External validation of BIA equations to estimate appendicular skeletal muscle mass in older adults: Importance of the bias analysis and derivation of correction factors to achieve agreement

Abstract

There are several equations based on bioelectrical impedance analysis (BIA) to estimate with high precision appendicular skeletal muscle mass (ASM). However, most of the external validation studies have reported that these equations are inaccurate or biased when applied to different populations. Furthermore, none of the published studies has derived correction factors (CFs) in samples of community-dwelling older adults, and none of the published studies have assessed the influence of the dual-energy X-ray absorptiometry (DXA) model on the validation process. This study assessed the agreement between six BIA equations and DXA to estimate ASM in non-Caucasian older adults considering the DXA model and proposed a CF for three of them. This analysis included 547 non-institutionalized subjects over 60 years old from the northwest of Mexico who were physically independent and without cognitive impairment: 192 subjects were measured using DXA Hologic, while 355 were measured by DXA Lunar. The agreement between each of the equations and DXA was tested considering the DXA model used as a reference method for the design of each equation, using the Bland and Altman procedure, a paired t test, and simple linear regression as objective tests. This process was supported by the differences reported in the literature and confirmed in a subsample of 70 subjects measured with both models. Only six published BIA equations were included. The results showed that four equations overestimated ASMDXA, and two underestimated it (p < 0.001, 95% CI for Kim's equation:−5.86-−5.45, Toselli's:−0.51-−0.15, Kyle's: 1.43–1.84, Rangel-Peniche's: 0.32–0.74, Sergi's: 0.83–1.23, and Yoshida's: 4.16–4.63 kg). However, Toselli's, Kyle's and Rangel-Peniche's equations were the only ones that complied with having a homogeneous bias. This finding allowed the derivation of CFs, which consisted of subtracting or adding the mean of the differences from the original equation. After estimating ASM applying the respective CF, the new ASM estimations showed no significant bias and its distribution remained homogeneously distributed. Therefore, agreement with DXA in the sample of non-Caucasian was achieved. Adding valid CFs to some BIA equations allowed to reduce the bias of some equations, making them valid to estimate the mean values of ASM at group level.

Impact of skeletal muscle mass evaluating methods on severity of metabolic associated fatty liver disease in non-elderly adults

“…Although BIA showed good correction with DXA, the BIA algorithms were developed by the manufacturer in specific population and adjustment equations remain required in clinical application ( 38 ) . Even if the use of BIA for the evaluation of ASM is considered acceptable in different documents (only if DXA is not available), most of the researchers suggest caution in clinical practice and even more in scientific research ( 54 , 55 ) . Over or under-estimation of ASM is frequent with BIA compared with DXA, and there is the necessity to adopt adjustment equations that need to be ethnicity-/age-/sex-/disease-specific.…”

Phase angle, muscle tissue, and resistance training

“…Beyond the extensive number of predicting equations allowing the estimation of whole-body tissues from the raw biophysical parameters and other variables such as weight, height, and age [9][10][11], there is now a growing research interest in understanding the implications of both impedance and admittance proprieties of different body regions (i.e., arms, legs, trunk), as these tend to differ in their shapes and biological constitutions [12]. Several equations predicting regional proportions of body composition have therefore been recently developed for healthy populations [9,[13][14][15], as well as for other special populations (e.g., athletes) [16], which has opened the possibility to further characterize tissue composition of each segment and determine limb asymmetries at both functional and structural levels.…”