Abstract 15025: Validation of the TRA2°P TIMI50 Risk Score for Predicting Cardiovascular Events in Real World Patients
Objective: To assess the discrimination of an RCT (TRA2°P TIMI-50) derived risk tool in predicting recurrent cardiovascular events in the real world.
Methods: Patients experiencing a myocardial infarction (MI) between 2005 and 2012 at the Cleveland Clinic (CC) and 2007 and 2012 at Geisinger Health Systems (GHS) were included. Characteristics were compared and outcomes measured from the first outpatient visit 2-52 weeks after an index MI, until the first of the composite event (recurrent MI, ischemic stroke, cardiovascular death), or the last known follow-up. The TRA2°P TIMI-50 3-year risk score was compared to the cumulative incidence of the composite event. The concordance index and the Brier score, both accounting for competing risk of non-cardiovascular death, measured risk score performance.
Results: 6,373 patients were included in the validation cohort from CC and 4,323 from GHS. These populations were on average, older, more often female, had more comorbidities, more (CC), and less (GHS) racially diverse. The risk score, an integer value from 0-9, is based on easily obtainable clinical characteristics and associated with long term atherothrombotic risk. Both populations had a greater proportion of patients with higher risk scores. Likewise, 3-year cumulative incidence of composite event was also higher than observed in the trial. 73 and 65 percent, respectively, of the CC and GHS populations, were event free after 3 years, compared to ~90% of the trial population. The risk score was observed to be well calibrated, with Brier scores of 0.192 (CC) and 0.233 (GHS) and well discriminated, with a concordance index of 0.661 (CC) and 0.684 (GHS).
Conclusions: The TRA2°P TIMI-50 derived-risk score maintained risk discrimination and was well calibrated to the actual cumulative incidence of cardiovascular events, while adjusting for competing risks. Additional validation of the risk score in diverse populations will make its application increasingly relevant.
Author Disclosures: K.M. Chagin: Other Research Support; Significant; Merck & Co., Inc. M.W. Kattan: Other Research Support; Significant; Merck & Co., Inc. B.A. Williams: Other Research Support; Significant; Merck & Co., Inc., Roche, Biosense Webster, Daiichi Sankyo. A. Milinovich: Other Research Support; Significant; Merck & Co., Inc. J.M. Bauman: Other Research Support; Significant; Merck & Co., Inc. M.D. Patel: Employment; Significant; Merck & Co., Inc, Symphony Health Solutions. L.D. Bash: Employment; Significant; Merck & Co., Inc..
- © 2016 by American Heart Association, Inc.