Is it appropriate to use EuroSCORE II for prediction of 30-day mortality in cardiac surgery?

I read with great interest the article by Carino et al. [1] who aimed to evaluate the discriminatory ability of the EuroSCORE II in predicting 30-day mortality in a large cohort of patients (2645 patients operated from January 2012 to December 2019) undergoing surgical mitral valve repair (MVr) in a high-volume centre. Authors should be commended for outstanding results, especially for very low 30-day mortality in patients undergoing MVr for degenerative mitral regurgitation (MR, −0.39%; 10/2530 patients). The main findings of their study are that in patients with degenerative MR, the EuroSCORE II showed low discrimination (area under the receiver operative characteristic curve—AUC of 0.68), low accuracy (Brier score 0.27) and low calibration due to the overprediction of 30-day mortality. In patients with secondary MR, EuroSCORE II showed confirmed good discrimination (AUC 0.88), good accuracy (Brier score 0.003) and good calibration in predicting 30-day mortality. We have recently published a paper [2], which validated EuroSCORE II performances in Serbian cardiac surgical population (10 048 patients, operated between January 2012 and March 2017). For MVr surgery, we have observed in-hospital mortality of 0.69% (3 out of 437 patients who underwent MVr—not divided by MR aetiology). Discrimination was at borderline, with AUC of 0.70 (95% confidence interval from 0.00 to 1.00), calibration using Hosmer–Lemeshow test was good (P = 0.51). Although observed-to-expected mortality ratio was 0.50, 95% confidence interval of −0.07 to 1.07 did not confirmed significant overprediction of mortality (EuroSCORE II predicted in-hospital mortality was 1.46%). Although operative mortality is the most commonly used outcome measure in cardiac surgery, the definition of an early or perioperative mortality death still remains contentious (whether it is in-hospital mortality, or 30-day mortality, or both combined, etc.). EuroSCORE II final database set contained 22 381 cases from 154 centres in 43 countries [3]. Although only 56.6% of units provided data on 30-day status, EuroSCORE II was launched for prediction of in-hospital mortality, only [3]. In another words, EuroSCORE II risk scoring system predicts in-hospital mortality in cardiac surgery. Although authors [1] defined operative mortality as death occurring up to 30 postoperative days, my dilemma is—are we allowed to validate risk model that predicts in-hospital mortality using data of 30-day mortality? Although the in-hospital and the 30-day mortalities do not refer to the same patients (20% of the patients counted in each measure are not included in the other measure), it was concluded that in-hospital mortality and 30-day mortality measure 2 different types of end-point and are not interchangeable [4]. Therefore, it would be of great interest if authors are able to complete their results using EuroSCORE II to test in-hospital mortality.