P4779High facility preference for rivaroxaban in atrial fibrillation increases risk of major bleeding compared to facility preference for apixaban

Abstract Background No randomized trial has compared efficacy and safety of non-vitamin K antagonist oral anticoagulants (NOACs) in atrial fibrillation (AF). Previous real-world comparisons could be biased by patient characteristics of importance for treatment selection, but instrumental variables could potentially account for measured and unmeasured confounders. Purpose To compare efficacy and safety of rivaroxaban and apixaban using facility preference for type of NOAC as instrumental variable. Methods AF patients started on apixaban or rivaroxaban were identified using nationwide registries. We categorized patients according to facility preference for type of NOAC, measured as percentage of the prior 20 AF patients started on rivaroxaban in the same facility. Occurrence of stroke/thromboembolism (TE), major bleeding, myocardial infarction and all-cause mortality during two years of follow-up were investigated using adjusted Cox regressions. To further examine general frailty according to facility preferences we also investigated occurrence of cancer, urogenital tract infection, dehydration and fracture. Results We analyzed 6264 AF patients initiated on rivaroxaban or apixaban. Compared with patients treated in facilities that used rivaroxaban in 0–20% of cases, the adjusted hazard ratio for bleeding was 1.05 when treated in a facility with 25–40% use; 1.40 with 45–60% use; 1.50 with 65–80% use; and 1.81 for 85–100% use (Ptrend=0.002). Higher facility level use of rivaroxaban was not associated with increased risk of stroke/TE (Ptrend=0.06), myocardial infarction (Ptrend=0.87) or all-cause mortality (Ptrend=0.91), and there was no association between facility preference for rivaroxaban and risk of cancer (Ptrend=0.83), urogenital tract infection (Ptrend=0.49), dehydration (Ptrend=0.91) or fracture (Ptrend=0.47). Characteristics by facility preference Percent of previous AF patients from facility started on rivaroxaban P for trend 0–20% 25–40% 45–60% 65–80% 85–100% No. of patients 1406 1421 1551 930 956 Received rivaroxaban, (%) 279, (19.8) 499, (35.1) 711, (45.8) 632, (68.0) 774, (81.0) <0.001 Standard dose, (%) 1216, (86.5) 1232, (86.7%) 1366, (88.1%) 793, (85.3%) 824, (86.2%) 0.62 Median age, (interquartile range) 70, (63.3–74) 69, (63–74) 70, (64–74) 70, (64–75) 70, (63–75) 0.11 Below median income, (%) 740, (52.6) 699, (49.2) 764, (49.3) 458, (49.3) 471, (49.3) 0.31 Prior stroke, (%) 99, (7.0) 115, (8.1) 134, (8.6) 69, (7.4) 74, (7.7) 0.56 Prior bleeding, (%) 136, (9.7) 141, (9.9) 163, (10.5) 91, (9.8) 97, (10.1) 0.51 Antiplatelet therapy, (%) 445, (31.7) 465, (32.7) 491, (31.7) 303, (32.6) 317, (33.2) 0.49 Rate of events according to instrument Conclusion High facility preference for rivaroxaban increases risk of major bleeding compared to facility preference for apixaban. Acknowledgement/Funding This study was funded by an unrestricted grant from the Capital Region of Denmark, Foundation for Health Research.