1986 Volume 23 Issue 2 Pages 163-171
In 1965, food frequency interview inquriing about usual intake, in terms of the frequency of consumption of various food items, was carried out in Hisayama, Fukuoka Prefecture, a rural area located in the south-western part of Japan. Data concerning frequency of taking rice, meat, fish, milk, green vegitable, “tukemono” (pickled vegetables), “misoshiru” (miso soup), cigarette smoking and alcohol drinking were obtained from 1, 724 (94 per cent) residents, aged 40 to 69 years. Long-term follow-up was completed for 1, 609 (726 men and 883 women) at the time of health examination performed almost every couple of years in the community and it was confirmed that 309 (181 men and 128 women) out of 1, 609 subjects died during the subsequent 15 year follow-up period. Survival rate for the women was nearly 10 per cent higher than that for the men after the follow-up period (chisquare=30.12, P=0.0001). To evaluate effects of frequency of foods, alcohol and tabacco on total mortality, analyses using Cox's proportional hazard model were executed with a computer software PHGLM in SAS. After controlling for age at the baseline in 1965 and sex, univariate analysis showed that meat, “tukemono” and cigarette smoking were significant (p<0.05) for mortality. When all of the food items were taken into the Cox's model, multivariate analysis showed that for the whole subjects, advancing age, male sex, cigarette smoking and “tukemono” were significant (p<0.05) for mortality, and that advancing age, cigarette smoking, low intake of “tukemono” and meat for the men and advancing age alone for the women. Blood pressure, body height, weight, serum protein and serum total cholesterol were determined for 1, 378 subjects at the time of examination in the summer of 1965, independent of the food frequency interview. After controlling for age and sex, univariate analysis showed that systolic and diastolic blood pressure were selected as significant (p<0.01) for mortality. When all of these data from 1, 378 subjects were added into the multivariate model, advancing age, elevated systolic blood pressure, cigarette smoking and male sex were selected as significant (p<0.05). More detailed information about incidence of cerebrovascular diseases (CVD) or malignant neoplasms was obtained from the subjects who were followed-up since 1961 as cohort of Hisayama Study, an ongoing longitudinal epidemiological study. As a result, it was confirmed that 111 out of 1, 111 subjects who belonged to the Hisayama cohort had newly developed CVD and other 59 died from malignant neoplasms verified by autopsy. Analyses based on these cases showed that elevated systolic blood pressure, male sex and advancing age were selected as significant (p<0.05) for development of CVD and cigarette smoking for mortality from malignant neoplasms. As the questionnaire did not record precisely what the respondent had eaten but revealed their usual pattern of consumption, it was strongly suggested that alcohol drinking precipitates development of CVD through the previously established effects of elevated blood pressure and that there is a highly close relationship between cigarette smoking and mortality from malignant neoplasms.
