The validity of the original EuroSCORE and EuroSCORE II in patients over the age of seventy

OBJECTIVES: EuroSCORE II, despite improving on the original EuroSCORE system, has not solved all the calibration and predictability issues. We investigated the sensitivity, specificity and predictability of original EuroSCORE and EuroSCORE II system in elderly patients. METHODS: The original logistic EuroSCORE and EuroSCORE II were assessed via receiver operator characteristic (ROC) and Hosmer-Lemeshow test probability analysis with regard to accuracy of predicting in-hospital mortality. Analysis was performed on isolated coronary artery bypass graft (CABG) (n = 2913), aortic valve replacement (AVR) (n = 814), mitral valve surgery (MVR) (n = 340), combined AVR and CABG cases (n = 517) and the above cases combined (n = 4584). Elderly was defined as ≥70 years old. Age <70 was used for comparative purposes. RESULTS: Institutional mortality was 2.9%, for all isolated CABG, AVR, MVR and combined AVR and CABG cases. In all patients aged ≥70 neither the original EuroSCORE nor EuroSCORE II had a ROC c-statistic above 0.7. For isolated CABG, the ROC c-statistic was not acceptable in patients ≥70 years of age, but was fine for patients under the age of 70 years. For isolated AVR the ROC c-statistic was good for patients aged less than 70 years of age for both risk models; however, the ROC was unacceptably low in patients aged ≥70 for both models. For isolated MVR, the ROC c-statistic and Hosmer-Lemeshow test probability was good for all patients regardless of age. For combined AVR and CABG, the ROC c-statistic was unacceptably low for all patients, regardless of age group using the original EuroSCORE, and in those aged ≥70 using the EuroSCORE II risk model. The original EuroSCORE had no issues with the Hosmer-Lemeshow test probability; however, EuroSCORE II had poor model predictability for all patients, P < 0.0001, and for isolated CABG, P = 0.05 and AVR, P = 0.06. CONCLUSIONS: The original EuroSCORE and the EuroSCORE II risk models should be used with caution in patients aged 70 or older undergoing cardiac surgery in the modern era. Below the age of 70, both models are sensitive, specific and have good predictive power. Our work needs validation by other large groups.