Background: Within the past 10 y, dual-energy X-ray absorptiometry (DXA) has become one of the most widely used methods of measuring human body composition. However, DXA has not been fully evaluated against an independent criterion method of measuring body fatness in young females.
Objective: Our objective was to determine the bias and agreement between DXA and a 4-compartment model in predicting the percentage of fat mass (%FM) in a multiethnic group of young females.
Design: The %FM values measured by DXA of 73 white, 43 African American, 14 Hispanic, and 11 Asian females with a mean (+/- SD) age of 13.0 +/- 1.9 y were compared with the 4-compartment values, which were based on measurements of body density, body water, and bone mineral content.
Results: The %FM values measured by the 2 methods were correlated at r = 0.90 with an SEE of 3.3%; Bland-Altman analysis indicated an average bias of 3.9%. After nullification of the average bias, an individual estimate of %FM by DXA could be underestimated or overestimated by 6.7% when compared with the 4-compartment value.
Conclusions: DXA is an appropriate method for estimating body composition in a group of young females because its bias and limits of agreement are independent of age, ethnicity, and body fatness. However, the limits of agreement of 6.7% could cause an individual FM value to be underestimated or overestimated by 28% relative to the 4-compartment value. Therefore, DXA may not be the optimal method of measuring the body fatness of young females.