1995 Volume 61 Issue 5 Pages 249-260
Risk factors of death were assessed from the personal data of the dead and the survivors, total 2, 934 subjects, who had undergone a public mass adult health examination ten years ago in the three villages in Kanto area in Japan. Age of the subjects at the time of the health examination was 40 years or more and averaged early fifties. Number of cases observed are 1, 299 males and 1, 705 females. One hundred and seven males (8.7%) and 73 females (4.3%) died during the ten years of observation period. Cox's proportional hazard model was applied to the data, which revealed the three factors: age at the health examination, systolic blood pressure, and smoking habit, were significantly related to the mortality risk of males. The hazard ratio increased to 1.7 at 110 mmHg of systolic blood pressure or more against 1 .0 at average blood pressure level. The BMI showed no significant relation to the mortality risk, though the lower BMI showed rather the higher mortality risk. When the cancer death in male was picked up, the hazard ratio was related significantly to the age at the health examination, drinking, and smoking habits. Hazard ratio of the smoker was 11.3 or over compared to the non-smoker, whereas that of the drinker was 3.3 or over compared to the non-drinker. The mortality risk of all causes of death in females was significantly related to the age at health examination, diastolic blood pressure and the BMI. The hazard ratio increased to 2.4 or more at 110 mmHg of systolic blood pressure or more against 1.0 at average blood pressure level. The lower the BMI, the higher was the mortality risk of females. The highest risk factor was the age at the health examination, in the three analysis: total cause of deaths in male and female, cancer deaths in male. Estimation of the death probability for the 10 years from the same data using multiple logistic model revealed similar results to those by Cox's proportional hazard model. The range of the estimated risk of death of the three death groups was classified into ten classes and calculated the number of the expected deaths, which were closely related to the observed number of deaths. The highest risk class of all causes of death in male produced 24% of death rate, and that of male cancer deaths and female all causes of death produced both 13% of death rate.