RESUMO
This study was conducted to establish a quantitative model to predict the risk of in-hospital mortality for patients undergoing cardiac valve replacement and to decrease mortality in patients with predicted high risk using prophylactic extracorporeal membrane oxygenation (PECMO). We retrospectively reviewed the medical records of 4482 patients who underwent cardiac valve replacement from January 1994 to December 2004, at Anzhen Hospital, Beijing, China. A total of 158 patients were going to receive heart valve replacement. Associations between mortality and the demographic, clinical, and laboratory variables of patients were first assessed using univariate analysis. Six of 7 variables in the univariate analysis were statistically significant and were included in the multivariate analysis: renal function; age; left ventricular ejection fraction (EF); coronary artery disease (CAD); pulmonary artery pressure (PAP); and left ventricular end-diastolic diameter (LVEDD). The area under the receiver operating characteristic (ROC) curve (AUC) was 73.58%. Observed mortality in the group with PECMO (5.45%, 3/55) was significantly lower (Pearson Chi2 = 4.314, P = 0.038, P < 0.05) than in the group without PECMO (24.27%, 25/103). With the use of our scoring model, the risk of postoperative mortality in patients planning to undergo valve replacement can be predicted before the procedure is performed. For patients with predicted mortality greater than 10%, the use of PECMO during surgery, in addition to extracorporeal circulation, was found to decrease mortality.