Percent Body Fat and Chronic Disease Risk
Factors in U.S. Children and Youth
Scott B. Going, PhD, Timothy G. Lohman, PhD, Ellen C. Cussler, MS,
Daniel P. Williams, PhD, John A. Morrison, PhD, Paul S. Horn, PhD
Background: The dramatic increase in pediatric obesity has renewed interest in accurate methods
and screening indexes for identifying at-risk children and youth. Whether age-specifıc standards are
needed is a factor that remains uncertain.
Purpose: This study was designed to describe the age-specifıc fatness–risk factor relationship in
boys and girls across a wide age range.
Methods: Data were from 12,279 white, black, and Mexican-American children and adolescents
from the National Health and Nutritional Examination Surveys (NHANES) III (1998 –1994) and IV
(1999 –2004). Children were grouped based on percent fat, estimated from subscapular and triceps
skinfolds, and the age-specifıc relationships between percent fat and chronic disease risk factors (e.g.,
blood pressure, lipids and lipoprotein levels, glucose, insulin, and circulating C-reactive protein
levels) were described in boys and girls, aged 6 –18 years.
Results: Percent fat was signifıcantly related to risk factor levels. At higher levels of percent fat,
the prevalence of adverse cardiovascular disease risk factors was higher, particularly above 20%
fat in boys and above 30% fat in girls. In boys and girls, the interaction term age by percent fat was
a signifıcant predictor of risk factors, whereas the percent fat by race interaction term was
nonsignifıcant.
Conclusions: The results demonstrate a strong relationship between chronic disease risk factors
and percent fat in children and youth that varies by age in boys and girls.
(Am J Prev Med 2011;41(4S2):S77–S86) © 2011 American Journal of Preventive Medicine
Introduction
C
hildhood and adolescent overweight and obesity
are important public health concerns. The most
recent U.S. national surveys indicate that 31.7% of
youth are overweight,1 and more than half of this group,
16.9%, are obese. The high prevalence of overweight and
obesity in children and youth, with its attendant health
risks,2–7 has renewed interest in the development of accurate methods for body composition assessment and
screening indexes for identifying children and youth at
risk for obesity-related comorbidities.
From the Department of Nutritional Sciences (Going), Department of
Physiology (Lohman, Cussler), University of Arizona, Tucson, Arizona;
Department of Health and Physical Education, Northern State University
(Williams), Aberdeen, South Dakota; and Children’s Hospital Medical
Center, Division of Cardiology (Morrison), Department of Mathematical
Sciences (Horn), University of Cincinnati, Cincinnati, Ohio
Address correspondence to: Scott B. Going, PhD, Department of
Nutritional Sciences, University of Arizona, Ina E. Gittings Building,
Room 3Q, 1713 E. University Boulevard #93, Tucson AZ 85721. E-mail:
[email protected].
0749-3797/$17.00
doi: 10.1016/j.amepre.2011.07.006
Common defınitions of pediatric overweight and
obesity are based on the BMI. Early recommendations
set the 85th percentile of age- and gender-specifıc BMI
as the level at which children and youth were considered at risk of overweight (now called overweight instead of at risk of overweight), and the corresponding
95th percentile of BMI was defıned as overweight (now
called obesity).8 These defınitions continue to be used
today, although present-day surveys compare children
and youth to BMI distributions that existed in the
1960s and 1970s.9
Although BMI is a practical, easy-to-obtain index, its
application assumes that differences in BMI among individuals reflect differences in adiposity and that individuals with identical BMI have identical body composition.
Clearly these assumptions are not valid,10,11 although it is
the degree of variation that matters when setting a screening standard. A more important limitation may be the
arbitrariness of a defınition based on the population distribution of the BMI rather than an a priori criterion, such
as disease risk.
© 2011 American Journal of Preventive Medicine • Published by Elsevier Inc.
Am J Prev Med 2011;41(4S2):S77–S86 S77
S78
Going et al / Am J Prev Med 2011;41(4S2):S77–S86
Although studies have shown that increasing levels of
BMI are associated with higher levels of disease risk especially in youth, some studies have shown that worrisome
levels of risk occur at percentiles higher than the 95th
percentiles of BMI,9 whereas only modest risk is associated with the current recommendations.6,12 Age-,
gender-, and race/ethnicity-related variation in the BMI–
body fat relationship potentially limit the widespread
use of BMI for assessing disease risk.13–15 An alternative approach to setting obesity standards is to study
the association of a more direct measure of adiposity
with risk factors to determine the level of fatness associated with high levels of risk factors, although the
challenges of measuring fatness directly outside of the
laboratory are acknowledged. Using this approach,
body fat levels ranging from 20%–25% in boys and
30%–35% in girls have been shown to be associated
with health risk.16 –19 These ranges are remarkably
similar, given the differences in samples, sample sizes,
and methodology in these studies.
In one of the largest studies of percent fat and risk
factors, Williams et al.19 reported that body fat levels
ⱖ25% in boys and ⱖ30% in girls were associated with
over-representation (excess risk) in the highest fıfth of
age-, gender-, and race-specifıc distributions of cardiovascular disease (CVD) risk factors. These fatness levels
have been used to differentiate between unhealthy and
healthier levels of body fat in children and youth ranging
from age 5–18 years (www.cooperinstitute.org/). The appropriateness of applying the same standard across such a
wide age range is a factor that has remained uncertain.
Developing standards against a surrogate for adiposity
such as BMI in absolute units of kg/m2 is confounded by
differential changes in fat and fat-free mass at different
stages of maturation that likely influence the relationships between BMI and risk factors, which is why ageand gender-specifıc BMI percentiles are used.9,20
When risk is defıned against more direct measures of
body fat, the level of body fatness that confers risk may be
very similar at different ages. However, due to the
changes in hormonal profıle and fat-free mass that accompany maturation, the relationships between body fat
and risk factors may also vary with age, thereby suggesting that health-related adiposity standards for boys and
girls should be defıned differently at different ages. Given
the uncertainty concerning the influence of age on the
fatness–risk factor relationship, the present analysis was
undertaken to describe the age-specifıc percent fat–risk
factor relationship in boys and girls across a wide age
range in a large sample representative of U.S. boys and
girls.
Methods
Participants
Data for a cohort of children (aged 6 –18 years) were selected
from the National Health and Nutritional Examination Surveys
(NHANES) III (1988 –1994; Series 11 data; n⫽8559) and IV
(1999 –2004; n⫽10,324). Standardized interviews, clinical examinations, and laboratory tests were conducted for both surveys.
Results were reviewed for the examination and laboratory portions
of the surveys, and based on the aim, children were excluded for the
following reasons: aged ⬍6 years (n⫽1625), for limited numbers of
outcome measurements and numerous implausible body weight
measures among younger children; race other than white, black, or
Mexican-American (n⫽944) because sample sizes from other racial groups were too small for meaningful analysis; missing percent
fat by skinfolds (triceps and subscapular) (n⫽865); evidence of
acute inflammation due to infection from self-report or diagnostic
coding (n⫽199); pregnancy (n⫽30); having a pacemaker (n⫽1);
and taking medication for high blood pressure and/or diabetes
(n⫽21).
Since missing data were not imputed, the number included in
each analysis varied somewhat. Analysis with glucose was limited
to boys and girls aged 9 –18 years since results for younger children
were not available. Analysis with insulin was limited to adolescents
since results for children aged ⬍12 years were not collected. The
fınal sample for analysis (n⫽12,279 children and adolescents) included 6222 boys and 6057 girls.
Anthropometry
For NHANES III (1998 –1994), subscapular and triceps skinfold
measurements were taken twice, by two different technicians, in
centimeters to the nearest 0.1 cm using Holtain skinfold calipers.
Standing height was measured in centimeters to the nearest 0.1 cm
using an electronic stadiometer. Weight was measured in kilograms to the nearest 0.1 kg using a Toledo self-zeroing weight scale.
The mean of repeated measures was used in the analyses. BMI was
calculated as weight (kg) divided by the square of standing height in
meters. In NHANES IV (1999 –2004), the same standardized protocols were used to measure height and weight. For greater accuracy, the measured heights and weights were fed directly into the
Integrated Survey and Information System (ISIS) computer system. Skinfold measurements in NHANES IV (1999 –2004) were
obtained by a single technician. Detailed training procedures, examination protocols, and procedures for all anthropometric measures can be found on the NHANES website.21 Quality control
checks are included throughout the data collection procedure.22,23
Percent Body Fat
Percent fat was calculated from triceps and subscapular skinfolds
using the equations of Williams et al.,19 which adjust for agerelated differences in the composition of the fat-free mass that
confound other skinfold equations in children and youth. In this
approach, gender- and race-specifıc regression equations are used
to estimate body density from age and the sum of skinfolds using
quadratic equations that adjust for the curvilinear nature of the
skinfold– body density relationship. Subsequently, body density is
converted to percent fat using age-adjusted, gender-specifıc regression equations that modify the youth-specifıc body density conwww.ajpmonline.org
Going et al / Am J Prev Med 2011;41(4S2):S77–S86
stants for individual differences in age. The details and the equations are published elsewhere.19
Blood Pressure
Blood pressure levels were measured by a physician using the
appropriate size cuff (based on measured arm circumference) so
that the lower edge was 1 inch above the antecubital space. Both
systolic and diastolic pressures were measured in millimeters of
mercury (mm Hg) three to four times after the participant rested
quietly for 5 minutes, and the means of the fınal two measures were
used in the analysis.24,25
Blood Specimens
Blood samples were drawn from survey participants, and serum
and/or plasma was analyzed according to standardized NHANES
protocols.26,27 Circulating lipid and lipoprotein cholesterol levels,
including total cholesterol, high-density lipoprotein cholesterol
(HDL-C), and triglycerides, were assessed in milligrams per deciliter. Low-density cholesterol (LDL-C) was calculated (in milligrams per deciliter) from total cholesterol, HDL-C, and triglycerides in those individuals without extremely high triglyceride levels.
Circulating C-reactive protein (CRP) levels were assessed in milligrams per liter. Fasting glucose (milligrams per deciliter) was measured only in children aged ⱖ12 years in both surveys, and insulin
(microunits per milliliter) was measured only in children aged ⱖ12
years in the NHANES IV (1999 –2004) survey.26,27
Statistical Analysis
For analysis, children were grouped based on percent body fat as
follows: for boys, ⬍10.0%, 10.0%–14.9%, 15.0%–19.9%, 20.0%–
24.9%, 25.0%–29.9%, and ⱖ30.0%, and for girls, ⬍20.0%, 20.0%–
24.9%, 25.0%–29.9%, 30.0%–34.9%, and ⱖ35.0%. Similar groupings have been used in past analyses.19 Although the aim in the
current study was to stratify by percent fat into as many genderspecifıc groups (5% fat intervals) as possible, sample sizes became
smaller as percent fat increased, and they became limiting above
30% fat in boys and 35% fat in girls.
Children and adolescents were grouped into age, gender, and
race-specifıc quintiles for each risk factor. Binary variables were
then constructed from quintiles so that numeral one (1) represented membership in the fıfth group (top 20%) for the risk factor,
and zero (0) represented membership in the bottom four fıfths
(lower 80%). HDL-C groups were reverse coded, since it is the
lower levels of HDL-C that are worrisome. The numeral one
(1) represented membership in the fırst group (bottom 20%) for the
low HDL-C risk factor, and zero (0) represented membership in
the top four fıfths (upper 80%).
Means and SDs for age, anthropometry, body composition,
CVD risk factors, and percentages for race were calculated for boys
and girls separately. Cross-tabulation was used to test for differences in the percentages of boys and girls at different ages in the
most adverse fıfth (or 20%) of each age-, gender-, and race-specifıc
risk factor distribution across percentage body fat levels. Chisquare tests were conducted using the null hypothesis that the
percentage of youth in the adverse fıfth of the risk factor distribution would be similar across all levels of body fat within each ageand gender-specifıc group. The alpha level for signifıcance was set
at ␣⫽0.05, two-sided. Chi-square tests were used to identify body
fatness levels within each age- and gender-specifıc group wherein
October 2011
S79
signifıcant under- (⬍20%) and over-representation (⬎ 20%) of the
most adverse fıfth of the risk factor distribution was present.
To further compare children in the most adverse fıfth to children
in the more favorable four fıfths of a specifıc risk factor distribution, multiple logistic regression models were developed with levels
of percent body fat as the independent variable of interest. Models
were adjusted for NHANES (III [1998 –1994] or IV [1999 –2004]),
the ratio of subscapular to triceps skinfolds, race, the interaction of
race by percent fat, and the interaction of age by percent fat. In
these analyses, the subscapular to triceps skinfold ratio adjusted for
the influence of truncal versus limb fat patterning, which may vary
with age and race.
Because of a wide range in reported fasting times prior to venipuncture, models predicting glucose and insulin outcomes were
compared with and without a covariate for fasting hours. ORs and
95% CIs for percent fat groups were computed showing the odds of
being in the most adverse fıfth for a risk factor at increasing levels of
percent body fat relative to the lowest percent body fat group
(referent). All analyses were stratifıed by gender and age group
because of the gender differences in the assignment to percent fat
groups and to describe any potential age-related differences in the
threshold levels of body fat associated with the most adverse fıfth of
the risk factor distributions. Analysis was conducted in PASW
Statistics Version 17.0.
Results
Approximately 36% of the sample was black, ⬃36% was
Mexican American, and ⬃28% was white (Table 1). The
race distribution was very similar in boys and girls. Using
the CDC age- and gender-specifıc 85th and 95th percentiles of BMI,9 approximately 30% of the boys and girls
were overweight or obese, and approximately 15% were
obese. Body fat averaged 17.6% of body weight in boys
and 23.9% of body weight in girls. The number of boys
and girls in each of the gender-specifıc body fat categories
(⬍10% to ⱖ30% in boys and ⬍20% to ⱖ35% in girls) are
reported in Table 2. Thirty percent of boys had body fat
ⱖ20% of body weight, and 19% had body fat ⱖ25% of
body weight, levels of percent fat that have been related to
elevated CVD risk factors.16,19 Twenty percent of girls
had body fat ⱖ30% of body weight—levels of percent fat
that have been related to elevated CVD risk factors.16,19
Odds ratios describing the chances of being in the
most adverse risk factor group (highest fıfth, except for
HDL-C, which was defıned as the lowest fıfth) are reported by increasing body fat levels in Table 2. All of the
ORs were adjusted for survey (NHANES III [1998 –1994]
or IV [1999 –2004]); race; the ratio of triceps to subscapular skinfold thickness (a measure of truncal versus limb
fat patterning); and the interactions of percent fat with
race and age. In children (aged 6 –11 years) and adolescents (aged 12–18 years), the odds of membership in the
most adverse risk factor group increased with increasing
levels of percent fat in boys and girls. However, there was
considerable variation in the magnitude of the ORs and in
Going et al / Am J Prev Med 2011;41(4S2):S77–S86
S80
Table 1. Characteristics of selected children
(n⫽12,279)a aged 6 –18 years who participated in the
NHANES III (1998 –1994) or NHANES IV (1999 –2004)
survey
Boys (n⫽6,222) Girls (n⫽6,057)
M⫾SD or n (%) M⫾SD or n (%)
Age (years)
12.2 (3.7)
12.1 (3.7)
White
1703 (27.4)
1734 (28.6)
Black
2254 (36.2)
2132 (35.2)
Mexican-American
2265 (36.4)
2191 (36.2)
Height (cm)
154.0⫾20.2
148.9⫾16.2
Weight (kg)
51.1⫾21.4
47.6⫾17.6
BMI
20.5⫾4.8
20.7⫾4.8
ⱖ85th percentile
1879 (30.2)
1764 (29.1)
ⱖ95th percentile
935 (15.0)
783 (12.9)
17.6⫾7.9
23.9⫾6.4
0.9⫾0.3
0.8⫾0.2
Race
Anthropometry
% fat by skinfoldsb
Ratio of subscapular
to triceps skinfold
Risk factors
a
Systolic blood
pressure (mm Hg)
107.0⫾11.3
103.1⫾9.7
Diastolic blood
pressure (mm Hg)
57.8⫾12.2
58.0⫾11.6
Total serum
cholesterol
(mg/dL)
162.9⫾29.6
165.9⫾30.6
LDL cholesterol
(mg/dL)
(n⫽3584)c
92.2⫾25.9
94.9⫾27.7
HDL cholesterol
(mg/dL)
52.0⫾12.6
53.2⫾12.2
Triglycerides (mg/dL)
89.6⫾54.0
91.0⫾51.5
C-reactive protein
(mg/L)
0.2⫾0.4
0.2⫾0.5
Glucose, fasting
(mg/dL)
(n⫽4157)c
91.1⫾12.5
87.3⫾13.0
Insulin (␮U/mL)
(n⫽1940)c
11.9⫾9.0
12.8⫾9.2
Unweighted Ms; race limited to white, black, Mexican-American with
measurements for triceps and subscapular skinfolds, not pregnant,
no pacemaker, no indication of infection, not using blood pressure
or diabetes (insulin) medications; body weight ⱖ10 kg
b
% fat calculated per Williams et al.18
c
Number of blood tests performed was substantially lower than
number with race and anthropometry
HDL, high-density lipoprotein; LDL, low-density lipoprotein
the level of percent fat at which the lower bounds of the
95% CI around the ORs were greater than 1.00.
In part, the variability in the ORs and in the width of
the 95% CIs about them reflects the variability in the
independent relationships between increasing percent fat
levels and the likelihood of membership in each of the
most-adverse risk factor groups. The variability in the
ORs and in the width of the 95% CIs about them also
reflects the variability in the post hoc sampling and sample sizes within the age, gender, and body fat–specifıc
groups, which likewise affect the magnitude and the precision of the OR estimates. The interaction term, percent
fat by race, included in the logistic regression analysis
with each risk factor, was always nonsignifıcant in both
boys and girls. In boys, there was a greater and more
reliable likelihood of being in the most adverse risk factor
group above 20% fat. In girls, there was a greater and
more reliable likelihood of being in the most-adverse risk
factor groups to some extent above 25% fat and, in particular, above 30% fat.
The prevalence of each of the adverse CVD risk factors
by age and body fat groups is illustrated in Figure 1 for
boys and in Figure 2 for girls. The influence of age and
percent fat on the prevalence of high blood pressures, low
HDL-C, and high triglyceride levels is shown because
these variables (in some combination) are usually included in youth-specifıc defınitions of the metabolic syndrome.28 The influence of age and percent fat on the
prevalence of elevated CRP, glucose, and insulin levels is
also shown, as systemic inflammation and insulin resistance may provide common mechanistic links among the
CVD risk factors that constitute the metabolic syndrome
in youth.
The fıgures clearly depict the rising prevalence of
adverse CVD risk factors at higher levels of percent fat.
The sharply rising prevalence of elevated CRP and
insulin levels are particularly evident above 20% fat in
boys (Figure 1) and above 30% fat in girls (Figure 2).
The influence of age and body fat on adverse CVD risk
factor prevalence are complex and undoubtedly affected by a variety of biological and behavioral factors,
including an effect of age on the prevalence of boys and
girls in the most adverse group of any given risk factor
at the various levels of percent fat. In multiple logistic
regression analyses, the multiplicative interaction
term for age ⫻ percent fat was a signifıcant predictor
(data not shown) of elevated systolic and diastolic
blood pressures, elevated total cholesterol levels, elevated triglycerides, elevated LDL-C/HDL-C ratios, elevated CRP levels, and of low HDL-C levels in boys.
The interaction term for age ⫻ percent fat was likewise
a signifıcant predictor (data not shown) of elevated
www.ajpmonline.org
October 2011
Table 2. ORsa for being in the most-adverse fifth of the risk-factor distribution by percent fat within gender- and age-specific groups
Boys (n⫽6222)
Risk factor by age
in years
10.0–14.9
(n⫽2377)
15.0–19.9
(n⫽1189)
20.0–24.9
(n⫽691)
Girls (n⫽6057)
25.0–29.9
(n⫽547)
ⱖ30.0
(n⫽651)
20.0–24.9
(n⫽1556)
25.0–29.9
(n⫽1194)
30.0–34.9
(n⫽838)
ⱖ35.0
(n⫽363)
Systolic BP (mm Hg)
1.64 (0.99, 2.71)
2.66 (1.53, 4.63)
3.03 (1.69, 5.46)
3.88 (2.14, 7.03)
7.77 (4.46, 13.6)
1.39 (1.03, 1.89)
1.92 (1.35, 2.72)
2.88 (1.98, 4.20)
3.14 (1.98, 4.98)
12–18
1.69 (1.19, 2.40)
2.72 (1.85, 4.01)
3.55 (2.35, 5.35)
6.28 (4.17, 9.46)
6.91 (4.58, 10.4)
1.20 (0.90, 1.60)
1.97 (1.46, 2.65)
2.98 (2.15, 4.13)
3.97 (2.59, 6.09)
6–11
1.39 (0.85, 2.28)
1.84 (1.07, 3.17)
2.07 (1.16, 3.67)
1.74 (0.94, 3.22)
3.15 (1.80, 5.52)
1.01 (0.74, 1.38)
1.05 (0.73, 1.52)
1.43 (0.96, 2.14)
1.76 (1.07, 2.88)
12–18
0.99 (0.75, 1.31)
1.05 (0.76, 1.45)
0.85 (0.58, 1.23)
1.26 (0.87, 1.83)
0.78 (0.52, 1.17)
0.94 (0.73, 1.22)
0.92 (0.69, 1.21)
1.28 (0.94, 1.75)
1.42 (0.92, 2.19)
6–11
1.62 (0.97, 2.71)
1.86 (1.05, 3.29)
2.39 (1.27, 4.48)
3.04 (1.63, 5.68)
4.31 (2.38, 7.79)
0.85 (0.60, 1.20)
1.63 (1.11, 2.39)
1.90 (1.23, 2.93)
2.30 (1.36, 3.89)
12–18
1.43 (0.94, 2.18)
1.80 (1.12, 2.91)
2.49 (1.49, 4.15)
4.66 (2.82, 7.71)
7.07 (4.30, 11.6)
1.22 (0.85, 1.75)
1.81 (1.25, 2.62)
2.17 (1.45, 3.26)
3.70 (2.19, 6.24)
6–11
1.05 (0.69, 1.61)
1.07 (0.66, 1.73)
1.68 (1.01, 2.81)
1.81 (1.06, 3.08)
3.99 (2.44, 6.53)
1.04 (0.77, 1.40)
1.58 (1.13, 2.20)
2.60 (1.81, 3.74)
2.27 (1.43, 3.58)
12–18
1.09 (0.78, 1.52)
1.18 (0.80, 1.74)
2.55 (1.73, 3.76)
3.76 (2.54, 5.57)
5.02 (3.39, 7.42)
0.89 (0.66, 1.19)
1.35 (1.01, 1.81)
1.90 (1.38, 2.63)
4.35 (2.89, 6.56)
Diastolic BP (mm Hg)
Triglycerides (mg/dL)
HDL cholesterol (mg/dL)
C-reactive protein (mg/L)
6–11
1.15 (0.56, 2.36)
1.18 (0.54, 2.56)
1.90 (0.86, 4.17)
3.27 (1.50, 7.11)
9.98 (4.83, 20.6)
1.12 (0.75, 1.66)
2.17 (1.44, 3.27)
3.35 (2.16, 5.19)
8.17 (4.98, 13.4)
12–18
1.28 (0.83, 1.98)
1.57 (0.96, 2.56)
2.17 (1.29, 3.62)
3.89 (2.37, 6.39)
8.76 (5.42, 14.2)
1.25 (0.86, 1.81)
2.04 (1.42, 2.93)
3.53 (2.40, 5.20)
6.67 (4.17, 10.7)
6–11
—
—
—
—
—
—
—
—
—
12–18
1.23 (0.84, 1.80)
1.12 (0.72, 1.74)
1.35 (0.84, 2.19)
1.67 (1.03, 2.73)
1.99 (1.22, 3.25)
1.05 (0.75, 1.46)
1.31 (0.92, 1.86)
1.50 (1.01, 2.23)
1.97 (1.15, 3.38)
6–11
—
—
—
—
—
—
—
—
—
12–18
1.33 (0.61, 2.91)
2.70 (1.17, 6.20)
3.93 (1.64, 9.42)
10.9 (4.73, 25.3)
26.1 (10.8, 62.6)
2.35 (1.19, 4.63)
4.63 (2.27, 9.42)
8.29 (3.86, 17.8)
34.1 (11.8, 98.5)
Glucose, fasting (mg/dL)
Going et al / Am J Prev Med 2011;41(4S2):S77–S86
6–11
Insulin (mU/mL)
Note: Comparison groups; boys, percent fat ⱕ10% (n⫽767); girls, percent fat ⱕ20% (n⫽2106). ORs with parenthetical 95% CIs that do not include 1.00 are shown in bold.
a
Adjusted for NHANES III (1998 –1994) versus IV (1999 –2004); race (white, black, Mexican-American); ratio of triceps to subscapular skinfold thickness; and interaction terms for percent fat with race and age in multiple
logistic regression
BP, blood pressure; HDL, high-density lipoprotein
S81
S82
Going et al / Am J Prev Med 2011;41(4S2):S77–S86
systolic blood pressure, elevated triglycerides, and elevated LDL-C/HDL-C ratios in girls.
Discussion
The results of the present analyses show signifıcant relationships between percent body fat estimated from skinfold thicknesses and various chronic disease risk factors
in children and adolescents across a wide age span. Although this is not the fırst study to examine these relationships,16,19 the majority of past studies have used BMI
as a surrogate for a more direct measure of body composition,6,7,29 which has previously identifıed limitations.10
Given that it is excess adiposity that creates risk rather
than excess weight for height per se, studies with more
direct measures of fatness are necessary to defıne risk,
especially in children and adolescents who are experiencing changes in both adipose and non-adipose tissues at
different rates depending on the stage of maturation.30 It
is acknowledged that the current approach—estimation of
body density from skinfolds, followed by conversion to
percent fat—is also indirect. Nevertheless, skinfolds provide a more accurate estimate of percent fat than does
BMI, and the conversion equations used are designed to
adjust for variation in the composition of fat-free mass
that confounds other equations used to estimate percent
fat from body density in children and youth.19
The current fındings, in a large sample representative
of U.S. children and adolescents, confırm fındings in
smaller samples, and in samples from other countries.16 –18 The relationships are particularly striking for
insulin and CRP, followed by HDL-C, triglycerides, and
blood pressures. Clearly, there is considerable heterogeneity across risk factors; nevertheless, across the age span
of the children and adolescents in the present study, there
is a clear relationship between higher estimated body fat
percentages and an increased risk for more adverse levels
of CVD risk factors in boys and girls.
Past attempts to defıne body composition standards
have been largely based on population distributions of
BMI.7,12,20,31,32 Although age- and gender-specifıc BMI
percentiles are correlated with CVD risk factors, thresholds should be derived using direct measures of adiposity
related to increased health risk rather than rely on a
population-based distribution of an indirect index of
adiposity. In one of the few studies using a criterionreferenced approach, Dwyer and Blizzard16 proposed
thresholds of 20% fat for boys and 30% fat for girls based on
at-risk groups for dyslipidemia and hypertension. In a large
U.S. sample, Williams et al.19 identifıed remarkably similar
thresholds of 25% fat in boys and 30% fat in girls that were
indicative of an increased risk of being in the highest quintile
for blood pressure and serum lipoprotein ratios in children
and adolescents. In a relatively large sample of Asian boys
and girls aged 9 –10 years, Washino et al.18 reported that
fatness above 23% of body weight was related to a greater
risk of an adverse atherogenic index. Although there is relatively good agreement across the age groups represented in
these studies, the effect of age on the percent fat–risk factor
relationships was not investigated directly. This limitation
notwithstanding, the results of Williams et al.19 were used
to develop thresholds that have been applied across a
wide age range in the FITNESSGRAM® program
(www.cooperinstitute.org).
A criticism of this approach has been that “static”
thresholds do not take into consideration the normal
changes in adiposity that occur with growth and maturation. As shown by Laurson et al.,33 there are predictable
age- and gender-specifıc changes in percent body fat that
may influence the percent fat–risk factor relationships.
Indeed, the results of the present analyses showed that age
modifıed the relationship for several of the risk factors in
both boys and girls, suggesting that age-specifıc adiposity
standards may be useful.
In contrast to age, race/ethnicity was not independently associated either by itself or in its interaction with
body fat with adverse CVD risk factor levels. Similarly, in
their work, Williams et al.19 found no effect of race on the
percent fat–risk factor relationship in black or white boys
and girls from the Bogalusa Heart Study. Their results
suggest percent fat standards are robust, and equivalent
thresholds can be used in different race/ethnic groups,
although more research on this question is warranted.
Whether this is true of the contemporary age- and genderbut not race-specifıc BMI standards1 for overweight and
obesity is uncertain. Many studies have shown the
BMI–percent fat relationship varies among race/ethnic
groups,13–15 which could potentially confound application of a single BMI cutpoint across these groups.
Although there is a clear relationship with percent fat,
there is considerable variability across risk factors (Figures 1 and 2). The challenge of creating effıcacious and
feasible thresholds is obvious. Although it is of interest to
defıne thresholds relative to individual risk factors to
better understand the relationships, having multiple
thresholds is clearly not feasible for screening programs.
In adults, the concept of metabolic syndrome as an integrated index that captures overall risk has widespread
support.34 More recently, defınitions have been proposed
for youth.29,35,36 The use of the metabolic syndrome as a
more integrated way to potentially capture CVD and type
2 diabetes risk may provide a feasible method for developing health-related obesity thresholds. Studies in adults
have clearly demonstrated a strong relationship between
obesity and the metabolic syndrome. Far fewer studies
www.ajpmonline.org
Going et al / Am J Prev Med 2011;41(4S2):S77–S86
Diast olic blood pressure
60
35
50
30
% in 5th quintile
% in 5th quintile
Syst olic blood pressure
40
30
20
10
25
20
15
10
5
0
0
<10
10.0–
14.9
15.0–
19.9
20.0–
24.9
25.0–
29.9
≥30.0
<10
10.0–
14.9
% fat by skinfold
50
20.0–
24.9
25.0–
29.9
≥30.0
Triglycerides
60
50
% in 5th quintile
40
% in 5th quintile
15.0–
19.9
% fat by skinfold
HDL cholest erol
30
20
10
40
30
20
10
0
0
<10
10.0–
14.9
15.0–
19.9
20.0–
24.9
25.0–
29.9
<10
≥30.0
10.0–
14.9
% fat by skinfold
15.0–
19.9
20.0–
24.9
25.0–
29.9
≥30.0
% fat by skinfold
C-react i ve protein
45
Glucose
35
40
30
% in 5th quintile
35
% in 5th quintile
S83
30
25
20
15
10
25
20
15
10
5
5
0
0
<10
10.0–
14.9
15.0–
19.9
20.0–
24.9
25.0–
29.9
≥30.0
% fat by skinfold
<10
10.0–
14.9
15.0–
19.9
20.0–
24.9
25.0–
29.9
≥30.0
% fat by skinfold
Insulin
80
% in 5th quintile
70
60
Age (years)
50
6–8
9–12
40
30
13–15
20
16–18
10
0
<10
10.0–
14.9
15.0–
19.9
20.0–
24.9
25.0–
29.9
≥30.0
% fat by skinfold
Figure 1. Relationships between cardiovascular disease risk factors and percent fat in boys at various ages
HDL, high-density lipoprotein
October 2011
Going et al / Am J Prev Med 2011;41(4S2):S77–S86
S84
Diastolic blood pressure
60
50
50
% in 5th quintile
% in 5th quintile
Systolic blood pressure
60
40
30
20
10
40
30
20
10
0
0
<20.0
20.0–24.9 25.0–29.9 30.0–34.9
≥35
<20.0
% fat by skinfold
20.0–
24.9
25.0–
29.9
30.0–
34.9
≥35
% fat by skinfold
Triglycerides
60
60
50
50
% in 5th quintile
% in 5th quintile
HDL cholest erol
40
30
20
10
40
30
20
10
0
0
20.0–24 .9 25.0–29.9 30.0–34.9
<20.0
≥35
<20.0
20.0–24.9 25.0–29.9
% fat by skinfold
C-react ive protein
60
≥35
Glucose
60
50
% in 5th quintile
50
% in 5th quintile
30.0–34.9
% fat by skinfold
40
30
20
10
40
30
20
10
0
0
<20.0
20.0– 24.9 25.0–29.9
30.0–34.9
≥35
% fat by skinfold
<20.0
20.0–
24.9
25.0–
29.9
30.0–
34.9
≥35
% fat by skinfold
Insulin
90
80
% in 5th quintile
70
60
Age (years)
50
6–8
9–12
40
30
13–15
20
16–18
10
0
<20.0
20.0– 24.9 25.0–29.9
30.0–34.9
≥35
% fat by skinfold
Figure 2. Relationships between cardiovascular disease risk factors and percent fat in girls at various ages
HDL, high-density lipoprotein
www.ajpmonline.org
Going et al / Am J Prev Med 2011;41(4S2):S77–S86
have been done in youth, although available reports support a similar relationship in adolescents.35,37
The present study has several strengths, including the
use of a large, representative sample of U.S. children and
adolescents across a broad age range. The inclusion of
multiple race and ethnic groups is also a strength, as is the
use of measures of skinfold thicknesses to obtain percent
fat estimates using equations that were developed to adjust for the age-related variation in the chemical composition of lean tissue that confounds many other skinfold
equations.38 Although sampling weights were not used to
adjust for the NHANES complex sampling procedure,
and this may be viewed as a limitation, the purpose for the
current paper was not to generalize to the U.S. population. Rather, the goal was to demonstrate whether age
modifıed the relationship between percent fat and risk
factors. Sampling variation and smaller sample sizes for
some risk factors undoubtedly contributed to some of the
variation in the relationships. Nevertheless, it is clear that
age is an important variable to consider in the development of percent fat thresholds. Future studies with large,
representative samples are needed to precisely defıne the
percent fat–risk factor relationships at different ages.
Conclusion
The current results demonstrate a strong relationship
between chronic disease risk factors and percent fat in
children and youth. The relationship varies with age for
most risk factors, and the results suggest that criterionreferenced body composition standards should vary by
age and gender in children and youth aged 6 –18 years.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
Publication of this article was supported by The Cooper Institute through a philanthropic gift from Lyda Hill.
No fınancial disclosures were reported by the authors of this
paper.
19.
20.
References
1. Flegal KM, Ogden CL, Yanovski JA, et al. High adiposity and high body
mass index-for-age in U.S. children and adolescents overall and by
race-ethnic group. Am J Clin Nutr 2010;91:1020 – 6.
2. Daniels SR. The consequences of childhood overweight and obesity.
Future Child 2006;16:47– 67.
3. Franks PW, Hanson RL, Knowler WC, Sievers ML, Bennett PH,
Looker HC. Childhood obesity, other cardiovascular risk factors, and
premature death. N Engl J Med 2010;362:485–93.
4. Kavey RE, Allada V, Daniels SR, et al. Cardiovascular risk reduction in
high-risk pediatric patients: a scientifıc statement from the American
Heart Association Expert Panel on Population and Prevention Science;
the Councils on Cardiovascular Disease in the Young, Epidemiology
and Prevention, Nutrition, Physical Activity and Metabolism, High
Blood Pressure Research, Cardiovascular Nursing, and the Kidney in
Heart Disease; and the Interdisciplinary Working Group on Quality of
October 2011
21.
22.
23.
24.
25.
26.
S85
Care and Outcomes Research: endorsed by the American Academy of
Pediatrics. Circulation 2006;114:2710 –38.
Lawlor DA, Martin RM, Gunnell D, et al. Association of body mass
index measured in childhood, adolescence, and young adulthood with
risk of ischemic heart disease and stroke: fındings from 3 historical
cohort studies. Am J Clin Nutr 2006;83:767–73.
Freedman DS, Khan LK, Dietz WH, Srinivasan SR, Berenson GS.
Relationship of childhood obesity to coronary heart disease risk factors
in adulthood: the Bogalusa Heart Study. Pediatrics 2001;108:712– 8.
Freedman DS, Mei Z, Srinivasan SR, Berenson GS, Dietz WH. Cardiovascular risk factors and excess adiposity among overweight children
and adolescents: the Bogalusa Heart Study. J Pediatr 2007;150:12–7.e2.
Himes JH, Dietz WH. Guidelines for overweight in adolescent preventive services: recommendations from an expert committee. Am J Clin
Nutr 1994;59:307–16.
Kuczmarski RJ, Ogden CL, Grummer-Strawn LM, Flegal KM, Guo SS,
Wei R, et al. 2000 CDC growth charts: United States. Advance data
from Vital Health Stat 314. Hyattsville MD: National Center for Health
Statistics, 2000. www.cdc.gov/nchs/data/ad/ad314.pdf.
Garn SM, Leonard WR, Hawthorne VM. Three limitations of the body
mass index. Am J Clin Nutr 1986;44:996 –7.
Prentice AM, Jebb SA. Beyond body mass index. Obes Rev 2001;2:
141–7.
Freedman DS, Wang J, Thornton JC, et al. Classifıcation of body fatness
by body mass index-for-age categories among children. Arch Pediatr
Adolesc Med 2009;163:805–11.
Deurenberg P, Deurenberg Yap M, Wang J, Lin FP, Schmidt G. The
impact of body build on the relationship between body mass index and
percent body fat. Int J Obes Relat Metab Disord 1999;23:537– 42.
Deurenberg P, Deurenberg-Yap M, Foo LF, Schmidt G, Wang J. Differences in body composition between Singapore Chinese, Beijing Chinese and Dutch children. Eur J Clin Nutr 2003;57:405–9.
Deurenberg P, Deurenberg-Yap M, Guricci S. Asians are different from
Caucasians and from each other in their body mass index/body fat per
cent relationship. Obes Rev 2002;3:141– 6.
Dwyer T, Blizzard CL. Defıning obesity in children by biological endpoint rather than population distribution. Int J Obes 1996;20:472– 80.
Higgins PB, Gower BA, Hunter GR, Goran MI. Defıning health-related
obesity in prepubertal children. Obes Res 2001;9:233– 40.
Washino K, Takada H, Nagashima M, Iwata H. Signifıcance of the
atherosclerogenic index and body fat in children as markers for future,
potential coronary heart disease. Pediatr Int 1999;41:260 –5.
Williams DP, Going SB, Lohman TG, et al. Body fatness and risk for
elevated blood pressure, total cholesterol, and serum lipoprotein ratios
in children and adolescents. Am J Public Health 1992;82:358 – 63.
Cole TJ, Bellizzi MC, Flegal KM, Dietz WH. Establishing a standard
defınition for child overweight and obesity worldwide: international
survey. BMJ 2000;320:1240 –3.
CDC, National Health and Nutrition Examination Survey. www.cdc.
gov/nchs/nhanes.htm.
CDC. National Health and Nutrition Examination Survey IV: Anthropometry Procedures Manual. Revised December 2000. www.cdc.gov/
nchs/data/nhanes/bm.pdf.
CDC. National Health and Nutrition Examination Survey III: Body
Measurements (Anthropometry). October 1998. www.cdc.gov/nchs/
data/nhanes/nhanes3/cdrom/nchs/manuals/anthro.pdf.
CDC. National Health and Nutrition Examination Survey III, Cycle 2:
Physician Examiner’s Training Manual. June 1991. www.cdc.gov/
nchs/data/nhanes/nhanes3/cdrom/nchs/manuals/phys.pdf.
CDC. National Health and Nutrition Examination Survey: Physician
Examination Procedures Manual. Revised August 2000. www.cdc.gov/
nchs/data/nhanes/pe.pdf.
Gunther EW, Lewis BG, Kincikowski SM. Laboratory Procedures Used
for the Third National Health and Nutrition Examination Survey
(NHANES III), 1988 –1994. Atlanta GA: DHHS, CDC, 1996. www.
S86
27.
28.
29.
30.
31.
32.
Going et al / Am J Prev Med 2011;41(4S2):S77–S86
cdc.gov/nchs/data/nhanes/nhanes3/cdrom/nchs/manuals/labman.
pdf.
CDC. National Health and Nutrition Examination Survey: Laboratory
Procedures. April 2001. www.cdc.gov/nchs/data/nhanes/lab1-6.pdf
and www.cdc.gov/nchs/data/nhanes/lab7-11.pdf.
Jolliffe CJ, Janssen I. Development of age-specifıc adolescent metabolic
syndrome criteria that are linked to the Adult Treatment Panel III and
International Diabetes Federation criteria. J Am Coll Cardiol 2007;
49:891– 8.
Lambert M, Delvin EE, Levy E, et al. Prevalence of cardiometabolic risk
factors by weight status in a population-based sample of Quebec children and adolescents. Can J Cardiol 2008;24:575– 83.
Maynard LM, Wisemandle W, Roche AF, Chumlea WC, Guo SS,
Siervogel RM. Childhood body composition in relation to body mass
index. Pediatrics 2001;107:344 –50.
McCarthy HD, Cole TJ, Fry T, Jebb SA, Prentice AM. Body fat reference curves for children. Int J Obes (Lond) 2006;30:598 – 602.
Taylor RW, Jones IE, Williams SM, Goulding A. Body fat percentages
measured by dual-energy X-ray absorptiometry corresponding to recently recommended body mass index cutoffs for overweight and
obesity in children and adolescents aged 3-18 y. Am J Clin Nutr
2002;76:1416 –21.
33. Laurson KR, Eisenmann JC, Welk GJ. Body fat percentile curves for
U.S. children and adolescents. Am J Prev Med 2011;41(4S2):
S87–92.
34. Zhu S, Wang Z, Shen W, Heymsfıeld SB, Heshka S. Percentage body fat
ranges associated with metabolic syndrome risk: results based on the
third National Health and Nutrition Examination Survey (1988 –
1994). Am J Clin Nutr 2003;78:228 –35.
35. Cook S, Weitzman M, Auinger P, Nguyen M, Dietz WH. Prevalence of
a metabolic syndrome phenotype in adolescents: fındings from the
third National Health and Nutrition Examination Survey, 1988 –1994.
Arch Pediatr Adolesc Med 2003;157:821–7.
36. Ford ES, Li C. Defıning the metabolic syndrome in children and adolescents: will the real defınition please stand up? J Pediatr 2008;152:
160 – 4.
37. Ruiz JR, Rizzo NS, Ortega FB, Loit HM, Veidebaum T, Sjostrom M.
Markers of insulin resistance are associated with fatness and fıtness in
school-aged children: the European Youth Heart Study. Diabetologia
2007;50:1401– 8.
38. Lohman TG. Applicability of body composition techniques and constants for children and youths. In: Pandolf K, editor. Exercise and
sports science reviews. Baltimore MD: Williams & Wilkins, 1986
Have you seen the AJPM website lately?
Visit www.ajpmonline.org today!
www.ajpmonline.org