Prospective and Multicentric Validation of the ArgenSCORE in Aortic Valve Replacement Surgery. Comparison with the EuroSCORE I and the EuroSCORE II

Introduction: In patients with aortic stenosis and planned aortic valve replacement, an accurate stratification of surgical risk is mandatory to offer the best individual option. Preoperative risk scores have recovered a leading role in the assessment of these patients. Objectives: The aim of this study was to perform a prospective, multicentric validation of the ArgenSCORE in patients with aortic valve replacement and compare its performance with the EuroSCORE I and the EuroSCORE II. Methods: A total of 250 adult patients undergoing aortic valve replacement at four centers ofthe City of Buenos Aires were included in the study from February 2008 to December2012. The ArgenSCORE was compared with the EuroSCORE I and the Euro-SCORE II, evaluating model discrimination with the area under the ROC curve andcalibration power comparing the relation between observed mortality and predictedmortality. Results: The mean age of the validation population (n = 250) was 68.62 ± 13.3 years and overall mortality of was 3.6 %. The ArgenSCORE showed good discrimination power (area under the ROC curve of 0.82) and a good predictive capacity to allocate risk (relation between observed mortality: 3.6 % vs. predicted mortality: 3.39%; p = 0.471).The EuroSCORE I showed poor discrimination power (area under the ROC curve of0.62) and risk overestimation (relation between observed mortality: 3.6 % vs. predictedmortality: 5.58 %; p < 0.0001). The EuroSCORE II showed an acceptable discrimination power (area under the ROC curve of 0.76), though lower than that of the ArgenSCORE, but a significant underestimation of predicted risk (relation betwee nobserved mortality: 3.6 % vs. predicted mortality: 1.64 %; p < 0.0001). Conclusions: The ArgenSCORE evidenced adequate excellent ability to predict mortality inpatients undergoing AVR aortic valve replacementsurgery. This local model demonstrated good discrimination power and better calibration compared to with the European models, as the EuroSCORE I overestimated and the EuroSCORE II underestimatedpredicted risk.