Validation of a Multielectrode Bioelectrical Impedance Analyzer With a Dual-Energy X-Ray Absorptiometer for the Assessment of Body Composition in Older Adults

• Published in: Journal of aging and physical activity 2020-08, Vol.28 (4), p.1-604
• Main Authors: Meier, Nathan F, Bai, Yang, Wang, Chong, Lee, Duck-Chul
• Format: Article
• Language: English
• Subjects: Bioelectrical impedance; Body composition; Body fat; Body mass index; Cancer; Measurement; Mobility; Older people; Older women; Sarcopenia; X-rays
• ISSN: 1063-8652; 1543-267X
• DOI: 10.1123/JAPA.2019-0211

Changes in body composition are related to mobility, fall risk, and mortality, especially in older adults. Various devices and methods exist to measure body composition, but bioelectrical impedance analysis (BIA) has several advantages. The purpose of this study was to validate a common BIA device with a dual-energy X-ray absorptiometer (DXA) in older adults and develop prediction equations to improve the accuracy of the BIA measurements. The participants were 277 older adults (162 women and 115 men; age 73.9 ± 5.8 years) without a history of cancer and without a history of severe medical or mental conditions. Individuals fasted 12 hr before BIA and DXA measurement. The correlations between the two methods for appendicular lean mass (ALM), fat-free mass (FFM), and percentage body fat (%BF) were .86, .93, and .92, respectively, adjusting for age and sex. The mean percentage error (DXA-InBody) and mean absolute percentage error were -12% and 13% for ALM, -13% and 13% for FFM, and 16% and 17% for %BF. The prediction equations estimated ALM, FFM, and %BF; sex was coded as 1 for male and 0 for female: DXAALM=0.0673+(0.6732×BIAALM)+(2.33507×sex)+(0.13349×BMI),R2=.94; DXAFFM=0.72323+(0.72384×BIAFFM)+(3.675012×sex)+(0.2816×BMI),R2=.97; and DXA%BF=15.8896+(0.64694×BIA%BF) -(3.99945×sex)+(0.13824×BMI),R2=.91 Although highly correlated, BIA overestimated FFM, and ALM and underestimated %BF compared with DXA. An application of prediction equations eliminated the mean error and reduced the range of individual error across the sample. Prediction equations may improve BIA accuracy sufficiently to substitute for DXA in some cases.