Overestimation of aortic valve replacement risk by EuroSCORE: implications for percutaneous valve replacement

The EuroSCORE has been proposed to identify patients at high risk for surgical aortic valve replacement (AVR) and estimate for them the risk-benefit of percutaneous valve replacement. The aim of our study was to investigate the validity of this proposal. From 1994 to March 2006, 1545 consecutive patients with aortic stenosis underwent isolated surgical AVR at the Department of Cardiac Surgery of Heidelberg. Both additive and logistic EuroSCOREs were calculated for each patient and summed for expected 30-day mortality. Expected and observed mortalities were compared, particularly with respect to 'high-risk' status and era of operation. Overall, 30-day mortality was low (34/1545, 2.2%) and substantially overestimated by both additive (6.1%) and logistic (9.3%) EuroSCOREs. Although both EuroSCOREs stratified patients monotonically with respect to mortality risk, high-risk patients had a 3.6% mortality (29/833), whereas additive and logistic EuroSCOREs predicted 8.3 and 14.8%. Indeed, none of the 71 patients with a EuroSCORE of 11-20 (extremely high risk) died. The more recent the era of operation, the more pronounced was the discrepancy between expected and observed mortalities. Although the EuroSCORE still successfully stratifies patients undergoing surgical AVR relative to 30-day mortality, it has become increasingly uncalibrated with absolute risk, resulting in overestimation of 30-day mortality. Inaccurately predicted mortality, especially in 'high-risk' patients, renders it unsuitable for assessing risk reduction of percutaneous valve replacement.