Body mass index, waist girth, and waist-to-hip ratio as indexes of total and regional adiposity in women: evaluation using receiver operating characteristic curves

Abstract

Receiver operating characteristic (ROC) curves were constructed to assess the value of body mass index (BMI) as a screening measure for total adiposity and to examine waist-to-hip ratio (WHR) and waist circumference as measures of central fat distribution. Body fat reference measurements were determined by dual-energy X-ray absorptiometry (DXA). The study population comprised 96 healthy white women aged 16-80 y. A positive reference test was defined as a result at or above the 75th percentile for our study population for all DXA measurements. Sensitivity and specificity were calculated at several percentile cutoffs for BMI, WHR, and waist girth. The areas under the ROC curves were calculated to compare the relative ability of each anthropometric technique to correctly classify subjects according to the reference measurement for that technique. BMI (our 75th percentile = 27.3) performed well as a screening measure of total adiposity, correctly identifying 83% of subjects with a high body fat mass while misclassifying only eight subjects [four false-negatives (subjects with high fat mass who were in the low BMI category) and four false-positives (subjects with a low fat mass who were in the high BMI category)]. The screening performance of WHR (our 75th percentile = 0.81) was lower, accurately categorizing 58% of subjects while misclassifying 28 subjects. By contrast, waist circumference (our 75th percentile = 86.9 cm) was significantly better than WHR at screening for regional fat distribution, accurately classifying 83% of subjects and misclassifying eight subjects (P < 0.05). We conclude that BMI and waist circumference provide simple yet sensitive methods for the estimation of total and central adiposity in groups of adult women.