Body fat by dual photon absorptiometry: Comparisons with traditional methods in Asians, blacks, and whites.

Measurement of body fat by dual photon absorptiometry (FatDPA ) is based on the physical principles of differential attenuation of discrete photons by fat and other soft tissue components. The three traditional methods of measuring fat are indirect, in that they depend on measuring the fat free mass (FFM) based on the assumed constancy of water, potassium, and density in the FFM. Comparison of FatDPA with fat measured by each of the indirect methods discloses systematic differences which shed light on the nature of the constancy assumptions. The purpose of this report is to study the effects of sex, age, and fatness, as well as race, on the traditional methods for estimating body fat%. In Asians (99 m, 109 f), blacks (64 m, 48 f), and whites (166 m, 212 f), 18 to 94 years, with body mass indices of 16 to 36 kg/m2 , body fat measured by DPA was compared with fat by three traditional methods: total body potassium (TBK; 40 K counting), total body water (TBW; HTO dilution), and underwater weighing (UWW). DPA showed the highest overall correlation with TBW and the lowest with UWW, among the three traditional methods. Asians showed the lowest correlations between FatDPA and FatUWW . On average, FatTBK was higher, while FatUWW was lower than FatDPA . The differences between FatDPA and FatOTHER by the three traditional methods were correlated with age with a negative slope in all groups, the steepest slope being for FatDPA - FatTBK in blacks. Correlations with fatness showed positive slopes for FatDPA - FatTBW and - FatUWW but not for - FatTBK . FatDPA could become a criterion method for body fat, particularly in the young, old or ill, where traditional "constants" become unreliable, if calibration can be accomplished throughout the range of subject size. Ethnicity, age, sex, and fatness as well as illness are potent variables affecting the composition of the FFM; direct measurements of fat are preferable, to avoid artifacts resulting from using indirect methods. © 1992 Wiley-Liss, Inc.