BACKGROUND AND AIMS: The relationship between body mass index (BMI) and mortality remains inconclusive in seniors. This study aimed at assessing this relationship in a community-dwelling elderly population in Japan. METHODS: The subjects were 371 Japanese elders, 65 years old and older, who lived in a geographically well-defined rural community and had participated in a general health screening program in 1995. Both height and weight of subjects were measured directly by medical staff. Subjects were classified into three groups according to their BMI values: low, <18.5; normal, 18.5-25.0; and high, >25.0. Univariate analysis was applied to explore potential associations between mortality and possible confounders. Multivariate Cox proportional hazards models were used to analyze the association between mortality and BMI, after adjusting for other risk factors. RESULTS: At baseline, 54 subjects (14.6%) had BMI values in the low range, 280 (75.5%) in the normal range, and 37 (10.0%) in the high range. All 371 subjects were followed prospectively for mortality. Over the next five years, 37 subjects had died. In univariate analysis, male sex, age, BMI and serum creatinine were associated with mortality. The mortality rate in the low BMI group was about twice that in the normal BMI group. No deaths were observed in the higher BMI group. In multivariate analysis, age and low BMI were associated with mortality. CONCLUSIONS: BMI may be a useful predictor of mortality among seniors living in the general, non-institutionalized population.
BACKGROUND AND AIMS: The relationship between body mass index (BMI) and mortality remains inconclusive in seniors. This study aimed at assessing this relationship in a community-dwelling elderly population in Japan. METHODS: The subjects were 371 Japanese elders, 65 years old and older, who lived in a geographically well-defined rural community and had participated in a general health screening program in 1995. Both height and weight of subjects were measured directly by medical staff. Subjects were classified into three groups according to their BMI values: low, <18.5; normal, 18.5-25.0; and high, >25.0. Univariate analysis was applied to explore potential associations between mortality and possible confounders. Multivariate Cox proportional hazards models were used to analyze the association between mortality and BMI, after adjusting for other risk factors. RESULTS: At baseline, 54 subjects (14.6%) had BMI values in the low range, 280 (75.5%) in the normal range, and 37 (10.0%) in the high range. All 371 subjects were followed prospectively for mortality. Over the next five years, 37 subjects had died. In univariate analysis, male sex, age, BMI and serum creatinine were associated with mortality. The mortality rate in the low BMI group was about twice that in the normal BMI group. No deaths were observed in the higher BMI group. In multivariate analysis, age and low BMI were associated with mortality. CONCLUSIONS: BMI may be a useful predictor of mortality among seniors living in the general, non-institutionalized population.
Authors: Hiroshi Murayama; Jersey Liang; Joan M Bennett; Benjamin A Shaw; Anda Botoseneanu; Erika Kobayashi; Taro Fukaya; Shoji Shinkai Journal: Am J Epidemiol Date: 2015-09-12 Impact factor: 4.897