Risk Stratification in Cardiac Surgery

Abstract

Risk stratification in cardiac surgery is based on different models. The great majority of these models derives from a logistic regression equation, and the resulting risk score attributes a specific value to a number of risk factors. However, the number, definition, and type of risk factors included in each risk score greatly varies. Some of the existing risk models include up to 15-20 risk factors. However, it has been demonstrated that for elective patients a simple risk model based on just three factors (age, creatinine, and ejection fraction) has the same level of accuracy and a much better calibration that more complex models. The main deficiency of all the risk models is that they cannot take into consideration the presence of extreme risk conditions which have a very low prevalence, but which are accompanied by a very high operative mortality rate.