Predicting risk for aortic and mitral valve surgery is important both for informed consent of patients and objective review of surgical outcomes. Development of reliable prediction rules requires large data sets with appropriate risk factors that are available before surgery.Methods
Data from eight Northern New England Medical Centers in the period January 1991 through December 2001 were analyzed on 8,943 heart valve surgery patients aged 30 years and older. There were 5,793 cases of aortic valve replacement and 3,150 cases of mitral valve surgery (repair or replacement). Logistic regression was used to examine the relationship between risk factors and in-hospital mortality.Results
In the multivariable analysis, 11 variables in the aortic model (older age, lower body surface area, prior cardiac operation, elevated creatinine, prior stroke, New York Heart Association [NYHA] class IV, congestive heart failure [CHF], atrial fibrillation, acuity, year of surgery, and concomitant coronary artery bypass grafting) and 10 variables in the mitral model (female sex, older age, diabetes, coronary artery disease, prior cerebrovascular accident, elevated creatinine, NYHA class IV, CHF, acuity, and valve replacement) remained independent predictors of the outcome. The mathematical models were highly significant predictors of the outcome, in-hospital mortality, and the results are in general agreement with those of others. The area under the receiver operating characteristic curve for the aortic model was 0.75 (95% confidence interval [CI], 0.72 to 0.77), and for the mitral model, 0.79 (95% CI, 0.76 to 0.81). The goodness-of-fit statistic for the aortic model was χ2 [8 df] = 11.88, p = 0.157, and for the mitral model it was χ2 [8 df] = 5.45, p = 0.708.Conclusions
We present results and methods for use in day-to-day practice to calculate patient-specific in-hospital mortality after aortic and mitral valve surgery, by the logistic equation for each model or a simple scoring system with a look-up table for mortality rate.