Comparison of DEXA-derived body fat measurement to two race-specific bioelectrical impedance equations in healthy Indians

Aim: Prevalence of obesity is increasing in Asian Indians. Reliable, precise and convenient methods to estimate body composition are required. This study aimed to test the accuracy of bioelectrical impedance analysis (BIA) estimates of body composition among Asian Indians according to two BIA equations--one developed for Asians, the other for Caucasians.

Methods: Two hundred apparently healthy Asian Indians (100 males, 100 females; mean age 36.6 ± 7.6 years; mean BMI 16.6-46.7 kg/m(2)) underwent BIA assessment of fat mass (FM), fat free mass (FFM) and percentage body fat (%BF) using Tanita Multi-Frequency Body Composition Analyzer MC-180MA (Tanita Corporation, Tokyo, Japan). One set of BIA values was automatically calculated by the analyzer (Caucasian figures), the other set by Tanita (Japanese Asian figures). Results were compared to dual X-ray absorptiometry (DEXA) as the standard measure.

Results: A moderate level of relative agreement was found between the DEXA-derived measurement of %BF and the estimate from both the Caucasian (r(2)=0.75; p<0.001) and Asian equation (r(2)=0.7; p<0.001). Despite this, the level of absolute agreement was poor, with large bias and wide limits of agreement. According to the Caucasian equation the mean difference between methods was -8.3 ± 3.9 (95% limits of agreement -20.10 to 9.40), for the Asian equations mean difference was -5.4 ± 4.3 (95% limits of agreement -20.63 to 11.41).

Interpretation and conclusions: When compared to DEXA, the current Asian and Caucasian Tanita formulae significantly under-estimate the %BF of Asian Indians.