Dose specific effectiveness and safety of DOACs in patients with non-valvular atrial fibrillation: A Canadian retrospective cohort study

Background: Direct oral anticoagulants (DOACs) have been proven to be effective and safe for prevention of ischemic stroke and systemic embolism in patients with non-valvular atrial fibrillation (NVAF). However, suboptimal adherence, variable dosing and use in patient populations that otherwise would have been excluded from clinical trials may impact the efficacy and safety profile of DOACs in a routine care setting. We compared stroke, bleeding, and mortality rates on and off therapy for standard and low-dose DOACs (apixaban, rivaroxaban, dabigatran) versus warfarin in a Canadian cohort. We also assessed persistence of DOACs compared to warfarin. Methods: We conducted six 1-1 propensity-score matched retrospective cohort analyses using Quebec health administrative databases (2011-2017). NVAF patients (>= 18 years) covered by the public medication insurance plan entered the cohort on the first OAC dispensation date. We excluded those with OAC use in the previous year or stroke or bleeding diagnoses in the previous two years. Follow-up ended at death, March 2017 or end of medication coverage by the public plan. Time-dependent Cox regression was applied. Results: We evaluated 10,893 patients initiated on apixaban (7206 standard, 3687 low-dose), 10,190 on rivaroxaban (7396 standard, 2794 low-dose), 5884 on dabigatran (2756 standard, 3128 low-dose), and propensity score-matched warfarin users. Across standard-dose DOACs, compared to warfarin, stroke risks were similar; bleeding risks were lower with apixaban (hazard ratio 0.63; 95% confidence interval 0.52-0.77) and dabigatran (0.47; 0.35-0.64) but not rivaroxaban (0.93; 0.79-1.10); death risks were lower with all DOACs. For low-dose DOACs, rivaroxaban demonstrated higher stroke (1.79; 1.21-2.64) and bleeding risks (1.37; 1.09-1.73); other agents had stroke risks similar to warfarin and bleeding risks lower than warfarin; only low-dose dabigatran had lower death risk (0.59; 0.52-0.68). Treatment discontinuation was lower with DOACs versus warfarin with the exception of low-dose rivaroxaban. The risks of stroke were 2-4 folds higher during time off any OAC versus time on warfarin. The risks of death were higher, while the risks of bleeding were generally lower during times off any OAC. Conclusions: Standard-dose DOACs had similar stroke, better persistence and mortality profiles than warfarin. Only standard dose apixaban and dabigatran had better bleeding profiles than warfarin. Low-dose rivaroxaban had worse persistence, stroke and bleeding profiles than warfarin, while low-dose apixaban and dabigatran had similar stroke and better bleeding profiles. Real-world use of DOACs may explain some of the differences observed in Canadian routine care versus the phase III clinical trials.