Comparison of Foot-to-Foot and Hand-to-Foot Bioelectrical Impedance Methods in a Population with a Wide Range of Body Mass Indices

Abstract

Background:

Several techniques are currently used for measurement of body composition. Bioelectrical impedance assessment (BIA) is a simple, noninvasive method of assessing body composition. We aimed to compare multifrequency hand-to-foot (HF-BIA) and foot-to-foot (FF-BIA) bioelectrical impedance analysis techniques to assess fat-free mass (FFM) in a population with a wide range of body mass indices (BMI).

Methods:

This was a cross-sectional study of 198 adult subjects. Anthropometric and BIA measures (HF-BIA with Hydra ICF/ECF, Xitron Technologies and FF-BIA with Tanita, model TBF-300A) were recorded after a 12-h fast.

Results:

Participants had a mean age of 42 years and BMI of 33.5 ± 0.7 (range, 17.7–65.6) kg/m². Mean FFM with HF-BIA (FFM_BIA/HF) and FF-BIA (FFM_BIA/FF) were 61.3 ± 1.3 kg and 58.1 ± 0.9 kg, respectively (P < 0.001). In subjects with BMI <25 kg/m², FFM_BIA/FF was not significantly different compared to FFM_BIA/HF (+0.2 kg; P = 0.8). However, FFM_BIA/FF was significantly lower in subjects with BMI 25–30 kg/m² (−2.0 kg; P = 0.009), 30–34 kg/m² (−1.8 kg; P = 0.04), 34–42 kg/m² (−4.7 kg; P < 0.001) and >42 kg/m² (−8.0 kg; P = 0.001). Pearson correlations between both methods were very high for FFM (r = 0.92), fat mass (r = 0.91), and % fat mass (r = 0.85), all P < 0.001. Correlation coefficients for FFM were high in each quintile of BMI. FFM_BIA/FF was the only significant independent predictor of FFM_BIA/HF (P < 0.001) in linear regression analyses using clinical and FF-BIA variables, but introducing BMI in the model added precision.

Conclusion:

FFM_BIA/FF correlates closely with FFM_BIA/HF across all quintiles of BMI, but FF-BIA gives lower FFM in overweight and obese subjects.

Introduction

S everal techniques are currently used for measurement of body composition. Underwater weighing, the reference method, is impractical for use in large clinical trials and uncomfortable for patients. Assessment of body composition by dual-energy X-ray absorptiometry (DXA) is more convenient, but costly. Consequently, bioelectrical impedance analysis (BIA), a simple, noninvasive, and less expensive method, has gained in popularity in the last decade, and is more practical for use in clinical and epidemiological studies.

BIA is based on two important principles. The first one is that the resistance of a length of homogeneous conductive material of uniform cross-sectional area is proportional to its length and inversely proportional to its cross-sectional area. Thus, the technique assumes that the body is a uniform cylinder. A second assumption is that water represents 73.8% of the fat-free mass (FFM). Instruments for BIA introduce a current of about 800 A in the body that flows between two electrodes at single or multiple frequencies throughout conducting material.¹ The measured impedance, which reflects tissue resistance to electrical current, provides an estimate of total body water (TBW). Various equations are used to derive FFM from TBW. FFM is probably the most accurate index recorded by BIA and is a useful parameter to follow in overweight and obese individuals, especially those undertaking restrictive diet programs. Indeed, without exercise, these diet approaches are associated with a decrease in FFM, accounting for about 25% of the total weight loss.^2,3

Conventional hand-to-foot supine BIA method (HF-BIA) has been reported to systematically overestimate FFM in obese individuals when compared with underwater weighing or other methods.^4

–9 However, a meta-analysis showed that multifrequency BIA (MF-BIA) may be more accurate than single frequency BIA (SF-BIA) to estimate TBW in obese subjects.¹⁰ The newer single-frequency standing foot-to-foot BIA (FF-BIA) method is simpler and less expensive than HF-BIA, but it has not been studied in depth in large cohorts of obese subjects. One study demonstrated that, in comparison with a four-compartment model that combines measurements of body weight, density, TBW, and bone mineral to assess body composition, FF-BIA and HF-BIA have similar accuracies for assessment of FFM in a cohort of subjects with a mean BMI of 25.9 kg/m².¹¹ Another study showed comparable results of FFM assessed by underwater weighing and FF-BIA in 98 moderately obese women (BMI 33.2 ±0.6 kg/m²).¹²

No previous study has compared measures of FFM assessed by FF-BIA with the widely used and well-accepted HF-BIA technique in populations with a large range of BMI. Therefore, the purpose of the present study is to compare foot-to-foot standing and multifrequency hand-to-foot supine BIA to estimate FFM in adults with a wide range of BMI, including severely obese individuals.

Materials and Methods

Subjects and methods

This cross-sectional study reports data from 198 adult subjects who participated in metabolic studies at the Centre de recherche clinique Etienne-Le Bel of the Centre hospitalier universitaire de Sherbrooke (CHUS). Study protocols were approved by the Institutional Review Board of the CHUS, and all participants gave written consent. Pregnant women and patients with a cardiostimulator were excluded. Subjects were asked to fast for 12 h before their arrival. Measurements were taken at approximately 8:00 a.m., in a room with controlled temperature (23°C). Body weight was measured without shoes and in light clothing to the nearest 0.1 kg, using a mechanical scale, and height to the nearest 0.1 cm, using a wall-mounted stadiometer. Waist circumference was measured to the nearest 0.5 cm using a tape placed at the midpoint between the lower rib and the upper iliac crest.

BIA measurements

Determination of FFM was performed according to manufacturer's instructions using a FF-BIA meter (Tanita Corporation, model TBF-300A, Tokyo, Japan) that delivers a current with a single frequency of 50 kHz. Subjects stood barefoot on the metal sole plates of the device. Height, age, and gender were entered manually, and weight was measured by the device to within 0.1 kg. FFM was estimated using the standard prediction equations provided by the manufacturer.

FFM by HF-BIA was measured with a Xitron Hydra ECF/ICF, model 4200 (Xitron Technologies, San Diego, CA), according to manufacturer's instruction. This device uses four gel electrodes placed over the ipsilateral wrist and ankle. It provides measurement of complex impedance at different frequencies between 5 and 1000 KHz. Subjects' height, weight, and gender were entered manually. Volume equations provided by the manufacturer were derived from the Hanai mixture theory.¹³ These equations have been previously validated.^14,15 The equation to estimate FFM was: FFM = (d_ECW V_ECW ) + (d_ICW V_ICW ), where d _ECW = 1.106 is the mean density of the extracellular water compartment, V _ECW is the total extracellular fluid volume, d _ICW = 1.521 is the mean density of the intracellular water compartment, and V _ICW is the total intracellular fluid volume.

Statistical analysis

Statistical analyses were performed using SPSS software version 10.0.5 (SPSS Inc., Chicago, IL). Data are presented as means ± standard error of the mean (SEM). Pearson correlation coefficients (r) were calculated to compare estimates of FFM by the two BIA methods. The influence of BMI on FFM estimates by the two BIA methods was examined by comparing these estimates in each quintile of BMI using a paired t-test. Differences in FFM by HF-BIA (FFM_BIA/HF) and FF-BIA (FFM_BIA/FF) between quintiles were analyzed using analysis of variance (ANOVA) with Tukey post hoc analysis (quintiles 1 to 5 respectively: <25, 25–30, 30–34, 34–2, and >42 kg/m²).

To determine independent predictive factors of FFM by HF-BIA, a forward stepwise multivariate analysis was performed using clinical, anthropometric, and FF-BIA variables that were significantly associated with FFM by HF-BIA in univariate analyses, starting with the variable displaying the lowest P value. At each step, parameters that did not contribute significantly to the model (partial P > 0.05) were excluded. On the basis of the results of this analysis, an equation to predict FFM by HF-BIA was generated. Because BMI affected the relationship between FFM_BIA/FF and FFM_BIA/HF on the basis of ANOVA, this variable was then forced in the model to create the best-fit equation.

To test this equation, the entire group was randomly divided in two equal groups using a list of 198 computer-generated random numbers. Regression models to maximize R ² were developed in the first group comprising one-half of the subjects. The performance of the equation derived from this model to estimate FFM by HF-BIA from FF-BIA was then tested in the second group for validation.

Results

Clinical characteristics of participants

The clinical characteristics of the study group are depicted in Table 1. Ninety-one percent of the participants were Caucasians aged between 18 and 78 years. The mean BMI of the cohort was 33.5 ± 0.7 (range, 17.7–65.6) kg/m² and 80% of the participants were overweight or obese.

Table 1.

Clinical Characteristics of Participants

		Quintiles of BMI
	All subjects	1	2	3	4	5
Number of subjects, n	198	40	40	40	39	39
Range of BMI	17.7–65.6	17.7–24.6	24.7–29.9	30.0–34.0	34.1–42.0	42.1–65.6
Gender, n (%)
Male	92 (46.5)	21 (52.5)	25 (62.5)	19 (47.5)	13 (33.3)	14 (35.9)
Female	106 (53.5)	19 (47.5)	15 (37.5)	21 (52.5)	26 (66.7)	25 (64.1)
Age, years	42.2 ± 1.0	28.5 ± 1.5	37.8 ± 2.2	48.7 ± 2.4	48.9 ± 2.3	47.4 ± 1.8
Height, meters	1.67 ± 0.0	1.71 ± 0.0	1.69 ± 0.0	1.67 ± 0.0	1.65 ± 0.0	1.65 ± 0.0
Weight, kg	93.6 ± 1.9	63.9 ± 1.4	78.6 ± 1.7	89.1 ± 1.7	103.2 ± 2.1	134.5 ± 3.5
BMI, kg/m²	33.5 ± 0.7	21.9 ± 0.3	27.3 ± 0.2	31.7 ± 0.2	38.0 ± 0.4	49.2 ± 0.8
Waist circumference, cm	104 ± 2	74 ± 1	89 ± 1	103 ± 1	116 ± 2	138 ± 3
FFM, kg
HF-BIA	61.3 ± 1.3	51.3 ± 1.9	58.6 ± 2.1	57.4 ± 1.9	62.7 ± 2.3	77.0 ± 4.0
FF-BIA	58.1 ± 0.9	51.5 ± 1.5	56.6 ± 1.7	55.6 ± 1.6	58.0 ± 1.8	69.0 ± 2.3
P value^a	<0.001	0.809	0.009	0.042	<0.001	0.001

Data are presented as means ± standard error of the mean (SEM).

Paired t test between mean FFM_BIA/HF and FFM_BIA/FF.

Abbreviations: BMI, Body mass index; FFM, fat-free mass; HF-BIA, hand-to-foot bioelectrical impedance analysis; FF-BIA, foot-to-foot bioelectrical impedance analysis

Comparison of FFM by HF-BIA and FF-BIA in the entire group

Pearson correlation between FFM_BIA/HF and FFM_BIA/FF was very high (r = 0.92; P < 0.001). In the whole group, FFM_BIA/FF was lower than FFM_BIA/HF by an average of 3.2 ± 0.6 kg (P < 0.001). Fat mass (r = 0.91) and % fat mass (r = 0.85) results by both methods were also highly correlated (P < 0.001). Fat mass estimated by FF-BIA was higher than fat mass by HF-BIA by 3.2 ± 0.1 kg (see Fig. 1).

FIG. 1.

Correlation between fat-free mass (FFM) estimates from hand-to-foot bioelectrical impedance analysis (HF-BIA) and foot-to-foot bioelectrical impedance analysis (FF-BIA) in the 198 subjects included in this study.

Comparison of FFM measurements by HF-BIA and FF-BIA by quintiles of BMI

Correlation coefficients for FFM were high and similar in each quintile of BMI. In subjects with the lowest BMI, the FFM_BIA/FF result was comparable to FFM_BIA/HF (+0.2 ± 0.7 kg; P = 0.8). However, in overweight and obese subjects, FFM was systematically lower when determined by FF-BIA than with HF-BIA. These differences in FFM between the two methods increased progressively with higher BMI, and there is a significant interaction between the method used to determine FFM and quintiles of BMI (P < 0.0001 by mixed-model repeated measures ANOVA). There is also a significant interaction between the method used to determine FFM and gender (p < 0.0001 by mixed-model repeated measures ANOVA), where the difference between methods is greater in men. However, correlations between both methods are very high in both genders.

Best-fit equation to predict FFM by HF-BIA

The entire group was then randomly divided into two equal groups. There was no significant difference between groups. Regression analyses were performed in the first group. Gender, height, weight, BMI, waist circumference, and FFM_BIA/FF were all predictive of FFM_BIA/HF in univariate analyses (all P < 0.001). In this population, age was not a predictive factor of FFM_BIA/HF (P = 0.4). Multivariate analysis found that only FFM_BIA/FF was a significant independent predictive factor of FFM_BIA/HF (adjusted R ² = 0.86). Because BMI affected the relationship between FFM_BIA/FF and FFM_BIA/HF (Table 2), this variable was then forced in the model to create the best-fit equation. The equation generated to predict FFM_BIA/HF using FFM_BIA/FF was: FFM_BIA/HF = 11.32 + 1.18 × FFM_BIA/FF + 0.098 × BMI.

Table 2.

Comparison of FFM by HF-BIA versus FF-BIA by Quintiles of BMI

		Quintiles of BMI
	All subjects	1	2	3	4	5
FFM, kg
HF-BIA	61.3 ± 1.3	51.3 ± 1.9	58.6 ± 2.1	57.4 ± 1.9	62.7 ± 2.3	77.0 ± 4.0
FF-BIA	58.1 ± 0.9	51.5 ± 1.5	56.6 ± 1.7	55.6 ± 1.6	58.0 ± 1.8	69.0 ± 2.3
Mean difference in FFM, kg	−3.2 ± 0.6	+0.2 ± 0.7^a,b	−1.9 ± 0.7	−1.8 ± 0.9^a	−4.7 ± 0.9	−8.0 ± 2.3
Pearson correlation, r ^c	0.92	0.94	0.95	0.90	0.94	0.88

FFM_BIA/FF minus FFM_BIA/HF.

Difference between the group and group 5 (ANOVA) P < 0.01.

Difference between the group and group 4 (ANOVA) P < 0.01.

P values for correlations, all <0.001.

Abbreviations: FFM, fat-free mass; HF-BIA, hand-to-foot bioelectrical impedance analysis; FF-BIA, foot-to-foot bioelectrical impedance analysis; BMI, body mass index; ANOVA, analysis of variance.

The second group was then used to validate the equation. Correlation between the estimated FFM using the formula and the FFM measured by HF-BIA was excellent (r = 0.92; P < 0.001) and similar for both genders. The mean difference in FFM between HF-BIA and the calculated FFM was reduced to 1.23 ± 0.8 kg, a difference that was not significant (P = 0.15). Differences between FFM_BIA/HF and calculated FFM were decreased in all quintiles of BMI and where not significant (quintiles 1–5, respectively: −0.8 kg; P = 0.4, +0.8 kg; P = 0.4, +0.1 kg; P = 0.9, +1.4 kg; P = 0.2, +4.8 kg; P = 0.2).

Discussion

In this large cross-sectional study, FFM by multifrequency HF-BIA and single-frequency FF-BIA was highly correlated across all quintiles of BMI. The difference in FFM estimates between both methods was minimal in lean subjects. Therefore, FF-BIA could be used with confidence in this population. This finding may have practical and economical issues because FF-BIA is cheaper, simpler, and more convenient for large-scale studies or clinical practice; thus, such use would be favored by the results of this study. On the other hand, there was a progressive and significant underestimation of FFM assessed by FF-BIA versus HF-BIA in overweight and obese subjects. The discrepancy between the two methods was more prominent in men and in subjects with a BMI over 34 kg/m², approaching a mean of 8 kg in subjects within the fifth quintile of BMI. Therefore, estimation of FFM with BIA methods should be interpreted with caution, especially in severely obese individuals. Our equation using FFM_BIA/FF and BMI was shown accurate in predicting FFM_BIA/HF in overweight and obese subjects. Thus, it could be used to minimize the difference in FFM between both methods in these individuals.

In obese subjects, there remains uncertainty as to which method best estimates FFM. Single-frequency HF-BIA has been shown in many studies to overestimate FFM of obese people when compared with reference methods.^4

–9 Interestingly, when MF-BIA was compared with deuterium oxide (D₂O) dilution, overestimation of TBW was not seen,¹⁰ suggesting that MF-BIA may be suitable for use in obese populations. However, severely obese subjects were not included in these studies, limiting the generalization of the results to this population.^15,16 The accuracy of FF-BIA has also been scarcely addressed in subjects with a BMI over 30 kg/m². To our knowledge, no study was performed in severely obese subjects. One study in 98 moderately obese women showed no significant difference in FFM between FF-BIA and underwater weighing.¹² However, it is difficult to draw conclusions on the accurateness of BIA methods in obese subjects due to the small sample size and heterogeneity of study populations and the variability of prediction equations used to assess TBW.

The higher results of FFM_BIA/HF in obese subjects when compared with FF-BIA (and other reference methods) may be explained partly by the distribution of water volumes. Expansion of the extracellular water (ECW) compartment has been reported in obese people,¹⁷ suggesting that they may have relatively more water in the trunk. Because the trunk is short and has a large cross-sectional area, it does not contribute much to the overall resistance. Second, the average amount of body fat found in obese men and women is 30% and 50%, respectively.¹⁸ In both genders, the major body fat depots are the upper body subcutaneous fat (44 vs. 53%), the intraabdominal (visceral) fat (25 vs. 18%), and lower body (gluteofemoral) body fat (32 vs. 36%). Therefore, the combination of these phenomena could account for the lower resistance values obtained when the electrical current is passing through this area. Other factors, like the type of current, electrodes, calibration, and interpretation algorithms, may also influence the results.

Conclusion

FFM by multifrequency HF-BIA and single-frequency FF-BIA were highly correlated in lean and obese subjects. There was no difference in FFM between both methods in lean individuals. Because FF-BIA is simpler, less expensive, and more practical for use in large clinical and epidemiological studies, there is no clear advantage of HF-BIA over FF-BIA in this population. Thus, FF-BIA methods may be useful and accurate in clinical and research settings. However, FF-BIA gave systematically lower FFM values than HF-BIA in overweight and obese subjects, particularly males. An equation using FFM_BIA/FF and BMI was validated to predict FFM_BIA/HF accurately in overweight and obese individuals and it can be applied in these settings. Our results support the need of larger trials comparing FF-BIA with reference methods in obese subjects to determine whether this practical method may be suitable to assess FFM in this population.

Footnotes

Acknowledgments

The present study was supported by grants from the Canadian Institutes of Health Research (CIHR) (Dr. Carpentier, MOP53094; Dr. Baillargeon, MOP62946; Dr. Ardilouze, MOP77729; from the Canadian Diabetes Association (to Dr. Carpentier in honor of the late Marion L. Monroe) and from the Ministère de la santé et des services sociaux du Québec (Programme de subventions en santé publique, to Dr Langlois). Dr. Gagnon is the recipient of a Novo-Nordisk Endocrine Fellow Research Grant. Dr. Ardilouze is a new investigator of the CIHR. Drs. Baillargeon, Carpentier and Langlois are Scholars of the Fonds de recherche en santé du Québec (FRSQ). This work was presented in part at the 2006 Annual Scientific Meeting of NAASO.

Author Disclosure Statement

All authors declare having no competing financial interests.

References

Kyle

, Bosaeus

, De Lorenzo

, Deurenberg

, Elia

, Gómez

, Heitmann

, Kent-Smith

, Melchior

, Pirlich

. Bioelectrical impedance analysis—part I: Review of principles and methods. Clin Nutr, 2004; 23:1226–1243.

Villareal

, Apovian

, Kushner

, Klein

, American Society for

Nutrition

. NAASO. Obesity in older adults: Technical review and position statement of the American Society for Nutrition and NAASO, The Obesity Society. Obes Res, 2005; 13:1849–1863.

Zamboni

, Mazzali

, Zoico

, Harris

, Meigs

, Di Francesco

, Fantin

, Bissoli

, Bosello

. Health consequences of obesity in the elderly: A review of four unresolved questions. Int J Obes, 2005; 29:1011–1029.

Research

, Carella

, Rodgers

, Anderson

, Gossain

. Serial measurements of body composition in obese subjects during a very-low-energy diet (VLED) comparing bioelectrical impedance with hydrodensitometry. Obes Res, 1997; 5:250–256.

Hendel

, Gotfredsen

, Hojgaard

, Andersen

, Hilsted

. Change in fat-free mass assessed by bioelectrical impedance, total body potassium and dual energy X-ray absorptiometry during prolonged weight loss. Scand J Clin Lab Invest, 1996; 56:671–679.

Heyward

, Cook

, Hicks

, Jenkins

, Quatrochi

, Wilson

. Predictive accuracy of three field methods for estimating relative body fatness of nonobese and obese women. Int J Sport Nutr, 1992; 2:75–86.

Paijmans

, Wilmore

. Use of skinfolds and bioelectrical impedance for body composition assessment after weight reduction. J Am Coll Nutr, 1992; 11:145–151.

McNeill

, Fowler

, Maughan

, McGaw

, Fuller

, Gvozdanovic

. Body fat in lean and overweight women estimated by six methods. Br J Nutr, 1991; 65:95–103.

Gray

, Bray

, Gemayel

, Kaplan

. Effect of obesity on bioelectrical impedance. Am J Clin Nutr, 1989; 50:255–260.

10.

Martinoli

, Mohamed

, Maiolo

, Cianci

, Denoth

, Salvadori

, Iacopino

. Total body water estimation using bioelectrical impedance: a meta-analysis of the data available in the literature. Acta Diabetolog, 2003; 40,Suppl 1:S203–S206.

11.

Jebb

, Cole

, Doman

, Murgatroyd

, Prentice

. Evaluation of the novel Tanita body-fat analyser to measure body composition by comparison with a four-compartment model. Br J Nutr, 2000; 83:115–122.

12.

Utter

, Nieman

, Ward

, Butterworth

. Use of the leg-to-leg bioelectrical impedance method in assessing body-composition change in obese women. Am J Clin Nutr, 1999; 69:603–607.

13.

Hanai

. Electrical properties of emulsions. Sherman

. Emulsion Science. London: Academic, 1968; 354–477.

14.

Patel

, Matthie

, Withers

, Peterson

, Zarowitz

. Estimation of total body and extracellular water using single- and multiple-frequency bioimpedance. Ann Pharmacother, 1994; 28:565–569.

15.

De Lorenzo

, Andreoli

, Matthie

, Withers

. Predicting body cell mass with bioimpedance by using theoretical methods: A technological review. [erratum appears in J Appl Physiol 1997 Dec;83(6):followi] J Appl Physiol, 1997; 82:1542–1558.

16.

Steijaert

, Deurenberg

, Van Gaal

, De Leeuw

. The use of multi-frequency impedance to determine total body water and extracellular water in obese and lean female individuals. Int J Obes Relat Metab Disord, 1997; 21:930–934.

17.

Waki

, Kral

, Mazariegos

, Wang

, Pierson

Jr , Heymsfield

. Relative expansion of extracellular fluid in obese vs. nonobese women. Am J Physiol, 1991; 261,Pt 1:E199–E203.

18.

Jensen

. Adipose tissue and fatty acid metabolism in humans. J Roy Soc Med, 2002; 95,Suppl 42:3–7.