Association of BMI With Mortality

Many of the observational epidemiologic studies of BMI and mortality have reported a 'U-' or 'J-shaped' relationship between BMI and mortality (28).  Mortality rates are elevated in persons with low BMI (usually below 20) as well as in persons with high BMI (28, 31, 32). In some studies, adjustment for factors that potentially confound the relationship between BMI and mortality, such as smoking status and pre-existing illness, tends to reduce the upturn in mortality rate at low BMIs (31), but in a meta-analysis the higher mortality at low BMIs was not eliminated after adjustment for confounding factors (32). It is unclear whether the elevated mortality observed at low BMIs is due to an artifact of incomplete control for confounding factors (285), inadequate body fat and/or inadequate body protein stores that result from unintentional weight loss (286), or individual genetic factors. Currently, there is no evidence that intentional weight gain in persons with low BMIs will lead to a reduction in mortality.