Validation of a three-frequency bioimpedance spectroscopic method for body composition analysis

Objective We assessed whether whole-body multiple frequency impedance (MFBIA) data obtained at a few discrete frequencies could be used to estimate accurately resistance at 0 (R0) and infinite (R∞) frequencies required for prediction of body composition by mixture theory. Methods Fat-free mass (FFM) was measured in 157 subjects (77 males, 80 females; body mass index [BMI] 17.8–41.7 kg/m2) by dual X-ray absorptiometry (DXA). Whole-body impedance was measured and R0 and R∞ were calculated by three different methods. FFM predicted using the different values of R0 and R∞ were compared with each other and with the reference DXA values for all subjects stratified according to BMI band (BMI <24.9 kg/m2, normal weight; BMI 25–29.9 kg/m2, overweight; BMI >30 kg/m2, obese). Results All BIA procedures predicted an FFM that was slightly but significantly different from DXA-derived values, underestimating by 0.24 to 1.4 kg in the normal-weight subjects and overestimating by 5.3 to 7.1 kg in the obese subjects. Although statistically significant, the different impedance procedures were highly correlated (r > 0.98), with small limits of agreement (approximately ±2%) when used to predict FFM. Predictive power was associated with BMI, worsening as BMI increased. Conclusion MFBIA can be used to estimate impedance parameters required for mixture theory prediction of body composition, but this approach requires adjustment for BMI to be accurate.