Predicting all causes mortality in the general population as a function of blood pressure and smoking habits measured in population samples: a pilot study. Research Group ATS-RF2.

Nine population samples of mean aged 40-59, for a total of 1777 individuals drawn from nine health units, were examined in 1978-79 for the measurement of some risk factors possibly related to total mortality. The analysis evaluated the possible relationship between blood pressure levels and smoking habits measured in the population samples and the official death rates covering the period 1980-82 in the health units from which the samples were drawn. The analysis was conducted in a direct way but also by the help of a risk function linking blood pressure and smoking habits as obtained from another Italian population sample. The cograduation test between mean blood pressure in the nine areas and total mortality was 0.58 (linear correlation 0.57); the cograduation test was -0.70 for non-smokers, 0.62 for ex-smokers, 0.50 for present smokers (linear correlation of -0.27, 0.29 and 0.14 respectively). The death rates estimated by the risk function (Cox model) tended to overestimate the death rates from 5 to 64% (average 36.3%). The cograduation test between expected and observed death rates was 0.56 (linear correlation coefficient 0.66). Some limitations of this pilot study are largely explained by the small size of the sample and by the need to use a risk function provided by another population sample.