Impact of body mass index and body weight variabilities on mortality: a nationwide cohort study

Background/objectives There is limited information regarding the impact of body mass index (BMI) and body weight (BWt) variabilities on mortality. This study aimed to investigate the association between BMI, BWt variabilities and subsequent mortality in the Korean population. Subjects/methods This study used a representative sample cohort enrolled in the national health examination program conducted by the Korean National Health Insurance Service, and 125,391 individuals were included and followed up until 2013 (mean follow-up period = 84 months). BMI and BWt variabilities were estimated as the standard deviation (SD) and coefficient of variation (CV) of serial measurements of BMI and BWt (BMI_SD, BWt_SD, BMI_CV, and BWt_CV). Cox proportional hazard regression models were used to evaluate the all-cause and cause-specific mortality according to variability indices. Results Baseline BMI showed a non-linear association with all-cause mortality. The highest quartile (Q4) groups of variability indices were associated with increased all-cause mortality risk compared to the lowest quartile (Q1) groups after adjusting for confounding factors (hazard ratio [95% confidence interval] = 1.23 [1.11-1.37] for BMI_SD; 1.25 [1.06-1.47] for BMI_CV; 1.33 [1.20-1.48] for BWt_SD; 1.42 [1.28-1.58] for BWt_CV). The hazard ratios of all-cause mortality increased from the Q2 to the Q4 groups of variability indices, in each sex group and among individuals aged >= 40 years. Furthermore, Q4 groups of variability indices were positively associated with cause-specific mortality compared to groups with Q1-Q3 of the indices (1.28 [1.05-1.56] for BWt_SD, 1.21 [1.001-1.47] for BMI_CV, 1.29 [1.06-1.56] for BWt_CV regarding cardiovascular diseases mortality; 1.18 [1.03-1.36] for BWt_SD, 1.21 [1.06-1.39] for BMI_CV, 1.26 [1.10-1.44] for BWt_CV regarding cancer mortality). Conclusions Our results suggest that BMI and BWt variabilities are independent risk factors for all-cause and cause-specific mortality.