Ethnic bias in anthropometric estimates of DXA abdominal fat: the TIGER study.

BACKGROUND/INTRODUCTION The purpose of this study was to examine the race/ethnicity bias of using waist circumference (WC) to estimate abdominal fat. METHODS A total of 771 females and 484 males (17-35 yr) were tested one to three times during a prescribed 30-wk aerobic exercise program. The race/ethnicity distribution for women was non-Hispanic white, 29%; Hispanic, 25%; African American (AA), 35%; Asian Indian, 3%; and Asian, 8%. The distribution for men was non-Hispanic white, 37%; Hispanic, 26%; AA, 22%; Asian Indian, 5%; and Asian, 10%. Abdominal fat (L1-L5) was estimated from whole-body scanning using dual-energy x-ray absorptiometry (DXA Abd-Fat). RESULTS DXA Abd-Fat varied by race/ethnicity after accounting for WC and height in both women and men. The increase in DXA Abd-Fat per increase in WC was lower in the Asian and Asian-Indian women than that in the other women. The increase in DXA Abd-Fat per increase in WC was higher in the AA men and lower in the Asian-Indian men than that in the other men. These differential race/ethnicity effects were most notable when WC exceeded ≍90 cm in the women and ≍100 cm in the men, values which are consistent with current definitions of abdominal obesity in the United States. CONCLUSIONS Prediction equations for abdominal fat using WC that do not account for race/ethnicity group provide biased estimates. These results may affect assessment of disease risk from abdominal obesity among racial/ethnic groups.