Page Banner

United States Department of Agriculture

Agricultural Research Service

Title: Accuracy of BMI to detect percent fat obesity in men and women, ages 17 to 39: The TIGER Study

item Jackson, Andrew
item Ellis, Ken
item Mcfarlin, Brian
item Sailors, Mary
item Turpin, Ian
item Bray, Molly

Submitted to: Medicine and Science in Sports and Exercise
Publication Type: Abstract Only
Publication Acceptance Date: 10/3/2006
Publication Date: 11/7/2006
Citation: Jackson, A.S., Ellis, K.J., McFarlin, B.K., Sailors, M.H., Turpin, I., Bray, M.S. 2006. Accuracy of BMI to detect percent fat obesity in men and women, ages 17 to 39: The TIGER Study [abstract]. Medicine and Science in Sports and Exercise. 38:S311.

Interpretive Summary:

Technical Abstract: A BMI of 30 kg/m2 is used to define obesity of men and women of all ages. Using variable samples, investigators have shown that age and gender account for percent fat (%Fat) variance independent of BMI. This age and gender bias can be traced to the inability of BMI to distinguish between the body's fat and fat-free weight components. PURPOSE: The purpose of this study was to determine the accuracy of BMI to detect DXA percent fat (%Fat) obesity in young men and women. METHODS: The subjects were 278 women (Mean age = 22.5 +/- 4.4) and 141 men (Mean age = 23.6 +/- 4.2) representative of an ethnically diverse population of college students. Percent body fat was determined with DXA. The gender-specific DXA %Fat obesity cut scores were > 25% and > 33% for men and women, respectively. Receiver operating curves (ROC) were used to examine test sensitivities and associated false positive rates (1 - specificity) for BMI cut points. RESULTS: The body composition characteristics were: women, BMI (Mean = 26.7 +/- 6.1), DXA %Fat (Mean = 33.6 +/- 7.4); and men, BMI (Mean = 28.1 +/- 5.8), DXA %Fat (Mean = 22.7 +/- 7.9). The proportion of subjects found to be obese by each method were: women, BMI 24%, DXA 53%; and men, BMI 30%, DXA 35%. The ROC analyses produced similar curves for women and men, the area under the curves were 91% and 86% respectively. Test sensitivities for a BMI of 30 kg/m2 were low, 45% and 66% for women and men, respectively. The associated false positive rates were: women, 1%; and men, 10%. Lowering the BMI cut point to 25 kg/m2 increased test sensitivity to 81% for women and 92% for men. The false positive rates for these test sensitivities were 13% and 55% respectively. Using a men's BMI cut of 28 kg/m2 lowered test sensitivity to 80% with a false positive rate to 27%. CONCLUSIONS: These results demonstrated that the accepted BMI obesity cut of 30 kg/m2 did not provide the most accurate estimates of %Fat-determined obesity. A women's BMI cut of 25 kg/m2 and men's cut of 28 kg/m2 provided more accurate estimates of DXA %Fat obesity in this population of young adults. The ROC analyses showed that selecting a BMI test Sen >80% markedly increased the rate of false positive tests.

Last Modified: 10/19/2017
Footer Content Back to Top of Page