Limitations and validation of bioelectrical impedance analysis in

morbidly obese patients

Luciana Z. Coppinia,b, Dan L. Waitzberga,b and Antonio Carlos L. Camposc

Purpose of review Abbreviations

Several factors limit the use of bioelectrical impedance BIA bioelectrical impedance analysis
BIS bioelectrical impedance spectroscopy
analysis as a valid predictor of the amount of body fat in FFM fat-free mass
morbidly obese individuals. The purpose of this review is to TBW total body water
ECW extracellular water
examine the theory and assumptions that may limit the use ICW intracellular water
of bioelectrical impedance analysis in such individuals. BMI body mass index
FFMI fat-free mass index
Recent findings FMI fat mass index
There is currently insufficient validation of bioelectrical
impedance analysis equations in obese individuals with # 2005 Lippincott Williams & Wilkins
body mass indices greater than 34 kg/m2. Several factors 1363-1950
limit the application of bioelectrical impedance analysis in
morbidly obese individuals. Obese individuals have a
relatively high amount of extracellular water and total body Introduction
water, which may overestimate fat-free mass and Obesity is defined as an excessive accumulation of body
underestimate fat mass. Central body fat will generally fat. In the young adult normal levels of body fat are
overestimate the percentage of fat-free mass and considered to be 12–20% of the body weight in men and
underestimate the percentage of fat mass in overweight and 20–30% of the body weight in women, whereas levels of
obese adults with the use of prediction formulas developed over 25% of body weight in men and over 33% in women
in normal weight individuals. can be considered obesity [1].
Summary
A relatively increased amount of total body water and a To measure fat in the body is difficult and expensive, and
relative increase in extracellular water will result in an no accurate method is easily available for routine clinical
underestimation of the percentage of body fat and an use. Traditionally, overweight and obesity have been
overestimation of fat-free mass in the morbid obesity state. defined by equations derived from anthropometric mea-
A different body build disposition (mainly in those with surements such as weight and height [2], which have
severe abdominal obesity) will result in an overestimation of serious limitations, especially in extremes of weight or
the percentage of body fat. New equations are needed to height.
validate bioelectrical impedance analysis in morbidly obese
patients. Among several methods to assess body fat, two are
available and accessible in clinical practice. These are
Keywords bioelectrical impedance analysis (BIA) and dual energy
bioelectrical impedance analysis, fat-free mass, fat mass, X-ray absorptiometry. BIA measures the body impedance
morbid obesity, total body water using an electrode in one leg and another electrode in the
arm or in the other leg. The method actually measures
Curr Opin in Clin Nutr Metab Care 8:329–332. # 2005 Lippincott Williams & Wilkins. body water, which is then used in equations to estimate
fat-free mass (FFM), and subtracting this value from
a
University of São Paulo Medical School, São Paulo, Brazil, bGrupo de Nutrição weight estimates body fat [3].
Humana (GANEP), São Paulo, Brazil, and cPost-Graduate Program in Surgery,
Federal University of Paraná, Curitiba, Brazil
For the average male and female adult weighing 70 kg,
Correspondence to Luciana Z. Coppini, GANEP, Rua Maestro Cardim 1175, São
Paulo, SP Brazil CEP 01323.001 water constitutes approximately 60 and 55% of body
Tel: +55 11 32846318; fax: +55 11 2897391; e-mail: [email protected] weight, respectively. The total body water (TBW) com-
Current Opinion in Clinical Nutrition and Metabolic Care 2005, 8:329–332 partment consists of the extracellular water (ECW)
(approximately 45%) and the intracellular water (ICW)
(approximately 55%). It has been shown that the ratio of
TBW to body weight differs greatly among individuals,
and that these differences are mostly attributable to
differences in the amount of adipose tissue [4]. As this
method measures water, it is inaccurate in individuals
with oedema or electrolyte disturbances. Changes in
329
330 Pharmaceutical aspects, devices and techniques

ambient temperature and in the placement of the elec- Body composition in obese individuals
trodes can also interfere with the results obtained. A Not only is the total amount of fat that an individual
number of different instruments are available for mea- carries important, but also where the fat is distributed in
suring fatness by BIA, and some instruments are much the body. Studies have shown that fat in a central or upper
superior to others in accuracy and reliability [5]. body (android) distribution is most related to health risk
[2].
Bioelectrical impedance analysis
BIA methods share many of the same advantages as other Although the best and most accurate way to measure
field-ready techniques. This is an accurate technique, is central obesity is by magnetic resonance imaging [10] or
non-invasive, portable, needs low maintenance and mini- computer-assisted tomography scanning [11], these
mal operator training, and is inexpensive for quantitatively approaches are too expensive for routine use. As a result,
estimating body compartments. BIA estimates total body simple anthropometric measurements can be used. The
weight, ICW, ECW, fat mass and FFM for both euvolemic waist–hip circumference has been widely used, particu-
adult healthy individuals and for those affected by several larly in epidemiological studies [12].
pathological conditions [6]. However, recent research
findings have shown that the application of the various Several factors limit the use of BIA as a valid predictor of
impedance models to individuals in some selected clinical the amount of body fat in extremely obese individuals.
populations may be limited [5,6,7]. The presence of increased body fat may affect the vali-
dity of BIA. The validity of BIA decreases with increas-
Single and multi-frequency BIA and bioelectrical impe- ing obesity and improves with weight loss, because the
dance spectroscopy (BIS) are used for estimating TBW, abdominal adiposity invalidates the assumption that the
ICW, ECW, fat mass and FFM on the basis of mathe- body is a cylindrical conductor [5]. Studies have docu-
matical formulae using measurements of resistance, reac- mented an overestimation of the percentage of FFM by
tance, and impedance. The single-frequency serial model approximately 3.4% of body weight or an underestima-
is the most common model used for in-vivo body water tion of the percentage of fat mass by 8.2% of the body
analysis [8]. weight in overweight and obese adults [12].

This BIA model is based on the assumption that the body Swan and McConnell [13] examined the accuracy of
behaves as a cylindrical conductor (two arms, two legs, several generalizable anthropometric and BIA regression
and a trunk) with a uniform cross-sectional area [5]. equations to estimate the percentage of body fat in
However, because resistance is inversely proportional to women with either upper body or lower body fat dis-
the cross-section area, it is clear that the resistance of the tribution patterns. They divided the subjects into two
whole body is largely determined by the resistance of the groups: overweight and obese women into two groups
limbs, which have the smallest cross-sectional areas. It is matched for age, body mass index (BMI), percentage fat
estimated that an arm is approximately 4% body weight mass, and those upper versus lower body obesity. The
and a leg is approximately 17% of body weight, but they authors used three BIA equations to predict the percen-
contribute to approximately 47% and 50%, respectively, tage of FFM, from the percentage of fat mass [14–16].
of the whole body resistance. The trunk contains 50% of BIA overestimated the percentage of fat mass in women
the body mass but contributes to only 5–12% of whole- with either upper body obesity by 5% using the Gray
body resistance. This phenomenon can be attributed to equation and in both groups by approximately 6% using
the fact that the trunk is relatively short and has a large the VanLoan and Mayclin [16] equation, whereas the
diameter (impedance is directly proportional to length Lohman [14] equation underestimated the percentage of
and inversely proportional to diameter) [5,8,9]. There- fat mass in the lower body by approximately 4%. There-
fore, changes in the trunk region are relatively insensitive fore, one equation significantly overestimated the per-
to detection by whole-body, single-frequency BIA [5]. centage of fat mass in both groups, another significantly
overestimated the percentage of fat mass in women with
This model is based on the assumption that the human central body obesity, and the third underestimated the
body has a homogenous composition, with a specific percentage of fat mass in women with lower body obesity.
resistance that is constant throughout the conductor In conclusion, the authors noted that anthropometric and
length and a resistance that is the same regardless of some BIA equations are accurate for predicting the per-
the direction of the current flow [5]. The specific centage of fat mass in lower body fat ‘shaped’ women, but
resistance of the body fluid depends on the kind and are not appropriate for women with primarily abdominal
amount of dissolved electrolytes as well as on tempera- fat patterning [5,13].
ture. However, the variable tissue composition of the
limbs and trunk results in regional differences in specific Some BIA instruments are based on leg-to-leg or hand-to-
resistance [5,9]. hand impedance to predict the percentage of fat mass
 Bioelectrical impedance analysis in morbidly obese patients Coppini et al. 331

directly. Although attractive and simple to use as alter- weight loss and as a change in triceps skinfold. Measured
natives to the whole-body BIA model, the random error changes in resistance values of ECW and resistance
resulting from these instruments may be large [5]. values of ICW were less than expected for an ideal
Most studies using these methods have focused on over- agreement between dilution methods and mixture equa-
weight and obese individuals to measure the percentage tions. In conclusion, the use of mixture equations over-
of fat mass. Deurenberg et al. [9] found that the measure- estimates fluid losses during weight loss. The error is
ment of the percentage of fat mass was underestimated in associated with the amount of fat loss. The large con-
the leg-to-leg model in boys more than in girls with tribution of the factor weight in the mixture equations is
android obesity. In another study [4], a leg-to-leg model likely to be responsible.
overestimated the percentage of fat mass in girls and
women, but not in boys and men. The authors noted that The relative ratio of extracelullar water to TBW is higher
hand-to-hand BIA measures the impedance of the upper in obese individuals. This effect is more pronounced at
body only and derives the estimation to the amount of the higher BMIs. Multi-frequency BIA shows that at the
whole body. In conclusion, equations in leg-to-leg and often-used frequency of 50 kHz the current does not
hand-to-hand BIA models take into account differences fully penetrate the cell membrane. This means that
in body fat distribution, and their validity remains limited measured impedance at 50 kHz is not a measure of
[5]. TBW but a measure of ECW plus partly ICW [5,20].

Body water distribution in obese individuals Bioelectrical impedance analysis

The major limitation of using TBW estimation by BIA is measurement in bariatric surgery patients
that this method assumes that the hydration status is Direct measurements of body fat may offer distinct
constant. However, physiological and pathological factors advantages to predict outcomes compared with the total
such as pregnancy, disease states, obesity, growth, mal- percentage of excess weight loss [21].
nutrition and ethnicity may interfere with the hydration
status [5,17]. Obese individuals are likely to have a Many studies have reported the use of BIA in the follow-
different body water distribution compared with lean up of bariatric surgery patients. Livingston et al. [22]
individuals [18]. studied changes in body composition with BIA after
Roux-en-Y gastric bypass. They found that fat mass loss
Bray et al. [19] found that the hydration of FFM was and lean body mass loss decreased in a linear manner after
significantly higher in obese versus lean children (79.2 surgery. Das et al. [23] performed a sophisticated valida-
versus 76.7%). Martinoli et al. [4] performed a meta- tion study of BIA in bariatric surgery patients. They
analysis including 10 studies conducted in obese adults. measured body density by air displacement plethysmo-
On the basis of the weighted mean difference, they found graphy and TBW by H2(18)O dilution and BIA to deter-
that those studies using multifrequency BIA only did not mine TBW and the percentage of fat mass. A comparison
significantly overestimate the TBW compared with the of this gold-standard technique with other technologies
H2(18)O dilution. However, those studies that used found that BIA was associated with reliable measures of
simple-frequency BIA or BIS significantly overestimated body composition [20]. The additional result from that
the TBW in relation to H2(18)O dilution. study was that the relationship between BMI and the
percentage of fat mass appeared to change between the
Cox-Reijven et al. [20] studied the effects of the compo- extremely obese and the weight-reduced states, perhaps
sition of weight loss on the accuracy of BIS to measure indicating a plateau in the relationship between the
changes in body fluid volumes during severe weight loss. percentage of fat mass and BMI at higher BMI values.
The effect of variable losses of fat mass was studied in 10 These findings indicate that further research is needed to
morbidly obese female subjects who underwent gastric examine the utility of BMI to predict the percentage of
reduction surgery. Changes in TBW and ECW were fat mass accurately in different subject groups. In the
monitored for one year by deuterium and bromide dilu- meantime, it should be noted that BIA (using any of
tion and by BIS. Extrapolated resistance values of ECW the tested equations) was more strongly associated with
and ICW were used in mixture equations for calculating the percentage of fat mass than BMI, and therefore could
the corresponding fluid volumes. After one year, weight be more widely used as a tool to measure body fatness in
decreased by 53 kg, TBW (deuterium) loss was 8.7 l, and clinical practice [22–24].
ECW (bromide) loss was 4.3 l. A comparison of BIS with
reference methods for measuring all possible changes The body composition as assessed by BIA permits the
over six time intervals revealed a mean overestimation calculation of the fat-free mass index (FFMI) and the fat
of TBW of 2.4  2.9 l and ECW of 0.74  2.6 l losses by mass index (FMI), which are calculated in the same
BIS. Overestimation increased significantly with increas- manner as BMI. These indices are more reliable indica-
ing fat losses, expressed as a percentage of fat of the tors of obesity and fat-associated health risk [25]. The use
332 Pharmaceutical aspects, devices and techniques

7 Campos ACL, Chen M, Meguid MM. Comparisons of body composition

of the BMI, FMI and FFMI may provide more complete derived from anthropometric and bioelectrical impedance methods. J Am Coll
information about weight loss after bariatric surgery than Nutr 1989; 8:484–489.
BMI or the percentage loss of preoperative weight alone. 8 Gudivaka R, Schoeller DA, Ho T, et al. Effect of body position, electrode
placement, and time on prediction of total body water by multifrequency
Outcome based on the loss of fat mass or changes in FMI bioelectrical impedance analysis. Age Nutr 1994; 5:111–117.
may be a more reliable measure of outcome than total 9 Deurenberg P, Andreoli A, Borg P, et al. The validity of predicted body fat
body weight loss or BMI, because the goal in bariatric percentage from body mass index and from impedance in samples of five
European populations. Eur J Clin Nutr 2001; 55:973–979.
surgery is fat loss and not weight loss. However, the use of
10 Albu JB, Murphy L, Frager DH, et al. Visceral fat and race-dependent health
FFMI and FMI require rigorous evaluations in the future risks in obese nondiabetic premenopausal women. Diabetes 1997; 46:456–
[23,24]. 462.
11 Abate N, Garg A, Peshock RM, et al. Relationship of generalized and regional
adiposity to insulin sensitivity in men with NIDDM. Diabetes 1996; 45:1684–
Conclusion 1693.
In the morbid obesity state several body-composition 12 Eisenkolbl J, Kartasurya M, Windhalm K. Understimation of percentage fat
characteristics that affect the validation of the interpreta- mass measured by bioelectrical impedance analysis compared to dual energy
X-ray absorptiometry method in obese children. Eur J Clin Nutr 2001;
tion of impedance measurements are changed. A rela- 55:429–432.
tively increased amount of TBW and a relative increase in 13 Swan PD, McConnell KE. Anthropometry and bioelectrical impedance incon-
ECW will result in an underestimation of the percentage sistently predict fatness in women with regional adiposity. Med Sci Sports
Exerc 1999; 31:1068–1075.
of body fat and an overestimation of the FFM. A different
14 Lohman TG. Advances in body composition assessment. Current issues in
body build (mainly in those with severe abdominal obe- exercise science series. Monograph 3. Champaign, IL: Human Kinetics
sity) will result in an overestimation of relative body fat Publishers; 1992.
percentage. More research is needed in order to deter- 15 Gray DS, Bray GA, Gemayel N, Kaplan K. Effect of obesity on bioelectrical
impedance. Am J Clin Nutr 1989; 50:255–260.
mine the validity of BIA in morbidly obese individuals.
16 Van Loan M, Mayclin P. Bioelectrical impedance analysis: is it a reliable estimator
of lean body mass and total body water. Hum Biol 1987; 59:299–309.
References and recommended reading 17 Jartti L, Hakanen M, Paakkunainen U, et al. Comparison of hand-to-leg and
Papers of particular interest, published within the annual period of review, have leg-to leg bioelectric impedance devices in the assessment of body adiposity
been highlighted as: in prepubertal children: the STRIP study: Special Turku Coronary Risk Factor
 of special interest Intervention Project. Acta Pediatr 2000; 89:781–786.
 of outstanding interest
18 Deurenberg P. Limitations of the bioelectrical impedance method for the
assessment of body fat in severe obesity. Am J Clin Nutr 1996;
1 Bray GA. Contemporary diagnosis and management of obesity. Newton, PA: 64(Suppl.):449S–452S.
Handbooks in Health Care; 1998.
19 Bray GA, DeLany JP, Harsha DW, et al. Evaluation of body fat in fatter and
2 Pi-Sunyer FX. Symposium on body weight regulation and obesity: metabolic leaner 10-y-old. African American and white children: the Baton Rouge
and clinical aspects, 1st plenary session: obesity. Proc Nutr Soc 2000; Children’s Study. Am J Clin Nutr 2001; 73:687–702.
59:505–509.
20 Cox-Reijven PEL, Van Kreel B, Soeters PB. Accuracy of bioelectrical
3 Segal KR, Gutin B, Presta E, et al. Estimation of human body composition by impedance spectroscopy in measuring changes in body composition during
electrical impedance methods: a comparative study. J Appl Physiol 1985; severe weight loss. J Parenter Enter Nutr 2002; 26:120–127.
58:1565–1571.
21 Ballantyne GH. Measuring outcomes following bariatric surgery: weight loss
4 Martinoli R, Mohamed EI, Maiolo C, et al. Total body water estimation using  parameters, improvement in co-morbid conditions, change in quality of life and
 bioelectrical impedance: a meta-analysis of the data available in the literature. patient satisfaction. Obesity Surg 2003; 13:954–964.
Acta Diabetol 2003; 40:S203–S206. This study discusses the validation of BIA after bariatric surgery.
This study shows that multi-frequency BIA is a more accurate method for
22 Livingston EH, Sebastian JL, Huerta S, et al. Bioexponential model for
estimating the TBW compartments for healthy and obese adult individuals.
predicting weight loss after gastric surgery for obesity. J Surg Res 2001;
5 Buchholoz AC, Bartij C, Schoeller A. The validity of bioelectrical 101:216–224.
 impedance models in clinical population. Nutr Clin Practice 2004; 19:
23 Das SK, Roberts SB, Kehayias JJ, et al. Body composition assessment in
433–446.
extreme obesity and after massive weight loss induced by gastric bypass
This paper reviews the theory and assumptions underlying BIA, how and why these
surgery. Am J Physiol Endocrinol Metab 2003; 284:E1080–E1088.
assumptions may be violated in clinical populations, and the effects these viola-
tions may have on the accuracy of BIA and BIS models. 24 Sartorio A, Conte G, Morini P, et al. Changes of bioelectrical impedance after
a body weight reduction program in highly obese subjects. Diabet Nutr Metab
6 Kyle UG, Piccoli A, Pichard C. Body composition measurements: interpreta-
2000; 13:186–191.
 tion finally made easy for clinical use. Curr Opin Clin Nutr Metab Care 2003;
6:387–393. 25 Schutz Y, Kyle UUG, Pichard C. Fat-free mass index and fat mass index
This study discusses the evaluation of nutritional status by using FFMI and FMI. percentiles of caucasians aged 18–98 y. Int J Obes 2002; 26:953–960.