementia. DESIGN: New-user retrospective cohort study using The Health Improvement Network database. SETTING AND PARTICIPANTS: A population-based sample comprising people with AF and dementia prescribed DOACs or warfarin from August 2011 to September 2017. METHODS: Risk of ischemic stroke (IS), ischemic stroke/transient ischemic attack/systemic embolism (IS/TIA/SE), all-cause mortality, intracranial bleeding (ICB), gastrointestinal bleeding (GIB), and other bleeding were compared for DOACs vs warfarin using propensity score-adjusted Poisson regression. Incidence rate ratios (IRRs) and absolute risk differences (ARDs) were calculated. RESULTS: Overall, 2399 people with AF and dementia initiated DOACs (42%) or warfarin (58%). Before propensity score adjustment, patients who initiated DOACs were older and had more comorbidities. After adjustment, DOAC initiators demonstrated similar risks of IS, TIA, or SE; IS alone; and other bleeding but reduced ICB risk (IRR 0.27, 95% CI 0.08, 0.86; ARD -5.2, 95% CI -6.5, -1.0, per 1000 person-years) compared with warfarin. Increased risk of GIB (IRR 2.11, 95% CI 1.30, 3.42; ARD 14.8, 95% CI 4.0, 32.4, per 1000 person-years) and all-cause mortality (IRR 2.06, 95% CI 1.60, 2.65; ARD 53.0, 95% CI 30.2, 82.8, per 1000 person-years) were observed in DOAC initiators compared with warfarin. CONCLUSIONS AND IMPLICATIONS Among people with AF and dementia, initiating treatment with DOACs compared with warfarin was associated with similar risks of IS, TIA, or SE and IS alone. DOAC-treated patients demonstrated reduced ICB risk but increased GIB and all-cause mortality risks. We cannot exclude the possible impact of residual confounding from channeling of DOACs toward older and sicker people, particularly for the outcome of all-cause mortality. Further safety data are urgently needed to confirm findings.