External validation of the ORACLE bleeding risk score using the database of the RECORD 3 registry

Abstract

Aim. To validate a domestic scale for assessment risk of bleeding ORACUL (ОРАКУЛ) based on an independent sample of patients with acute coronary syndrome (ACS). Materials and methods. External validation of the ORACUL score was carried out using database of an independent observational study RECORD-3 which comprised data from all patients hospitalized for 1 month (march-april 2015) in 47 centers of 37 cities in 21 regions of Russia. Total number of included patients was 2370, mean age 64.2±11.96 years, 821 patients (34.6%) had ST-elevation, other patients – non-ST elevation ACS. Results. The following bleeding events were registered in RECORD-3: bleedings during hospitalization (n=34, 1.43%), inhospital bleedings requiring withdrawal of antithrombotic treatment (n=16, 0.68%), inhospital bleedings, which required drug or surgical treatment or hemotransfusion (n=16, 0.68%). Forty eight hemorrhagic complications were registered during 6 months of observation after hospital discharge. Diagnostic value of the ORACUL score for estimation of risk of bleedings during index hospitalization was good (C-criterion 0.691±0.050; р<0.001), sensitivity of the model was 58.1%, specificity 79.9%. Earlier on the cohort of patients of the ORACUL study diagnostic value of the score for inhospital bleedings was found to be 0.777±0.046. Difference of diagnostic values was inessential. For estimation of the bleeding risk during 6 months of post discharge observation area under the ROC curve (C-criterion) was 0.628±0.045 (р=0.003), sensitivity and specificity of the model were 53.9 and 73.7%, respectively. On the ORACUL study cohort AUC 0.748±0.048 (р=0.071). Conclusion. External validation confirmed that statistical power of the OCACUL score is sufficient for prediction of bleedings during both periods of hospitalization and after hospital discharge.