Logo image
Back
Precision, least significant change and validity of body composition estimates from bioelectrical impedance analysis, 3D-optical scanning, and dual X-ray absorptiometry
Journal article   Open access   Peer reviewed

Precision, least significant change and validity of body composition estimates from bioelectrical impedance analysis, 3D-optical scanning, and dual X-ray absorptiometry

Chris Oliver, Luke Del Vecchio, Mike Climstein, Nedeljka Rosic, Shelley Robinson and Stephen Myers
Clinical nutrition ESPEN, Vol.First online
07/02/2026
DOI: https://doi.org/10.1016/j.clnesp.2026.102959
PMID: 41662990
url
Precision, least significant change and validityView
Published (Version of record) Open CC BY V4.0

Related links

Metrics

13 Record Views

Abstract

Bland-Altman analysis clinical decision-making risk assessment
Background: Accurate assessment of body composition beyond BMI is necessary in clinical care. Bioelectrical impedance analysis (BIA) and 3D scanning (3DS) are practical two-compartment (2C) methods estimating % body fat (%BF), fat mass (FM), and fat-free mass (FFM). Precision and validity are essential components of all body composition devices, yet BIA and 3DS have rarely been compared simultaneously against a reference method. Objectives: To evaluate the precision of BIA and 3DS and their validity versus dual-energy X-ray absorptiometry (DXA). Methods: Forty-four healthy women (40–65 y) underwent body composition assessment by DXA, BIA, and 3DS. Outcomes were whole-body 2C estimates for all devices and regional estimates for DXA and BIA. Precision was derived from duplicate measures; validity was assessed using Bland-Altman analyses between devices. Results: Precision error (%CV) for %BF, FM, FFM was: DXA 1.17, 1.14, 0.72; BIA 1.25, 1.26, 0.57; 3DS 2.78, 2.76, 1.08. All met preset acceptability limits (%BF 2%, FM 3%, FFM 2%) except 3DS %BF. Bland–Altman analyses showed poor individual-level agreement across devices. Despite only minor biases between devices in some cases, there were wide limits of agreement that crossed zero, had proportional bias, or a combination of the two. Conclusion: BIA and 3DS exhibited acceptable precision for most whole-body 2C outcomes, yet measurements were not interchangeable with DXA at the individual level. These findings caution against applying cross-device normative data or clinical cut-points and underscore the limitations of substituting BIA or 3DS for DXA in general practice.

Details

Logo image