Abstract P232: Influence of Individualized Risk Prediction and Patient Preferences on Selection of Prasugrel Versus Clopidogrel at the Time of Percutaneous Coronary Intervention for an Acute Coronary Syndrome

Introduction: Given the need to weigh prevention of ischemia against bleeding risk with prasugrel vs. clopidogrel at the time of PCI for an acute coronary syndrome (ACS), we developed risk models for both outcomes from TRITON-TIMI 38. We then applied these models to each TRITON patient and examined how preferences for outcomes could influence thienopyridine selection. Methods: We built separate multivariable regression models for ischemia (cardiac death, MI, stroke) and bleeding (TIMI major/minor) from 12,579 patients in TRITON-TIMI 38 with no history of stroke or TIA. For each patient, we calculated the probability of ischemic and bleeding events with prasugrel vs. clopidogrel and the associated benefit:risk ratio (predicted difference in ischemic events/predicted difference in bleeding). We then examined the impact of alternative outcome weights - benefit:risk preference thresholds - on individualized decision making. Results: Based on individualized risk predictions, the majority of ACS patients treated with PCI (66%) may choose prasugrel when preventing ischemia is considered equally important as avoiding bleeding (Figure), but this proportion varied from 32% to 80% when the benefit:risk preference threshold was varied from 3 (reflecting a 3-times greater preference for avoiding bleeding) to 0.25 (reflecting a 4-times greater preference for avoiding ischemia). Conclusions: Based on empirical analyses, the expected absolute benefits and risks of prasugrel vs. clopidogrel depend highly on patient characteristics and preferences. Presenting individualized predictions of benefits and risks with competing treatments may improve shared decision making.