Withholding of oral anticoagulants in patients with dementia and atrial fibrillation by frailty status

Background: While oral anticoagulants (OAC) are effective in stroke prevention in patients with atrial fibrillation (AF), dementia and frailty are among the leading reasons for withholding OAC when it is indicated.

Objectives: To assess outcomes of withholding OACs in patients with dementia and AF by frailty status.

Methods: We used the US Medicare claims data (2013-2020) and Optum Clinformatics® Data Mart (2013-2023) to identify patients aged ≥65 years with non-valvular AF and dementia, who have a CHA2DS2-VASc score ≥2 (male) or ≥3 (female). We established 3 comparative cohorts of new users (no use in the preceding year) vs. non-users of OAC for apixaban, rivaroxaban, warfarin, respectively, using incidence density sampling (time zero was the dispensing date of OAC for the users and the sampling date for the non-users). We used 1:1 propensity score (PS) matching to balance 89 covariates. The outcome was the composite endpoint of hospitalization due to ischemic stroke, major bleeding, and mortality (“net clinical benefit”). We used Cox proportional hazard models to estimate hazard ratios (HR) and generalized linear models to estimate rate differences (RD) and their 95 % confidence intervals (CI), by frailty status determined by a validated frailty index.

Results: Our study cohorts included 36,093 PS-matched apixaban user vs. non-user pairs [mean age 83.5, female 65.0%], 13,578 matched rivaroxaban user vs. non-user pairs [mean age 82.8, female 64.7%]), and 15,826 matched warfarin user vs. non-user pairs [mean age 83.1, female 65.0%]. Compared to non-users of OAC, OAC use was associated with a reduced risk of the composite outcome for apixaban (HR [95% CI], 0.60 [0.47, 0.77]; RD per 1,000 person-years (PY), -171.66 [-233.12, -110.19), rivaroxaban (HR, 0.71 [0.68, 0.75]; RD per 1,000 PY, -107.87 [-124.01, -91.73]), and warfarin users (HR 0.66 [0.43, 1.02]; RD per 1,000 PY, -118.06 [-240.49, 4.37]). The absolute net clinical benefits associated with OAC use were did not differ significantly by frailty levels (e.g., RD per 1,000 PY was -143.84 [-156.33, -131.34] and -129.90 [-155.31, -104.49] in those with vs. without frailty among apixaban users, p for heterogeneity=0.1; a similar pattern was observed for rivaroxaban and warfarin users).

Conclusions: Based on two US national cohorts of older adults with AF and dementia, we found that use of apixaban, rivaroxaban, or warfarin was associated with a net reduction in a composite outcome of ischemic stroke, major bleeding, and death, compared to non-use of any OAC, regardless of frailty status, lending support for prescribing of OAC despite presence of dementia and frailty.