Body mass index and measures of body fat for defining obesity and underweight: a cross-sectional, population-based study

BACKGROUND The body mass index (BMI) is commonly used as a surrogate marker for adiposity. However, the BMI indicates weight-for-height without considering differences in body composition and the contribution of body fat to overall body weight. The aim of this cross-sectional study was to identify sex-and-age-specific values for percentage body fat (%BF), measured using whole body dual energy x-ray absorptiometry (DXA), that correspond to BMI 18.5 kg/m(2) (threshold for underweight), 25.0 kg/m(2) (overweight) and 30.0 kg/m(2) (obesity) and compare the prevalence of underweight, overweight and obesity in the adult white Australian population using these BMI thresholds and equivalent values for %BF. These analyses utilise data from randomly-selected men (n = 1446) and women (n = 1045), age 20-96 years, who had concurrent anthropometry and DXA assessments as part of the Geelong Osteoporosis Study, 2001-2008. RESULTS Values for %BF cut-points for underweight, overweight and obesity were predicted from sex, age and BMI. Using these cut-points, the age-standardised prevalence among men for underweight was 3.1% (95% CI 2.1, 4.1), overweight 40.4% (95% CI 37.7, 43.1) and obesity 24.7% (95% CI 22.2, 27.1); among women, prevalence for underweight was 3.8% (95% CI 2.6, 5.0), overweight 32.3% (95% CI 29.5, 35.2) and obesity 29.5% (95% CI 26.7, 32.3). Prevalence estimates using BMI criteria for men were: underweight 0.6% (95% CI 0.2, 1.1), overweight 45.5% (95% CI 42.7, 48.2) and obesity 19.7% (95% CI 17.5, 21.9); and for women, underweight 1.4% (95% CI 0.7, 2.0), overweight 30.3% (95% CI 27.5, 33.1) and obesity 28.2% (95% CI 25.4, 31.0). CONCLUSIONS Utilising a single BMI threshold may underestimate the true extent of obesity in the white population, particularly among men. Similarly, the BMI underestimates the prevalence of underweight, suggesting that this body build is apparent in the population, albeit at a low prevalence. Optimal thresholds for defining underweight and obesity will ultimately depend on risk assessment for impaired health and early mortality.