American Journal of Epidemiology
Copyright O 1996 by The Johns Hopkins University School of Hygiene and Public Health
All rights reserved
Vol. 143, No. 9
Printed in U.S.A
Sex Differences in Measures of Body Fat and Body Fat Distribution in the
Elderly
Deborah Goodman-Gruen and Elizabeth Barrett-Connor
This study describes sex differences in obesity and body fat distribution using commonly used assessment
methods in 140 men and 245 women aged 65-96 years from Rancho Bernardo, California. Significant
correlations were shown among all obesity measures. The waist/hip ratio was more strongly correlated with
the truncal fat/leg fat ratio in women than men. The waist/hip ratio correlated significantly with the subscapular/
triceps skinfold ratio in women only. In both sexes, waist circumference was more strongly correlated with
body mass index and the percentage of body fat by bioelectric impedance analysis and dual-energy X-ray
absorptiometry than with the waist/hip ratio. In those aged over 80 years, age stratification showed that the
waist/hip ratio was not correlated with any other measurement of obesity or fat distribution in men and
correlated only with subscapular skinfolds in women. Waist circumference, however, correlated significantly
with almost ail other measures of central obesity in older and younger men and women. Estimates of upper
body (central) fat distribution appear to be age specific. After age 80, the waist/hip ratio is a poor method of
assessing central or visceral adiposity, and waist circumference is a better measure of body fat distribution.
Am J Epidemiol 1996; 143:898-906.
absorptiometry, photon; body constitution; electric
scanned projection
Obesity and body fat distribution have been shown
to be independent risk factors for a number of chronic
diseases (1-10). Upper body (or central) fat distribution is thought to reflect visceral adiposity (11) and has
been shown to be associated with insulin resistance
and dyslipidemia (12, 13). Historically, anthropometric techniques have been used to estimate body fat and
body fat distribution in epidemiologic studies, because
more precise measurement methods, such as computerized tomography for visceral fat or body water by
isotope dilution, whole-body counting of potassium40, or neutron activation analysis for total body fat, are
more expensive and difficult to administer.
Although several studies among adults have found a
strong association between less invasive estimates of
obesity, such as bioelectric impedance, dual-energy
X-ray absorptiometry, and anthropometric measures
(14-17), the generalizability of these results may be
limited by age. Few have examined these associations
in an elderly population, when age-related changes in
human anatomy and physiology may alter the way in
impedance; obesity; radiography,
dual-energy
which these tools reflect body composition (18, 19).
Decreases in skin elasticity (20), alterations in hydration or bone mineral content (21), kyphosis, and relaxation of the abdominal musculature may lead to
measurement error in older persons. The aim of this
study was to compare the commonly used estimates of
obesity and body fat distribution in an elderly, largely
nonobese population of healthy men and women.
MATERIALS AND METHODS
The subjects for this study were the first 385 men
and women aged 65 years and older who participated
in the 1992-1994 evaluation of the Rancho Bernardo
Heart and Chronic Disease Study (22). All were community dwelling and ambulatory. The history of cigarette smoking, alcohol use, physical activity, and
postmenopausal estrogen use were determined by a
structured questionnaire. Height and weight were measured with the participants wearing light clothing without shoes. Body mass index (weight (kg)/height (m)2)
was calculated. Height from the baseline visit approximately 20 years earlier was used as a proxy of "true"
height to examine the possible effect of height loss or
kyphosis (often associated with osteoporosis) on measures of central obesity. Waist and hip girth were
measured in centimeters over single-thickness clothing
Received for publication February 8, 1995, and in final form
February 13, 1996.
From the Department of Family and Preventive Medicine, University of California, San Diego, La Jolla, CA.
Reprint requests to Dr. Elizabeth Barrett-Connor, Department of
Family and Preventive Medicine, University of California, San Diego,
9500 Gilman Drive, La Jolla, CA 92093-0628.
898
Sex Differences in Body Fat Measures
with the participant standing in an erect position with
feet together. The waist was measured at the bending
point (point marked where the participant naturally
bends forward and measured after the participant has
returned to the upright position) and the narrowest
circumference. Hip circumference was measured at
the iliac crest and at the largest circumference. Both
waist and both hip circumferences were highly correlated; for consistency with our previous reports, the
bending point/iliac crest ratio was used as the waist/
hip ratio for these analyses. Using Harpenden calipers,
subscapular skinfold thickness was measured inferior
to the inferior angle of the scapula, and triceps skinfold thickness was measured at the midpoint between
the acromial process and inferior border of the ulnar
olecranon process, with the elbow flexed 90 degrees.
Bioelectric impedance analysis was used to measure
the percentage of fat mass (model 1990B; Valhalla
Scientific, Inc., San Diego, California) by assessing
the voltage drop between two pairs of electrodes using
an alternating current of 500 /JLA at a frequency of 50
kHz. Total body fat and regional body fat measurements, including truncal and leg fat, were measured
using total body dual-energy X-ray absorptiometry
(model QDR-2000 X-ray bone densitometer; Hologic,
Inc., Waltham, Massachusetts), which uses a single
beam scanning mode. Truncal fat boundaries were
determined 1) superiorly, a line bisecting the glenoid
fossas; 2) laterally, lines extending to above the iliac
crests, and 3) interiorly, oblique lines bisecting the
femoral necks. Measurement of leg fat was delineated
by the area inferior to the oblique lines passing
through the femoral necks. Truncal fat and leg fat
measurements were expressed as a percentage of fat in
the truncal region and the leg region, respectively. The
waist/hip ratio, subscapular skinfolds/triceps skinfolds
ratio, and the truncal fat/leg fat ratio were used to
estimate upper body obesity (central obesity and upper
body obesity are used interchangeably in this paper).
Waist circumference (at the bending point) is used as
an integrated measure of obesity and fat distribution,
because small studies have shown that waist circumference is highly correlated with both total and visceral
body fat measured by computed tomography and magnetic resonance imaging (23, 24).
Data were analyzed using SAS and SAS/STAT software (SAS Institute, Inc., Cary, North Carolina) (25).
Student's t tests were used to test for significant sex
differences in mean body fat using each measurement
technique. Pearson's partial correlation coefficients
were used to compare the associations between the
measures of total body fat and body fat distribution,
and the strength of associations was compared using
Fischer's Z transformation. Frequency tables by tertile
Am J Epidemiol
Vol. 143, No. 9, 1996
899
of body fat and body fat distribution were used to
assess concordance between the measurement methods. Logarithms of body fat measurements were used
to account for slightly skewed distributions. All p
values are two tailed. No adjustment was made for
multiple comparisons; instead, exact p values are
shown.
RESULTS
The average age of both men and women was 80
years (range, 65-96 years). Measures of total body fat
and body fat distribution for the 245 women and 140
men are shown in table 1. Significant differences in
body composition between men and women were
shown by each method of measurement. Although the
body mass index and subscapular skinfolds were
greater in men than women, the percentage of total
body fat by bioelectric impedance analysis or dualenergy X-ray absorptiometry was about 8 percent
greater in women than men. Men had significantly
greater upper body obesity than did women based on
the waist circumference, waist/hip ratio, subscapular/
triceps ratio, or truncal fat/leg fat ratio.
As shown in tables 2 and 3, the age-adjusted correlations among obesity measurements by body mass
index, bioelectric impedance analysis, dual-energy Xray absorptiometry, and subscapular and triceps skinfolds were strong and significant in both men and
women (rs > 0.43; ps > 0.0001). In contrast, the
strength of associations among different measures of
upper body obesity was weaker and differed between
the sexes. In women, the waist/hip ratio was significantly correlated with the dual-energy X-ray absorptiometry truncal fat/leg fat ratio (r = 0.30; p =
0.0001) and subscapular/triceps skinfold ratio (r =
0.21; p - 0.001), and the truncal fat/leg fat ratio was
associated with the subscapular/triceps ratio (r = 0.34;
p — 0.0001). In men, the waist/hip ratio was weakly
correlated with the truncal fat/leg fat ratio (r = 0.19;
p = 0.03) and was not associated with the subscapular/
triceps ratio (r = 0.05; p = 0.60); the truncal fat/leg
fat ratio and the skinfold ratio were weakly inversely
associated (r = -0.17; p = 0.07). In both men and
women, waist circumference was more strongly correlated with body mass index (r > 0.81; p = 0.0001)
and the percentage of body fat estimated by bioelectric
impedance analysis (r = 0.60; p = 0.0001) and dualenergy X-ray absorptiometry (r S: 0.64; p = 0.0001)
than with the waist/hip ratio (r > 0.56; p = 0.0001),
and these differences were statistically significant
(p < 0.0001). Stratification by physical activity, cigarette smoking, alcohol use, and postmenopausal estrogen use in women did not materially alter these
results (data not shown).
8
CD
p
!
27.8
22.9
0.0001
8.0
6.4
40.0
25.3
0.0001
Leg (at
(
Truncalfat
Mean
24.0
25.6
Mean
0.0002
SO
7.2
6.8
SDt
7.0
5.5
SD
26.9
19.7
3.9
3.9
0.0001
BIAf
5.8
5.7
0.07
0.04
0.87
0.96
0.0001
SO
Mean
0.81
1.32
Mean
0.90
0.69
Mean
0.0001
0.14
0.17
SO
7.5
7.1
SO
80.34
95.01
Mean
• DEXA, dual-energy X-ray absorptiometry.
t Numbers In parentheses, p value.
0.44 (0.0001)
-0.27 (0.002)
0.86 (0.0001)
0.59 (0.0001 ) t
0.73(0.0001)
0.67 (0.0001)
0.68 (0.0001)
0.13 (0.14)
Body
mass
Index
0.57 (0.0001)
-0.30 (0.007)
0.60 (0.0001)
0.76(0.0001)
0.43 (0.0001)
0.51 (0.0001)
0.15 (0.10)
Body fat
percentage
(bioelectric
Impedance)
0.57 (0.0001)
-0.29 (0.0009)
0.78 (0.0001)
0.55 (0.0001)
0.57(0.0001)
0.22 (0.02)
Body tat
percentage
(DEXA*)
0.36 (0.0001)
-0.02 (0.86)
0.64(0.0001)
0.71 (0.0001)
0.26 (0.004)
Subscapular
sWnfokte
0.34 (0.0001)
-0.66 (0.0001)
0.59(0.0001)
0.11 (0.21)
Triceps
sWntoJds
0.19 (0.03)
0.05 (0.60)
0.43(0.0001)
Waist/hip
ratio
-0.17 (0.07)
0.72(0.0001)
ratio (DEXA)
ra
-0.17(0.0001)
0.0001
11.14
10.73
SO
Waist circumference
(cm)
7.6
7.3
17.9
13.2
0.0001
SD
Mean
Triceps sWntotds
(mm)
Age-adjusted correlations between measures of total body fat and body fat distribution in men from the Rancho Bernardo Study, 1992-1994
Body fat percentage (bioelectric
Impedance)
Body fat percentage (DEXA)
Subscapular sklnfolds
Triceps stonfokJs
Waist/hip ratio
Truncal fat (%)/leg fat (%) ratio
(DEXA)
SubscapularAriceps ratio
Waist circumference
TABLE 2.
0.0001
0.38
0.51
SD
Truncaltatf
leg fat ratio*
13.9
15.5
Mean
Subscapular/triceps
ratio
6.8
5.3
SD
Subscaputar sWntokls
(mm)
0.04
DEXAf
0.0001
32.5
24.0
Mean
Body tat (%)
SO
Walst/hlp ratio
Mean
SO
Body mass Index
(kg/m»)
0.23
Mean
79.4
80.3
Mean
Age
(years)
Sex differences in mean age, body fat, and body fat distribution by various measures in 245 women and 140 men from the Rancho Bernardo Study, 1992-1994
* Dual-energy X-ray absorptiometry.
t SD, standard deviation; BIA, bioelectric impedance analysis; DEXA, dual-energy X-ray absorptiometry.
p value
Women
Men
p value
Women
Men
TABLE 1.
Sex Differences in Body Fat Measures
o o
ou
S
d d
o i-
p
S88
o
2-2- £
•Si
co
"g
S
C\J
in
odd
I
to
o
o
p
o_
I
CO
CM
8
2-S-2.
co o en
•rf r ; U)
odd
2-2.S-
S
I
CO
co
d d
in oo
in (£)<£>
CD
odd
"D
3
o
•o
c
a
3
II
o o ,-.
o ,— o
O
8.88
o8 d52d 3
Q CVJ
pop
odd
o d d
f
o
o
p
d
o o co
o o o
p p p
odo
co co
CVJ
oo
N U) (O i -
o o o o
o o o o
3
T-
8
8S8S
o o o o o
Socq
N m (vi
to <o cJ
d d d d d
p cvi p
odd
S
CO O
P "!
odd
8
(M r
r
IO CM 0 0
o d d
s
o
S
CO
3
<
o
! •
If
— 3
D 5.
3.
CO
it
_U 3
00
Am J Epidemiol
QZ
Vol. 143, No. 9, 1996
901
Table 4 shows the sex-specific, adjusted correlations
between the waist/hip ratio and the other measures of
obesity and body fat distribution dichotomized by age.
In the 65- to 79-year-old group, significant correlations were seen for both men and women between the
waist/hip ratio and most measures of obesity and body
fat distribution. However, in the group aged 80 years
and older, the waist/hip ratio was not associated with
any other measurement method, except for a significant skinfold-waist/hip ratio association in the oldest
women.
Table 5 shows the age- and sex-specific correlations
between the waist circumference and the other measures of obesity and body fat distribution. As with the
waist/hip ratio, in the 65- to 79-year-old group, significant correlations were seen for both men and
women between the waist circumference and most
measures of obesity and body fat distribution. In contrast to the waist/hip ratio, the association between
waist circumference and the other measures of body
fat distribution was significant in both the younger and
older age groups.
To exclude possible confounding effects by height
loss with aging, the age-adjusted correlations were
compared between each measurement technique of
obesity and the body mass index calculated using
height from the 1992-1994 visit and the body mass
index using height from the 1972-1974 visit. Although women had lost more height than had men
between baseline (1972-1974) and follow-up (19921994) visits (an average loss of 3.7 cm and 2.7 cm,
respectively; p — 0.0002), in women the correlations
between measures of body fat and body fat distribution
did not differ between the two methods of calculating
body mass index. In contrast, in men the correlation
between body mass index and both dual-energy X-ray
absorptiometry and bioelectric impedance analysis decreased, and the correlation between the truncal fat/leg
fat ratio and skinfold ratio and body mass index increased, when height from the 1972-1974 visit was
used in the body mass index calculation. The correlation between body mass index and waist circumference did not differ between the two methods of calculating body mass index (data not shown).
The percentage of concordance among tertiles for
different .methods of estimating obesity is shown in
figure 1. For both sexes, the greatest agreement was
between bioelectric impedance analysis and dualenergy X-ray absorptiometry (66 percent for men; 75
percent for women). The smallest concordance rate
was found for the dual-energy X-ray absorptiometry
and skinfolds (subscapular: 47 percent for men, 53
percent for women; triceps: 47 percent for men, 54
902
Goodman-Gruen and Barrett-Connor
TABLE 4. Correlation coefficients between walst/hip ratio and measures of total body fat and body fat
distribution for men and women stratified by age, Rancho Bernardo Study, 1992-1994
Men
65-79 years
(n» 64)
Body mass Index
Bioelectric impedance analysis
Dual-energy X-ray absorptlometry
Subscapular skinfotds
Triceps skinfolds
Truncal tat (%)/leg tat (%) ratio
Subscapular sklnfotd/trtceps skinfcrtd
ratio
Women
280 years
(n=76)
0.35 (0.002)'
0.45(0.0001)
0.46(0.0001)
0.50(0.0001)
0.27 (0.02)
0.44(0.0001)
-0.03(0.81)
0.02 (0.85)
0.005 (0.97)
0.08 (0.52)
0.06 (0.63)
0.05 (0.66)
0.12(0.27)
0.03 (0.84)
65-79 years
(n=122)
280 years
(0-123)
0.23 (0.006)
0.15 (0.07)
0.17(0.05)
0.30 (0.0003)
0.16(0.06)
0.26 (0.002)
0.18 (0.06)
0.16 (0.10)
0.10 (0.30)
0.32 (0.0006)
0.26 (0.006)
0.13(0.19)
0.24 (0.003)
0.19 (0.05)
* Numbers in parentheses, p value.
TABLE 5. Correlation coefficients between waist circumference and measures of total body fat and
body fat distribution for men and women stratified by age, Rancho Bernardo Study, 1992-1994
Men
65-79 years
(n»64)
0.87(0.0001)*
Body mass index
0.74(0.0001)
Bioelectric impedance analysis
Dual-energy X-ray absorptlometry
0.78(0.0001)
0.67(0.0001)
Subscapular skinfolds
0.61 (0.0001)
Triceps skinfolds
0.52(0.0001)
Waist/hip ratio
0.73(0.0001)
Truncal fat (%)/leg tat (%) ratio
Subscapular skJnfokl/triceps skinfoW
ratio
-0.18(0.19)
Women
280 years
(n=76)
65-79 years
(n=122)
280 years
(r>=123)
0.86(0.0001)
0.56(0.0001)
0.76(0.0001)
0.57(0.0001)
0.57(0.0001)
0.32 (0.008)
0.69(0.0001)
0.82(0.0001)
0.55(0.0001)
0.64(0.0001)
0.66(0.0001)
0.51 (0.0001)
0.62(0.0001)
0.75(0.0001)
0.79 (0.0001)
0.65(0.0001)
0.61 (0.0001)
0.68(0.0001)
0.66(0.0001)
0.51 (0.0001)
0.68 (0.005)
0.37(0.0001)
0.27(0.0001)
-0.24 (0.05)
* Numbers in parentheses, p value.
percent for women). For all comparisons but one, the
concordance rate was greater in women than men.
The percentage of concordance among tertiles for
different methods of estimating upper body fat distribution is shown in figure 2. Women were more concordant than were men for the waist/hip ratio-skinfold
ratio, skinfold ratio-dual-energy X-ray absorptiometry
ratio, waist/hip ratio-waist circumference, skinfold
ratio-waist circumference, and dual-energy X-ray
absorptiometry ratio-waist circumference comparisons; men were more concordant than were women for
the waist/hip ratio-dual-energy X-ray absorptiometry
association. The range of concordance rates was narrower in women (42-56 percent) than in men (28-50
percent).
DISCUSSION
In this cohort of community-dwelling, ambulatory,
elderly Caucasians, the average levels of obesity and
upper body obesity in men were similar to national
values and those of other community-based studies,
while the women were thinner than women reported in
the literature (26-28). All measurement techniques
(waist circumference, waist/hip ratio, subscapular/
triceps ratio, and the dual-energy X-ray absorptiometry truncal fat/leg fat ratio) confirmed that upper body
obesity is more common among men than women
even in old age (29). In accord with a previous report
(17), women had a greater percentage of leg fat (dualenergy X-ray absorptiometry) and percentage of truncal fat (dual-energy X-ray absorptiometry) than had
men. Although obesity estimated from the body mass
index was greater among men than women, the percentage of body fat estimated from either bioelectric
impedance analysis or dual-energy X-ray absorptiometry was greater among women than men. This discrepancy is consistent with the literature and has been
explained by the greater lean body mass among men
influencing the body mass index calculation (30).
Significant correlations were found among obesity
measures, including body mass index, bioelectric impedance analysis, dual-energy X-ray absorptiometry,
and subscapular and triceps skinfolds, for both men
and women. These results are in agreement with the
literature, which has found strong correlations between skinfolds and both bioelectric impedance analysis and dual-energy X-ray absorptiometry measureAm J Epidemiol
Vol. 143, No. 9, 1996
Sex Differences in Body Fat Measures
BMI
BIA
BMI
BMI
SSF
DEXA
BMI
TSF
903
SSF
TSF
FIGURE 1. Percentage of concordance among tertiles for different measurement estimations of obesity for men (•) and women (•), Rancho
Bernardo Study, 1992-1994. BMI, body mass index; BIA, bioelectric impedance analysis; DEXA, dual-energy X-ray absorptiometry; SSF,
subscapular skinfold; TSF, triceps skinfold.
WHR
SFR
WHR
WHR
SFR
SFR
DEXA
waist
DEXA
waist
DEXA
waist
FIGURE 2. Percentage of concordance among tertiles for different measurement estimations of upper body fat distribution for men (•) and
women (Q). Rancho Bernardo Study, 1992-1994. WHR, waist/hip ratio; SFR, subscapularAriceps skinfold ratio; DEXA, dual-energy X-ray
absorptiometry; waist, waist circumference.
ments (14, 15) and between the body mass index and
bioelectric impedance analysis (16).
Previous studies have found fairly strong correlations between the waist circumference and total body
fat, subcutaneous abdominal fat, and intraabdominal
fat derived from computed tomography (23, 24), suggesting that the waist circumference is an integrated
Am J Epidemiol
Vol. 143, No. 9, 1996
estimate of obesity and fat distribution. In Rancho
Bernardo men and women, there was a significantly
stronger correlation between the waist circumference
and the obesity measures than between the waist circumference and waist/hip ratio.
All measures of body fat distribution were correlated with each other in women. In contrast, the waist
904
Goodman-Gruen and Barrett-Connor
circumference in men was strongly correlated with the
dual-energy X-ray absorptiometry ratio and the skinfold ratio, but the waist/hip ratio and dual-energy
X-ray absorptiometry ratio were weakly correlated,
suggesting that the waist circumference may be a more
useful anthropometric measure of upper body obesity
than the waist/hip ratio in men 65 years and older. This
is in agreement with several small studies of young
adults (31-34). The strong correlation between the
waist circumference and most other measures of obesity and body fat distribution was seen in both men and
women and was present in those whose age was
greater than 80 years. It is difficult to determine
whedier the higher correlation between waist circumference and the dual-energy X-ray absorptiometry ratio was due to subcutaneous fat or intraabdominal fat,
since both measure both components of abdominal fat
(33).
In this cohort, the waist/hip ratio and the body mass
index were significantly correlated only in women.
This sex differential was not explained by age differences (see table 1) or height loss; the association
between the waist/hip ratio and body mass index was
still absent in men when the body mass index was
calculated using height from a visit 20 years earlier.
Although a smaller range of waist/hip ratio may explain the absent association in men (range for men =
0.85-1.11; range for women = 0.72-1.07), it is
equally plausible that relatively lean old men have
androgen deficiency-driven loss of lean body mass
coupled with upper body obesity (35).
It has been suggested that a decrease in skin elasticity with increasing age could lead to an underestimation of total body fat by skinfold measurement (20),
but skinfold measurements correlated quite well with
the body mass index (rs > 0.67; p = 0.0001) in this
older population. Measurement of body fat by bioelectric impedance analysis, a technique based on the
principle that body fluids act as electrical conductors
and cell membranes act as electrical capacitors (36,
37), may be influenced by age-related changes in
physiology, such as alterations in hydration or bone
mineral content (21). Dual-energy X-ray absorptiometry assumes a constant and fixed hydration of lean
body mass (38, 39), a property which is sometimes
absent among the very elderly (40). Nevertheless, in
this healthy elderly cohort, both bioelectric impedance
analysis and dual-energy X-ray absorptiometry measurements were strongly associated with all other measures of obesity. Height loss, kyphosis, and relaxation
of the abdominal musculature could lead to a larger
waist/hip ratio and waist circumference secondary to
abdominal protuberance without central obesity. This
may be why an association between the waist/hip ratio
and all measures of obesity and body fat distribution
was present in the younger (65-79 years) but not in the
older (2:80 years) men, but, if this is so, it is surprising
that the age difference was not more dramatic in
women. It is possible that this effect is not seen in
these Rancho Bernardo women because they are relatively lean. The similarly of correlations between the
waist circumference and most other measures of obesity and fat distribution stratified by age group argues
against this possibility.
This study did not evaluate the validity of measurement methods, but instead it compared the concordance of various methods of estimating obesity and
body fat distribution and quantitated the potential for
differential classification. For both men and women,
the comparisons that included skinfold measurements
had the lowest rate of concordance, with greater than
50 percent of observations potentially classified into
discordant tertiles. Although the dual-energy X-ray
absorptiometry to bioelectric impedance analysis comparisons had the least discordance for both sexes,
between 25 and 34 percent of observations were differentially classified. A small effect of obesity on a
given outcome could easily be obscured by this among
variation in categorical classification. Measures of
body fat distribution showed even greater rates of
discordance than did the obesity measures for both
sexes. This suggests that the chance of missing an
effect of central obesity would be even larger. It is also
possible that the measurement methods estimate different types of body fat distribution (26) and would not
be expected to classify observations into concordant
categories.
Although the body mass index has been criticized
for its lack of ability to separate the weight of fat from
the fat-free mass (41, 42), this method has been demonstrated to be relatively insensitive to intra- or interobserver errors (43). In this study, this inexpensive and
readily obtainable measure was highly correlated with
obesity by bioelectric impedance analysis or dualenergy X-ray absorptiometry, suggesting it is an adequate surrogate for the percentage of body fat in men
and women for use in epidemiologic research. Although skinfold measures were also highly correlated
with the other measures of body fat, the poorer interobserver reliability (44) decreases their usefulness
and, as shown here, the potential for discordant classification is greatest with this measurement technique.
The advantage of dual-energy X-ray absorptiometry
versus bioelectric impedance analysis for the estimation of fat mass is unclear. Dual-energy X-ray absorptiometry is more expensive and involves a small
amount of radiation. Bioelectric impedance analysis is
less expensive, does not involve radiation, and may be
Am J Epidemiol
Vol. 143, No. 9, 1996
Sex Differences in Body Fat Measures
used at the bedside in nonambulatory patients (36).
Both methods are noninvasive, rapid, and simple to
use. However, two small studies comparing the use of
dual-energy X-ray absorptiometry and bioelectric impedance analysis for body fat estimation have found
conflicting results (45, 46), and we are aware of no
other large studies of elderly persons.
The association among measures of body fat distribution was weaker than that among obesity measures,
varied more by age, and was sex specific in the oldest
subjects. The waist/hip ratio was clearly an inadequate
method of assessing upper body obesity in men and
women over 80 years old. These data suggest that the
waist circumference could be the preferred anthropometric measure of body fat distribution, especially in
the very old. It is necessary to evaluate the total body
fat, abdominal obesity, and abdominal visceral fat
contribution to the waist circumference measure. Although one small study in men and women aged <50
years found that waist circumference was consistently
and strongly associated with several metabolic variables (31), further studies are needed in the elderly to
compare the association between the various measurement methods and outcomes of interest. The "best"
anthropometric measure may be dependent on the risk
factor of interest. At the present time, several measures
should be used.
ACKNOWLEDGMENTS
This research was supported by National Institute of
Diabetes and Digestive and Kidney Diseases grant DK
31801 and Weight Watchers Foundation grant WWF93148. Dr. Goodman-Gruen is supported by PSA /AB0035307.
REFERENCES
1. Thompson CJ, Ryu JE, Crave TE, et al. Central adipose
distribution is related to coronary atherosclerosis. Arterioscler
Thromb 1991;11:327-33.
2. Larsson B, Svardsudd K, Welin L, et al. Abdominal adipose
tissue distribution, obesity, and risk of cardiovascular disease
and death: 13 year follow-up of participants in the study of
men bom in 1913. Br Med J 1984;288:1401-4.
3. Ducimetiere P, Richard J, Cambien F. The pattern of subcutaneous fat distribution in middle-aged men and the risk of
coronary heart disease: the Paris Prospective Study. Int J Obes
1986;10:229-40.
4. Vague J. The degree of masculine differentiation of obesities:
a fact for determining predisposition to diabetes, atherosclerosis, gout, and uric calculous disease. Am J Clin Nutr 1956;
18:478-86.
5. Kissebah AH, Vydelingum N, Murray R, et al. Relation of
body fat distribution to metabolic complications of obesity. J
Clin Endocrinol Metab 1982;54:254-60.
6. Haffner SM, Diehl AK, Stem MP, et al. Central adiposity and
gallbladder disease in Mexican Americans. Am J Epidemiol
Am J Epidemiol
Vol. 143, No. 9, 1996
905
1989:129:587-95.
7. Ballard-Barbash R, Schatzkin A, Carter CL, et al. Body fat
distribution and breast cancer in the Framingham Study. J Natl
Cancer Inst 1990;82:286-90.
8. Lapidus L, Helgesson O, Merck C, et al. Adipose tissue
distribution and female carcinomas: a 12 year follow-up of
participants in the population study of women in Gothenburg,
Sweden. Int J Obes 1988; 12:361-7.
9. Maggino T, Pirrone F, Velluti F, et al. The role of the endocrine factors and obesity in hormone-dependent gynecological
neoplasias. Eur J Gynaecol Oncol 1993;14:119-26.
10. Folsom AR, Kaye SA, Prineas RJ, et al. Increased incidence of
carcinoma of the breast associated with abdominal adiposity in
postmenopausal women. Am J Epidemiol 1990;131:794-803.
11. Weits T, Van der Beek EF, Wedel M, et al. Computed tomography measurements of abdominal fat deposition in relation to
anthropometry. Int J Obes 1988; 12:217-25.
12. Kissebah AH, Evans DJ, Peiris A, et al. Endocrine characteristics in regional obesities: role of sex steroids. In: Vague J, et
al., eds. Metabolic complications of human obesities: proceedings of the 6th international symposium, Marseille, 30 May-1
June 1985. New York: Elsevier Science Publishing Co, 1985:
115.
13. Fujioka S, Matsuzawa Y, Tokunaga K, et al. Contribution of
intra-abdominal fat accumulation to the impairment of glucose
and lipid metabolism in human obesity. Metabolism 1987;36:
54-9.
14. Fuller NJ, Elia M. Potential use of bioelectrical impedance of
the "whole body" and of body segments for the assessment of
body composition: comparison with densitometry and anthropometry. Eur J Clin Nutr 1989;43:779-91.
15. Stewart SP, Bramley PN, Heighton R, et al. Estimation of
body composition from bioelectrical impedance of body
segments: comparison from bioelectrical impedance of body
segments: comparison with dual-energy X-ray absorptiometry. Br J Nutr 1993;69:645-55.
16. McNeill G, Fowler PAZ, Maughan RJ, et al. Body fat in lean
and overweight women estimated by six methods. Br J Nutr
1991;65:95-103.
17. Fuller NJ, Laskey MA, Elia M. Assessment of the composition of major body regions by dual-energy X-ray absorptiometry, with special reference to limb muscle mass. Clin Physiol
1992; 12:253-66.
18. Steen B. Body composition and aging. Nutr Rev 1988;46:
45-51.
19. Deurenberg P, van der Kooij K, Evers P, et al. Assessment of
body composition by bioelectrical impedance in a population
aged > 60 y. Am J Clin Nutr 1990;51:3-6.
20. Kuczmarcki RJ. Need for body composition information in
elderly subjects. Am J Clin Nutr 1989;50:l 150-7.
21. Lohman TG. Skinfolds and body density and their relation to
body fatness: a review. Hum Biol 1981;53:181-225.
22. Wingard DL, Sinsheimer P, Barrett-Connor E, et al. Community-based study of prevalence of NIDDM in older adults.
Diabetes Care 1990;13:3-8.
23. Borkan G ^ Hults DE. Relationships between computed tomography tissue areas, thicknesses, and total body composition. Ann Hum Biol 1983; 10:537-46.
24. Ross R, Shaw KD, Rissanen J, et al. Sex differences in lean
and adipose tissue distribution by magnetic resonance
imaging: anthropometric relationships. Am J Clin Nutr 1994;
59:277-85.
25. SAS Institute, Inc. SAS/STAT user's guide, version 6 ed.
Cary, NC: SAS Institute, Inc, 1989.
26. Najjar MF, Rowland M. Anthropometric reference data and
prevalence of overweight Vital Health Stat [11] 1987;238:
1-73.
27. Wellens R, Chumlea WC, Guo S, et al. Body composition in
white adults by dual-energy X-ray absorptiometry, densitometry, and total body water. Am J Clin Nutr 1994;59:547-55.
28. Haffner SM, Stem MP, Hazuda HP, et al. Do upper-body and
centralized adiposity measure different aspects of regional
906
Goodman-Gruen and Barrett-Connor
body-fat distribution? Diabetes 1987;36:43-51.
29. Dumin JVGA, Womersley J. Body fat assessed from total
body density and its estimation from skinfold thickness: measurements on 481 men and women aged from 16 to 72 years.
Br J Nutr 1974;32:77-97.
30. Ley CJ, Lees B, Stevenson JC. Sex- and menopause-associated changes in body-fat distribution. Am J Clin Nutr 1992;
55:950-4.
31. Poulit M, Despres J, Lemieux S, et al. Waist circumference
and abdominal sagittal diameter best simple anthropometric
indexes of abdominal visceral adipose tissue accumulation and
related cardiovascular risk in men and women. Am J Cardiol
1994;73:460-8.
32. Ferland M, Despres JP, Tremblay A, et al. Assessment of
adipose tissue distribution by computed axial tomography in
obese women: association with body density and anthropometric measurements. Br J Nutr 1989;61:139-48.
33. Despres JP, Prud'homme D, Pouliot MC, et al. Estimation of
deep abdominal adipose-tissue accumulation from simple anthropometric measurements in men. Am J Clin Nutr 1991;54:
471-7.
34. Seidell JC, Oosterlee A, Deurenberg P, et al. Abdominal fat
depots measured with computed tomography: effects of degree of obesity, sex, and age. Eur J Clin Nutr 1988;42:805-15.
35. Khaw K-T, Barrett-Connor E. Lower endogenous androgens
predict central adiposity in men. Ann Epidemiol 1992;2:
675-82.
36. Lukaski HC, Bolonchuk WW, Hall CB, et al. Validation of
tetrapolar bioelectrical impedance method to assess human
body composition. J Appl Physiol 1986;60:1327-32.
37. Wedgewood RJ. Inconstancy of the lean body mass. Ann N Y
AcadSci 1%3;110:141-52.
38. Roubenoff R, Kehayias JJ. The meaning and measurement of
lean body mass. Nutr Rev 1991;46:163-75.
39. Heymsfield SB, Waki M. Body composition in humans: advances in the development of multicompartment chemical
models. Nutr Rev 1991;49:97-108.
40. Forbes GL. Human body composition: growth, aging, nutrition, and activity. New York: Springer-Verlag, 1987.
41. Pollock ML, Jackson AS. Research progress in validation of
clinical methods of assessing body composition. Med Sci
Sports Exerc 1984;16:6O6-13.
42. Gam SM, Leonard WR, Hawthorne V. Reply to letter by
Miccozi and Albanes. (Letter). Am J Clin Nutr 1987;46:377.
43. Deurenberg P, Weststrate JA, Seidell JC. Body mass index as
a measure of body fatness: age-and sex-specific prediction
formulas. Br J Nutr 1991;65:105-14.
44. Hassager C, Gotfredson A, Jensen J, et al. Prediction of body
composition by age, height, weight, and skinfold thickness in
normal adults. Metabolism 1986;35:1081-4.
45. Valero MA, Leon-Sanz M, Gomez I, et al. A comparison
between double-photon absorptiometry (DEXA), impedance,
and anthropometry in the study of the body composition of
obese subjects. Nutr Hosp 1994;9:12-17.
46. Pritchard JE, Nowson CA, Strauss BJ, et al. Evaluation of dual
energy X-ray absorptiometry as a method of measurement of
body fat. Eur J Clin Nutr 1993;47:216-28.
Am J Epidemiol
Vol. 143, No. 9, 1996