Regression analysis of late survival following mitral valve replacement
Grunkemeier, G.L.; Macmanus, Q.; Thomas, D.R.; Starr, A.
Journal of Thoracic and Cardiovascular Surgery 75(1): 131-137
1978
ISSN/ISBN: 0022-5223 PMID: 619175 Document Number: 128340
This series of patients with the Starr-Edwards Model 6000 prosthesis extends back to 1960 and includes the first to have successful mitral valve replacement. Its position as the longest mitral replacement series, and the consequent high percentage of late deaths, makes this series a rich statistical resource for the analysis of factors affecting late survival. Preoperative clinical and catheterization variables (17) were examined for their influence on late survival. The operative survivors were divided into groups on the basis of each preoperative variable, and the survival curves were compared statistically. Only 2 variables (age and aortic stenosis) produced a difference in late survival which was significant at the 5% level. To assess the influence of several preoperative variables acting simultaneously and to provide a survival projection for an individual on the basis of the composite profile of his preoperative characteristics, a multiple regression approach to survival estimation was utilized. The variables which remained in the regression equation (at the 5% level) were age, prior heart surgery, aortic regurgitation and cardiac index. The survival curves corresponding to patients possessing various combinations of values of these 4 variables were constructed, with a range of survival at 15 yr from 78-5%, compared to 51% for the group as a whole.