Risk scores for predicting outcomes in valvular heart disease: how useful?

Risk scoring tools have been developed from clinical databases to predict expected patient mortality from cardiac surgical procedures. The risk algorithms have been developed and validated from variables that have been demonstrated to be predictive of mortality. At least six risk models have been developed from different databases measuring outcomes of cardiac surgery. These algorithms have then been used to select very high risk patients for conventional aortic valve replacement (AVR) who would be appropriate candidates for transcatheter aortic valve implantation (TAVI). The two most common risk models used for TAVI selection are the logistic EuroSCORE (LES) and the Society of Thoracic Surgeons Predicted Risk of Mortality (STS-PROM) algorithms. Although both models are accurate in predicting mortality in low-risk patients, the LES has been clearly demonstrated to overpredict expected mortality by a factor of three in high-risk candidates for AVR. Various factors that also impact mortality but are not included in either algorithm include liver disease, frailty, porcelain aorta, and previous radiation. Despite these shortcomings, risk algorithms are effective models for predicting risk, with the STS-PROM being more accurate in high-risk patients. Ultimately, a new risk algorithm specific for TAVI will need to be developed once sufficient databases are developed.