Durand et al are to be commended for their interesting study on assessing the performance of the European System for Cardiac Operative Risk Evaluation (EuroSCORE) II for predicting 30-day mortality in patients who undergo transcatheter aortic valve implantation. After analyzing 250 patients, the investigators found that this was a well-calibrated model but the discriminatory power was suboptimal, which would make this system unreliable to predict risk mortality after transcatheter procedures. The inclusion of irrelevant variables for percutaneous procedures included in the EuroSCORE II (such as previous cardiac surgery) and the absence of common and relevant parameters normally associated with these interventions (such as liver dysfunction or porcelain aorta) very likely weaken this predictive model because of its inability to discriminate the risk. We therefore agree with these investigators when claiming for a new and reliable risk score for transcatheter procedures.

However, creating predictive risk models in cardiac surgery seems to be difficult. In 1995, the EuroSCORE I was a risk model derived from data collected from 14,799 consecutive patients in 100 European centers. Despite its thoroughness, this system overpredicted the mortality risk for patients who underwent cardiac surgery. A reason for this may be that a highly heterogenous patient group was chosen for cardiac procedures, but most had coronary artery disease, thus favoring the model in patients undergoing coronary artery surgery. Another reason may be that the model accounted for only the most common and prevalent risk factors for 30-day mortality but not for rare ones. Lastly, the surgical techniques and the postoperative patient management are constantly changing and improving over the years, which currently make the EuroSCORE I a redundant model.

The EuroSCORE II was published 13 years later to overcome the previous limitations. This model is based on the results of >22,000 patients who underwent cardiac surgery in 2010 in 154 hospital centers. The core of risk factors is almost the same but 2 variables were removed, 3 were added, and 4 were defined more precisely and categorized. However, the promising results published by the investigators of the score (discrimination tested by area under the receiver operating characteristic curve was 0.81) proved again to be disappointing when the external validation of the model was assessed by independent researchers. The reasons why the updated scoring system have failed may be because the EuroSCORE II developers included variables that were not significantly associated with mortality by multivariate regression, they removed one of the major risk factors of EuroSCORE I (postinfarction ventricular septal rupture) because of its low incidence, and they failed to analyze other variables that could have had a major weight in an updated score.

Thus, a reliable risk score for transcatheter aortic valve implantation techniques seems to have to be based on recent data of a homogenous sample formed by a large number of patients and should include the most clinically relevant variables. Ultimately, this predictive model is a utopic desire by cardiologists and cardiac surgeons who could identify patients who may benefit from transcatheter techniques. However, it seems to be a titanic undertaking.