Performance of euroscore ii compared with its predecessors in octogenarian patients undergoing aortic valve replacement

All octogenarians who underwent aortic valve replacement between 2009 and 2015 were analyzed. EuroSCORE II was assessed calculating discrimination using the area under the ROC curve (AROC), and calibration using the Hosmer-Lemeshow (HL) test and the risk adjusted mortality ratio (RAMR) of the model. The same analysis was performed by risk quartiles. 482 octogenarian patients underwent aortic valve replacement in our center during the study period. Discrimination of EuroSCORE II was excellent, AROC=0,91, and better than its previous versions, AROC for logistic EuroSCORE =0,81 and for additive EuroSCORE=0,80. Calibration was poor due to underestimation of the mortality risk. (p for Hosmer-Lemeshow test of 0,03 and risk adjusted mortality ratio=8,9/6,1=1,46). However, analyzing by quartiles, EuroSCORE II has almost perfect calibration and discrimination for all patients except for those with the highest risk. In the fourth quartile of risk, EuroSCORE II markedly underestimates the mortality risk. In these high-risk patients, the best calibration is reached using the logistic EuroSCORE (p for HL test 0,23 and RAMR =29,2/24,91= 1,17).