Satoshi Seino1, Akihiko Kitamura2, Takumi Abe2, Yu Taniguchi3, Yuri Yokoyama2, Hidenori Amano2, Mariko Nishi2, Yu Nofuji2, Miki Narita2, Tomoko Ikeuchi4, Yoshinori Fujiwara2, Shoji Shinkai5. 1. Research Team for Social Participation and Community Health, Tokyo Metropolitan Institute of Gerontology, Tokyo, Japan. Electronic address: seino@tmig.or.jp. 2. Research Team for Social Participation and Community Health, Tokyo Metropolitan Institute of Gerontology, Tokyo, Japan. 3. Research Team for Social Participation and Community Health, Tokyo Metropolitan Institute of Gerontology, Tokyo, Japan; Center for Health and Environmental Risk Research, National Institute for Environmental Studies, Ibaraki, Japan. 4. Research Team for Human Care, Tokyo Metropolitan Institute of Gerontology, Tokyo, Japan. 5. Research on Social and Human Sciences, Tokyo Metropolitan Institute of Gerontology, Tokyo, Japan.
Abstract
OBJECTIVES: We examined the dose-response relationships of body composition indices with mortality and identified the best predictor. DESIGN AND SETTING: Kusatsu Longitudinal Study and Hatoyama Cohort Study, Japan. PARTICIPANTS: In total, 1977 community-dwelling Japanese adults age ≥65 years (966 men and 1011 women) participated. MEASUREMENTS: Body mass index (BMI), fat mass index (FMI), fat-free mass index (FFMI), and skeletal muscle mass index (SMI) were determined by segmental multifrequency bioelectrical impedance analysis. The main outcome was all-cause mortality. We determined multivariate-adjusted hazard ratios for mortality relative to sex-specific medians of each body composition index and examined the association shapes. RESULTS: During the median follow-up of 5.3 years, 128 (13.3%) men and 75 (7.4%) women died. Compared with median BMIs (23.3 kg/m2 in men and 22.8 kg/m2 in women), a BMI >23.3 and ≤26.1 kg/m2 was associated with significantly lower mortality risk in men, and a BMI <22.8 kg/m2 was associated with significantly higher mortality risk in women. The inverse dose-response relationship with mortality was clearer for FFMI [hazard ratios (95% confidence interval) of 10th and 90th percentiles: 1.58 (1.23-2.03) and 0.58 (0.44-0.79), respectively, in men and 1.56 (1.12-2.16) and 0.68 (0.51-0.91), respectively, in women] and SMI [1.57 (1.22-2.01) and 0.60 (0.45-0.80), respectively, in men and 1.45 (1.05-2.01) and 0.77 (0.61-0.96), respectively, in women] than for BMI [1.30 (0.92-1.83) and 0.65 (0.41-1.03), respectively, in men and 1.87 (1.18-2.95) and 0.88 (0.54-1.42), respectively, in women]. FMI was not associated with mortality in either sex. CONCLUSIONS AND IMPLICATIONS: FFMI and SMI were more definitive predictors of mortality than were BMI and FMI. The lower mortality risk with higher FFMI, regardless of FMI, may explain the age-related weakening of the association between higher BMI and mortality (the "obesity paradox"). FFMI and SMI evaluation should be introduced to clinical assessments of older adults because mortality risk might be reduced by maintaining muscle mass.
OBJECTIVES: We examined the dose-response relationships of body composition indices with mortality and identified the best predictor. DESIGN AND SETTING: Kusatsu Longitudinal Study and Hatoyama Cohort Study, Japan. PARTICIPANTS: In total, 1977 community-dwelling Japanese adults age ≥65 years (966 men and 1011 women) participated. MEASUREMENTS: Body mass index (BMI), fat mass index (FMI), fat-free mass index (FFMI), and skeletal muscle mass index (SMI) were determined by segmental multifrequency bioelectrical impedance analysis. The main outcome was all-cause mortality. We determined multivariate-adjusted hazard ratios for mortality relative to sex-specific medians of each body composition index and examined the association shapes. RESULTS: During the median follow-up of 5.3 years, 128 (13.3%) men and 75 (7.4%) women died. Compared with median BMIs (23.3 kg/m2 in men and 22.8 kg/m2 in women), a BMI >23.3 and ≤26.1 kg/m2 was associated with significantly lower mortality risk in men, and a BMI <22.8 kg/m2 was associated with significantly higher mortality risk in women. The inverse dose-response relationship with mortality was clearer for FFMI [hazard ratios (95% confidence interval) of 10th and 90th percentiles: 1.58 (1.23-2.03) and 0.58 (0.44-0.79), respectively, in men and 1.56 (1.12-2.16) and 0.68 (0.51-0.91), respectively, in women] and SMI [1.57 (1.22-2.01) and 0.60 (0.45-0.80), respectively, in men and 1.45 (1.05-2.01) and 0.77 (0.61-0.96), respectively, in women] than for BMI [1.30 (0.92-1.83) and 0.65 (0.41-1.03), respectively, in men and 1.87 (1.18-2.95) and 0.88 (0.54-1.42), respectively, in women]. FMI was not associated with mortality in either sex. CONCLUSIONS AND IMPLICATIONS: FFMI and SMI were more definitive predictors of mortality than were BMI and FMI. The lower mortality risk with higher FFMI, regardless of FMI, may explain the age-related weakening of the association between higher BMI and mortality (the "obesity paradox"). FFMI and SMI evaluation should be introduced to clinical assessments of older adults because mortality risk might be reduced by maintaining muscle mass.