The Annals of Thoracic Surgery, Vol 30, 240-246, Copyright © 1980 by The Society of Thoracic Surgeons
The use of time-interrelated covariates to predict survival following aortic valve replacement
GL Grunkemeier, Q Macmanus, DR Thomas, JM Luber, LE Lambert, YF Suen and A Starr
One hundred four patients survived isolated aortic valve replacement with
the model 1200 prosthesis between 1965 and 1968, with a 12-year survival of
64%. Multiple regression survival analysis was employed in an attempt to
determine which of 26 preoperative variables affected late survival and to
devise a formula to predict survival for a given individual. The most
important variables in the regression equation were right atrial mean
pressure, etiology, and sex. The effect of the last two were found to vary
with time over the 12-year post-operative period. An extension of the
standard regression analysis technique was developed to incorporate
time-related cofactors into the model. Based on the multiple regression
model, 12-year survival was estimated to range from 92% to 14% for the best
and worst combinations, respectively, of the three significant variables.
The advantages of the regression method are outlined and the findings of
other studies with regard to factors affecting survival after aortic valve
replacement are summarized and discussed.
