The logistic EuroSCORE in cardiac surgery: how well does it predict operative risk?

Objectives: To study the ability of the logistic EuroSCORE to predict operative risk in contemporary cardiac surgery. Design: Retrospective analysis of prospectively collected data. Setting: All National Health Service centres undertaking adult cardiac surgery in northwest England. Patients: All patients undergoing cardiac surgery between April 2002 and March 2004. Main outcome measures: The predictive ability of the logistic EuroSCORE was assessed by analysing how well it discriminates between patients with differing observed risk by using the area under the receiver operating characteristic (ROC) curve and studying how well it is calibrated against observed in-hospital mortality. The performance of the EuroSCORE was examined in the following surgical subgroups: all cardiac surgery, isolated coronary artery surgery, isolated valve surgery, combined valve and coronary surgery, mitral valve surgery, aortic valve surgery and other surgery. Results: 9995 patients underwent surgery. The discrimination of the logistic EuroSCORE was good with a ROC curve area of 0.79 for all cardiac surgery (range 0.71-0.79 in the subgroups). For all operations, the predicted mortality was 5.7% and observed mortality was 3.3%. The logistic EuroSCORE overpredicted observed mortality for all subgroups but by differing degrees (p = 0.02) Conclusions: The logistic EuroSCORE is a reasonable overall predictor for contemporary cardiac surgery but overestimates observed mortality. Its accuracy at predicting risk in different surgical subgroups varies. The logistic EuroSCORE should be recalibrated before it is used to gain reassurance about outcomes. Caution should be exercised when using it to compare hospitals or surgeons with a different operative case mix.