Abstract 17115: Bayesian Analysis of Risk Stratification Strategies in Non-ST Elevation Acute Coronary Syndromes

Background: Early stratification of patients with acute coronary syndromes is a critical aspect of contemporary cardiology. Risk scores are often used to help identify patients likely to benefit from early interventional strategies, based on expected short term prognosis. Additionally, understanding long term benefit from utilization of risk scores to guide intervention, as well as differential benefit of particular risk scoring strategies can help guide clinical application and management. To investigate this, we compared the PURSUIT, GRACE and TIMI risk scores for NSTEMI, evaluating mortality and non-fatal MI at one year, using Bayesian statistical methods. Methods : Using published retrospective data of 460 patients, Bayesian posterior probabilities were calculated. Risk scoring systems were stratified into high and low risk groups, based on either PURSUIT score of 14, TIMI score of 3, or a GRACE score of 133. Within these groups, early revascularization or conservative management was compared based on mortality or non-fatal MI events at 1 year. Posterior probabilities as well as probabilities of greater than 20% benefit of using a risk scoring system were calculated. Results: Based on the study results, the Bayesian posterior probability for the primary combined endpoint of death and nonfatal MI for the low risk group are 59.3% for PURSUIT, 68.8% for TIMI, and 70.1% for GRACE. The Bayesian posterior probabilities for the high risk group are 73.1% for PURSUIT, 58.6% for TIMI, and 74.0% for GRACE. Probabilities of >20% predictive benefit and sensitivity analysis was calculated, and have similar trends to the posterior probabilities. Conclusions: Bayesian analysis suggests that the using low GRACE and TIMI scores for conservative management have a greater likelihood of clinical benefit than the PURSUIT risk scale. The GRACE and PURSUIT scores have a clinically relevant benefit in predicting patients deemed to be high risk undergoing early revascularization compared to the TIMI score. Overall, the GRACE score was demonstrated to be the most robust scoring system for guiding both interventional and conservative management strategies for 1 year outcomes, and should be applied more widely.