[A prospective study on the relationship between body mass index and mortality from all causes in 15,827 middle aged...

The association between Body Mass Index (BMI) and mortality from all causes was investigated in a cohort study of 15827 citizens (aged 35 and over) in Shanghai. During about 3-year follow up the person-year mortality for men and women were 1311.78/100000 and 1036.87/100000, respectively. Cox regression analysis was conducted in a lower and an upper range of BMI, respectively. The inverse association between BMI and mortality in the lower range was statistically significant for men and women. The positive association between BMI and mortality in the upper range was highly statistically significant for women but not for men. Logistic regression analysis conducted over the entire range of BMI showed that BMI and all-cause mortality association was significantly "L"-shaped distribution for men and women. The mortality or relative hazard curve can be fit with quadratic model (Y = a-bX + cX2) in different sex.