Validity of Six Field and Laboratory Methods
for Measurement of Body Composition in Boys
Lisa Parker,* John J. Reilly,† Christine Slater,† Jonathan C.K. Wells,‡ and Yannis Pitsiladis*
Abstract
PARKER, LISA, JOHN J. REILLY, CHRISTINE
SLATER, JONATHAN C.K. WELLS, AND YANNIS
PITSILADIS. Validity of six field and laboratory methods
for measurement of body composition in boys. Obes Res.
2003;11:852– 858.
Objective: To determine the validity of the following six
body composition methods against a reference method
(three-component model): air displacement plethysmography (BODPOD); estimation from body density using BODPOD; skinfold thickness using the Slaughter equations;
bioelectrical impedance, both leg-leg (TANITA) and hand–
foot (Bodystat) approaches; and total body water.
Research Methods and Procedures: Forty-two healthy
white 10- to 14-year-old boys (mean age, 12.9 ⫾ 1.0 years)
were enrolled in this study. Measures of body fat percentage
and body fat mass derived from the three-component model
were used as the reference method. Validity of all of the
other methods was assessed by comparison against the
reference by calculation of biases and limits of agreement.
Results: Mean body fatness measured using the reference
method was 16.4 ⫾ 11.6% and 8.7 ⫾ 7.0 kg. Estimates of
fatness from total body water had the narrowest limits of
agreement relative to the reference (⫹0.9 ⫾ 5.0% body fat;
⫹0.5 ⫾ 2.9 kg fat mass). For all other methods tested, we
observed large biases and very wide limits of agreement.
Discussion: This study suggests that the validity of newer
field and laboratory methods for estimation of body composition is poor in adolescent boys. For applications where
high accuracy of estimation at the individual level is essential, only reference methods would be acceptable.
Received for review October 30, 2002.
Accepted in final form May 1, 2003.
*Centre for Exercise Science and Medicine, West Medical Building, University of Glasgow,
Glasgow, Scotland; †University of Glasgow, Division of Developmental Medicine, Royal
Hospital for Sick Children, Glasgow, Scotland; and ‡Childhood Nutrition Research Centre,
Institute of Child Health, University of London, London, England.
Address correspondence to Dr. John J. Reilly, University of Glasgow, Division of Developmental Medicine, Royal Hospital for Sick Children, Dalnair Street, Glasgow G3 8SJ,
Scotland.
E-mail: [email protected]
Copyright © 2003 NAASO
852
OBESITY RESEARCH Vol. 11 No. 7 July 2003
Key words: total body water, body density, bioelectric
impedance, air displacement plethysmography, anthropometry
Introduction
Pediatric body composition methodology is a subject of
increasing interest as a result of increased awareness of its
importance (1–3) and the advent of novel field and laboratory methods that are particularly suitable for pediatric use.
The main research focus is the development of more accurate methods for both field/epidemiological and clinical use
(1,2,4). Newer methods of particular note are bioelectrical
impedance using hand–foot (5,6) and leg–leg systems (1,7)
and air displacement plethysmography using the BODPOD
(3,8). However, evidence on the validity of these newer
methods is scarce for two principal reasons. First, establishing validity requires comparison against reference or criterion methods (1,4), and these are not widely available. Most
methodological studies in this area to date have simply
compared nonreference methods against each other, which
does not establish validity (1,9,10). Second, an important
recent review (3) concluded that most previous studies have
been too small (typically ⬍15 subjects per age/sex group) to
test validity adequately for any given population. In addition, most of these compared only one method against a
reference method, and there is a paucity of evidence from
studies where the validity of several methods has been
tested simultaneously against a reference method (3). In
summary, more validation studies are required.
We have previously demonstrated that the three-component model is a reference method in children (4). The aim of
this study was to assess simultaneously the validity of six
popular field and laboratory methods for estimation of body
fatness using the three-component model as the reference.
Research Methods and Procedures
Subjects
We recruited 56 apparently healthy 10- to 14-year-old
boys from the Glasgow area. Subjects were excluded if they
Validity of Body Composition Measurement Methods, Parker et al.
were suffering from acute illness or took medication that
would have produced abnormalities in body composition,
but we included subjects with mild and stable chronic
disease (e.g., mild asthma). Subjects were asked to abstain
completely from consuming food and drink in the 3 hours
before visiting the laboratory, and this was achieved by
scheduling laboratory sessions in the mid-morning (after an
early breakfast) or just after school (at least 3 hours after
lunch). A total of 42 subjects, all Scottish and white, attended for measurement and complied with all of the necessary procedures. The others dropped out or failed to
comply adequately with the study protocol (in relation to
fasting, timing, or provision of urine samples). All measures
were carried out by the same investigator and were performed in the same order. All procedures were completed
during a single visit to the laboratory. Subjects and parents
gave informed consent before participation, and the study
was approved by the University of Glasgow Research Ethics Committee.
Experimental Design, Analysis, and Power
Whereas no gold standard exists for pediatric body composition measurement in the absence of chemical analysis
(1,4,9), the three-component model based on measurement
of total body water (TBW)1 and body density is acceptable
as a reference method in children (1,4,9). Therefore, we
used this as the reference method in the present study and
assessed the validity of all of the simpler “field” and “laboratory” methods against it. Assessments were made by the
“Bland-Altman” method, using calculations of biases and
limits of agreement relative to the reference (1,11). We
plotted individual differences in fat mass (kilograms) between each method and the reference. Heteroscedasticity of
the between-method differences was systematically examined by plotting the absolute differences (ignoring the sign
or direction of difference) against the means and calculating
correlation coefficients (11). If differences were normally
distributed, we calculated the mean error (bias) and limits of
agreement (1.96 times the SD of the between-method differences). Where differences were not normally distributed
(only skinfold thickness in this study), this approach was
inappropriate (11), and data were transformed by taking the
natural logs of both methods before calculating the limits of
agreement (11). In this case, bias and limits of agreement
were calculated on the log scale and expressed as a ratio
after conversion to antilogs (11). We also tested the significance of differences between each method and the reference using one-sample Wilcoxon tests. Errors for each
method are presented largely in the form of absolute fat
1
Nonstandard abbreviations: TBW, total body water; ABV, actual body volume; RBV, raw
body volume; LV, lung volume; TGV, thoracic gas volume; SAA, surface area artifact; SA,
surface area; FRC, functional residual capacity; TV, tidal volume; FM, fat mass; BW, body
weight; FFM, fat-free mass.
mass (kilograms) but are also expressed as a percentage of
body mass to illustrate the magnitude of the error in relative
terms. In this study, the conventional statistical significance
of differences between methods was of limited value because this does not address the question of agreement between methods (1,11). Assessment of agreement between
methods required an alternative approach, calculation of
biases and limits of agreement (1,11). Studying fewer than
15 subjects per age/sex group has been the norm in previous
methodological studies that have calculated biases and limits of agreement (3). We aimed to recruit ⬎40 subjects to
the study and to recruit a sample that was the same sex
(male) and fairly homogeneous in terms of age.
Reference Method: Measurement of Fat Mass by ThreeComponent Model
Measurement of body composition by the three-component reference method requires measurement of body density and TBW. The reference method used was described
previously (4). Briefly, this involved calculation of actual
body volume (ABV) using the BODPOD (Life Measurement Instruments, Concord, CA), after adjustment for predicted lung volume (LV) and surface area artifact (SAA)
(8). We measured ABV in duplicate or in triplicate, when
the initial two measures differed by ⬎150 mL (8), and used
the mean value in subsequent calculations. We recorded the
“raw” body volume (RBV) that was measured on each
occasion (which appears transiently on the instrument display after a measurement) and applied corrections for LV
[the component of LV used is thoracic gas volume (TGV)]
and SAA, using the equations of Rosenthal et al. (12) and
Zapletal et al. (13), respectively, to obtain ABV as follows:
ABV ⫽ RBV ⫹0.4 TGV ⫺ SAA
Surface area (SA) ⫽ 71.84 weight0.425 ⫻ (height0.725)
SAA ⫽ (k ⫻ SA)
where k is – 0.467 ⫻ 10⫺5 (8).
Calculation of functional residual capacity (FRC), dependent on height and gender, and tidal volume (TV) was as
follows:
Males ⬍ 1.626 m: FRC
⫽ [(0.02394 ⫻ height(cm)] ⫹ (⫺1.716)
Males ⬎ 1.626 m: FRC
⫽ [0.05918 ⫻ height (cm)] ⫹ (⫺7.036)
logTV ⫽ [1.8643 ⫻ logheight (cm)] ⫺ 1.3956
TGV ⫽ FRC ⫹ 0.5TV
We measured TBW as deuterium dilution space/1.04, to
correct for nonaqueous exchange of the isotope in vivo, as
OBESITY RESEARCH Vol. 11 No. 7 July 2003
853
Validity of Body Composition Measurement Methods, Parker et al.
Table 1. Body fatness estimates by each of the seven methods
Body fat percentage
Body fat mass (kg)
Method
Mean
SD
Median
Range
Mean
SD
Median
Range
Reference
Hand–foot impedance
Leg–leg impedance
BODPOD software
Body density
TBW
Skinfolds
16.4
18.8
20.5
21.2
19.0
17.3
15.0
11.6
7.9
7.8
11.7
11.5
11.7
8.2
12.8
16.2
16.6
18.3
16.2
13.3
13.1
2.0–46.0
8.2–40.7
7.0–40.5
1.1–49.7
⫺1.3–46.6
3.4–45.9
6.0–40.6
8.7
10.1
10.9
10.9
10.0
9.2
8.1
7.0
5.6
5.8
7.1
7.1
7.2
5.8
6.4
8.3
8.9
9.9
8.7
7.5
6.5
0.9–30.9
3.8–24.3
3.4–27.3
0.5–33.5
⫺0.6–31.4
1.6–30.9
2.5–30.3
previously described (14). In summary, a “baseline” urine
sample was collected from each subject, and this was followed by a dose of deuterium oxide (99 atom % deuterium;
Sigma-Aldrich, Poole, UK; 0.05 g/kg body mass, diluted
1:10 with tap water) administered orally. After consumption
of the dose, the volume of the next three voids was measured, and the urine produced over a minimum period of 5
hours or for three voids was retained, whichever was the
longer. We calculated deuterium dilution space after correction for loss of tracer in the urine. The enrichment of
deuterium in body water was determined by continuous
flow isotope ratio mass spectrometry (Hydra; PDZ Europa,
Crewe, UK) after equilibration with a reference gas (5% hydrogen in helium) over a platinum catalyst. The mass spectrometer was calibrated using gravimetric standards of known
deuterium content that were prepared and analyzed with each
batch of samples, as previously described (15–17).
After calculation of TBW and ABV, we derived fat mass
(FM) as follows (4,8):
FM (kg) ⫽ [(2.22 ⫻ ABV) ⫺ (0.764 ⫻ TBW)]
⫺ (1.465 ⫻ BW)
where BW is body weight (kilograms). Body weight was
measured to 0.1 kg, using calibrated electronic scales (Alpha Model 770; SECA Ltd., Birmingham, UK), while the
child wore swimming trunks. Height was measured to 0.1
cm using a stadiometer (Leicester Height Measure; Child
Growth Foundation, London, UK).
Nonreference Methods Used to Estimate Body
Composition
Two-Component Estimates of FM from TBW. After measuring TBW as described above, we derived “two-component” fat-free mass (FFM) estimates from the TBW measures by applying the age- and sex-specific hydration
“constant” for FFM provided by Lohman et al. (9), i.e., that
hydration of FFM was 0.764. In all cases, FFM was subtracted from BW to calculate FM.
854
OBESITY RESEARCH Vol. 11 No. 7 July 2003
Estimates of FM from Skinfold Thickness. Skinfold thickness was measured at triceps and subscapular sites as previously described (4,18). We used the equations of Slaughter et al. (19) to estimate body fatness, on the grounds that
these provided greatest accuracy relative to reference (twocomponent) models in previous studies (18):
Subjects with sum of two skinfolds ⬍ 35 mm:
Percent body fat ⫽ 1.21 (sum of two skinfolds)
⫺ 0.008 (sum of two skinfolds)2 ⫺ 1.7
Subjects with sum of two skinfolds ⬎ 35 mm:
Percent body fat ⫽ 0.783 (sum of two skinfolds) ⫹ 1.6
Estimates of FM from Bioelectrical Impedance. Bioelectrical impedance was measured in all subjects, using both
hand–foot (1500 MDD; Bodystat, Isle of Man, UK) and
leg–leg (TBF 521; TANITA, Uxbridge, UK) systems at 50
kHz. For the impedance estimates, subjects were asked to
fast for 3 hours as described above, to avoid strenuous
exercise in the 24-hour period before the procedure, and to
be rested for 5 minutes in the appropriate position (standing
or lying down as appropriate) before measurements. We
used estimates of body fat percentage provided by the
manufacturer’s software in each case. For the TANITA
system, this prediction equation is unknown. Most impedance software is confidential because of its commercially
sensitive nature; therefore, the equation used cannot be
presented. The Bodystat is unusual in that the technical
manual states that it uses the equation of Houtkooper et al.
(20) as a default. This equation was used with the Bodystat
system because it has previously been shown to be the most
accurate of the available impedance predictions in Scottish
children, relative to a two-component model (6):
FFM (kg) ⫽ 0.61 (height2/impedance in ohms)
⫹ 0.25(BW) ⫹ 1.31
Validity of Body Composition Measurement Methods, Parker et al.
Figure 1: Biases (mean errors, kilograms fat mass, solid horizontal line) and limits of agreement (1.96 ⫻ SD of the errors, broken horizontal
line) for the six methods plotted against the mean of the reference plus the method [Bland/Altman method (11)].
In all cases, FFM estimates were subtracted from body
mass to obtain estimates of FM.
Two-Component Estimates of FM from Body Density
Measurement. The BODPOD (Life Measurement Instruments) was used to estimate body fatness after measurement
of body density (as described above) in two ways. First, the
manufacturer’s (adult) software was used to generate estimates directly, using calculations that are unpublished and
undisclosed. Second, the “raw” measurements of body density were combined with age- and sex- specific “constants”
OBESITY RESEARCH Vol. 11 No. 7 July 2003
855
Validity of Body Composition Measurement Methods, Parker et al.
Table 2. Biases and limits of agreement* (method minus reference) for 6 methods relative to the reference
Body fat percentage
Fat mass (kg)
Method
Bias (mean)
SD
Limits (ⴙ/ⴚ)
Bias (mean)
SD
Limits (ⴙ/ⴚ)
Hand-foot impedance
Leg-leg impedance
BODPOD software
Body density
Total body water
Skinfolds
⫹2.4
⫹4.1
⫹4.7
⫹2.6
⫹0.9
⫺1.4
6.0
7.2
4.9
4.8
2.6
6.0
11.8
14.1
9.6
9.4
5.0
11.8
⫹1.4
⫹2.3
⫹2.3
⫹1.2
⫹0.5
⫺0.6
3.9
4.0
2.5
2.5
1.5
3.1
7.6
7.8
4.9
4.9
2.9
6.0
* Limits of agreement calculated as 1.96 ⫻ SD of the differences.
for the density of FFM provided by Lohman et al. (9) to
estimate body composition using a “two-component” approach from body density. This involved density of FFM
values that were assumed to be 1.084 kg/m2 (10 year olds),
1.087 kg/m2 (11 to 12 year olds), and 1.094 (13 to 14 year
olds) (9). For the estimates provided from the “raw” body
density, we used the raw body volume and adjusted this for
LV and SAA as described above. Measurement procedures
for the BODPOD were as previously described (8), with
measurements in triplicate and with subjects in swimming
trunks, wearing a swimming cap, and with all jewelry
removed.
Results
Characteristics of Subjects
Mean age of the 42 subjects was 12.9 ⫾ 1.0 years
(median, 13.0 years; range, 10.1 to 14.5 years). Mean height
was 1.59 ⫾ 0.11 m (median, 1.59 m; range, 1.36 to 1.86 m),
and mean weight was 52.5 ⫾ 12.9 kg (median, 49.4 kg;
range, 35.6 to 74.6 kg). The BMI SD score was calculated
relative to UK 1990 population reference data (21) as a
simple index of under-/overweight in the sample: mean
BMI SD score was ⫹0.50 ⫾ 0.54 (median, 0.53; range,
⫺1.75 to ⫹2.38). Summary data for each method of estimating/measuring body composition are shown in Table 1.
Validity of the Various Methods
Correlations with the Reference. Estimates of both FM
and percentage body fat from all of the methods were
positively and significantly correlated with the reference, as
expected. Data are not shown here because the correlations
are of limited value in this context and can be misleading
(1,11).
Paired Comparisons with the Reference Method.
FM Differences. With the exception of TBW (p ⫽ 0.05)
and skinfolds (p ⫽ 0.34), differences were significant (one856
OBESITY RESEARCH Vol. 11 No. 7 July 2003
sample Wilcoxon) in all cases: hand–foot impedance (p ⫽
0.02); leg–leg impedance (p ⫽ 0.001); BODPOD, using
manufacturer’s software (p ⫽ 0.001); and body density,
using Lohman’s age- and sex-specific constants for density
of FFM (p ⫽ 0.002).
Body Fat Percentage Differences. With the exception of
skinfolds (p ⫽ 0.21) and TBW (p ⫽ 0.08), differences were
significant (one-sample Wilcoxon) in all cases: hand–foot
impedance (p ⫽ 0.04); foot–foot impedance (p ⫽ 0.001);
BODPOD, using manufacturer’s software (p ⫽ 0.001); and
body density, using Lohman’s constants (p ⫽ 0.001).
Biases and Limits of Agreement Relative to the Reference. Biases were generally large, and limits of agreement
between each method and the reference were generally
wide, although the two-component approach based on TBW
had the smallest bias and narrowest limits of agreement
(Figure 1). Summary statistics for the biases and limits of
agreement are shown in Table 2. For reasons described
above, this analysis is inappropriate for the assessment of
skinfold errors, so these are presented separately in Figure
2, where it is clear that agreement between skinfolds and the
reference is also poor at the level of the individual. In all
cases except skinfolds, errors were not significantly correlated with the size of the FM.
Discussion
Main Findings and Implications
This study was unique in that it provided a comprehensive assessment of the validity of six of the most widely
used body composition methods against a reference method.
The recent pediatric body composition literature is dominated by comparisons among nonreference methods (7,22–
24), which cannot address the issue of methodological validity.
This study suggests that validity, at the level of the
individual child, is poor for all of the methods tested, with
wide limits of agreement and large biases. Of the “two-
Validity of Body Composition Measurement Methods, Parker et al.
and limits of agreement of 0.5 ⫾ 2.9 kg. For impedance and
air displacement plethysmography, the results of this study
are not encouraging. The greatest bias and widest limits of
agreement (Table 2) were found using TANITA: mean error
was 2.3 kg for FM; limits of agreement were 2.3 ⫾ 7.8 kg.
Newer methods such as the BODPOD and TANITA systems are very practical in pediatrics, but our study suggests
that estimates of body composition obtained from them
should be interpreted cautiously, and this is consistent with
other studies. For example, in 14 boys and 11 girls, 9 to 14
years old, Fields and Goran (10) reported wide limits of
agreement for BODPOD relative to a four-component reference method. A comprehensive analysis of possible
sources of error is beyond the scope of this report. Detailed
treatment of this issue is available in recent reviews (1–
4,10).
Figure 2: Errors for fat mass (kilograms) using skinfold method
with direction of error (sign) retained (A) or removed (B). (C)
Ratio of fat mass by skinfolds to reference, all plotted against mean
fat mass by skinfolds and reference [Bland/Altman method (11)].
component” methods tested, TBW had the highest accuracy,
and this is consistent with our previous study (4). However,
TBW errors were also fairly large at the individual level.
Table 2 illustrates a 0.5-kg bias for FM (mean FM, 8.7 kg)
Limitations
The use of a three-component model as reference might
be seen as a disadvantage (relative to a four-component
model), but our earlier study found empirical evidence that
the three-component approach, as used here, was acceptable
(4). Our concern with using a four-component model was
that the measurement of body mineral is likely to be hardware- and software-dependent, and large errors have been
reported with some combinations of hardware and software
(1,4). Until this issue is resolved, the three-component
model seems to be the most appropriate reference. We could
not carry out pubertal staging of subjects. Inclusion of
information on pubertal status might have improved accuracy of prediction if variation in maturational state affects
the composition of the FFM (and hence the validity of the
assumptions made by most of the techniques) to any great
extent (19). We could not include all available body composition methods (DXA was not included). This is potentially important because DXA can provide fairly precise
body composition estimates, and accuracy is potentially
good (25), although it has recently been reported as poor
relative to a four-component model in female youth (26).
Finally, our relatively homogenous sample (same sex, narrow age range; n ⫽ 42) was a strength in the sense that it
permitted reasonable confidence in our conclusions at least
for this sex/age group. Extrapolation of our results to girls,
to subjects of different ages, or to extremes of body fatness
should be carried out cautiously. Our pessimistic findings
may best be regarded as providing testable hypotheses for
other populations. However, the (limited) literature comparing methods with reference methods (1,2,3,4,6,10,18,26)
has generally produced similarly negative conclusions as to
their validity.
Conclusions
This study found that the accuracy of most of the popular
field and laboratory methods for assessment of body comOBESITY RESEARCH Vol. 11 No. 7 July 2003
857
Validity of Body Composition Measurement Methods, Parker et al.
position was poor in 10- to 14-year-old boys. Accuracy is
likely to be even poorer in disease states (1,4), and we,
therefore, suggest that all existing (nonreference) methods
be used with caution. Further research aimed at improving
the accuracy of simple field or clinical methods of body
composition estimation is essential. Alternatively, the relatively poor accuracy of the existing methods may have to be
accepted. The present study confirms that, if high accuracy
of body composition estimates in individuals is essential,
multicomponent methods must be used.
Acknowledgments
This study was funded in part by the John Robertson
Bequest. We thank Sakkie Meeuwsen (Bodystat Ltd., Isle of
Man, UK) for excellent technical assistance. The cooperation of the subjects was greatly appreciated.
References
1. Reilly JJ. Assessment of body composition in infants and
children. Nutrition. 1998;14:821–5.
2. Ellis KJ. Human body composition: in vivo methods. Physiol
Rev. 2000;80:649 – 80.
3. Fields DA, Goran MI, McRory MA. Body composition
assessment via air displacement plethysmography in adults
and children: a review. Am J Clin Nutr. 2002;75:453– 67.
4. Wells JC, Fuller NJ, Dewit O, Fewtrell MS, Elia M, Cole
TJ. Four compartment model of body composition in children: density and hydration of fat-free mass and comparison
with simpler models. Am J Clin Nutr. 1999;69:904 –12.
5. Houtkooper LB, Lohman TG, Going SB, Hall MC. Validity of bioelectric impedance for body composition assessment
in children. J Appl Physiol. 1989;66:814 –21.
6. Reilly JJ, Wilson J, Carmichael JM, McColl JH, Durnin
JVGA. Ability of bioelectric impedance to predict fat free
mass in pre-pubertal children. Pediatr Res. 1996;39:176 –9.
7. Tyrell VJ, Richards G, Hofman P, Gillies GF, Robinson E,
Cutfield WS. Foot-foot bio-electrical impedance analysis: a
valuable tool for the measurement of body composition in
children. Int J Obes Relat Metab Disord. 2001;25:273– 8.
8. Dewit O, Fuller NJ, Fewtrell MS, Elia M, Wells JCK.
Whole body air displacement plethysmography compared
with hydrodensitometry for body composition analysis. Arch
Dis Child. 2000;82:159 – 64.
9. Lohman TG. Advances in Body Composition Assessment.
Current Issues in Exercise. Monograph No. 3, 1993. Champaign, IL: Human Kinetics; 1993.
10. Fields DA, Goran MI. Body composition techniques and the
four compartment model in children. J Appl Physiol. 2000;89:
613–20.
11. Bland JM, Altman DG. Statistical methods for assessing
agreement between two methods of clinical measurement.
Lancet. 1986;1:307–12.
858
OBESITY RESEARCH Vol. 11 No. 7 July 2003
12. Rosenthal M, Cramer D, Bain SH, Denison D, Bush A,
Warner JO. Lung function in white children aged 4 to 19
years II: single breath analysis and plethysmography. Thorax.
1993;48:803– 8.
13. Zapletal A, Paul T, Samanek M. Normal values of static
pulmonary volumes and ventilation in children and adolescents (translation). Czech Pediatrics 1976;31:532–9.
14. Du Bois D, Dubois EF. A formula to estimate the approximate surface area if height and weight be known. Arch Intern
Med. 1916;17:863–71.
15. Reilly JJ, Montgomery C, Jackson D, MacRitchie J, Armstrong J. Energy intake by multiple pass 24-hour recall and
total energy expenditure: a comparison in a representative
sample of 3– 4 year olds. Br J Nutr. 2001;86:601–5.
16. Prosser SJ, Scrimgeour CM. High precision determination
of 2H/1H in H2O by continuous flow isotope ratio mass
spectrometry. Anal Chem. 1995;67:1992–7.
17. Scrimgeour CM, Rollo MM, Mudambo S, Handley L,
Prosser SJ. A simplified method for deuterium/hydrogen
isotope ratio measurements on water samples of biological
origin. Biol Mass Spectrom. 1993;22:383–7.
18. Reilly JJ, Wilson J, Durnin JVGA. Determination of body
composition from skinfolds: a validation study. Arch Dis
Child 1995;73:305– 8.
19. Slaughter MH, Lohman TG, Boileau RA. Skinfold equations for estimation of body fatness in children and youth.
Hum Biol. 1988;60:709 –23.
20. Houtkooper LB, Going SB, Lohman TG, Roche AF, Van
Loan M. Bioelectrical impedance estimation of fat free body
mass in children and youth: a cross validation study. J Appl
Physiol. 1992;72:366 –73.
21. Cole TJ, Freeman JV, Preece MA. Body mass index reference curves for the UK, 1990. Arch Dis Child. 1995;73:25–9.
22. Nunez C, Kovera AJ, Pietrobelli A, et al. Body composition
in children and adults by air displacement plethysmography.
Eur J Clin Nutr. 1999;53:382–7.
23. Nicholson JC, McDuffie JR, Bonat SH, et al. Estimation of
body fatness by air displacement plethysmography in African
American and white children. Pediatr Res. 2001;50:467–73.
24. Sung RYT, Lau P, Yu CW, Lam PKW, Nelson EAS.
Measurement of body fat using leg-leg bioimpedance. Arch
Dis Child. 2001;85:263–7.
25. Pintauro SJ, Nagy TR, Duthie CM, Goran MI. Crosscalibration of fat and lean measurements by DEXA to pig
carcass analysis in the pediatric body weight range. Am J Clin
Nutr. 1996;63:293– 8.
26. Wong WW, Hergenroeder AC, Stuff JE, Butte NF,
O’Brian Smith E, Ellis KJ. Evaluating body fat in girls and
female adolescents: advantages and disadvantages of dual
energy x-ray absorptiometry. Am J Clin Nutr. 2002;76:384 –9.