Performance of EuroSCORE II in Hungary: A Single-centre Validation Study

Abstract

The efficacy of the updated cardiac surgical risk stratification system, EuroSCORE II, needs widespread assessment in the cardiac surgical centres where it is intended to be used. The present paper is a single-centre validation study carried out in Hungary. An adult cardiac surgical cohort of 2287 patients was investigated. The general levels of performance of the logistic EuroSCORE and that of EuroSCORE II were compared using the Hosmer-Lemeshow test, ROC analysis and calculation of the Brier score. The calibrations were visualised by smoothed curves derived with the help of local polynomial regression. The efficacy of EuroSCORE II was analysed in different operation types and urgency subgroups. The old EuroSCORE over-estimated the risk (O:E ratio: 0.66, HL test, p < 0.01), while EuroSCORE II slightly under-predicted mortality (O:E ratio:1.19, HL test, p = 0.0084). Comparing the ROC AUCs, we did not find a significant difference between the accuracy of the old and new versions of EuroSCORE (0.8017, 95% CI: 0.7596-0.8438 vs. 0.8177 95% CI: 0.7786-0.8569). EuroSCORE II performed well among CABG patients (O:E ratio: 0.75, HL test, p = 0.5789) and in those who underwent elective surgery (O:E ratio: 1.1, HL test, p = 0.1396), but failed in the emergency (O:E ratio: 1.71, HL test, p = 0.0055) and salvage (O:E ratio:1.36, HL test, p = 0.0245) categories. EuroSCORE II proved to be more suitable for cardiac surgical risk prediction compared with its previous version, but its reliability can be questioned among patients who need emergency and salvage surgery, as well as in the case of combined operations.