Comparative validation of three contemporary bleeding risk scores in acute coronary syndromes

Abstract

Background: Hemorrhagic complications are strongly linked with subsequent adverse outcomes in acute coronary syndrome (ACS) patients. Various risk scores (RS) are available to estimate the bleeding risk in these patients. Aims: To compare the predictive accuracy of the three contemporary bleeding RS in ACS. Methods: We studied 4500 consecutive patients with ACS. For each patient, the ACTION, CRUSADE, and Mehran et al bleeding RS were calculated. We assessed their performance either for the prediction of their own major bleeding events or to predict the TIMI serious (major and minor) bleeding episodes in the overall population, in patients with non-ST elevation ACS (NSTEACS) and in those with ST-elevation myocardial infarction (STEMI) patients. Calibration (Hosmer-Lemeshow test) and discrimination (c-statistic) for the three RS were computed and compared. We used the concept of net reclassification improvement (NRI) to compare the incremental prognostic value of using a particular RS over the remaining scores in predicting the TIMI serious bleeding. Results: The best predictive accuracy was obtained by the CRUSADE score either for the prediction of its own major bleeding events (c-statistic=0.80, 0.791, and 0.81 for the entire sample, for STEMI, and for NSTEACS patients, respectively) or to predict the TIMI serious bleed occurrence (c-statistic=0.741, 0.738,and 0.745 for the whole population, for STEMI and NSTEACS patients, respectively). The lowest bleeding rates observed in patients classified as low risk corresponded to the CRUSADE RS. All scores performed modestly in patients who did not undergo coronariography (all c-statistic <0.70). The CRUSADE score was significantly superior to the ACTION model in predicting the TIMI serious bleeding occurrence in terms of NRI overall and by ACS subgroups (p<0.05). Overall, the CRUSADE RS exhibited better calibration for predicting the TIMI serious bleeding compared to the ACTION and Mehran et al scores (Hosmer-Lemeshow p-values of 0.26, 0.13, and 0.07, respectively). Conclusion: The CRUSADE score represents, among the more contemporary bleeding RS, the most accurate and reliable quantitative clinical tool in STEACS and STEMI patients. We encourage the utilization of the CRUSADE index for bleeding risk stratification purposes in daily clinical practice and in ACS outcome studies. The performance of the three more contemporary bleeding RS is modest in those patients who received conservative management.