Survival rates were determined for a group of 598 patients in whom severe coronary disease was demonstrated by arteriography; initially they were treated medically and were followed-up for 15 years. Deaths due to noncoronary causes were uncommon (5% of total) in the first 5 year period but were frequent (36%) in the third period. Survival No. 5, 986-997, 1983. BECAUSE OF PROBLEMS in the study of the natural history of coronary disease, many of which are insurmountable, no completely satisfactory investigation of survival is possible. The present objective is to update studies in a large group of patients proved to have severe coronary disease by arteriography and to examine the association between clinical and invasively obtained variables and survival.
MethodsThe methods and terms used in the study have been described. 1-3 At 15 year follow-up, three of 601 patients who had severe coronary disease were excluded from the series on the basis of the original criteria for selection: two were found to have had poudrage operations less than 1 year after catheterization, and one had had a Vineberg operation at 3 years. One of the three had single-artery disease, and two had involvement of two arteries; two had functional class II angina, and the other had experienced prolonged attacks of pain. Two additional patients had had bypass operations at 85 and 88 months, so these were considered dropouts at those dates instead of 10 year survivors. Fifteen year follow-up was obtained in all of the remaining 598 patients except for one, a criminal who was incarcerated at the time of the previous follow-up. Cardiac survival rates have been reported, although total survival rates (all deaths counted) were studied for some subsets.' Total survival is used in the present study, with a few exceptions specifically noted. Patients were considered dropouts for actuarial survival studies at the middle of the year in which coronary operation occurred. For cardiac survival, deaths due to noncardiac causes were considered dropouts in the year of occurrence. Survival rates were calculated by the actuarial method.5 6 Equality of survival curves was tested by a nonparametric rank test, the Breslow statistic.' Multivariate statistical analyses used the Cox proportional hazards model, permitting a stepwise selection of covariates significantly associated with survival; at each step the removal criterion was p > .15, and the entry criterion was p < .10.8 9 The test statistic was adjusted for variables that had already been entered in the model. Details of statistical methods will be supplied on request.Sixty of the original 598 patients had operations for coronary disease during the period of the study.
ResultsDuring the first 5 years, 95% of all deaths were due to coronary disease, but the percentages were 84% and