Background and purpose: The comparative effectiveness of direct-acting oral anticoagulants, compared with warfarin, for risks of stroke/systemic embolism, major bleeding, or death have not been studied in Medicare beneficiaries with atrial fibrillation and nondialysis-dependent chronic kidney disease.
Methods: Medicare data from 2011 to 2017 were used to identify patients with stages 3, 4, or 5 chronic kidney disease and new atrial fibrillation who received a new prescription for warfarin, apixaban, rivaroxaban, or dabigatran. We estimated marginal hazard ratios with 95% CIs for the association of each direct-acting oral anticoagulant, compared with warfarin, for the outcomes of interest using inverse-probability-of-treatment weighted Cox proportional hazards models in as-treated and intention-to-treat analyses.
Results: A total of 22 739 individuals met criteria (46.3% warfarin, 29.6% apixaban, 17.2% rivaroxaban, 6.9% dabigatran). Across the groups of anticoagulant users, mean age was 78.4 to 79.0 years; 50.3% to 51.4% were women, and 80.3% to 82.8% had stage 3 chronic kidney disease. In the as-treated analysis, for stroke/systemic embolism, hazard ratios, all compared with warfarin, were 0.70 (0.51-0.96) for apixaban, 0.80 (0.54-1.17) for rivaroxaban, and 1.15 (0.69-1.94) for dabigatran. For major bleeding, analogous hazard ratios were 0.47 (0.37-0.59) for apixaban, 1.05 (0.85-1.30) for rivaroxaban, and 0.95 (0.70-1.31) for dabigatran. There was no difference in the risk of all-cause mortality between the direct-acting oral anticoagulants and warfarin. Results of the intention-to-treat analysis were similar.
Conclusions: Apixaban, compared with warfarin, was associated with decreased risk of stroke/systemic embolism and major bleeding; risks for both outcomes with rivaroxaban and dabigatran did not differ from risks with warfarin.
Keywords: atrial fibrillation; dabigatran; edoxaban; rivaroxaban; warfarin.