Risk index for predicting in-hospital mortality for cardiac valve surgery

Background. Numerous studies have developed a severity score or risk index for short-term mortality associated with coronary artery bypass graft (CABG) surgery, but very few studies have developed risk indices derived from statistical models to predict outcomes for cardiac valve replacement patients. Methods. Data from New York's Cardiac Surgery Reporting System in 2001 to 2003 were used to develop statistical models that predict mortality for valve surgery and for valve/CABG surgery. These models were used to develop risk indices based on the type of valve surgery performed and several patient risk factors. The fit of each index was tested by examining the correspondence of expected and observed mortality rates for various risk score ranges using New York data between 1998 and 2000. Results. There were a total of 11 risk factors for valve patients without CABG surgery and 12 risk factors for patients with both valve and CABG surgery. Risk factors represented measures of demographics, type of valve surgery, previous open heart surgery, ventricular function, hemodynamic state, and various comorbidities. Possible variable scores ranged from 0 to 7 in the isolated valve model and 0 to 5 in the valve/CABG model. The highest overall risk scores possible for the two models were 49 for isolated valve surgery and 35 for valve/CABG surgery, and the highest scores observed for any patient were 32 and 26, respectively. Conclusions. These valve surgery risk indices will enable providers to estimate patients' short-term mortality risk and allow for comparisons of valve surgery outcomes with other regions.