P285Comparison of analysis methodologies for net outcome with edoxaban vs warfarin in patients with atrial fibrillation in the ENGAGE AF-TIMI 48 trial

Abstract

Background: Trials comparing vitamin K antagonists (VKAs) with non‐VKA oral anticoagulants (NOACs) have evaluated efficacy and safety separately; there is no consensus on how to weigh components of a net clinical outcome (NCO). Methods: We compared higher‐dose (HDE) and lower‐dose (LDE) edoxaban regimens with warfarin using the Win Ratio approach of a NCO consisting of death, efficacy, and safety endpoints in 21,105 patients enrolled in ENGAGE AFTIMI 48. The priority of endpoints was assigned by the fatality rate of 4 groups of events: death (100%); hemorrhagic stroke (45%); ischemic stroke, subdural/ epidural hematoma, systemic embolism (18%); other major bleed (5%). All possible patient pairs between treatment groups were compared (>49 million per treatment‐group comparison), and a loss was defined as the patient who first had an event, starting with death and if neither died, then moving on to the other events sequentially. The Win Ratio was calculated as wins/losses, with a Win Ratio >1 favoring the edoxaban regimen and a Win Ratio of <1 favoring the warfarin regimen. Results: Both HDE and LDE were significantly better than warfarin with Win Ratios of 1.12 for HDE and 1.19 for LDE vs warfarin (Figure). There was no significant difference between HDE and LDE. The endpoints that contributed the most to the favorable Win Ratios were death (for both HDE and LDE) and other major bleeds (LDE, Table). Conclusions: Using a hierarchical approach to evaluate a NCO that combined death, efficacy, and safety events in ENGAGE AF‐TIMI 48, both regimens of edoxaban were superior to warfarin.