[Evaluation of the EuroSCORE Model in the Patient Population of Our Department of Cardiac Surgery].

INTRODUCTION:The objective of current study is the evaluation of the accuracy and precision of EuroSCORE in the population of those patients undergone cardiac surgery in our department.MATERIAL AND METHODS:We have analyzed the data of 1839 consecutive patients who had their operations between 1/January 2003 and 31/December 2005. We have compared the mortality rates predicted preoperatively by additive and logistic EuroSCORE with the actual 30-day mortality figures. On statistical analysis the discriminative accuracy of the methods has been defined with the use of C-statistics. The calibration and precision of the methods have been checked by the Hosmer-Lemeshow statistics.RESULTS:The overall mortality rate in the above period was 3.3%. Additive and logistic EuroSCORE predicted 4.1 +/- 2.8 and 4.5 +/- 6.1%, consecutively. Based on the C-statistics the area below the Receiver Operating Characteristic curve has measured 0.699 (0.629-0.769) and 0.711 (0.642-0.779). The kappa 2 value for the Hosmer-Lemeshow statistics has proved 6.5 ( p = 0.475) and 12.5 ( p = 0.131).CONCLUSION:Logistic EuroSCORE has appropriate discriminative power and satisfactory precision, whilst the accuracy of additive EuroSCORE only comes near to the acceptable level, but at the same time it has an adequate calibration value. These results suggest that logistic EuroSCORE is more suitable for the preoperative risk assessment of these patients.