Body mass index and all-cause mortality in a nationwide cohort of U.S. women and men

Abstract

This prospective occupation-based cohort study was designed to assess the effects of body mass index (BMI) on mortality by gender and age. Using Cox proportional hazards regression models with age as the time-scale, stratified at baseline for birth cohorts in 5-year intervals, we evaluated mortality risk associated with BMI (based on self-reported weight and height) in 64,733 female and 19,011 male radiologic technologists who responded to a questionnaire administered in 1983 to 1989 and who were free of cancer and myocardial infarction at that time. During an average of 14.7 years of follow-up or 1.25 million person-years, there were 2278 female and 1495 male deaths. The analysis focused on never-smoking subjects, excluding the first 5 years of follow-up, to limit the potential confounding effect of low BMI associated with smoking and illness-related weight loss. Risks were assessed in male and female subjects less than 55 years old and 55 or older. Mortality risks were highest in both women and men in the two age groups (<55 and ≥55 years) at a BMI greater than 30.0. In never-smoking women under age 55 years (excluding an initial follow-up period), risk rose as BMI increased above 21.0, whereas in older women, risk increased beginning at a higher BMI (≥25.0). Among younger men who never smoked (also excluding an initial follow-up period), risk began to rise with a BMI above 23.0; in older men, risk did not begin to rise until BMI exceeded 30.0. In both women and men, all-cause mortality risks appear to increase at a lower BMI in younger than in older adults.