Online Program

Return to main conference page
Saturday, February 17
PS3 Poster Session 3 and Continental Breakfast Sat, Feb 17, 8:00 AM - 9:15 AM
Salons F-I

Win Ratio Application for a Composite Outcome in a Randomized Cardiovascular Trial (303630)

View Presentation View Presentation

Robert Giugliano, TIMI Study Group 
Erica L. Goodrich, TIMI Study Group 
*Rose A. Hamershock, TIMI Study Group 
Estella Kanevsky, TIMI Study Group 

Keywords: Win Ratio, Composite outcome, Survival analysis

Cardiovascular clinical trials often use survival analysis to compare composite outcomes (CO) between treatment groups. If the clinical severity of event types included in a CO differ, or there is a temporal relationship among event types, then there is a risk of underrepresenting more severe or later-occurring events. Most notably, death is usually considered the most severe event type and can only occur in isolation from or after other nonfatal events. Therefore, analyzing a CO including an event type like death using a Cox proportional hazards model may underrepresent the most severe events by taking the first event and censoring the rest.

The Win Ratio, as described by Pocock et al., may be a preferable approach. With this method, event types are compared in order of clinical severity among pairs of cases and controls. This poster aims to explain the methodology used in the Win Ratio and compare its results with those from a Cox model for a CO that combines cardiovascular death, disabling stroke, and life-threatening bleeding, using data from the randomized cardiovascular clinical trial, ENGAGE AF-TIMI 48, which compared two anticoagulants in patients with atrial fibrillation.