An International Publisher for Academic and Scientific Journals
Author Login 
Scholars Journal of Applied Medical Sciences | Volume-8 | Issue-06
Performance of EuroSCORE and EuroSCORE II in Institut Jantung Negara (IJN), Kuala Lumpur, Malaysia
M. Azizan Petra, M. Ezani Taib, M. F. Ramli, N. Jaafar, I. F. Gaafar
Published: June 8, 2020 | 185 110
DOI: 10.36347/sjams.2020.v08i06.001
Pages: 1390-1396
Downloads
Abstract
Objective: Assessing the performance of European System for Cardiac Operative Evaluation (EuroSCORE) and EuroSCORE II. Method: 4145 patients who underwent cardiac surgery between 1st January 2015 to 31st December 2016 in Institut Jantung Negara (IJN) were included. The entire cohort and isolated coronary bypass graft (CABG) patients were analyzed by measuring the area under the receiver operating characteristic (ROC) curve for model discrimination and Hosmer-Lemeshow Chi-squared test for model calibration. Performance of both models was compared. Result: For the entire cohort, ROC curve for EuroSCORE was 0.679; EuroSCORE II was 0.615. For isolated CABG patients, ROC curve for EuroSCORE was 0.670; EuroSCORE II was 0.609. For the entire cohort, Hosmer-Lemeshow test showed no significant difference between expected and observed mortality according to EuroSCORE model (Chi-square = 5.284, P = 0.508) and EuroSCORE II model (Chi-square = 15.828, P = 0.050). For the isolated CABG patients, Hosmer-Lemeshow test showed no significant difference between expected and observed mortality according to EuroSCORE model (Chi-square = 5.365, P = 0.498) and EuroSCORE II model (Chi-square = 9.839, P = 0.276). For the entire cohort (Table 7), the observed and predicted mortality were 4.56% and 3.7% respectively for EuroSCORE; observed and predicted mortality were similar at 4.56% for EuroSCORE II. For isolated CABG patients (Table 8), the observed and predicted mortality were 3.62% and 3.36% respectively for EuroSCORE; the observed and predicted mortality were 3.62% and 3.97% respectively for EuroSCORE II. Conclusion: Despite poor discrimination under the ROC, the calibration of both models was good and acceptable to be used for risk prediction tools in our centre.