1. The University of Alabama, Tuscaloosa, AL
VALIDITY OF SELECTED BIOIMPEDANCE EQUATIONS FOR ESTIAMATING BODY FAT PERCENTAGE: A FOUR-COMPARTMENT MODEL COMPARISON
Brett S. Nickerson, Bailey A. Welborn, Kelsey A. Pezzuti, Phillip A. Bishop, and Michael R. Esco
The University of Alabama, Tuscaloosa, AL
Table 1. Comparison of BF% values between the BIA equations and 4-C model (n = 82)
Abstract
Methods
Value
Mean ± SD P
Cohen’s d R
SEE CE
Limits
4-C BF%
21.8 ± 7.8 -
BIA-Ch
24.9 ± 7.6
<0.001
-0.40
0.93
3.0
3.1
9.0, -2.8
BIA-De
25.5 ± 6.9
-0.50
0.87
3.8
3.7
11.1, -3.8
BIA-Kyle
24.1 ± 7.2
-0.31
0.90
3.4
2.3
8.8, -4.3
BIA-Sun
21.7 ± 7.3
0.702
0.01
0.91
3.3
-0.1
6.2, -6.5
The assessment of body fat percentage (BF%) with bioelectrical impedance analysis (BIA) is advantageous in field settings due to its non-invasive nature, quick administration, and relatively low cost. Numerous BIA equations have been developed and are available for practitioners. However, the validity of each when compared to a four-compartment model (4-C) model is unknown. PURPOSE: The purpose of this study was to compare BF% values derived from published BIA equations to the 4-C model. METHODS: Eighty-two adults (42 men and 40 women) volunteered to participate (age = 22.6 ± 4.9 years). The impedance measurements (i.e., resistance and reactance) were determined with a hand-to-foot BIA (single-frequency), which was then used to calculate BF% from previously developed equations by Chumlea et al. (BIA-Chumlea), Deurenberg et al. (BIA-Deurenberg), Kyle et al. (BIA-Kyle), and Sun et al. (BIA-Sun). Criterion BF% was calculated with the 4-C model from body mass, body volume (BV), total body water (TBW), and bone mineral content (BMC). Underwater weighing with simultaneous residual volume was used to determine BV, bioimpedance spectrscopy for TBW, and dual energy x-ray absorptiometry for BMC. RESULTS: No significant difference (p = 0.70) in mean BF% was seen for BIA-Sun (21.7 ± 7.3%) versus the 4-C model (21.8 ± 7.8%), while BIA-Chumlea (24.9 ± 7.6%), BIA-Deurenberg (25.5 ± 6.9%) and BIA-Kyle (24.1 ± 7.2%), were significantly higher (p < 0.05) compared to the criterion. The standard error of estimate for BIA-Chumlea, BIA-Deurenberg, BIA-Kyle, and BIA-Sun were 3.0%, 3.8%, 3.4%, and 3.3% with a 95% limits of agreement of (±5.9%), (±7.4%), (±6.6%), and (±6.4%), respectively. CONCLUSIONS: The BIA-Chumlea, BIA-Deurenberg, and BIA-Kyle do not appear to be acceptably valid with the 4-C model. However, the non-significant mean difference, low standard error of estimate and narrow limits of agreement found in the BIA-Sun equation suggests it is valid, when compared to the 4-C model, for estimating BF% in adult males and females. PRACTICAL APPLICATIONS: The Sun et al. equation is recommended when estimating BF% with hand-to-foot BIA. However, caution should be used when predicting BF% with BIA using the Chumlea et al., Deurenberg et al., and Kyle et al. equations.
Eighty-two adults (42 men and 40 women) volunteered to participate (age = 22.6 ± 4.9 years).
Prior to BIA measurements, all participants were required to be hydrated and provide a urine specific gravity <
Once hydration was confirmed, subjects had their height and weight measured.
Following height and weight measurements, subjects were instructed to lie on a gurney for 5 min and impedance values (i.e., resistance and reactance) were determined with a single-frequency hand-to-foot BIA device.
After determining impedance values with the single-frequency BIA device, criterion BF% was calculated with the 4-C model. Underwater weighing with simultaneous residual volume was used to determine BV, bioimpedance spectrscopy for TBW, and dual energy x-ray absorptiometry for BMC.
4-C = four-compartment model; BF% = body fat percentage; BIA-Ch = BIA-BF% prediction model from Chumlea et al. (2002); BIA-De = BIA-BF% prediction model from Deurenberg et al. (1991); BIA-Kyle = BIA-BF% prediction model from Kyle et al. (2001); BIA-Sun = BIA-BF% prediction model from Sun et al. (2003); SEE = standard error of estimate; CE = constant error
Results
Effect sizes trivial to small for all BIA equations
SEE < 4.0% for all BIA equations
Correlations were very large (r = 0.87) to near perfect (r = 0.93).
Conclusions * * *
BIA-Sun is a valid equations for estimating individual estimates of BF%
Caution should be employed when using the BIA equations from Chumlea et al. Deurenberg et al. and Kyle et al.
Practical Applications
The estimation of BF% via 4-C model can be time consuming, expensive, and impractical.
Practitioners in the field can utilize a single-frequency hand-to-foot BIA device for measuring impedance values and accurately predict 4-C model BF% with the BIA equation derived from Sun et al.
Intro & Purpose
The impedance values (i.e., resistance and reactance) from BIA devices can be used to estimate BF%
Many equations have been developed from criterions (i.e., UWW and DXA) other than the 4-C model
There is limited information available evaluating the validity between BIA equations derived from various criterions to the 4-C model
The purpose of this study was to compare BF% values derived from published BIA equations to the 4- C model
Figure 1. Mean between the BIA equations and 4-C model (n = 82)
*BIA equation was significantly different than 4-C model p < 0.001