612

REVIEW

Measuring body composition

J C K Wells, M S Fewtrell
...............................................................................................................................

Arch Dis Child 2006;91:612–617. doi: 10.1136/adc.2005.085522

Several aspects of body composition, in particular the TECHNIQUES FOR MEASURING BODY
COMPOSITION
amount and distribution of body fat and the amount and Simple measurements or indices
composition of lean mass, are now understood to be Traditionally, skinfold thickness measurements
important health outcomes in infants and children. Their have been used to rank individuals in terms of
relative ‘‘fatness’’ or to assess the size of specific
measurement is increasingly considered in clinical practice; subcutaneous fat depots.2 Measurements are
however, paediatricians are often unsure as to which quick and simple to obtain in most age groups
techniques are appropriate and suitable for application in including young infants. In general, intraobser-
specific contexts. This article summarises the pros and cons ver and interobserver error are low compared to
between-subject variability, but in obese children
of measurement technologies currently available for accuracy and precision are poorer.
paediatric application. Simple techniques are adequate for The best use of skinfold thickness data is as
many purposes, and simple regional data may often be of raw values, where they act as reliable indices of
regional fatness. They can be converted into
greater value than ‘‘whole body’’ values obtained by more standard deviation score (SDS) format for long-
sophisticated approaches. itudinal evaluations. However, publication of
........................................................................... contemporary children’s skinfold reference data
remains a current research priority for assess-
ment of relative fatness in patients. Although the

B
ody composition and growth are key com- contemporary epidemic of obesity presents chal-
ponents of health in both individuals and lenges for body composition reference data,
populations. The ongoing epidemic of obe- whether individual patients are becoming more
sity in children and adults has highlighted the or less fat over time can only be assessed through
importance of body fat for short term and long comparison with a reference population. Such
term health.1 However, other components of data would therefore represent a ‘‘reference’’
body composition also influence health out- (what exists), but not a ‘‘standard’’ (what should
comes, and its measurement is increasingly exist).
considered valuable in clinical practice. Body mass index (BMI, calculated as weight/
The gold standard for body composition analysis height2) is also widely used as an index of
is cadaver analysis, so no in vivo technique may be relative weight, often expressed as SDS to take
considered to meet the highest criteria of accuracy. into account age and sex.3 In adults, BMI is
As discussed below, only multicomponent models predictive of clinical outcomes such as type 2
are now considered sufficiently accurate to act as diabetes; however, its predictive value for chil-
reference or criterion methods for the molecular dren and adolescents is less clear. BMI is a global
approach to measuring body composition (distin- index of nutritional status—used, for example, to
guishing fat and fat-free masses), against which categorise both overweight/obesity,4 and eating
other methods should be evaluated. Several disorders in combination with psychological
techniques are available, varying in complexity criteria5—but its relation with body composition
and ease of use, and each making assumptions per se is controversial.
that may affect its suitability for different condi- Although correlated with per cent fat,6 7 BMI
tions. A single technique is unlikely to be optimal cannot distinguish fat and lean masses, and
in all circumstances. A further important issue is there is a twofold range of variation in fatness for
that of the difficulty of validating techniques in a given BMI value in individual children.8 BMI
humans. In vivo techniques do not measure body may be particularly misleading in hospital
composition directly, but rather predict it from patients, where children apparently ‘‘malnour-
measurements of body properties. Thus all tech- ished’’ in terms of BMI actually have an increase
niques suffer from two types of error: methodo- in relative body fat and a severe decrease in lean
See end of article for logical error when collecting raw data, and error in
authors’ affiliations tissue (fig 1).9 This may be important for their
....................... the assumptions by which raw data are converted nutritional management, as the low BMI may
to final values. The relative magnitude of these lead to inappropriate overfeeding.
Correspondence to: errors varies between techniques.
Dr Jonathan Wells, MRC Waist circumference (WC) provides a simple
Childhood Nutrition The aim of this review is to discuss the measure of central fatness, which may be more
Centre, Institute of Child theoretical basis, assumptions, and advantages predictive of adverse outcomes such as lipid
Health, 30 Guilford Street, and disadvantages of available techniques. For
London WC1N 1EH, UK; more information, readers are strongly encour-
[email protected] Abbreviations: BIA, bioelectric impedance analysis; BMI,
aged to read the more detailed e-version of this body mass index; DXA, dual energy x ray
Accepted 24 March 2006 article on the journal website (http://www. absorptiometry; FFM, fat-free mass; FM, fat mass; TBW,
....................... archdischild.com/supplemental). total body water; WC, waist circumference

www.archdischild.com
Measuring body composition 613

profile or insulin resistance than total fat. In adults, waist– applied subsequently to predict TBW, which is converted to
hip ratio is independently associated with morbidity after FFM as described below.
adjustment for relative weight, such that the use of relative BIA incorporates various assumptions. The simplest model,
weight and body shape simultaneously gives a better involving hand–foot or foot–foot measurements made at 50
estimate of risk of morbidity than either alone.10 11 Similar KHz, relies most heavily on these assumptions, and therefore
findings are now emerging in children.12 Reference data on provides the crudest values for body composition. The relation
WC were obtained from a nationally representative sample of between bioelectrical data and TBW is influenced by the age
United Kingdom children in 1988.13 range investigated and other characteristics of the population.
Studies investigating the relation of WC with measures of Published BIA equations are population specific and perform
abdominal fatness obtained from magnetic resonance ima- poorly in healthy individuals, with errors in individuals
ging (MRI) have shown correlations consistently in the range typically ¡8% fat.23 Paediatric equations have been derived
of 0.5 to 0.8, although the associations with total abdominal for disease states such as obesity,24 cystic fibrosis,25 and HIV26;
fat tend be higher than those with intra-abdominal fat.14–16 In however, accuracy in individuals remains poor and measure-
contrast, studies reporting the association the between waist– ments may be confounded by clinical status, for example the
hip ratio and abdominal fat are inconsistent, and some find presence of oedema (see below). Foot–foot measurements,
no significant relation.14 17 though easier to obtain, have slightly poorer accuracy than
whole body measurements.27
Predictive techniques Despite these limitations, more sophisticated approaches to
Both skinfold thickness and bioelectric impedance measure- BIA have the potential both to improve accuracy and to
ments can be used to predict body composition. A generic increase the specificity of outcomes, and such progress is
problem with this approach is that it involves not one but two desirable given the ease with which measurements can be
predictions. First, raw measurements are used to predict a made in most age groups. For example, segmental measure-
body component or property using regression equations; ments of limbs or torso avoid the confounding effect of
second, this value is converted to final body composition data variability in body build, while multifrequency measure-
using further theoretical assumptions. In this section, we ments provide additional information about water distribu-
address the first of these issues. The second is then addressed tion. These are active areas of research but knowledge is
in the following section. currently insufficient to justify routine clinical application.
Several equations have been derived for the prediction of Conventional BIA analysis is inherently unsuited to the
per cent fat or body density from skinfold thickness prediction of body fatness in individuals, given that it
measurements.18–21 Such equations may not be valid in measures properties of the FFM only. Conventional single
populations other than those from which they were derived. frequency BIA does have high precision, providing electrode
Accuracy in individuals is invariably poor (limits of agree- placement is consistent, and hence could be suitable for
ment, ¡9% fat), and also varies in relation to body assessing short term changes in TBW within individuals.
fatness,22 23 making this approach unsuitable for longitudinal Given that body build is relatively consistent over short time
comparisons. periods even in growing children, such measurements could in
Prediction equations inevitably confound accurate raw turn indicate the direction of changes in FFM, but are unlikely
values with predictive error (standard error of the estimate). to quantify their magnitude with accuracy. In obesity clinics,
Hence, for assessment of fatness, it is better to leave skinfolds BIA might find a role in indicating whether changes in lean
in raw form or as SDS, where they act as reliable indices of mass are in the same direction as body weight, but should not
regional fatness, than to attempt prediction of total fat mass. be used to estimate change in fat mass. These issues should be
For assessment of total fat-free mass (FFM), an approach borne in mind as reference data for BIA data become available.
based on skinfold equations is particularly inappropriate, as At present, therefore, the value of BIA is primarily as an
no index of this component of weight is directly measured. epidemiological technique, where it is the only predictive
Bioelectric impedance analysis (BIA) measures impedance of technique that estimates lean mass.
the body to a small electric current. The generic theoretical
model treats the body as a single cylinder, with measure- Two-component techniques and models
ments made between electrodes placed manually on the wrist Two-component methods divide the body into fat mass (FM)
and ankle. Adjustment of bioelectrical data for height allows and FFM. They avoid two of the difficulties mentioned above,
estimation of total body water (TBW). In practice, this addressing both components of weight and avoiding the need
requires the empirical derivation of regression equations to predict total masses from regional or superficial proxies.
relating height2/impedance to TBW. These equations are then However, they remain dependent on theoretical assumptions,
such as constancy of the composition of FFM.
Between birth and adulthood, chemical maturation of lean
mass occurs, whereby the relative proportions of the three
BMI
main components (water, protein, and mineral) change with
Lean mass age and pubertal status. In two-component techniques, these
changes are addressed by assuming constant lean mass
Fat mass
characteristics for a given age and sex. Such assumptions
Subscapular may largely hold true for healthy subjects but are more
problematic in patients who may have deranged body
Triceps composition or hydration. Various methods are available.
0
–5 –4 –3 –2 –1 0 1 2 3 4 5 Dual energy x ray absorptiometry
z Score Dual energy x ray absorptiometry (DXA) was developed to
measure bone mineral mass, which is calculated from the
Figure 1 Data from infant patients with congenital myasthenia, a differential absorption of x rays of two different energies.
condition in which the development of connective tissue is impaired.
Despite extremely low BMI SDS, the patients have body fat levels higher
Because this calculation requires allowance for (and hence
than the average in healthy children. This paradox can be attributed to quantification of) overlying soft tissue, values of fat and FFM
extremely low levels of lean mass.9 Though underweight, energy intake is are also calculated for whole body scans, using instrument
not itself constraining their growth. specific algorithms.

www.archdischild.com
614 Wells, Fewtrell

plethysmograph has also become available, allowing mea-

surement of body volume during the first 6 months of life.33 34
In general, densitometry is unsuitable for application as a
two-component technique in patients where the composition
of lean mass may be abnormal. Typical effects of disease are
excess fluid retention and undermineralisation; both
decrease the density of lean mass and hence lead to the
overestimation of fatness. However, in relative terms,
densitometric errors are smaller in larger individuals. Thus,
densitometry may prove useful for monitoring changes over
time in overweight or obese individuals, and its accuracy is
less likely to be confounded by longitudinal changes in
fatness than DXA.

Isotope dilution (hydrometry)

Deuterium dilution can be used to measure TBW, allowing
estimation of FFM. A dose of water labelled with deuterium
is given and, following equilibration, enrichment of the body
water pool measured using samples of either saliva, urine, or
blood.35 Samples are generally analysed by isotope ratio mass
Figure 2 A child sitting inside the measurement chamber of the whole
body air displacement plethysmograph, known as the Bodpod (Life spectrometry; however, clinical services could be based on a
Measurements Inc, California). The measurement involves air being substantially cheaper but more labour intensive spectro-
gently blown around the subject during two 1 minute measurement photometric technique.36
periods, while the subject breathes normally. Written permission was Estimation of FFM from TBW requires an assumed value
obtained from the child’s parents for the publication of this photograph. for FFM hydration. Published reference values are relatively
consistent with measured values in healthy infants and
DXA is quick and acceptable for children down to about 4 children,37 with between-individual variability also relatively
years, and can also be used in small infants. It uses ionising low.23 38 However, in disease states variability in FFM
radiation but the effective dose equivalents of contemporary hydration may be substantially higher, owing to either
instrumentation are below background levels. However, overhydration or underhydration. In the future, BIA may be
although widely used, particularly in the USA, problems used to assess whether hydration is abnormal or not, but for
with accuracy have received insufficient attention. Bias varies now isotope dilution is best used in populations where the
with age and fatness and, in some cases, underlying disease normality of hydration is known or can be assumed.
state.28 Comparisons between groups are likely to identify the Isotope dilution is simple to carry out and requires minimal
direction of changes, but fail to quantify the difference subject cooperation. It has proved particularly valuable in
accurately, while the accuracy with which changes can be infants and toddlers because of the low compliance required,
quantified within individuals gaining or losing weight is and can easily be used in field studies. It could prove a useful
likewise liable to be confounded by the change in weight clinical tool for individuals where normal hydration can be
status. Thus, although DXA studies increasingly contribute to assumed, with rapid provision of results possible using
the evidence base for clinical paediatrics, such studies should photospectrometric analysis.36
be interpreted with caution, and the technique does not
represent a reference method. Magnetic resonance imaging
The limitations of DXA vary according to body shape and MRI is an imaging technique that estimates the volume rather
outcome. Trunk composition involves substantial prediction than the mass of adipose tissue. By analysing the absorption
rather than measurement, and soft tissue estimation in this and emission of energy in the radio frequency range of the
area is less accurate than in the limbs. DXA may provide electromagnetic spectrum, the technique produces images
useful information on relative fat and lean masses as a single
based on spatial variations in the phase and frequency of the
measurement in an individual, particularly with respect to
energy absorbed and emitted. It primarily addresses hydrogen
limb lean mass. However, such assessments are currently
nuclei, located either in water or fat, and uses these data to
hampered by the lack of normal reference data, an issue that
discern tissue types in ‘‘imaging slices’’ which can then be
is currently being addressed. In contrast to soft tissue
summed to calculate regional tissue volumes.
outcomes, bone data have good accuracy.
Despite the high quality of imaging data obtained by MRI,
Densitometry there are difficulties in comparing results with those obtained
This approach distinguishes FM and FFM, assuming specific by other techniques. First, in order to derive fat mass, it is
densities of these two tissues,29 and therefore requires necessary to assume the fat content of adipose tissue and the
measurement of total body density (body mass/body density of fat. While the latter is relatively invariable, the
volume). While the density of fat is indeed relatively former is not. A second problem is that FM discerned by MRI
constant, that of FFM varies according to its composition. is only that present in adipose tissue. Thus techniques such as
This variability is partly explained by the process of chemical densitometry, hydrometry, or multicomponent models quan-
maturation that occurs before adulthood,30 but interindivi- tify a different entity from MRI, total FM versus adipose
dual variability is also significant, even in healthy children.23 tissue mass. Also MRI has relatively high cost and limited
Traditionally, body volume was measured by hydrodensi- availability.
tometry; however, this approach is clearly unsuitable for The main advantage of MRI over other techniques is its
many children, especially patients. A new alternative is air capacity for estimation of regional body composition, and it is
displacement plethysmography. The technique measures the currently the only accurate and viable approach for the
volume of air displaced by the subject (fig 2). The technique estimation of intra-abdominal adipose tissue. However, recent
has better precision than hydrodensitometry in children,31 32 studies support the use of waist circumference as a robust
and is acceptable in children as young as 4 years. An infant index of abdominal fat, and a useful index of visceral fat.39

www.archdischild.com
Measuring body composition 615

Table 1 Summary of assumptions underlying different techniques, their availability, and their main advantages and
disadvantages
Technique Assumptions made Reference data Availability* Advantages/disadvantages

Skinfolds – raw Constant skin protein content Y +++ For: simple measure of regional fat
Against: no information on lean mass

Skinfolds – equations Skinfolds 3 whole body N +++ For: simple and quick
fat Against: population specific, poor accuracy in individuals and groups

Body mass index Var weight = var fat Y +++ For: simple and quick
Against: measures nutritional status not body composition

Waist circumference Waist 3 central fat Y +++ For: simple, quick, robust measure of abdominal fat
Against: not so accurate as measure of internal visceral fat

Impedance Conductivity 3 body water N ++ For: simple and quick

Against: population specific; poor accuracy in individuals and groups

DXA Constant attenuations of N ++ For: accurate for limb lean and fat
FFM and F Against: radiation exposure; whole body bias ?size, sex, fatness

Densitometry
Constant Dffm and Df N + For: acceptable two-component technique

Against: effects of disease on lean mass reduce accuracy

Isotope dilution Constant Hffm N + For: only technique acceptable in all age groups
Against: delayed results; inaccurate if disease affects Hffm

MRI Electromagnetic properties N + For: accurate for regional AT

Against: expensive, limited availability, measures AT not fat

TOBEC Conductivity 3 body water N + For: acceptable two-component technique

Against: rarely available, accuracy unknown

TBK Constant K in cell mass N + For: measures functional component of body composition
Against: rarely available, poor accuracy for fatness

Multicomponent models Constant Dprot and Dmin, N + For: most accurate approach, all measurements acceptable
constant mineral composition Against: expensive, specialist research approach

*Availability: +, low; ++, medium; +++, high.

Densitometry by air displacement plethysmography (Bodpod)
AT, adipose tissue; Df, density of fat; Dffm, density of fat-free mass; Dmin, density of mineral; Dprot, density of protein; F, fat; FFM, fat-free mass; Hffm, hydration of
fat-free mass; K, potassium; Var, variability.

Other techniques measurements of body weight, body water by hydrometry,

Other techniques include total body electrical conductivity and body volume by densitometry. The four component
(TOBEC) and whole body potassium scanning (TBK). Both model further divides fat-free dry tissue into protein and
techniques suffer from three main limitations. First, neither mineral, and requires the same data plus measurement of
is widely available. Second, their outputs show poor bone mineral by DXA.23 36
agreement with those of other methods, although this The advantages and disadvantages of the various techni-
applies less to TOBEC than TBK. Third, there are no UK ques, their availability, and the existence of reference data are
reference data, hindering their clinical application. summarised in table 1.
Multicomponent models
Multicomponent models minimise the assumptions made in RECOMMENDATIONS
simpler models and are hence best placed to address As highlighted above, measurement of body composition in
variability in the composition of lean mass. TBW and bone vivo is an imperfect process, subject to various constraints,
mineral are measured by techniques specifically designed for yet the outcome has clinical value. We offer the following
those purposes, improving accuracy of those outcomes. suggestions for the measurement of body composition in
Whereas two-component models assume key body proper- paediatric clinical practice.
ties, multicomponent models measure them, and can provide
data on the hydration, density, and mineralisation of FFM. N Simple techniques should not be rejected because they
appear unsophisticated. Skinfold measurements and WC
This is particularly important in paediatric patients, in whom
body composition derangement is often extreme. In the provide a simple, easy, and quick yet highly informative
absence of carcass analysis, these models are regarded as the assessment of fatness in most patients. DXA is unlikely to
gold standard for in vivo measurement: more importantly, improve substantially on this simple approach unless there
they are the only reference or criterion methods against are difficulties in obtaining skinfold data, and it has poor
which less accurate methods may be compared. Because of accuracy for trunk fatness. For assessment of whole body
the diverse equipment required, they are only suitable for lean mass, both skinfold measurements and BIA have
research, and along with MRI for fat distribution should be limitations. Application of the combination of these
regarded as the optimum approach for acquiring the evidence methods may reduce the likelihood of misdiagnosis of
base to underpin clinical practice. high or low lean mass, but will be least successful when
The three-component model divides body weight into fat, applied to disease states where regional tissue distribu-
water, and remaining fat-free dry tissue, and requires tions differ from those of healthy children.

www.archdischild.com
616 Wells, Fewtrell

Table 2 Illustration of the advantages and disadvantages of different techniques for measuring body composition in clinical
practice. The patient is a 14 year old girl referred for assessment and management of obesity with associated type 2 diabetes.
Which measurements might be useful in terms of baseline assessment and monitoring of body composition during weight loss?
Technique Advantages Disadvantages

Simple measurements
Skinfold thickness – Poor accuracy and precision in obesity
BMI Useful as a simple baseline and longitudinal measurement of Will not allow assessment of fat and lean masses and changes
relative weight in fat which may be more relevant in terms of metabolic risk
Waist circumference Useful baseline and longitudinal measurement. Centile charts
are available, and central fatness is of greater relevance to
metabolic risk

Predictive measurements
Skinfold thickness equations – Poor precision and accuracy—any error will be magnified by
predicting FM and FFM the use of prediction equations not derived from a
comparable (obese) population
Whole body BIA Could provide information on the direction of longitudinal Poor accuracy in absolute terms. Changes in body weight, the
changes in lean mass relative proportions of trunk and limbs, or FFM hydration with
treatment could introduce further errors

Two-component techniques
DXA Could be used to measure regional (limb) lean mass which Limited use for measuring baseline fat mass or longitudinal
could provide information on whether weight loss is changes in fat mass with weight loss, since measurements are
accompanied by changes in lean mass as well as fat mass known to be biased by body size (thickness)
Densitometry (BodPod) Could provide longitudinal data on both lean and fat mass Does not provide regional data.
since its accuracy is less likely to be affected by changes in
fatness.
Deuterium dilution Could estimate whole body lean mass Results may be affected by the small differences in FFM
hydration in obesity, but these have been quantified and could
be adjusted for. Small differences in FFM hydration would
introduce minor errors in the short term

BIA, bioelectric impedance analysis; BMI, body mass index; DXA, dual energy x ray absorptiometry; FFM, fat-free mass; FM, fat mass.
Conclusions: In practical terms, the techniques most useful in this patient are BMI and waist circumference for monitoring nutritional status and central fat
distribution. If available, DXA could be used for assessing changes in limb lean mass, and densitometry or deuterium could also provide longitudinal data for both
fat and lean mass.

Table 3 Illustration of the advantages and disadvantages of different techniques for measuring body composition in clinical
practice. The patient is a 10 year old boy with chronic renal failure on peritoneal dialysis. Which measurements might be useful
in terms of baseline assessment and monitoring of body composition? In this child, measurement of lean mass would be
potentially useful as a guide for basing nutritional requirements or drug doses, but measurements may be complicated by
oedema or variation in FFM hydration or both
Technique Advantages Disadvantages

Simple measurements
Skinfold thickness Useful index of regional fatness Poorer accuracy and precision in obese/oedematous
individuals
BMI Useful as a simple baseline and longitudinal measurement of Will not allow assessment of fat and lean masses and changes
relative weight in fat which may be more relevant in terms of nutritional/drug
requirements and metabolic risk
Waist circumference Useful baseline and longitudinal measurement. Centile Value may be limited in this child who is on peritoneal dialysis
charts are available, and central fatness is of greater
relevance to metabolic risk

Predictive measurements
Skinfold thickness equations – Poor precision and accuracy—any error will be to magnified
predicting FM and FFM by the use of prediction equations not derived from a
comparable population
BIA Could provide information on the direction of longitudinal Poor accuracy in absolute terms. Changes in body weight, the
changes in lean mass relative proportions of trunk and limbs, or FFM hydration will
introduce further errors*

Two-component techniques
DXA Could be used to measure regional (limb) lean mass which Limited use for measuring baseline fat mass or longitudinal
could provide information on which to base nutritional changes in fat mass, since measurements are known to be
requirements or drug doses biased by body size (thickness). In this child, skinfold
thicknesses may already provide information on regional
fatness
Densitometry (BodPod) Could provide longitudinal data on both lean and fat mass Variations in the hydration (and therefore density) of FFM in
since its accuracy is less likely to be affected by changes in this child will invalidate assumptions of constant density
fatness
Deuterium dilution Could estimate whole body lean mass if hydration known, Results for both TBW and FFM may be affected by variation in
and could potentially aid calculation of dialysis fluids hydration in this child

*Multifrequency BIA could provide informative data on hydration in the future.

BIA, bioelectric impedance analysis; BMI, body mass index; DXA, dual energy x ray absorptiometry; FFM, fat-free mass; FM, fat mass; TBW, total body water.
Conclusions: In practical terms, the techniques most useful in this patient are BMI ¡ skinfold thickness measurement for monitoring fat mass and distribution. If
available, DXA could be used for assessing changes in (limb) lean mass, which might be more relevant for nutritional requirements and drug doses. This technique
will be of greater use in practice once reference data are available.

www.archdischild.com
Measuring body composition 617

N Though BIA has potential utility in measuring regional

body composition, its value in routine clinical paediatrics
5 World Health Organization. ICD-10 classication of mental and behavioural
disorders: clinical descriptions and diagnostic guidelines. Geneva: WHO, 1992.
6 Chan YL, Leung SSF, Lam WWM, et al. Body fat estimation in children by
is at present limited. Predictive error in individuals is high, magnetic resonance imaging, bioelectrical impedance, skinfold and body
even when using population specific equations. If used to mass index: a pilot study. J Paediatr Child Health 1998;34:22–8.
7 Pietrobelli A, Faith MS, Allison DB, et al. Body mass index as a measure of
monitor individuals over time, it can indicate the direc- adiposity among children and adolescents: a validation. J Pediatr
tion, but not the magnitude, of changes in lean mass. 1998;132:204–10.
However, such data are difficult to interpret in the absence 8 Wells JCK. A Hattori chart analysis of body mass index in infancy and
childhood. Int J Obesity 2000;24:325–9.
of reference data. 9 Wells JCK, Mok Q, Johnson AW. Nutritional status in children. Lancet
N Whole body data may appear optimal, but in practice
regional data may be more informative about clinical
2001;357:1293.
10 Rimm AA, Hartz AJ, Fischer ME. A weight shape index for assessing risk of
disease in 44,820 women. J Clin Epidemiol 1988;41:459–65.
condition, as well as more accurate. Many disease states 11 Walton C, Lees B, Crook D, et al. Body fat distribution, rather than overall
exert disproportionate effects on particular body regions. adiposity, influences serum lipids and lipoproteins in healthy men
In obesity, the main concern is central adiposity, so independently of age. Am J Med 1995;99:459–64.
12 Savva SC, Tornaritis M, Savva ME, et al. Waist circumference and waist-to-
monitoring of WC may provide a better indication of height ratio are better predictors of cardiovascular disease risk factors in
health risk, and response to treatment, than whole body children than body mass index. Int J Obes 2000;24:1453–8.
fatness. Whole body lean mass is difficult to measure with 13 McCarthy HD, Jarrett KV, Crawley HF. The development of waist
circumference percentiles in British children aged 5.0–16.9 y. Eur J Clin Nutr
simple methods, but DXA is a relatively accurate technique 2001;55:902–7.
for quantifying limb lean mass. Its value in this role will 14 de Ridder CM, de Boer RW, Seidell JC, et al. Body fat distribution in pubertal
increase once reference data for these outcomes are girls quantified by magnetic resonance imaging. Int J Obes 1992;16:443–9.
15 Fox K, Peters D, Armstrong N, et al. Abdominal fat deposition in 11-year-old
published. Whole body data are useful for estimating children. Int J Obes 1993;17:11–16.
energy requirements, or the dose size for drug/therapy 16 Owens S, Litaker M, Allison J, et al. Prediction of visceral adipose tissue from
treatments and dialysis. simple anthropometric measurements in youths with obesity. Obes Res
1999;7:16–22.
N Two-component models are ideal where the aim is to
quantify fat or FFM with greater accuracy than that
17 Brambilla P, Manzoni P, Sironi S, et al. Peripheral and abdominal adiposity in
childhood obesity. Int J Obes 1994;18:795–800.
permitted by the simplest methods. For example, whole 18 Brook CGD. Determination of body composition of children from skinfold
measurements. Arch Dis Child 1971;46:182–4.
body lean mass is sometimes used to calculate the quantity 19 Johnston JL, Leong MS, Checkland EG, et al. Body fat assessed from body
of treatment required. However, this approach is only density and estimated from skinfold thickness in normal children and children
justified if the technique can be assumed robust to likely with cystic fibrosis. Am J Clin Nutr 1988;48:1362–6.
20 Slaughter MH, Lohman TG, Boileau RA, et al. Skinfold equations for
derangements in FFM composition. Furthermore, while estimation of body fatness in children and youth. Hum Biol 1988;60:709–23.
two component techniques are of great value in research, 21 Deurenberg P, Pieters JJL, Hautvast JGA. The assessment of the body fat
they often may have less value in routine clinical percentage by skinfold thickness measurements in childhood and young
adolescence. Br J Nutr 1990;63:293–303.
management as they conceal important regional varia- 22 Reilly JJ, Wilson J, Durnin JVGA. Determination of body composition from
bility within a global whole body value. skinfold thickness: a validation study. Arch Dis Child 1995;73:305–10.
N Multicomponent models and MRI are ideal for detailed
analyses but remain unfeasible in most contexts. Their
23 Wells JCK, Fuller NJ, Dewit O, et al. Four-component 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.
main value lies in the quality of their evidence in 24 Wabitsch M, Braun U, Heinze E, et al. Body composition in 5–18-y-old obese
supporting treatment approaches, rather than in routine children and adolescents before and after weight reduction as assessed by
deuterium dilution and bioelectrical impedance analysis. Am J Clin Nutr
practical application. 1996;64:1–6.
N The value of any approach is greatly enhanced by the
availability of reference data. The acquisition of such data
25 Puiman PJ, Francis P, Buntain H, et al. Total body water in children with cystic
fibrosis using bioelectrical impedance. J Cystic Fibrosis 2004;3:243–7.
26 Arpadi SM, Wang J, Cuff PA, et al. Application of bioimpedance analysis for
is a current priority, being addressed by several research estimating body composition in prepubertal children infected with human
groups. immunodeficiency virus type 1. J Pediatr 1996;129:755–7.
27 Parker L, Reilly JJ, Slater C, et al. Validity of six field and laboratory methods
These recommendations are illustrated using two case for measurement of body composition in 10–14 year old boys. Obes Res
2003;11:852–8.
studies (tables 2 and 3). 28 Williams JE, Wells JCK, Wilson CM, et al. Evaluation of Lunar Prodigy dual-
energy X-ray absorptiometry for assessing body composition in healthy
individuals and patients by comparison with the criterion four-component
The extended version of this paper is available on the model. Am J Clin Nutr, (in press)..
journal website (http://www.archdischild.com/ 29 Siri WE. Body composition from fluid spaces and density: analysis of methods.
supplemental) In:Brozek J, Henschel A, editors. Techniques for measuring body composition.
Washington DC: National Academy of Sciences NRC, 1961:223–44.
30 Fomon SJ, Haschke F, Ziegler EE, et al. Body composition of reference
children from birth to age 10 years. Am J Clin Nutr 1982;35:1169–75.
31 Dewit O, Fuller NJ, Fewtrell MS, et al. Whole-body air-displacement
..................... plethysmography compared to hydrodensitometry for body composition
analysis. Arch Dis Child 2000;82:159–64.
Authors’ affiliations 32 Nunez C, Kovera AJ, Pietrobelli A, et al. Body composition in children and
J C K Wells, M S Fewtrell, MRC Childhood Nutrition Centre, Institute of adults by air displacement plethysmography. Eur J Clin Nutr 1999;53:382–7.
Child Health, London WC1, UK 33 Urlando A, Dempster P, Aitkens S. A new air displacement plethysmograph
for the measurement of body composition in infants. Pediatr Res
Competing interests: Dr Wells has received equipment gratis from Tanita 2003;53:486–92.
UK for childhood body composition studies. 34 Ma G, Yao M, Liu Y, et al. Validation of a new paediatric air-displacement
plethysmograph for assessing body composition in infants. Am J Clin Nutr
Informed consent was obtained for publication of figure 2.
2004;79:653–60.
35 Davies PSW, Wells JCK. Calculation of total body water in infancy. Eur J Clin
Nutr 1994;48:490–5.
REFERENCES 36 Jennings G, Bluck L, Wright A, et al. The use of infrared spectrophotometry for
1 Reilly JJ, Methven E, McDowell ZC, et al. Health consequences of obesity. measuring body water spaces. Clin Chem 1999;45:1077–81.
Arch Dis Child 2003;88:748–52. 37 Wells JCK, Fuller NJ, Wright A, et al. Evaluation of air-displacement
2 Tanner JM, Whitehouse RH. Revised standards for triceps and subscapular plethysmography in children aged 5–7 years using a three-component model
skinfolds in British children. Arch Dis Child 1975;50:142–5. of body composition. Br J Nutr 2003;90:699–707.
3 Cole TJ, Freeman JV, Preece MA. Body mass index reference curves for the 38 Butte NF, Hopkinson JM, Wong WW, et al. Body composition during the first
UK, 1990. Arch Dis Child 1995;73:25–9. 2 years of life: an updated reference. Pediatr Res 2000;47:578–85.
4 Cole TJ, Bellizzi MC, Flegal KM, et al. Establishing a standard definition for 39 Brambilla P, Bedogni G, Moreno LA, et al. Crossvalidation of anthropometry
child overweight and obesity worldwide: international survey. BMJ against magnetic resonance imaging for the assessment of visceral and
2000;320:1240–3. subcutaneous adipose tissue in children. Int J Obes 2006;30:23–30.

www.archdischild.com