Prediction of all causes of death as a function of some factors commonly measured in cardiovascular population surveys.

A 20-year follow-up study for all causes of death has been conducted on two groups of men aged 40-59 at entry in two rural communities of northern and central Italy on a total of 1,712 subjects representing 98.9% of defined demographic samples. Personal characteristics or risk factors usually studied for coronary heart disease (CHD), as measured at entry, have been related to the risk of dying in 20 years (600 cases). By means of multivariate analysis, 11 out of 33 considered characteristics were shown to be significant predictors of any cause of death. These were mean blood pressure, age, arm circumference (protective), early death of parents, cigarette smoking, forced expiratory volume (protective), CHD, arcus senilis, vital capacity (protective), xanthelasma, and serum cholesterol. Discrimination between cases and noncases was satisfactory with about 40% of deaths in the upper quintile of the estimated distribution of risk and less than 7% in the lowest quintile (relative risk = 6.20).