International Journal of Obesity (1998) 22, 878±884
ß 1998 Stockton Press All rights reserved 0307±0565/98 $12.00
http://www.stockton-press.co.uk/ijo
Gender differences in fat mass of 5±7-year old
children
M Mast1, I KoÈrtzinger1, E KoÈnig2 and MJ MuÈller1
1
Institut fuÈr HumanernaÈhrung und Lebensmittelkunde, Christian-Albrechts-UniversitaÈt zu Kiel and 2SchulaÈrztlicher Dienst der Stadt Kiel,
Kiel, Germany
OBJECTIVE: Studying gender differences in fat mass and distribution in a homogeneous group of children.
DESIGN: Crosssectional study.
SUBJECTS: 610 children aged 5 ± 7 y in Kiel, Germany.
METHODS: Anthropometric measures, bioelectrical impedance analysis (BIA).
RESULTS: Although boys had increased body weights (P<0.05), body mass indexes (BMI's) (P<0.001) and waist=hip
ratios (WHRs) (P<0.001), the %fat mass as assessed by BIA (P<0.05) was increased in girls. Although the increased
BMI in boys was independent of the percentile used, gender differences (that is, lower values for boys than for girls at
the same age) in WHR, the sum of four skinfolds and %fat were seen up to the 90th percentile. By contrast, above the
90th percentile there were no differences in skinfold thickness and %fat between boys and girls. Studying 42 BMImatched pairs (boys and girls) also showed that the %fat estimated by BIA (P<0.001) was increased in girls. Plotting
the average of %fat as obtained from skinfold- and BAI-measurements against the difference between data obtained
by the use of the two methods shows that BIA %fat overestimates skinfold %fat at low or normal percent fat mass
(that is, up to 20%) in both genders. By contrast, at increased fat mass, BIA %fat seems to underestimate skinfold %fat
in both genders.
CONCLUSIONS: Gender differences in fat mass and fat distribution are obvious in children aged 5 ± 7 y. These
differences are independent of gender differences in body weight. However, the nutritional state has an in¯uence
and gender differences cannot be detected in overweight and obese children. Our data also suggest that a childrenspeci®c formula used to calculate %fat from skinfold measurements is inappropriate.
Keywords: fat mass; fat distribution; obesity; children; bioelectrical impedance; skinfold measurements
Introduction
Gender differences in fat mass are generally considered to become manifest after puberty. Before puberty
boys and girls have similar average heights, weights
and body mass indices (BMI).1 However three recent
papers using detailed body composition analysis in
children, found differences in fat mass between prepubertal boys and girls.2 ± 4 Using dual-energy X-ray
absorptiometry (DEXA) or bioelectrical impedance
analysis (BIA) in 403 healthy Dutch children and
adolescents (age 4±20 y), percentage body fat (%fat)
was higher in girls than in boys at all ages.2 In another
study on 40 children aged 3±8 y, DEXA-measurements gave strong evidence for gender differences in
fat mass: the percentage body fat was 13.5% vs 20.4%
in boys and girls, respectively.3 In a third study,
Correspondence: Prof Dr MJ MuÈller, Institut fuÈr
HumanernaÈhrung und Lebensmittelkunde, Christian-AlbrechtsUniversitaÈt zu Kiel, DuÈsternbrooker Weg 17 ± 19, 24105 Kiel,
Germany.
E-mail: [email protected]
Received 26 November 1997; revised 2 April 1998; accepted
23 April 1998
DEXA-derived fat mass was signi®cantly increased
in a total of 31 African±American and eight Caucasian girls aged 5±10 y when compared to boys.4 All
these data were contrary to the results of previous
studies, where DEXA5 or DEXA plus anthropometry6
had been used:In the ®rst study, 48 perpubertal individuals did not show gender differences in fat mass,
which became manifest in peri- and postpubertal
subjects.5 This ®nding was supported by another
crossectional study on 265 subjects aged 4±26 y6
Before the age of 10 y, values were similar for both
genders.
DEXA is a novel technique, which offers the
potential of precise measurements of soft tissue composition,7,8 for example, the coef®cient of variation
for the in vivo (DEXA-) assessment of fat mass was
reported to be 2.52%.3 However, the precision of
measures of fat mass, differ between different
DEXA-absorptiometers and values between 1.9±
6.9% were reported, suggesting inter-machine differences.7,8 In addition, signi®cant differences between
the machines by the same manufacturer were found,
despite high standards of quality control within an
individual machine: the intra-machine differences in
the estimation of fat mass may reach about 3 kg.7
Gender and fat mass in young children
M Mast et al
Since different DEXA-machines have been used in
the studies cited above,2 ± 5 methodological problems
limit the comparison between studies and may explain
some of the observed discrepancies. Interestingly,
Bioelectrical impedance analysis BIA2 or anthropometry5 have been used in combination with DEXA in
two of the studies. There was a close correlation
between DEXA and BIA (r ˆ 0.88) and the mean
difference in percentage fat mass as assessed by the
two methods was only 5%.2 Higher values were found
by BIA than DEXA. The percentage body fat (%fat)
as estimated by DEXA, also correlated with skinfold
measurements (r ˆ 0.82).5 Comparing fat measurements of DEXA with anthropometry, showed no
differences between the means for male subjects, but
a difference of about 3 kg (that is, DEXA overestimated skinfold measurements) was found for the
female subjects.5 Thus, despite differences in absolute
values, BIA or anthropometry seem to support
DEXA-measurements.
Taken together, some discrepancies exist regarding
gender differences in fat mass of prepubertal children.
Some of the discrepancies may result from methodological problems. Other problems come from the
different study designs, for example, none of the
above mentioned studies took into account different
body weights or the percentiles of the different parameters measured. In addition, all the studies used
children from a wide age range, thus, conclusions
regarding speci®c age groups are not possible. In Kiel,
we had the opportunity to analyse the data of the 1996
cohort of the `Kiel Obesity Prevention Study' (KOPS)
to determine whether gender differences in fat mass,
as assessed by ®eld measures of body composition,
are present in healthy prepubertal children aged 5±
7 y. We used BIA, as well as anthropometric measurements, because KOPS is a large epidemiological
trial. It was designed to reduce the prevalence of
obesity in children and adolescents in order to
decrease obesity and obesity-related diseases in later
life.9 ± 11 In KOPS, about 600±1000 children aged 5±
7 y were=are assessed every year for a total of six
years. These children are followed until puberty. An
integrated intervention program was designed and
structured for obese children, non-obese children
with obese parents, families with obese children or
children at risk of to becoming obese, and their
teachers. The aim of the intervention is to encourage
the attractiveness of a healthy lifestyle, as well as to
increasing personal autonomy.
Faced with the discrepancies of the above mentioned results, observed in the fat mass of prepubertal
children, we took the opportunity to analyze our 1996
data. These data were used to determine whether
gender differences in fat mass, as assessed by ®eld
measures of body composition, are present in healthy
prepubertal children aged 5±7 y. The 1996 cohort of
KOPS, consists of 610 children aged 5±7 y. The
nutritional data were analyzed with respect to the
different percentiles of body weight. Regarding nutri-
tional measures our data show that in children aged 5±
7 y, gender differences exist. They are seen in normal
weight children, but are camou¯aged by overweight
and obesity.
Methods
Subjects
Between January±June 1996 610 healthy children
(316 boys and 294 girls) in different parts of the
town of Kiel, North Germany, were randomly sampled
and examined in 16 of the 29 primary schools. Nutritional data (age, gender, body weight and height), as
well as sociodemographic data, suggest that our study
population is representative of the total population of
children in Kiel aged 5±7 y. The procedures had been
explained to all parents. The 610 children represent
30.6% of the total population examined by the school
physicians. All parents had given their informed written consent. The assessment of the nutritional state
included bodyweight and height, skinfold measurements (triceps, biceps, abdominal, suprailiac, subscapular), measurements of waist, hip and mid-arm
circumferences and a BIA. The study was approved
by the local ethical committee.
Body composition
We decided to use anthropometric and BIA-measurements in our study for two reasons. First, anthropometric measurements are used because there is already
a considerable data base in Schleswig-Holstein which
can be compared with our data. Second, BIA is used
because it can be applied in epidemiological studies
(see National Health and Nutrition Examination
Survey (NHANES)) and may provide more accurate
data when compared with anthropometric data. Since
BIA and DEXA data also showed a very close
correlation in children,2 we feel justi®ed in taking
BIA data as a measure of %fat mass in children. We
were not allowed to take DEXA measurements for a
subgroup of children, for ethical reasons (that is,
radiation exposure).
Anthropometry Data were assessed for weight and
height of the children, as part of the medical school
test. Weight was measured to the nearest 0.1 kg on a
calibrated balance-beam scale with subjects wearing
light underwear on and height was estimated to the
nearest 0.5 cm. All other anthropometric measurements were performed by trained observers, according to standard techniques.12 Skinfold thickness was
determined to the nearest 0.2 mm at the right biceps
(BSF), triceps (TSF), subscapular (SSF) and suprailiac (SIF) sites, with a lafayette skinfold caliper,
(Model 01127, Lafayette Instrument Company, Indiana 47903) calibrated to exert a constant pressure of
879
Gender and fat mass in young children
M Mast et al
880
10 g=mm2. The coef®cients of variation forrepeated
(n ˆ 3) measurements of BSF, TSF, SSF and SIF in 10
children aged 5±7 y were 6.0%, 1.9%, 4.0% and
4.3%, respectively. Fat mass (FM) was determined
by age- and gender-speci®c formulas, according to
Lohmann,13 involving weight and the log-transformed
sum of BSF, TSF, SSF, and SIF skinfolds for subjects
aged<12.0 y.
FM[kg] ˆ weight*((5.28/D)74.86), where (D ˆ
body density).
D, boys [g/ml] ˆ 1.16907 0.0788*(log (sum of
four skinfolds))
D, girls [g/ml] ˆ 1.206370.0999* (log (sum of
four skinfolds))
Fat free mass (FFM) was calculated as the difference between weight and body fat. Waist circumference was measured midway between the lower rib
margin and the iliac crest. Hip circumference was
measured at the point yielding the maximum circumference over the buttocks. The WHR was calculated to
obtain an index for the pattern of body fat distribution.
groups): group 1: BMI<25. percentile, group 2: BMI
between 25±75 percentile, group 3: BMI>75 percentile; and group 4: all BMI matched pairs). In addition to
the standard procedure, a Bland and Altman plot17 was
used to compare the difference between the two
methods with the mean values. A positive difference
indicated a relative overestimation of %fat mass,
whereas a negative difference suggested a relative
underestimation of %fat mass.
Results
Figure 1A and B, show the association between height
and weight of our study population. When compared
with girls, boys aged 5±7 y had an increased body
Bioelectrical impedance analysis (BIA) Measurement of whole body bioimpedance in children, has
been described previously.14,15 The BIA measurements presented here were performed at a single
frequency
(50 kHz)
(Multi-Frequency-Analyzer
2000-M, Data Input GmbH, Frankfurt=M, Germany)
with one pair of electrodes appropriately placed on the
dorsal surfaces of the right hand and a second pair of
electrodes placed on the right foot.15 With the subjects
lying in a supine position, measurements were
performed while the hands and feet were extended
from the side of the trunk. The equation used for
children was based on body density and underwater
measurements of speci®c gravity, as reported by
Houtkooper et al.16 The equation provides an estimate
for fat mass (FM) [%] ˆ (71.11)*(ht)2/R‡1.04*wt
‡15.16, where R is the body's resistance (O), ht is
the subject's height (cm), and wt is the subject's body
weight (kg). FFM is then calculated as the difference
body weight and FM. The coef®cients of variation for
repeated (n ˆ 3) estimations of R and XC (reactance
(O)) in 10 children aged 5±7 y were 1.0 and 2.1%,
respectively, resulting in a cv of 1.5% in %fat mass.
Statistical analyses
All statistical analyses (calculation of percentiles,
regression analyses, statistics) were performed using
Excel 5.0. The data are presented as mean values and
range. Different percentiles of the various measurements were calculated. The basic statistics included the
Student t-Test, for comparison between gender groups,
correlation analyses for the relation of different assessment methods for fatmass and paired t-Test for comparison between methods in BMI matched groups 42
BMI matched pairs of boys and girls (divided into four
Figure 1 Correlation between height and body weight in 610
children aged 5 ± 7 y. (A) boys ( P<0.001) and (B) girls ( P<0.001).
Gender and fat mass in young children
M Mast et al
881
Table 1 Characteristics of the children
Mean (range)
Boys (n ˆ 316)
Age (y)
Height [m]
Weight [kg]
BMI [kg=m2]
MAC [cm]
WHR
TSF [mm]
BSF [mm]
SIF [mm]
SSF [mm]
Trunk: extremity skinfold ratio
Sum of four skinfolds [mm]
FM, A [%]
FM, A [kg]
FM, BIA [%]
FM, BIA [kg]
6.1
1.21
23.6
16.5
18.2
0.89
11.1
6.2
8.7
5.8
0.97
31.7
15.5
3.86
17.9
4.34
(5.0 ± 7.8)
(1.00 ± 1.38)
(15.0 ± 45.0)*
(11.2 ± 32.1)***
(12.0 ± 28.5)
(0.74 ± 1.07)***
(2.6 ± 25.3)
(1.3 ± 18)
(1.6 ± 36.3)
(1.3 ± 28.7)
(0.35 ± 2.87)
(8.8 ± 100.3)
(2.7 ± 36.1)***
(0.5 ± 15.1)***
(4.9 ± 34.3)*
(1.4 ± 15.4)
Girls (n ˆ 294)
6.1
1.2
22.8
15.6
18
0.87
11.4
6.3
8.9
6.2
1.00
32.8
13.6
3.24
18.6
4.31
(5.0 ± 7.3)
(1.07 ± 1.38)
(15.0 ± 40.7)
(11.5 ± 24.5)
(14.0 ± 24.0)
(0.7 ± 1.11)
(3.3 ± 27.0)
(2.3 ± 18.3)
(2.6 ± 37.3)
(1.6 ± 31.3)
(0.37 ± 2.35)
(11.9 ± 111.7)
(0.2 ± 41.1)
(0.03 ± 16.7)
(5.1 ± 29.1)
(1.2 ± 11.8)
BMI ˆ body mass index; MAC ˆ mid-arm circumference; WHR ˆ waist ± hip ratio;
TSF ˆ Triceps skinfold thickness; BSF ˆ Biceps skinfold thickness; SIF ˆ Subrailiacal
skinfold thickness; SSF ˆ Subscapular skinfold thickness; FM ˆ fat mass; FM, A ˆ fat mass
from Anthropometry; FM, BIA ˆ fat mass from bioelectrical impedance analysis (BIA); the
data are given as mean values and range. Gender differences were analysed by unpaired ttest: *P<0.05; ***P<0.001.
weight, BMI and WHR (Table 1). The sum of four
skinfolds tended to be greater in girls, whilst %fat
mass calculated from BIA-measurements was
increased in girls. Calculating %fat from skinfold
measurements according to Lohman13 resulted in a
%fat mass, which was different from BIA-data (Table
1). Table 2 shows the different percentiles for body
weight, height, BMI, triceps skinfolds, sum of four
skinfolds, FM as estimated from anthropometry, as
well as from BIA-measurements, and also includes the
WHR and the middle arm circumference (MAC). The
sum of four skinfolds and the %fat from BIA-measurements, were greater for girls up to the 80th and
90th percentiles, respectively (Table 3). Signi®cant
gender differences were observed up to the 80±90th
percentile (Table 3). Skinfold and BIA-measurements
showed an increased fat mass for girls up to the 80th
or 90th percentile, respectively (Table 3). By contrast,
overweight boys showed an increased fat mass when
compared with overweight girls (Table 3). There was
a correlation between the sum of four skinfolds and
BIA-derived %fat mass in boys as well as in girls
(Figure 2A and B). Calculating %fat from skinfold
measurements showed an increased FM in boys at
each percentile studied (Table 3).
Plotting the average %fat, as obtained by skinfoldand BIA-measurements against the difference
between %fat derived from the two methods for
boys (Figure 3A) and girls (Figure 3B) shows that
BIA %fat overestimates skinfold %fat at low or
normal percent fat mass (that is, up to 20%) in both
genders (Figure 1A and B). By contrast, at increased
values, BIA %fat seems to underestimate skinfold
%fat in both genders (Figure 3A and 3B).
As (i) body weight affects the calculation of FM
and (ii) differs between both genders, we paired a
subgroup of 42 BMI-matched pairs (boys and girls).
The data are shown in Table 4. Studying the 42 BMImatched pairs (boys and girls) showed that %fat BIA
(Table 4) and the sum of the skinfolds (Table 4) were
both increased, but only the BIA-values in girls
reached statistical signi®cance. Gender differences in
BIA %fat were signi®cant at the different BMI-percentiles studied (Table 4). Comparing three pairs at a
BMI >90th percentile showed a mean BIA %fat of
22.4 and 19.1% in girls and boys, respectively. By
contrast, calculating %fat from skinfold measurement
again showed opposite results for boys (Table 4).
Discussion
Gender differences in FM are evident in children aged
5±7 y (Tables 1±3). These differences are independent of BMI (Table 4). Our study con®rms and
extends three recent studies2 ± 4 in a larger sample
(that is, 610 children in our study vs 40±403 children).
Our study population was homogenous with respect to
age (that is, only children aged 5±7 y were studied,
whereas children aged 3±20 y were assessed by the
other authors. It is evident from our data that gender
differences in fat distribution and %fat mass were
signi®cant up to the 90th percentile of the nutritional
state (Table 3). No differences in %fat mass and fat
distribution could be observed in overweight and
obese children (Table 3). Thus overweight camou¯ages gender differences in FM and fat distribution. It
is tempting to speculate that the relative proportion of
overweight children may contribute to the discrepancies between the studies cited above.2 ± 6 It is evident
that the different percentiles of the nutritional parameters have to be taken into account in future studies.
17.76
1.12
13.06
15.7
0.77
7
18.2
0.8
3.19
15.84
75.71
2.74
14.86
14.76
76.89
10.81
56.28
18
1.11
13.56
16
0.8
6.3
17.2
1.5
6.91
15.68
72.97
2.82
14.31
14.93
76.05
10.93
55.67
girls
1.8
8.82
16.46
76.43
3.01
14.95
15.64
78.8
11.45
57.68
19
1.13
14.1
16
0.82
7.3
19.5
boys
1.07
5.35
16.64
77.8
3.08
15.55
15.65
78.2
11.45
57.24
19
1.14
13.8
16
0.8
7.7
20.8
girls
10 percentile
2.19
10.69
17.5
80.37
3.29
15.71
16.87
80.67
12.35
59.05
20
1.16
14.61
17
0.84
8.32
21.94
boys
1.75
8.26
17.55
80.47
3.45
16.43
16.21
79.58
11.87
58.25
19.96
1.16
14.34
16.86
0.82
9
24.08
girls
20 percentile
2.34
11.42
17.76
81.49
3.41
16.12
17.14
81.21
12.54
59.44
20.5
1.17
14.9
17
0.85
9
22.97
boys
1.88
9.09
17.99
81.93
3.53
16.83
16.57
80.05
12.13
58.6
20.1
1.17
14.6
17
0.83
9.6
25.1
girls
25 percentile
3.32
14.84
19.35
85.16
3.86
17.5
18.78
82.5
13.75
60.39
22.5
1.21
15.8
18
0.89
10.3
28.5
boys
2.86
12.84
19.4
87.16
4.03
18.48
18.44
81.52
13.5
59.67
22.5
1.21
15.5
18
0.87
11
30.2
girls
50 percentile
4.31
18.51
21.59
88.68
4.62
18.79
21.06
83.88
15.42
61.4
25.4
1.24
17.4
19
0.93
12.92
35.5
boys
4.13
18.07
21.16
90.91
4.84
19.95
20.02
83.17
14.65
60.88
24.5
1.24
16.38
19
0.91
13.33
38.9
girls
75 percentile
4.79
19.63
22
89.31
4.85
19.33
21.59
84.29
15.8
61.7
26
1.25
17.93
19.5
0.93
14
37.98
boys
4.56
19.53
21.41
91.74
5.02
20.42
20.36
83.57
14.9
61.17
25
1.25
16.76
19
0.92
14.12
41.71
girls
80 percentile
6.91
23.57
23.54
91.18
6.02
21.2
23.8
85.05
17.42
62.26
29.8
1.28
19.8
20.2
0.95
15.3
48.3
boys
5.82
22.2
22.34
94.65
5.88
21.8
21.87
84.45
16.01
61.82
27
1.27
17.8
20
0.95
15
47.3
girls
90 percentile
8.57
27.03
25.08
93.09
7.97
23.95
26.23
85.69
19.2
62.73
32.8
1.3
22.2
22.5
0.97
17.6
59.4
boys
7.04
24.29
23.76
96.81
6.7
23.11
22.95
85.14
16.8
62.32
29.7
1.29
18.6
21.1
0.97
16.3
52.2
girls
95 percentile
882
boys
18.6
1.14***
13.5
15.6
0.79
19.7***
girls
boys
girls
boys
21.4
1.2
15
17.4
0.85
27.6***
girls
25 to 50
percentile
21.6*
1.19
15.4***
17.3
0.83*** 0.81
0.84*** 0.83
0.87***
20.8
22.5*** 22.5
24.7*** 25.9
girls
20 to 25
percentile
19.1
19.2** 20.4*** 20
1.15
1.15
1.17
1.17
14.4*** 14.1
14.8*** 14.5
boys
10 to 20
percentile
girls
boys
18.2
21.5**
20.2*** 20
31.0***
1.3
20.6***
22.8***
0.96
52.7**
boys
28.5
1.28
18.2
21.5
0.96
49.4
girls
90 to 95
percentile
girls
1
73.7
28.5
25.4
0.99
65.1
37.7*** 33
1.34
1.32
24.7*** 20.4
boys
>95
percentile
20.7
25.0*** 23.3 30.5
21.0*** 22.2
22.2 27.5*
26.2
1.26***
17.2
19
0.93
44.2*
girls
80 to 90
percentile
27.7***
1.27
18.7***
19.1
0.94***
40.0*** 42.7
girls
25.9*** 25
1.25
1.25
17.7*** 16.6
boys
75 to 80
percentile
0.91*** 0.89
31.7
34.1*** 36.4
23.9
23.7
1.23
1.23
16.4*** 15.9
boys
50 to 75
percentile
5.4*** 1.6
7.9***
4.4
9.8***
6.9
11.0***
8.8
13.3*** 10.9
16.5*** 15.2
18.8
12.6
12.2 14.7
15.3*** 15.3
16.1*** 15.9
16.7*** 16.8
17.9*** 18.1
19.2*** 19.1
16.5 18.8
1.11 1.12
12.5 13.9***
15.5
0.78*
0.72 0.81***
15.3
16.3 18.4
17.2*
1.1
13.0*
girls
5 to10
percentile
Mean values within each percentile band are shown. Gender differences were analysed by unpaired t-Test. *P<0.05; **P<0.01; ***P<0.001; BMI ˆ body mass index; WHR ˆ waist ± hip ratio;
MAC ˆ middle arm circumference; FM, A=BIA ˆ fat mass from Anthropometry=bioelectrical impedance analysis (BIA).
Weight [kg]
Height [m]
BMI [kg=m2]
MAC [cm]
WHR
Sum of four
skinfolds [mm]
FM, A [%]
FM, BIA [%]
boys
0 to 5
percentile
Table 3 Body composition in different percentile groups
BMI ˆ body mass index; MAC ˆ mid-arm circumference; WHR ˆ waist ± hip ratio; TSF ˆ Triceps skinfold thickness; FM ˆ fat mass; FM, A ˆ fat mass from Anthropometry; FFM ˆ fat free mass;
TBW ˆ total body water; FM, BIA ˆ fat mass from bioelectrical impedance analysis (BIA).
Weight (kg)
Height [m]
BMI [kg=m2]
MAC [cm]
WHR
TSF [mm]
Sum of four
skinfolds [mm]
FM, A [kg]
FM, A [%]
FFM, A [kg]
FFM, A [%]
FM, BIA [kg]
FM, BIA [%]
FFM, BIA [kg]
FFM, BIA [%]
TBW, BIA [l]
TBW, BIA [%]
boys
05 percentile
Table 2 Percentiles of characteristics of the children (316 boys and 294 girls)
Gender and fat mass in young children
M Mast et al
Gender and fat mass in young children
M Mast et al
Our data also point to some methodological issues.
Anthropometric measures and BIA were used in our
study to assess FM. Comparing the two methods
skinfold thickness and %fat BIA went in parallel
(Table 1±4, Figure 2A and B). However, calculating
Figure 2 Correlation between %fat mass as derived from bioelectrical impedance analysis (BIA)-measurements and the sum
of four skinfolds in 610 children aged 5 ± 7 y (A) boys ( P<0.001)
and (B) girls ( P<0.001).
%fat from skinfolds according to Lohman,13 led to
opposite results (Table 1±4). This was also seen in
BMI-matched pairs (Table 4). The discrepancy in
%fat, as calculated from skinfold and BIA-measure-
Figure 3 Altman and Bland22 plot comparing the difference of
%fat as derived from anthropometric or bioelectrical impedance
analysis (BIA)-measurements and the mean of the results
obtained by two methods. The relative bias (BIA minus skinfolds)
is plotted against the size of measurement. Each data point
represents a single measurement. ( n ˆ 610 children; (A) 316
boys, (B) 294 girls).
Table 4 Body composition characteristics of children (n ˆ 42 body mass index (BMI) matched pairs)
BMI matched,
<25 percentile
girls (n ˆ 8)
Age (y)
BMI (kg=m2)
FM, BIA (%)
FM, A (%)
Sum of 4 skinfolds (mm)
6.0
13.9
18**
11.6
28.7
BMI matched,
between 25 and 75 percentile
boys (n ˆ 8) girls (n ˆ 28) boys (n ˆ 28)
6.0
13.8
16.5
14.2
28
6.2
15.7
18.2**
11.8
29.6
6.0
15.7
16.9
14.2*
28.2
BMI matched,
BMI>75 percentile
girls (n ˆ 6)
6.4
18.7
22.2*
21.4
46.8
BMI matched
boys (n ˆ 6) girls (n ˆ 42) boys (n ˆ 42)
6.2
18.8
18.1
20.9
42
6.2
15.8
18.8***
13.1
31.9
6.0
15.8
17
15.2*
30.2
Gender differences were analysed by paired t-test: *P<0.05; **P<0.01; ***P<0.001. Data are given as mean values. FM, BIA: fat mass
from bioelectrical impedance analysis (BIA); FM, A: fat mass from Anthropometry.
883
Gender and fat mass in young children
M Mast et al
884
ments. becomes obvious using a Bland and Altman
plot (Figure 3A and B). Our results suggest that the
calculation from skinfolds according to Lohman13
underestimate BIA data at low and normal %fat,
whereas an overestimation is seen at increased %fat.
These differences were seen in boys (Figure 3A) as
well as in girls (Figure 3B). When compared to BIA or
DEXA, skinfold measurements are generally considered to be less sensitive. Our data provide some
evidence that the Lohman-formula used to assess
%fat from skinfold measurements13 is inappropriate
for children aged 5±7 y. Comparing BIA with DEXA,
some authors have proposed that the body-composition data from DEXA could be a reference method
and thus could replace underwater weighing (comp.
18). However a recent metanalysis of 54 papers,
published between 1985±1996 on the measurement
of body composition in adult caucasians, various
methods (for example, DEXA and BIA) in comparison with underwater weighing, showed no systematic
over- or underestimation of %fat mass by DEXA or
BIA and both methods had similar differences in
bias.18 In children, there was a close agreement
between BIA and DEXA data, and the mean difference in %fat mass as assessed by the two methods was
only 5%.2 In addition, the accuracy of BIA-measurements in children (that is, a CU of 1.5%, see Methods)
is in the order of DEXA-measurements (comp.
Above). Thus, we feel that BIA can be used with
some con®dence for adults as well as children.
There are only few data on the prevalence of central
body obesity in children.19 Although the WHR was
shown to be only a poor predictor of intra-abdominal
adipose tissue in children19 and adults,20 our data may
add some useful information. First, the WHR had a
wide variance in children aged 5±7 y (that is, from
0.70±1.11; Table 1). Second, the mean values were
increased in boys up to the 90th percentile. Third, no
differences were found between overweight and obese
boys and girls. As there is ®rm evidence that central
body fat in childhood is a risk for hyperlipidaemia or
hypertension, for example, in adults,21,22 we are going
to follow the subgroup of overweight and obese children with an increased WHR (that is, >90th percentile)
in our obesity prevention study (i.e. KOPS).
Acknowledgements
This work was supported by grants from Verein zur
FoÈrderung der Rehabilitationsforschung in SchleswigHolstein e.V., LuÈbeck; Wirtschaftliche Vereinigung
Zucker, Bonn; Else KroÈner-Fresenius Stiftung, Bad
Homburg; Bad Schwartau Werke, Bad Schwartau;
team success, Selent.
References
1 Siervogel RM, Roche AF, Guo S, Mukherjee D, Chumlea WC.
Patterns of change in weight=stature2 from 2 to 18 years:
®ndings from long-term serial data for children in Fels longitudinal growth study. Int J Obes 1991; 15: 479 ± 485.
2 Boot AM, Bouquet J, de Ridder MAJ, Krenning EP, deMuinck
Keizer-Schrama SMPF. Determinants of body composition measured by dual energy X-ray absorptiometry in Dutch
children and adolescents. Am J Clin Nutr 1997; 66: 232 ± 238.
3 Taylor RW, Gold E, Manning P, Goulding A. Gender differences in body fat content are present well before puberty. Int.
J. Obes 1997; 21: 1082 ± 1084.
4 Nagy TR, Gower BA, Trownbridge CA, Dezenberg C, Shewchuk M, Goran M. Effects of gender, ethnicity, body composition, and fat distribution on serum leptin concentrations in
children. J Clin Endo Metab 1997; 82:
2148 ± 2152.
5 Rico H, Revilla M, Villa LF, Hernandez ER, Alvarez de
Buergo M, Villa M. Body composition in children and
Tanner's stages: A study with Dual-Energy X-ray Absorptiometry. Metabolism 1993; 42: 967 ± 970.
6 Ogle GD, Allen JR, Humphries IRJ, Lu PW, Briody JN,
Howard-Giles R, Cowell CT. Body-composition assessment
by dual-energy X-ray absorptiometry in subjects aged 4 ± 26 y.
Am J Clin Nutr 1995; 61: 746 ± 753.
7 Jebb SA. Measurement of soft tissue by dual-energy X-ray
absorptiometry. Br J Nutr 1997; 77: 151 ± 163.
8 Fogelholm M and von Marken Lichtenbelt W. Comparison of
body composition methods: a literature analysis. Eur J Clin
Nutr 1997; 51: 495 ± 503.
9 Mast M, KoÈrtzinger I, Bartrow A, Hunte V, Neite S, KoÈnig
E, MuÈller MJ. Kiel obesity prevention study (KOPS): De®nition of obesity in children. Int J Obes 1997; 21: Suppl. 2,
S121.
10 KoÈrtzinger I, Mast M, Bumbe A, Grund A, MuÈller MJ. Schooloriented intervention for the prevention of obesity as part of
KOPS (Kiel obesity prevention study). Intern J Obesity 1997;
21 (Suppl. 2): S30.
11 KoÈrtzinger I, Mast M, MuÈller MJ. PraÈvention der Adipositas
bei Kindern und Jugendlichen. ErnaÈhrungs-Umschau 1996;
43: 455 ± 460.
12 Lohman TG. Anthropometric Standardization Reference
Manual. A division of Human Kinetics Publishers: Champaign, 1988.
13 Lohman TG. Applicability of body-composition techniques
and constants for children and youths. In: Pandolph, KB (ed.),
Exercise and Sport Sciences Review, Macmillan: New York
1986, pp 325 ± 357.
14 Kushner R. Bioelectrical impedance analysis: A review of
principles and applications. J Am Coll Nutr 1992; 11: 199 ±
209.
15 MuÈller MJ, von zur MuÈhlen A, Lautz HU, Schmidt W, Daiber
M, HuÈrter P. Energy expenditure in children with type I
diabetes: evidence for increased thermogenesis. BMJ 1989;
299: 487 ± 491.
16 Houtkooper LB, Lohmann TG, Going SB, Hall MC. Validity
of bioelectrical impedance for body composition assessment in
children. J Appl Physiol 1989; 66: 814 ± 821.
17 Bland JM, Altman DG. Statistical methods for assessing
agreement between two methods of clinical measurement.
Lancet 1986, 307 ± 310.
18 Fogelholm M, van Marken Lichtenbelt W. Comparison of
body composition methods: a literature analysis. Eur J Clin
Nutr 1997; 51: 495 ± 503.
19 Goran MI, Kaskoun M, Shuman WP. Intra-abdominal adipose
tissue in young children. Int J Obesity 1995; 19: 279 ± 283.
20 Van der Koy K, Leenen R, Seidell JC, Deurenberg P, Droop
A, Bakker CJG. Waist ± hip ratio is a poor predictor of changes
in visceral fat. Am J Clin Nutr 1993; 57: 327 ± 333.
21 Freedman DS, Srinivasan SR, Harsha DW, Webber LS,
Berenson GS. Relation of body fat patterning to lipid and
lipoprotein concentrations in children and adolescents: the
Bogalusa Heart Study. Am J Clin Nutr 1989; 50: 930 ± 939.
22 Shear CL, Freedman DS, Burke GL, Harsha DW, Berenson
GS. Body fat patterning and blood pressure in children and
young adults. Hypertension 1987; 9: 236 ± 243.