Professional Documents
Culture Documents
Jappl 2000 89 2 465 PDF
Jappl 2000 89 2 465 PDF
Janssen, Ian, Steven B. Heymsfield, Richard N. in evaluating the influence of physical training on
Baumgartner, and Robert Ross. Estimation of skeletal muscle mass.
muscle mass by bioelectrical impedance analysis. J Appl The use of bioelectrical impedance analysis (BIA) in
Physiol 89: 465–471, 2000.—The purpose of this study was to the study of human body composition has grown rap-
develop and cross-validate predictive equations for estimat-
ing skeletal muscle (SM) mass using bioelectrical impedance
idly in the past two decades. BIA is a noninvasive,
analysis (BIA). Whole body SM mass, determined by mag- portable, quick, and inexpensive method for measuring
netic resonance imaging, was compared with BIA measure- body composition. BIA is based on the relation between
ments in a multiethnic sample of 388 men and women, aged the volume of a conductor and its electrical resistance.
18–86 yr, at two different laboratories. Within each labora- Because SM is the largest tissue in the body (27) and
tory, equations for predicting SM mass from BIA measure- because it is an electrolyte-rich tissue with a low resis-
ments were derived using the data of the Caucasian subjects. tance (1, 13, 28), muscle is a dominant conductor.
These equations were then applied to the Caucasian subjects Previous studies have shown that there is a strong
from the other laboratory to cross-validate the BIA method. correlation between BIA resistance and SM measure-
Because the equations cross-validated (i.e., were not differ- ments in the arms (8, 25) and legs (24, 25). However,
ent), the data from both laboratories were pooled to generate
the final regression equation
these studies were limited in that a criterion method
for measuring SM was not employed, whole body mus-
SM mass 共kg兲 ⫽ 关共Ht2ⲐR ⫻ 0.401兲 ⫹ 共gender ⫻ 3.825兲 cle mass was not measured, and small sample sizes
were used.
⫹ 共age ⫻ ⫺0.071兲兴 ⫹ 5.102
Magnetic resonance imaging (MRI) provides precise
where Ht is height in centimeters; R is BIA resistance in and reliable measurements of SM (6, 11, 23), is consid-
ohms; for gender, men ⫽ 1 and women ⫽ 0; and age is in ered a reference method for measuring SM in vivo (16,
years. The r2 and SE of estimate of the regression equation 21), and is ideally suited for measuring whole body SM
were 0.86 and 2.7 kg (9%), respectively. The Caucasian- mass. Therefore, the aim of this study was to develop
derived equation was applicable to Hispanics and African- and cross-validate a predictive equation for estimating
Americans, but it underestimated SM mass in Asians. These MRI-measured SM mass by using a conventional BIA
results suggest that the BIA equation provides valid esti-
analyzer in a large and heterogeneous sample of men
mates of SM mass in healthy adults varying in age and
adiposity. and women.
body composition; prediction equation METHODS
http://www.jap.org 8750-7587/00 $5.00 Copyright © 2000 the American Physiological Society 465
the Hispanic, African-American, and Asian subjects to deter- and between the medial and lateral malleoli at the ankle
mine whether the equation was valid for all races. (22). All resistance measurements were adjusted for stature
Subjects. Subjects consisted of healthy adult men (n ⫽ 230) [height (Ht)2/R; cm2/⍀].
and women (n ⫽ 158) who had participated in a variety of Body mass was measured to the nearest 0.1 kg, with the
body composition studies at Queen’s University (Kingston, subjects dressed in light clothing. Barefoot standing height
ON, Canada) and St. Luke’s/Roosevelt Hospital (New York, was measured to the nearest 0.1 cm by using a wall-mounted
NY). The subjects varied in age (18–86 yr), body mass index stadiometer.
(16–48 kg/m2), and race (269 Caucasian, 53 African-Ameri- Reliability of MRI and BIA measurements. Our laboratory
can, 40 Asian, 26 Hispanic). Subjects were recruited from recently determined the reproducibility of MRI-determined
among hospital employees, students at local universities, and SM measurements by comparing the intra- and interobserver
the general public through posted flyers and the local media. estimates for MRI measurements of one series of seven im-
All participants gave informed consent before participation, ages taken in the legs in three male and three female subjects
in accordance with the ethical guidelines of the respective (23). The interobserver difference was 1.8 ⫾ 0.6%, and the
institutional review boards. intraobserver difference was 0.34 ⫾ 1.1% (23). The intraob-
SM measurement by MRI. At both institutions, the MRI server difference was calculated by comparing the analysis of
images were obtained with a General Electric 1.5-T scanner
two separate MRI acquisitions in a single observer, whereas
(Milwaukee, WI). A T1-weighted, spin-echo sequence with a
the interobserver difference was determined by comparing
210-ms repetition time and a 17-ms echo time was used to
two observers⬘ analyses of the same images. We determined
obtain the MRI data. The MRI protocol is described in detail
the reproducibility of MRI-determined SM measurements
elsewhere (26). Briefly, the subjects lay in the magnet in a
prone position with their arms placed straight overhead. across the laboratories by comparing the two laboratories’
With the use of the intervertebral space between the fourth analysis of the same whole body images for five subjects. The
and fifth lumbar vertebrae as the point of origin, 10-mm- interlaboratory difference was 2.0 ⫾ 1.2%.
thick transverse images were obtained every 40 mm from Previous studies evaluating the reliability of BIA measure-
hand to foot, resulting in a total of ⬃41 images for each ments indicate that the coefficients of variation range from
subject. The total time required to acquire all of the MRI data 1.8 to 2.9% (18).
for each subject was ⬃25 min. All MRI data were transferred Statistical analysis. Differences between the Caucasian
to a computer workstation (Silicon Graphics, Mountain View, subject characteristics from the two laboratories were tested
CA) for analysis with specially designed image-analysis soft- for significance by a paired t-test (Table 1). Racial differences
ware (Tomovision, Montreal, PQ, Canada). were tested for significance by an analysis of variance (Table
Segmentation and calculation of tissue area, volume, and 1). Significant differences (P ⬍ 0.05) were analyzed by using
mass. The model used to segment the various tissues is fully a Scheffé’s post hoc comparison technique.
described and illustrated elsewhere (23, 26). Briefly, a mul- Multiple-regression analysis was applied to the data of the
tiple-step procedure was used to identify tissue area for a Caucasian subjects within each laboratory to derive the best-
given MRI image. In the first step, one of two techniques was fitting regression equations to predict MRI-measured SM
used. Either a threshold was selected for adipose tissue and mass (Table 2). The regression analyses were conducted in a
lean tissue on the basis of the gray-level histograms of the stepwise manner, using the independent variables of Ht2/R,
images (26), or a filter distinguished between different gray- age, gender, and weight. Multiple-regression analysis and
level regions on the images and lines were drawn around the analysis of variance were used to determine the equality of
different regions using a Watershed algorithm (23). Next, the regression slopes and intercepts for the relationships be-
observer labeled the different tissues by assigning them tween SM mass measured by MRI and the BIA prediction
different codes. Each image was then reviewed by an inter- equation (19) (Fig. 1).
active slice-editor program that allowed for verification and, A cross-validation of the equations for predicting SM from
where necessary, correction of the segmented results. The BIA was performed wherein the best-fitting equation derived
original gray-level image was superimposed on the binary from the Caucasian subjects within each laboratory was
segmented image by using a transparency mode to facilitate applied to the Caucasian subjects in the other laboratory.
the corrections. The area of the respective tissues in each Multiple-regression analysis and analysis of variance were
image were computed automatically by summing the number used to determine the equality of regression slopes and in-
of pixels and multiplying by the individual pixel surface area.
tercepts for the relationships between SM mass measured by
The volume of each tissue in each slice was calculated by
MRI and the BIA prediction equation derived from the other
multiplying the tissue area by the slice thickness of 10 mm.
laboratory (19) (Fig. 2).
The volume of the 40-mm space between two consecutive
slices was calculated by using a mathematical algorithm The difference between MRI-measured and BIA-predicted
given elsewhere (23). Volume units were converted to mass SM mass within the Caucasians was tested for significance
units by multiplying the volumes by the assumed constant by a paired t-test. The difference between MRI-measured and
density for adipose tissue-free SM (1.04 kg/l) (27). BIA-predicted SM within the Caucasians was also plotted
Bioelectrical resistance and anthropometric variables. Bio- against the mean of MRI-measured and BIA-predicted SM to
electrical resistance (R) was measured with a model 101B explore for systematic differences, as suggested by Bland and
BIA analyzer (RJL Systems, Detroit, MI) with an operating Altman (Fig. 3B) (7).
frequency of 50 kHz at 800 A. Whole body measurements To determine whether race influenced the BIA prediction
were taken with the subject in a supine position on a non- equation, the final equation derived from the Caucasian
conducting surface, with the arms slightly abducted from the subjects was applied to the Hispanic, African-American, and
trunk and the legs slightly separated. Surface electrodes Asian subjects (Fig. 5). Multiple-regression analysis and
were placed on the right side of the body on the dorsal surface analysis of variance were used to determine the equality of
of the hands and feet proximal to the metacarpal-phalangeal regression slopes and intercepts for the relationships be-
and metatarsal-phalangeal joints, respectively, and also me- tween SM mass measured by MRI and the BIA prediction
dially between the distal prominences of the radius and ulna equation derived from the Caucasians (19).
Age, yr 41.9 ⫾ 14.5 41.3 ⫾ 12.1 41.5 ⫾ 12.8 36.6 ⫾ 11.6 31.8 ⫾ 9.8† 33.5 ⫾ 11.1†
Gender, women/men 44/39* 55/131 99/170 29/24 17/23 13/13
Body mass, kg 72.1 ⫾ 15.1* 93.9 ⫾ 16.2 86.8 ⫾ 18.7 76.1 ⫾ 14.7† 62.0 ⫾ 10.4†‡ 72.5 ⫾ 15.9†
Body mass index,
kg/m2 24.2 ⫾ 3.5* 31.1 ⫾ 4.9 28.9 ⫾ 5.5 26.3 ⫾ 4.4† 22.0 ⫾ 2.5†‡¶ 26.1 ⫾ 3.9†
Skeletal muscle
mass, kg 26.4 ⫾ 7.6* 31.1 ⫾ 6.6 29.6 ⫾ 7.2 28.0 ⫾ 6.8 23.6 ⫾ 5.3†‡ 27.9 ⫾ 9.4
Resistance index
(Ht/R), cm2/⍀ 57.9 ⫾ 13.0* 64.5 ⫾ 11.9 62.5 ⫾ 12.6 57.8 ⫾ 12.8 52.5 ⫾ 10.4† 56.1 ⫾ 15.1
Values are as group means ⫾ SD; n, no. of subjects. Ht, height; R, bioelectrical resistance. * Laboratory A Caucasians significantly different
from laboratory B Caucasians, P ⬍ 0.05. † Significantly less than Caucasians (laboratories A ⫹ B), P ⬍ 0.05. ‡ Significantly less than
African-Americans, P ⬍ 0.05. ¶ Significantly less than Hispanics, P ⬍ 0.05.
Data are expressed as group means ⫾ SD. The 0.05 level of not statistically (P ⬎ 0.1) different from the lines of
significance was used for all data analysis. Data were ana- identify.
lyzed by using SYSTAT (Evanston, IL). Because the r2 and SEE values of the validation and
cross-validation equations were similar (Figs. 1 and 2),
RESULTS and because the slopes and intercepts of the regression
lines derived from the cross-validated equations were
Subject characteristics. The characteristics for the
not different from one and zero, respectively, we pooled
subjects within the two laboratories are given in Table
the Caucasian subjects (laboratories A ⫹ B, n ⫽ 269) to
1. With the exception of age, the Caucasian subjects in
generate the final regression equation
laboratory A were different (P ⬍ 0.05) from the Cau-
casian subjects in laboratory B for all characteristics SM mass 共kg兲 ⫽ 关共Ht2ⲐR ⫻ 0.401兲 ⫹ 共gender ⫻ 3.825兲
listed in Table 1. There were also a number of racial
⫹ 共age ⫻ ⫺ 0.071兲] ⫹ 5.102
differences for age, weight, BMI, SM mass, and Ht2/R
(Table 1). where Ht is in centimeters; R is in ohms; for gender,
Derivation of the BIA regression equations. The rela- men ⫽ 1 and women ⫽ 0; and age is in years. Analysis
tionship between SM mass as the dependent variable revealed that the slope and intercept were not signifi-
and Ht2/R, gender, age, and weight as independent cantly (P ⬎ 0.9) different from one and zero, respec-
variables was analyzed for the Caucasian subjects tively (Fig. 3A). The r2 and SEE values of the regres-
within laboratory A and laboratory B. As shown in sion equation were 0.86 and 2.7 kg or 9%, respectively.
Table 2, Ht2/R explained 85 and 79% of the variance in As shown in Fig. 4, when the BIA method was used
SM mass in laboratories A and B, respectively. Within to predict SM mass, the difference between BIA-pre-
both laboratories, gender and age were also significant dicted and MRI-measured SM mass was within 5, 10,
(P ⬍ 0.01) contributors to the multiple-regression and 15% of the MRI-measured SM mass for 43, 74, and
model (Table 2). Although weight also contributed sig-
nificantly (P ⬍ 0.05) to the model, it explained ⬍1% of
Table 2. Multiple-regression model for predicting
the variance in MRI-measured SM in laboratories A
MRI-measured skeletal muscle mass from BIA (Ht 2/
and B; thus weight was excluded from the final regres-
R), age, and gender within the Caucasian subjects
sion equations. The relationship between BIA-pre-
dicted and MRI-measured SM for each laboratory is Significance
shown in Fig. 1. In both cases, analysis indicated that Variable r2 SEE (P Value)
the slopes and intercepts were not different (P ⬎ 0.9) Laboratory A*
from one and zero, respectively. X1 Ht2/R, cm2/⍀ 0.85 2.82 ⬍0.001
Cross-validation of BIA regression equations. The X2 Gender, man ⫽ 1
prediction equation derived from the Caucasian sub- and woman ⫽ 0 0.88 2.69 ⬍0.001
X3 Age, yr 0.89 2.58 ⬍0.01
jects in laboratory B was used to predict SM mass in Laboratory B†
the Caucasian subjects in laboratory A (Fig. 2A), and X1 Ht2/R, cm2/⍀ 0.79 3.03 ⬍0.001
the prediction equation derived for the Caucasian sub- X2 Gender, men ⫽ 1
jects in laboratory A was used to predict SM mass in and women ⫽ 0 0.81 2.88 ⬍0.001
the Caucasian subjects in laboratory B (Fig. 2B). The r2 X3 Age, yr 0.83 2.73 ⬍0.001
and SE of estimate (SEE) values of the validation (Fig. MRI, magnetic resonance imaging; BIA, bioelectrical impedance
1) and cross-validation (Fig. 2) analysis were similar. analysis; SEE, SE of estimate; X1 –X3, variables of regression equa-
As shown in Fig. 2, analysis revealed that the slopes tion. * MRI skeletal muscle mass (kg) ⫽ [(Ht2/R ⫻ 0.385) ⫹ (gen-
der ⫻ 4.464) ⫹ (age ⫻ ⫺0.057)] ⫹ 4.395. † MRI skeletal muscle mass
and intercepts were not significantly (P ⬎ 0.05) differ- (kg) ⫽ [(Ht2/R ⫻ 0.396) ⫹ (gender ⫻ 3.443) ⫹ (age ⫻ ⫺0.078)] ⫹
ent from one and zero and that the plotted lines were 6.313.
Fig. 5. Prediction of skeletal muscle mass in the Hispanic (A), African-American (B), and Asian (C) subjects by
using the regression equation derived from the Caucasian subjects. Solid lines, regression lines; dotted lines, lines
of identity.
20). Because there is an equilibrium between total demographics varied, the model should be applicable to
body water and fluid volume with lean body mass, a large proportion of the adult population. Because BIA
there is a strong relationship between R and fat-free is simple, noninvasive, reliable, repeatable, and rela-
body mass (2, 18). Limited evidence indicates that tively inexpensive, the BIA method should be practical
appendicular BIA measurements are also highly corre- in epidemiological and clinical settings. However, to
lated to appendicular SM, as estimated by a single ensure that reliable BIA measurements are obtained,
computerized tomography image (8) and dual-energy several factors such as hydration status, food intake,
X-ray absorptiometry (24, 25). Our findings extend and exercise must be controlled (18).
these observations and indicate that whole body mus- Although the Bland-Altman plot indicates there was
cle mass is highly related to whole body resistance. a systematic error with the BIA method, this error was
When current flows through the body, it is parti- small. For example, the BIA method would underesti-
tioned among different tissues according to their indi- mate SM mass by an average of 3.0% in an individual
vidual resistances and volumes. Because SM has both with a SM mass of 20 kg and overestimate SM mass by
a large volume and a low resistance, most of the BIA ⬃2% in an individual with a SM mass of 40 kg. More-
current flows through SM (13). Conversely, adipose over, although MRI-measured SM mass was related to
tissue only influences R when the volume of adipose adipose tissue and lean tissues other than SM, it ap-
tissue exceeds that of SM, and this influence is small pears that the influence of these tissues on R did not
(5). The influence of bone and organ on R is also small account for the systematic error of the BIA SM predic-
because bone has an extremely high resistance (13) tion formula.
and because the trunk is of little concern in whole body In this study, we found that the BIA equation for
BIA measurements (3, 5, 14). Indeed, we found that predicting SM mass derived from the Caucasians un-
82% of the variance in R within the Caucasians was derpredicted SM within the Asian cohort. This sug-
explained by SM mass and that adipose tissue, organ, gests there are biological differences between these
and bone only explained an additional 4% of the vari- races that influence the relationship between R and
ance in R. SM mass. For example, the height and weight of the
Numerous studies have developed equations for es- Asian subjects was significantly less than that of the
timating lean body mass from BIA measurements (2, Caucasian subjects in this study. Others have also
18). In a review of these studies, Houtkooper et al. (18) shown that body composition prediction equations are
report that the r2 values ranged from 0.73 to 0.98 and influenced by differences in body build between Asians
that the SEEs ranged from 1.7 to 4.1 kg. In general, and Caucasians (9). We also observed that age and
these studies found BIA to be a valid method for esti- gender had a small influence on the relationship be-
mating lean body mass. In this study, we demonstrate tween R and SM mass, a finding that may be partially
that BIA is also an effective method for estimating SM explained by age and gender differences in SM mass,
mass. The r2 value for SM mass in the present study height, and weight.
(r2 ⫽ 0.86) is similar to those previously observed for In summary, BIA prediction equations for whole
lean body mass (2, 18). The magnitude of the error body SM mass were developed and cross-validated in
(SEE ⫽ 2.7 kg or 9%) in predicting SM mass from BIA Caucasian subjects in two laboratories. The cross-val-
within the Caucasians, African-Americans, and His- idation of the BIA equations for predicting SM mass
panics is encouraging. The BIA method was within a was successful, and the magnitude of the error in
5% error for 43% of the subjects and within a 10% error predicting SM mass from BIA was small. These obser-
for 74% of the subjects. These results suggest that the vations are encouraging and suggest that BIA can
BIA method may provide meaningful estimates of SM provide rapid and accurate estimates of SM in adult
mass within adult populations. Because the model de- populations. Our results indicate that the derived
velopment size was large, and because the subject equation is applicable for Caucasian, African-Ameri-
can, and Hispanic populations but not for Asian popu- 12. Evans WJ. Functional and metabolic consequences of sarcope-
lations. The validity of the BIA method in individuals nia. J Nutr 127: 998S–1003S, 1997.
13. Foster KR and Lukaski HC. Whole body impedance—what
whose hydration status may be altered, such as ath- does it measure? Am J Clin Nutr 64, Suppl 3: 388S–396S, 1996.
letes, extreme elderly, and diseased individuals, re- 14. Fuller NJ and Elia M. Potential use of bioelectrical impedance
quires investigation. Studies are also needed to deter- of the “whole body” and of body segments for the assessment of
mine the sensitivity of BIA to detect changes in SM body composition: comparison with densitometry and anthro-
mass in response to nutritional and exercise interven- pometry. Eur J Clin Nutr 43: 779–791, 1989.
tions and to develop a race-specific equation for pre- 15. Goodpaster BH and Kelley DE. Role of muscle in triglyceride
metabolism. Curr Opin Lipidol 9: 231–236, 1998.
dicting SM from BIA measurements in Asians. 16. Heymsfield SB, Gallagher D, Visser M, Nuñez C, and
This work was supported by Medical Research Council of Canada Wang Z-M. Measurement of skeletal muscle: laboratory and
Grant MT 13448 and Natural Sciences and Engineering Council of epidemiological methods. J Gerontol A Biol Sci Med Sci 50:
Canada Grant OGPIN 030 (to R. Ross) and by National Center for 23–29, 1995.
Research Resources Grant RR-0064 and National Institute of Dia- 17. Heymsfield SB, McManus C, Stevens V, and Smith J. Mus-
betes and Digestive and Kidney Diseases Grant DK-42618 (to S. B. cle mass: reliable indicator of protein energy malnutrition sever-
Heymsfield). I. Janssen is supported by a Natural Sciences and ity and outcome. Am J Clin Nutr 35: 1192–1199, 1982.
Engineering Research Council of Canada Postgraduate Scholar- 18. Houtkooper LB, Lohman TG, Going SB, and Howell WH.
ship B. Why bioelectrical impedance should be used for estimating adi-
posity. Am J Clin Nutr 64, Suppl 3: 436S–448S, 1996.
REFERENCES 19. Kleinbaum DG and Kupper LL. Comparing two straight-line
regression models. In: Applied Regression Analysis and Other
1. Barber DC and Brown BH. Applied potential tomography.
Multivariate Methods. North Scituate, MA: Duxbury, 1978, p.
Journal of Physics E Scientific Instruments 17: 723–733, 1984.
95–112.
2. Baumgartner RN. Electrical impedance and TOBEC. In: Hu-
man Body Composition, edited by Roche AF, Heymsfield SB, and 20. Kushner RF and Shoeller DA. Estimation of total body water
Lohman TG. Champaign, IL: Human Kinetics, 1996. by bioelectrical impedance analysis. Am J Clin Nutr 44: 417–
3. Baumgartner RN, Chumlea WC, and Roche AF. Estimation 424, 1986.
of body composition from bioelectrical impedance of body seg- 21. Lukaski HC. Estimation of muscle mass. In: Human Body
ments. Am J Clin Nutr 50: 221–225, 1989. Composition, edited by Roche AF, Heymsfield SB, and Lohman
4. Baumgartner RN, Koehler KM, Gallagher D, Romero L, TG. Champaign, IL: Human Kinetics, 1996.
Heymsfield SB, Ross R, Garry PJ, and Lineman RD. Epi- 22. Lukaski HC, Johnson PE, Bolonchuk WW, and Lykken GI.
demiology of sarcopenia among the elderly in New Mexico. Am J Assessment of fat-free mass using bioelectrical impedance mea-
Epidemiol 147: 755–763, 1998. surements of the human body. Am J Clin Nutr 41: 363–366,
5. Baumgartner RN, Ross R, and Heymsfield SB. Does adipose 1985.
tissue influence bioelectric impedance in obese men and women? 23. Mitsipoulos N, Baumgartner RN, Heymsfield SB, Lyons W,
J Appl Physiol 84: 257–262, 1998. Gallagher D, and Ross R. Cadaver validation of skeletal mus-
6. Beneke R, Neuerburg J, and Bohndorf K. Muscle cross- cle measurement by magnetic resonance imaging and comput-
sectional measurements by magnetic resonance imaging. Eur erized tomography. J Appl Physiol 85: 115–122, 1998.
J Appl Physiol 63: 424–429, 1991. 24. Nuñez C, Gallagher D, Grammes J, Baumgartner RN, Ross
7. Bland JM and Altman DG. Statistical method for assessing R, Wang ZM, Thornton J, and Heymsfield SB. Bioimpedance
agreement between two methods of clinical measurement. Lan- analysis: potential for measuring lower limb skeletal muscle
cet 8: 307–310, 1986. mass. JPEN J Paranter Enternal Nutr 23: 96–103, 1999.
8. Brown BH, Karatzas T, Hakielny R, and Clarke RG. Deter- 25. Petrobelli A, Morini P, Battistini N, Chiumello G, Nuñez
mination of upper arm muscle and fat areas using electrical
C, and Heymsfield SB. Appendicular skeletal muscle mass:
impedance measurements. Clin Phys Physiol Meas 9: 47–55,
prediction from multiple frequency segmental bioimpedance
1988.
9. Deurenberg P, Deurenberg Yap M, Wang J, Lin FP, and analysis. Eur J Clin Nutr 52: 507–511, 1998.
Schmidt G. The impact of body build on the relationship be- 26. Ross R, Rissanen J, Pedwell H, Clifford J, and Shragge P.
tween body mass index and percent body fat. Int J Obes 23: Influence of diet and exercise on skeletal muscle and visceral
537–542, 1999. adipose tissue in men. J Appl Physiol 81: 2445–2455, 1996.
10. Dutta C and Hadley EC. The significance of sarcopenia in the 27. Snyder WS, Cooke MJ, Manssett ES, Larhansen LT, How-
elderly. J Gerontol A Biol Sci Med Sci 50: 1–4, 1995. ells GP, and Tipton IH. Report of the Task Group on Reference
11. Engstrom CM, Loeb GE, Reid JG, Forrest WJ, and Avruch Man. Oxford, UK: Pergamon, 1975.
L. Morphometry of the human thigh muscles. A comparison 28. Zheng E, Shao S, and Webster JG. Impedance of skeletal
between anatomical sections and computer tomography and muscle from 1 Hz to 1 MHz. IEEE Trans Biomed Eng 31:
magnetic resonance images. J Anat 176: 139–156, 1991. 477–481, 1984.