Modeling the impact of real-world adherence to once-daily (QD) versus twice-daily (BID) non-vitamin K antagonist oral anticoagulants on stroke and major bleeding events among non-valvular atrial fibrillation patients

Objectives: To estimate the real-world (RW) impact of adherence to once-daily (QD: rivaroxaban and edoxaban) and twice-daily (BID: apixaban and dabigatran) non-vitamin K antagonist (NOACs) on the risk of stroke and major bleeding (MB) among non-valvular atrial fibrillation (NVAF) patients. Methods: First, claims from the Optum Clinformatics Data Mart database (July 2012-December 2016) were analyzed. Adult NVAF patients with >= 2 NOAC dispensings (index date) were included. The relationship between NOAC adherence (proportion of days covered >= 80%) and stroke/MB 1-year post-index was evaluated using adjusted Cox proportional hazards models. Second, the natural logarithm of hazard ratios (HRs) was multiplied to a literature-derived mean adherence difference between QD and BID NOACs yielding stroke and MB rates. Third, these rates were multiplied by 1-year Kaplan-Meier rates of stroke and MB which yielded the number of strokes prevented and MBs caused. Annual cost savings were evaluated using literature-based stroke ($81,414/patient) and MB ($63,905/patient) cost estimates. Results: In total, 54,280 patients were included. HRs for adherent vs non-adherent patients were 0.67 (p < .001) for stroke and 1.09 (p = .179) for MB. The claims-derived 1-year Kaplan-Meier rates were 3.0% and 3.4% for strokes and MBs, respectively. For 100,000 AF patients, 64 strokes were prevented (p < .001), and a non-significant number of MBs (n = 15, p < .191) were caused by QD vs BID NOACs annually, which leads to cost savings estimated at $58 million for QD NOACs. Conclusion: QD NOACs prevented a significant number of strokes and caused no significant increase in MBs compared to BID NOACs, which leads to significant net cost savings for NVAF patients in the US.