Abstract Background Multi-frequency bioelectrical impedance analysis (MFBIA) methods offer reliable and moderately accurate estimates of body composition in tightly controlled conditions (prandial and hydration status, recent exercise, time of day). Objective This study examined MFBIA reliability and validity in a real-world environment where these factors were not controlled. Methods Regional and total body composition estimates by MFBIA (InBody 770) were compared to dual-energy X-ray absorptiometry (DXA) in 1000 healthy adults (667 men; 333 women), including fat mass (FM), percent body fat (%BF), fat-free mass (FFM), and visceral adipose tissue (VAT). In subsets, reliability was determined from duplicate MFBIA and DXA obtained within 1 week, and total body water (TBW) was compared to single-frequency BIA (SFBIA). Results MFBIA demonstrated modest population-level agreement with DXA for total body FM (men, r = 0.93, bias −3.7 ± 2.6 kg; women, r = 0.96, bias, −1.9 ± 1.8 kg), %BF (men, r = 0.89, bias, −4.2 ± 3.0%; women, r = 0.92, bias, −2.8 ± 2.6%), and FFM (men, r = 0.95, bias, 3.4 ± 2.8 kg; women, r = 0.94, bias, 2.0 ± 2.2 kg). Regional correlations were highest for trunk FM (men, r = 0.92, CCC = 0.86; women r = 0.93, CCC = 0.93) and lowest for VAT (men, r = 0.74, CCC = 0.68; women, r = 0.74, CCC = 0.34). DXA and MFBIA regional and total assessments were highly reliable (DXA, ICC 0.990–0.998) and (MFBIA, ICC 0.987–0.995). TBW by MFBIA and SFBIA showed moderate agreement (men, r = 0.73, bias, −1.89 ± 3.31; women, r = 0.82, bias, −1.74 ± 2.01). Conclusion This MFBIA system was shown to have high retest reliability and, when compared to laboratory methods, provides a moderately accurate method for measuring TBW and body composition (except for VAT) in uncontrolled conditions.
Potter et al. (Tue,) studied this question.