- Email: [email protected]
Comparison in man of total body electrical conductivity and lean body mass derived from body density: Validation of a new body composition method
Comparison in Man of Total Body Electrical Conductivity and Lean Body Mass Derived from Body Density: Validation of a New Body Composition Method Elio Presta, Karen R. Segal, Bernard Gutin, Gail G. Harrison,
and Theodore
B. Van ltallie
This article reports a study in which total body electrical conductivity (TOBEC) measurements and lean body mass (LBM) estimated from hydrostatic weighing in human subjects were compared. The TOBEC method provides a new approach to assessment of human body composition that is based on the principle that the electrical conductivity of lean tissue is far greater than that of fat. In a sample of 32 men and women varying widely in age (20 to 53 years). body weight (45 to 155 kg), and adiposity (9.5 to 53.0% body fat), the TOBEC measurement was found to be extremely reliable (r = 0.999) and to correlate highly with hydrostatically estimated LBM (r = 0.903, P < 0.0001). When the TOBEC scores were transformed to provide a single variable; namely, the subject’s height times the square root of the TOBEC score, a higher correlation with LBM was obtained (r = 0.943). Taking gender into account further enhanced the prediction of LBM from TOBEC (r - 0.951). These observations strongly reinforce the results of a previous investigation in which high correlations were found between TOBEC and both total body potassium and total body water. Accordingly, this new method promises to provide a useful technique for the evaluation of body composition that is at once simple, rapid, objective, and noninvasive.
B
ODY
COMPOSITION ESTIMATION is an part of nutritional assessment, both in individuals and in populations. A variety of techniques for body composition estimation exist; however, there is no one technique that is at once reliable, safe, rapid, noninvasive, and applicable to hospitalized patients, and that requires neither a high level of technical skill on the part of the operator nor special cooperation from the subject. Measurement of total body electrical conductivity (TOBEC) provides a new approach to the assessment of human body composition that appears to meet all these criteria. The principle underlying the TOBEC method is that lean tissue is far more electrically conductive than fat, owing to the much greater content of electrolytes in most of the fat-free portions of the body.’ The TOBEC instrument is a large solenoidal coil driven by a ~-MHZ oscillating radiofrequency current. The difference between impedance when the subject is inside the coil and when the coil is empty is an index of the total electrical conductivity of the body,’ which, in turn, is proportional to the lean body mass (LBM) of the subject. This report describes validation of the TOBEC method through comparison with densitometry.
From The Departmeni of Medicine, College of Physicians and Surgeons, Columbia University at St Luke’s Hospital, The Applied Physiology Laboratory, Teachers College, Columbia University, New York, NY, and The Departments of Family and Community Medicine, Pediatrics and Nutrition and Food Science, College of Medicine, University of Arizona, Tucson, AZ. Supported in part by NIH Research Grants AM-26687 and AM 26688. Acceptedfor publieation December I. 1982. Address reprint requests to Dr. E. Presta. Medical Service, St. Luke’s_Roosevelt Hospital Center, Amsterdam Ave. at 114th St., New York, NY 10025. o I983 by Grune & Stratton, Inc. 0026/0495/83/3205~18$01.00/0 524
METHODS
important
Subjects Thirty-two healthy adults (16 men and 16 women) volunteered to participate in this study. Data on the subjects are given in Table 1. After the nature of the study and all procedures were explained, the written informed consent of the subjects was obtained. Both the consent form and the protocol for the study were reviewed and approved by the Institutional Review Board of the St. Luke’sRoosevelt Institute for Health Sciences. The underwater weight, residual volume, and TOBEC of each subject were measured on the same day, usually within the space of three hours.
TOBEC Total body electrical conductivity (TOBEC) was measured by means of a prototype instrument designed specifically for use in human subjects (The EMME Co., Dickey-john Corp., Auburn, IL). Safety standards for exposure to radio waves in the range of 5 x I O6 cycles/second used by the TOBEC device have not yet been established in the US, however, the maximal energy flux. as measured in the TOBEC device, is 0.01 mW/cm’, which is l/l000 of the US limit for microwave exposure over an eight-hour period. Since the exposure time in the TOBEC device is only a small fraction of this period (usually less than one minute), there is a very large margin of safety. The TOBEC instrument was calibrated daily with a standard designed to produce a constant signal. The calibration device consisted of a small rectangular wooden frame wound around which were several loops of copper wire in series with a precision transistor. The device was placed in the center of the magnetic field within the instrument with the wire loops coaxial to the coil, and the instrument was then adjusted to yield the appropriate constant signal. The subjects were weighed on an electronic piezoelectric scale accurate to *O.lO kg. The subjects lay comfortably on a stretcher on rollers connected to the instrument, which was inserted into the device at the time of measurement. Each individual reading was displayed digitally by the instrument’s computer and represented the mean of ten determinations obtained automatically within a losecond period. Ten successive readings were secured for each subject, and the average value was taken as the TOBEC score. Each individual trial was accomplished in roughly 15 seconds; thus, the ten measurements were completed in less than three minutes. Metabolism, Vol. 32, No. 5 (May), 1983
BODY ELECTRICAL CONDUCTIVITY
525
IN MAN
Table 1. Characteristics
-
Women
of the Subjects
(n = 16)
MenIn=
P value
16)
Age
31.5
? 6.2
32.3
+ 10.4
“.S.
Weight (kg)
63.9
& 17.3
83.7
? 24.9
.-0.05
Height (ml
1.63 ? 0.05
1.77 f; 0.09
BMI (weight/height’)
23.8
t 6.0
26.5
i 6.6
“.s.
+ 10.9
21.8
* 11.9
“.s.
25.47
Percentage of Fat*
1.0416
Body density (gm/ml)t
f 0.0234
1.0498
+ 0.0260
LBM (kg)
45.2
+ 6.7
63.1
i 9.3
TOBEC
14.4 * 5.6
24.2
r 7.6
6.1 + 1.2
TOBEC 5 x height
-:O.OOl
n,s. -0.001 \O.OOl
8.7 r 1.6
io.001
Values are mean + SD. *From hydrostatic weighing: % Fat = (4.95/D
- 4.5) x 100.
t From hydrostatic weighing: LBM = Weight -
(Weight x % Fat).
Densitonletry flydrostatic
weighing was used to determine
body density. The
subjects were tested in the morning after an overnight weight in ambient
air was measured
Tht, subjects wore a 5-kg diver’s counterweight submersion.
An additional
2-kg
fast. Body
before hydrostatic
weight
weighing.
belt to facilitate
was used for the obese
subjects. A swing seat was suspended in the water from the Chatillon 15.xg
scale used for determining
weight was determined weight
of the subject.
maximally remain
The
subjects
weight.
tare
were
instructed
to expire
as they submerged beneath the surface of the water and
five seconds while the underwater
time was allotted
was repeated
exhalation
was recorded.
for the subjects to practice,
the procedure
ten times. The highest weight that was reproduced
temperature
maintained
for
weight
three times was used as an estimate of the underwater water
The
from the total underwater
as strll as possible at the point of maximal
roughly After
underwater
and subtracted
was recorded
after
each
weight.j The
weighing
and
was
between 28 and 30 C.
Residual
lung volume was determined
out of the water while the
suqtects assumed a similar position to &hat used during the hydrostat c weighing. The closed circuit oxygen dilution technique4 was followed using a Collins 9-liter Venek BTI’S
spirometer
and a Hewlett
Packard
series nitrogen analyzer. The mean of two trials, corrected to (body temperature,
pressure. and saturation)
conditions, was
obtained.
Body density was calculated according to the equation of
Goldman
and Buskirk’
the Siri equation.6 manner: LBM
and the percentage of fat was derived using
Lean body mass was derived
in the following
= weight ~ (weight x % fat).
RESULTS
Descriptive statistics for the sample and the results of the TOBEC measurement and densitometry are Table 2.
-.-
Correlations
shown in Table 1. The sample was heterogeneous with respect to body composition, with the range extending from extreme leanness to extreme obesity. Unpaired t-tests indicated that the mean age, body mass index (BMI), body density, and percentage of fat were similar for the men and women. Body weight, height, LBM, and TOBEC were significantly higher for the men. The reliability of the TOBEC and residual volume measurements was tested using the intraclass correlation method.’ The reliability of the TOBEC method was excellent: for the mean of ten trials the reliability coefficient was r = 0.9999 and for a single trial the reliability coefficient was r = 0.999 1. The test-retest reliability for the two residual volume trials was r = 0.980. Multiple regression analyses (SPSS New Regression computer program) were carried out to derive equations to predict LBM from the TOBEC score.’ The dependent variable was LBM determined densitometrically. The contributions of the addition of variables such as gender, height, and age into the prediction equation were examined by means of stepwise regression. Gender was entered into the regression equation as a dummy variable (men = 0, women = I). Separate equations were tested using: ( 1) the untransformed TOBEC score, and (2) the square root of the TOBEC score times height, in order to maximize the Among Selected
Variables
I__
_____
BMI
Weight Vanable
-Age
Age
Weight
0.292
Height
- 0.066
TOBEC’
% Fat*
LBM’
TOBEC
x Height
___-
1.000 0.601 t
1 .ooo 0.271
8MI
0.422
0.924t
% Fat
0.275
0.650t
LBM TOBEC
0.155 0.315
0.803t 0.942t
TOBEC’ x height
0.22 1
0.895t
< 0.05.
(m)
(weight/ height’)
1.000
’ ! lstermined by hydrostatic weighing. tr
(kg)
Helghf
1.000 0.790t
1.000
0.841 t
0.584t
0.092
1.000
0.718t
0.812t
0.435t
0.903t
1.000
0.842t
0.703t
0.329
0.943t
0.978
-0.000
1.000
526
PRESTA ET AL
Table 3. Multiple Regression Equations for Predicting LBM from TOBEC Predictor Variables
R
I72
S.E.E.
F
F change
Equation 1: Xl
TOBEC
0.903
0.815
5.115
132.5
X2
Height
0.944
0.891
3.981
20.5
x3
(Sex)
0.951
0.905
3.800
3.9
LBM = -42.216
+ 0.87891X1)
P < 0.000 1 P < 0.001 n.s.
+ 46.9256(X2)
Equation 2: Xl
TOBEC?’ x hei&t
x2
Sex
0.943
0.889
3.968
0.951
0.903
3.763
L8M = 18.323 + 5.1721(X1) -
239.9 4.4
P < 0.0001 P < 0.05
3.8158(X2)
coefficient of determination and minimize the standard error of estimate S.E.E.).9 The product-moment correlations among selected variables are shown in Table 2. The transformed TOBEC score was a better predictor of LBM (r = 0.943) than the raw TOBEC score (r = 0.903), yielding a coefficient of determination of 0.89 instead of 0.82. The addition of gender into the prediction equation increased the fraction of explained variance by 1.5% (P < 0.05). When the untransformed TOBEC score and height were entered as separate independent variables, a similar but slightly smaller coefficient of determination was obtained, and the addition of gender at the third step did not achieve statistical significance (Table 3). The first set of predictors, however, yielded a smaller standard error of estimate and thus might be considered both more parsimonious and precise.g No systematic pattern was found in the residuals. DISCUSSION
The high correlation between the TOBEC scores and LBM estimated from body density suggests that this new approach to the estimation of body composition in humans is extremely promising. The rapidity and convenience of this method are currently unequalled by any of the other available techniques for body composition assessment. Some of the advantages of this technique for evaluating body composition are:
(1) no special skill is required of either the operator of the instrument or the subject; (2) it is suitable for hospitalized patients; (3) it is safe and noninvasive; and (4) the rapidity of measurement makes it appropriate for nutritional surveys and epidemiologic studies. The high correlation between TOBEC and densitometry is encouraging if one keeps in mind that aside from the error in measurement of body density and residual volume, the estimation of LBM from body density rests on assumptions about the biologic constancy of variables such as the density of lean tissue, the proportional content of water and bone in LBM, and the cell water content.” These assumptions are certainly questionable and, as Siri points out, even if the determination of body density were free of any measurement error, some uncertainty about the accuracy of the estimation of body composition would still exist, owing to variability in the composition of body constituents.6 The high correlations obtained between TOBEC and both TBK (r = 0.86) and TBW (r = 0.87), in a previous investigation,” further attest to the validity of the TOBEC method. These latter methods have their own errors of measurement and, moreover, are based upon average values for the potassium and water content of LBM, which may vary among individuals. It is particularly encouraging that different techniques, based on different assumptions, have yielded similar results. Studies are currently under way to compare TOBEC with direct analysis of body constituents in animals. The correlation between TOBEC and LBM estimated from direct analysis may be even higher since the error associated with the individual variability in the density and composition of lean body mass would be eliminated. Additional research is needed to examine the effect of changes in the degree of hydration and osmolality of LBM on TOBEC and to determine the sensitivity of TOBEC to small changes in LBM and total body fat, as would occur during dynamic states of weight gain or loss. If the TOBEC method proves capable of detecting relatively small changes in the total LBM of patients, then it would be an invaluable method for monitoring the progress of patients whose nutritional status is of clinical concern.
REFERENCES I. Pethig R: Dielectric and Electronic Properties of Biological Materials. Chichester, England, John Wiley and Sons, 1979, pp 225-235 2. US Patent 3735247: Method and apparatus for measuring fat content in animal tissue either in vivo or in slaughtered and prepared form. May 22,1973, W.H. Harker, inventor. The EMME Company, Assignee. US Patent Office 3. Behnke AR, Wilmore JH: Evaluation of Body Build and Composition. Englewood Cliffs, NJ, Prentice-Hall, 1974
4. Wilmore JH: A simplified method for determination ual lung volumes. J Appl Physiol 27:96-100, 1969
of resid-
5. Goldman RF, Buskirk ER: Body volume measurement by underwater weighing: description of a method, in Brozek J, Henschel A (eds): Techniques for Measuring Body Composition. Washington, DC, National Academy of Science, 1961, pp 78-89 6. Siri WE: Body composition from fluid spaces and density: Analysis of methods, in Brozek J, Henschel A (eds): Techniques for
BODY ELECTRICAL C0NDUCTlVll-Y
Measuring
Body Composition.
of Science,
1961, pp 2233244
IN MAN
Washington,
527
DC, National
Academy
T’. Winer BJ: Statistical Principles in Experimental Design, ed 2. New York, McGraw-Hill, 1962, pp 283-289 R. Nie NH, Hull CH (eds): SPSS Update 7-9: New Procedures and Facilities for Releases 7-9. New York, McGraw-Hill, 198 I, pp 94-120 9. Kerlinger FN. Pedhazur ES: Multiple Regression in Behav-
ioral Research. 40,66-72
New York, Holt, Rinehart
10. Wedgewood JR: Inconstancy Acad Sci 1 IO: 140-l 52. 1963
and Winston,
1973, pp
of lean body mass. Ann NY
11. Presta E, Wang J, Harrison GG, Bjijrntorn P. et al: Measurement of total body electrical conductivity: A new method for the estimation of human body composition. Am J Clin Nutr (in press)
and Theodore
B. Van ltallie
This article reports a study in which total body electrical conductivity (TOBEC) measurements and lean body mass (LBM) estimated from hydrostatic weighing in human subjects were compared. The TOBEC method provides a new approach to assessment of human body composition that is based on the principle that the electrical conductivity of lean tissue is far greater than that of fat. In a sample of 32 men and women varying widely in age (20 to 53 years). body weight (45 to 155 kg), and adiposity (9.5 to 53.0% body fat), the TOBEC measurement was found to be extremely reliable (r = 0.999) and to correlate highly with hydrostatically estimated LBM (r = 0.903, P < 0.0001). When the TOBEC scores were transformed to provide a single variable; namely, the subject’s height times the square root of the TOBEC score, a higher correlation with LBM was obtained (r = 0.943). Taking gender into account further enhanced the prediction of LBM from TOBEC (r - 0.951). These observations strongly reinforce the results of a previous investigation in which high correlations were found between TOBEC and both total body potassium and total body water. Accordingly, this new method promises to provide a useful technique for the evaluation of body composition that is at once simple, rapid, objective, and noninvasive.
B
ODY
COMPOSITION ESTIMATION is an part of nutritional assessment, both in individuals and in populations. A variety of techniques for body composition estimation exist; however, there is no one technique that is at once reliable, safe, rapid, noninvasive, and applicable to hospitalized patients, and that requires neither a high level of technical skill on the part of the operator nor special cooperation from the subject. Measurement of total body electrical conductivity (TOBEC) provides a new approach to the assessment of human body composition that appears to meet all these criteria. The principle underlying the TOBEC method is that lean tissue is far more electrically conductive than fat, owing to the much greater content of electrolytes in most of the fat-free portions of the body.’ The TOBEC instrument is a large solenoidal coil driven by a ~-MHZ oscillating radiofrequency current. The difference between impedance when the subject is inside the coil and when the coil is empty is an index of the total electrical conductivity of the body,’ which, in turn, is proportional to the lean body mass (LBM) of the subject. This report describes validation of the TOBEC method through comparison with densitometry.
From The Departmeni of Medicine, College of Physicians and Surgeons, Columbia University at St Luke’s Hospital, The Applied Physiology Laboratory, Teachers College, Columbia University, New York, NY, and The Departments of Family and Community Medicine, Pediatrics and Nutrition and Food Science, College of Medicine, University of Arizona, Tucson, AZ. Supported in part by NIH Research Grants AM-26687 and AM 26688. Acceptedfor publieation December I. 1982. Address reprint requests to Dr. E. Presta. Medical Service, St. Luke’s_Roosevelt Hospital Center, Amsterdam Ave. at 114th St., New York, NY 10025. o I983 by Grune & Stratton, Inc. 0026/0495/83/3205~18$01.00/0 524
METHODS
important
Subjects Thirty-two healthy adults (16 men and 16 women) volunteered to participate in this study. Data on the subjects are given in Table 1. After the nature of the study and all procedures were explained, the written informed consent of the subjects was obtained. Both the consent form and the protocol for the study were reviewed and approved by the Institutional Review Board of the St. Luke’sRoosevelt Institute for Health Sciences. The underwater weight, residual volume, and TOBEC of each subject were measured on the same day, usually within the space of three hours.
TOBEC Total body electrical conductivity (TOBEC) was measured by means of a prototype instrument designed specifically for use in human subjects (The EMME Co., Dickey-john Corp., Auburn, IL). Safety standards for exposure to radio waves in the range of 5 x I O6 cycles/second used by the TOBEC device have not yet been established in the US, however, the maximal energy flux. as measured in the TOBEC device, is 0.01 mW/cm’, which is l/l000 of the US limit for microwave exposure over an eight-hour period. Since the exposure time in the TOBEC device is only a small fraction of this period (usually less than one minute), there is a very large margin of safety. The TOBEC instrument was calibrated daily with a standard designed to produce a constant signal. The calibration device consisted of a small rectangular wooden frame wound around which were several loops of copper wire in series with a precision transistor. The device was placed in the center of the magnetic field within the instrument with the wire loops coaxial to the coil, and the instrument was then adjusted to yield the appropriate constant signal. The subjects were weighed on an electronic piezoelectric scale accurate to *O.lO kg. The subjects lay comfortably on a stretcher on rollers connected to the instrument, which was inserted into the device at the time of measurement. Each individual reading was displayed digitally by the instrument’s computer and represented the mean of ten determinations obtained automatically within a losecond period. Ten successive readings were secured for each subject, and the average value was taken as the TOBEC score. Each individual trial was accomplished in roughly 15 seconds; thus, the ten measurements were completed in less than three minutes. Metabolism, Vol. 32, No. 5 (May), 1983
BODY ELECTRICAL CONDUCTIVITY
525
IN MAN
Table 1. Characteristics
-
Women
of the Subjects
(n = 16)
MenIn=
P value
16)
Age
31.5
? 6.2
32.3
+ 10.4
“.S.
Weight (kg)
63.9
& 17.3
83.7
? 24.9
.-0.05
Height (ml
1.63 ? 0.05
1.77 f; 0.09
BMI (weight/height’)
23.8
t 6.0
26.5
i 6.6
“.s.
+ 10.9
21.8
* 11.9
“.s.
25.47
Percentage of Fat*
1.0416
Body density (gm/ml)t
f 0.0234
1.0498
+ 0.0260
LBM (kg)
45.2
+ 6.7
63.1
i 9.3
TOBEC
14.4 * 5.6
24.2
r 7.6
6.1 + 1.2
TOBEC 5 x height
-:O.OOl
n,s. -0.001 \O.OOl
8.7 r 1.6
io.001
Values are mean + SD. *From hydrostatic weighing: % Fat = (4.95/D
- 4.5) x 100.
t From hydrostatic weighing: LBM = Weight -
(Weight x % Fat).
Densitonletry flydrostatic
weighing was used to determine
body density. The
subjects were tested in the morning after an overnight weight in ambient
air was measured
Tht, subjects wore a 5-kg diver’s counterweight submersion.
An additional
2-kg
fast. Body
before hydrostatic
weight
weighing.
belt to facilitate
was used for the obese
subjects. A swing seat was suspended in the water from the Chatillon 15.xg
scale used for determining
weight was determined weight
of the subject.
maximally remain
The
subjects
weight.
tare
were
instructed
to expire
as they submerged beneath the surface of the water and
five seconds while the underwater
time was allotted
was repeated
exhalation
was recorded.
for the subjects to practice,
the procedure
ten times. The highest weight that was reproduced
temperature
maintained
for
weight
three times was used as an estimate of the underwater water
The
from the total underwater
as strll as possible at the point of maximal
roughly After
underwater
and subtracted
was recorded
after
each
weight.j The
weighing
and
was
between 28 and 30 C.
Residual
lung volume was determined
out of the water while the
suqtects assumed a similar position to &hat used during the hydrostat c weighing. The closed circuit oxygen dilution technique4 was followed using a Collins 9-liter Venek BTI’S
spirometer
and a Hewlett
Packard
series nitrogen analyzer. The mean of two trials, corrected to (body temperature,
pressure. and saturation)
conditions, was
obtained.
Body density was calculated according to the equation of
Goldman
and Buskirk’
the Siri equation.6 manner: LBM
and the percentage of fat was derived using
Lean body mass was derived
in the following
= weight ~ (weight x % fat).
RESULTS
Descriptive statistics for the sample and the results of the TOBEC measurement and densitometry are Table 2.
-.-
Correlations
shown in Table 1. The sample was heterogeneous with respect to body composition, with the range extending from extreme leanness to extreme obesity. Unpaired t-tests indicated that the mean age, body mass index (BMI), body density, and percentage of fat were similar for the men and women. Body weight, height, LBM, and TOBEC were significantly higher for the men. The reliability of the TOBEC and residual volume measurements was tested using the intraclass correlation method.’ The reliability of the TOBEC method was excellent: for the mean of ten trials the reliability coefficient was r = 0.9999 and for a single trial the reliability coefficient was r = 0.999 1. The test-retest reliability for the two residual volume trials was r = 0.980. Multiple regression analyses (SPSS New Regression computer program) were carried out to derive equations to predict LBM from the TOBEC score.’ The dependent variable was LBM determined densitometrically. The contributions of the addition of variables such as gender, height, and age into the prediction equation were examined by means of stepwise regression. Gender was entered into the regression equation as a dummy variable (men = 0, women = I). Separate equations were tested using: ( 1) the untransformed TOBEC score, and (2) the square root of the TOBEC score times height, in order to maximize the Among Selected
Variables
I__
_____
BMI
Weight Vanable
-Age
Age
Weight
0.292
Height
- 0.066
TOBEC’
% Fat*
LBM’
TOBEC
x Height
___-
1.000 0.601 t
1 .ooo 0.271
8MI
0.422
0.924t
% Fat
0.275
0.650t
LBM TOBEC
0.155 0.315
0.803t 0.942t
TOBEC’ x height
0.22 1
0.895t
< 0.05.
(m)
(weight/ height’)
1.000
’ ! lstermined by hydrostatic weighing. tr
(kg)
Helghf
1.000 0.790t
1.000
0.841 t
0.584t
0.092
1.000
0.718t
0.812t
0.435t
0.903t
1.000
0.842t
0.703t
0.329
0.943t
0.978
-0.000
1.000
526
PRESTA ET AL
Table 3. Multiple Regression Equations for Predicting LBM from TOBEC Predictor Variables
R
I72
S.E.E.
F
F change
Equation 1: Xl
TOBEC
0.903
0.815
5.115
132.5
X2
Height
0.944
0.891
3.981
20.5
x3
(Sex)
0.951
0.905
3.800
3.9
LBM = -42.216
+ 0.87891X1)
P < 0.000 1 P < 0.001 n.s.
+ 46.9256(X2)
Equation 2: Xl
TOBEC?’ x hei&t
x2
Sex
0.943
0.889
3.968
0.951
0.903
3.763
L8M = 18.323 + 5.1721(X1) -
239.9 4.4
P < 0.0001 P < 0.05
3.8158(X2)
coefficient of determination and minimize the standard error of estimate S.E.E.).9 The product-moment correlations among selected variables are shown in Table 2. The transformed TOBEC score was a better predictor of LBM (r = 0.943) than the raw TOBEC score (r = 0.903), yielding a coefficient of determination of 0.89 instead of 0.82. The addition of gender into the prediction equation increased the fraction of explained variance by 1.5% (P < 0.05). When the untransformed TOBEC score and height were entered as separate independent variables, a similar but slightly smaller coefficient of determination was obtained, and the addition of gender at the third step did not achieve statistical significance (Table 3). The first set of predictors, however, yielded a smaller standard error of estimate and thus might be considered both more parsimonious and precise.g No systematic pattern was found in the residuals. DISCUSSION
The high correlation between the TOBEC scores and LBM estimated from body density suggests that this new approach to the estimation of body composition in humans is extremely promising. The rapidity and convenience of this method are currently unequalled by any of the other available techniques for body composition assessment. Some of the advantages of this technique for evaluating body composition are:
(1) no special skill is required of either the operator of the instrument or the subject; (2) it is suitable for hospitalized patients; (3) it is safe and noninvasive; and (4) the rapidity of measurement makes it appropriate for nutritional surveys and epidemiologic studies. The high correlation between TOBEC and densitometry is encouraging if one keeps in mind that aside from the error in measurement of body density and residual volume, the estimation of LBM from body density rests on assumptions about the biologic constancy of variables such as the density of lean tissue, the proportional content of water and bone in LBM, and the cell water content.” These assumptions are certainly questionable and, as Siri points out, even if the determination of body density were free of any measurement error, some uncertainty about the accuracy of the estimation of body composition would still exist, owing to variability in the composition of body constituents.6 The high correlations obtained between TOBEC and both TBK (r = 0.86) and TBW (r = 0.87), in a previous investigation,” further attest to the validity of the TOBEC method. These latter methods have their own errors of measurement and, moreover, are based upon average values for the potassium and water content of LBM, which may vary among individuals. It is particularly encouraging that different techniques, based on different assumptions, have yielded similar results. Studies are currently under way to compare TOBEC with direct analysis of body constituents in animals. The correlation between TOBEC and LBM estimated from direct analysis may be even higher since the error associated with the individual variability in the density and composition of lean body mass would be eliminated. Additional research is needed to examine the effect of changes in the degree of hydration and osmolality of LBM on TOBEC and to determine the sensitivity of TOBEC to small changes in LBM and total body fat, as would occur during dynamic states of weight gain or loss. If the TOBEC method proves capable of detecting relatively small changes in the total LBM of patients, then it would be an invaluable method for monitoring the progress of patients whose nutritional status is of clinical concern.
REFERENCES I. Pethig R: Dielectric and Electronic Properties of Biological Materials. Chichester, England, John Wiley and Sons, 1979, pp 225-235 2. US Patent 3735247: Method and apparatus for measuring fat content in animal tissue either in vivo or in slaughtered and prepared form. May 22,1973, W.H. Harker, inventor. The EMME Company, Assignee. US Patent Office 3. Behnke AR, Wilmore JH: Evaluation of Body Build and Composition. Englewood Cliffs, NJ, Prentice-Hall, 1974
4. Wilmore JH: A simplified method for determination ual lung volumes. J Appl Physiol 27:96-100, 1969
of resid-
5. Goldman RF, Buskirk ER: Body volume measurement by underwater weighing: description of a method, in Brozek J, Henschel A (eds): Techniques for Measuring Body Composition. Washington, DC, National Academy of Science, 1961, pp 78-89 6. Siri WE: Body composition from fluid spaces and density: Analysis of methods, in Brozek J, Henschel A (eds): Techniques for
BODY ELECTRICAL C0NDUCTlVll-Y
Measuring
Body Composition.
of Science,
1961, pp 2233244
IN MAN
Washington,
527
DC, National
Academy
T’. Winer BJ: Statistical Principles in Experimental Design, ed 2. New York, McGraw-Hill, 1962, pp 283-289 R. Nie NH, Hull CH (eds): SPSS Update 7-9: New Procedures and Facilities for Releases 7-9. New York, McGraw-Hill, 198 I, pp 94-120 9. Kerlinger FN. Pedhazur ES: Multiple Regression in Behav-
ioral Research. 40,66-72
New York, Holt, Rinehart
10. Wedgewood JR: Inconstancy Acad Sci 1 IO: 140-l 52. 1963
and Winston,
1973, pp
of lean body mass. Ann NY
11. Presta E, Wang J, Harrison GG, Bjijrntorn P. et al: Measurement of total body electrical conductivity: A new method for the estimation of human body composition. Am J Clin Nutr (in press)