Aim: Renal dysfunction is an important variable in the EuroSCORE (European System for Cardiac Operative Risk Evaluation) model and is currently defined as creatinine >200 mmol/L. The aim of this study was to examine whether using other definitions of renal dysfunction could improve the predictive ability of the EuroSCORE.
Methods: Between January 2004 and January 2006, 1 205 patients underwent cardiac surgery. Their preoperative glomerular filtration rate and EuroSCORE were calculated. Four recalibrated EuroSCORE models were constructed using 1) creatinine as a binary variable; 2) creatinine as a continuous variable; 3) glomerular filtration rate as a categorical variable; or 4) glomerular filtration rate as a continuous variable. The predictive ability of these models was assessed using receiver operating characteristic curve analysis.
Results: Hospital mortality was 4% (N.=47). Receiver operating characteristic curve values were: 0.78 for the original EuroSCORE, 0.80 for the recalibrated binary creatinine model, 0.83 for the continuous creatinine model, 0.83 for the categorical glomerular filtration rate model, and 0.82 for the continuous glomerular filtration rate model.
Conclusions: The use of creatinine as a continuous variable or glomerular filtration rate as a categorical or continuous variable improves the predictive accuracy of the EuroSCORE model for hospital mortality. Given the increasing incidence of preoperative renal dysfunction and its impact on hospital mortality, future risk stratification models should include continuous creatinine or glomerular filtration rate rather than creatinine as a binary variable.