[Social integration and mortality in Japanese elderly].

The prospective association of social integration reported in 1987 with mortality over three years was examined in a cohort of 2200 Japanese elderly, aged 60 years and over. Social integration data was obtained from a national survey which measured both the quantity and frequency of attachment. Social ties was classified into five types; spouse, child(ren) living together, child(ren) living separately, friend, neighbor, and community group. Ten potential confounding variables-age, educational attainment, address, existence of chronic diseases, activities of daily living, subjective well-being, drinking, sleeping hours, cigarette smoking, and exercise- were controlled using logistic regression analysis for each sex. Men who belong to more community organizations or attend meeting of these groups more frequently had significantly less mortality during the follow-up period. Trends for women were not significant. No association was observed between mortality and other variables of social integration.