Accuracy of Commercial Body Composition Scales Versus Clinical Dual-Energy X-ray Absorptiometry and Bioelectrical Impedance Analysis

Abstract

BACKGROUND: Body composition (BC) analysis is increasingly available to consumers in the form of wifi-connected bioelectrical impedance analysis (BIA) bathroom scales. However the accuracy of current generation scales to criterion methods is not known. PURPOSE: To determine the accuracy and precision of BC measurements using scales for men and women compared to dual-energy X-ray absorptiometry (DXA). In addition, we investigate the use of scales for quantifying resting metabolic rate (RMR). METHODS: We recruited a sample of convenience of healthy adults to undergo a series of BC measurements. Height and weight was assessed using a Seca 284. BC was accessed using whole-body DXA (Horizon A), a clinical tetrapolar multifrequency BIA (InBody770), and 7 bathroom scales: Tanita BF-684W, Weightwatchers WW701Y, Taylor 7226SFC, Withings Smart Body Analyzer WS-50 and Body Cardio WBS04, Fitbit Aria, and QardioBase. Duplicate measures were taken on the scales to evaluate the test-retest precision. RMR was calculated using Nelson RMR equations for DXA and BIA measures, and Harris-Benedict (HB) using weight. RESULTS: In total, 22 participants were recruited (12 male) with an average age of 29.6 (±7.3 years), weight of 70.0 (±13.4 kg) and height of 170.0 (±10.1 cm). PBF was highly correlated between the 770 and DXA for both men and women (r=0.9) but less correlated between the scales and DXA (0 < r < 0.65). However, all BIA devices underestimated PBF by 2 to 11 units. There was poor obesity classification between the scales and DXA as well (kappa < 0.2). RMR was highly correlated between DXA and scales for both men and women (r=0.50 to 0.99). RMR was overestimated by all scales compared to DXA from 21 to 184 kcal but in all cases were improvements over the HB RMR estimate (HB-DXA 260 kcal). The test-retest precisions (% coef of var, standard deviation) for PBF and RMR varied by model for men and women (0.2 to 0.5 %, NA) and (0.2 to 0.6 %, 2 to 9 kcal) respectively. Overall, we found the Tanita to have the best correlation to DXA measures, Taylor and Weight Watchers to have the best precision. CONCLUSIONS: We conclude that the use of bathroom scales may provide a more accurate assessment of RMR than the HB equation, and that the precision is comparable to previously-reported DXA values. However, our results varied substantially by make.