Coronary risk factors and survival probability from coronary and other causes of death.

The analysis concerns the two rural Italian cohorts of the Seven Countries Study and includes 1,712 men who, at the entry examination in 1960, were aged 40-59 years and whose 20-year follow-up examinations were complete for life status, dates, and causes of death. Excluded were 175 men because they lacked one or more of the risk factors selected for the study. A total of 517 deaths occurred in the remaining 1,537 men. The Cox proportional hazards regression model was applied with a forward procedure to include the entry risk factors in the model. First, total mortality was used as the endpoint. Then the risk factors identified as being related to total mortality were used to predict specific causes of death, i.e., coronary heart disease, stroke, cancer, violent death, and other causes. Blood pressure appeared as a nonspecific risk factor for all causes, including cancer and violence, forced expiratory volume and arm circumference appeared as nonspecific risk factors, smoking habits could not be easily classified, and cholesterol was definitely specific only for coronary deaths. Models for estimating survival probabilities from total mortality and from any of several causes of death are provided.