EuroSCORE-predicted mortality and surgical judgment for interventional aortic valve replacement

Transcatheter aortic valve implantation (TAVI) is used as an alternative to conventional surgery in high-risk patients. The EuroSCORE quantifies the risk, and patients are selected for this procedure if their predicted mortality is high. The study aim was to evaluate the risk discrimination and calibration of the EuroSCORE, as well as the surgical judgment that is used when selecting the aortic valve replacement (AVR) approach.

Between 2003 and 2006, a total of 2757 patients underwent isolated, open AVR. The EuroSCORE was compared to the observed in-hospital mortality. A survey among surgeons assessed the relevance of patient-specific factors towards the approach of TAVI versus open AVR.

Survivors of AVR had a lower additive EuroSCORE than non-survivors (7.2 +/- 3.34 versus 12.0 +/- 3.34; p < 0.001). The EuroSCORE demonstrated reasonable risk stratification (c-statistic 0.843, p < 0.001). However, none of the 67 non-survivors was correctly predicted (Hosmer-Lemeshow test; p < 0.001), and most patients with a high EuroSCORE survived AVR. The actual mortality was 2.8% as compared to a EuroSCORE-predicted mortality of 12.5 +/- 14.2% (p < 0.001). This poor calibration was worse with higher EuroSCOREs (Cusum test; p < 0.001). Surgeons identified a porcelain aorta as the only factor that warranted TAVI. Most EuroSCORE variables had either no relevance for treatment selection, or prompted open AVR.

Patient selection for TAVI cannot be based on the EuroSCORE, because it lacks discrimination and center-specific calibration. Thus, individual, surgical judgment that weighs institutional expertise for high-risk patients against a possible reduction of mortality by using interventional techniques, is recommended.