1 Faculty of Kinesiology, University of Calgary, Calgary, AB, Canada.
2 INDI Department, Concordia University, Montreal, QC, Canada.
3 Hospital Research Center of Montreal University (CRCHUM), Montreal, QC, Canada.
4 PERFORM Center, Concordia University, Montreal, QC, Canada.
5 Department of Mathematics and Statistics, Concordia University, Montreal, QC, Canada.
6 PERFORM Center, Concordia University, Montreal, QC, Canada - lisa.kakinami@concordia.ca.

Background: Assessment of percent body fat (%BF) is typically measured with bioelectrical impedance (BIA) as a proxy for dual-energy X-ray absorptiometry (DXA). Notably, poorer agreement between BIA and DXA among persons who are overweight or obese has been reported. The use of electrical impedance myography (EIM) as a proxy for DXA has not been validated. The objective was to evaluate an EIM device and two multi-frequency BIA devices with the reference standard (DXA) stratified by weight status and gender.

Methods: In a convenience sample of 82 adults, %BF assessed by EIM and two BIA devices was compared to DXA. Agreement between devices was tested with intra-class correlation coefficients (ICC) and Bland-Altman plots.

Results: Agreement between DXA and EIM (ICC=0.77) was poorer than the agreement between either BIA device with DXA (ICC>0.87). Stratified by sex, agreement between EIM and DXA was greater for men than women (ICC=0.81 and ICC=0.61, respectively). Stratified by BMI, agreement between EIM and DXA was best for normal-weight individuals (ICC=0.89) and progressively poorer for overweight (ICC=0.80) and obese (ICC=0.67) individuals. Bland-Altman plots revealed wide limits of agreement and an increase in EIM mean difference as average %BF increased. Similar trends were seen in BIA assessments.

Conclusions: EIM and BIA substantially underestimate %BF in overweight and obese individuals. Wide limits of agreement coupled with variable ICC limit device interchangeability with one another and limit clinical utility.