This article was downloaded by: [Memorial University of Newfoundland]On: 17 July 2014, At: 18:38Publisher: RoutledgeInforma Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House,37-41 Mortimer Street, London W1T 3JH, UK

Journal of Sports SciencesPublication details, including instructions for authors and subscription information:http://www.tandfonline.com/loi/rjsp20

The accuracy and precision of DXA for assessing bodycomposition in team sport athletesJohann Christopher Bilsboroughab, Kate Greenwayc, David Oparc, Steuart Livingstonec,Justin Cordyb & Aaron James Couttsab

a University of Technology Sydney (UTS), Lindfield, Australiab Carlton Football Club, Carlton North, Australiac Royal Melbourne Institute of Technology (RMIT), Bundoora, AustraliaPublished online: 10 Jun 2014.

To cite this article: Johann Christopher Bilsborough, Kate Greenway, David Opar, Steuart Livingstone, Justin Cordy & AaronJames Coutts (2014): The accuracy and precision of DXA for assessing body composition in team sport athletes, Journal ofSports Sciences, DOI: 10.1080/02640414.2014.926380

To link to this article: http://dx.doi.org/10.1080/02640414.2014.926380

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) containedin the publications on our platform. However, Taylor & Francis, our agents, and our licensors make norepresentations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of theContent. Any opinions and views expressed in this publication are the opinions and views of the authors, andare not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon andshould be independently verified with primary sources of information. Taylor and Francis shall not be liable forany losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoeveror howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use ofthe Content.

This article may be used for research, teaching, and private study purposes. Any substantial or systematicreproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in anyform to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://www.tandfonline.com/page/terms-and-conditions

The accuracy and precision of DXA for assessing body compositionin team sport athletes

JOHANN CHRISTOPHER BILSBOROUGH1,2, KATE GREENWAY3, DAVID OPAR3,STEUART LIVINGSTONE3, JUSTIN CORDY2 & AARON JAMES COUTTS1,2

1University of Technology Sydney (UTS), Lindfield, Australia, 2Carlton Football Club, Carlton North, Australia and3Royal Melbourne Institute of Technology (RMIT), Bundoora, Australia

(Accepted 17 May 2014)

AbstractThis study determined the precision of pencil and fan beam dual-energy X-ray absorptiometry (DXA) devices for assessingbody composition in professional Australian Football players. Thirty-six professional Australian Football players, in twogroups (fan DXA, N = 22; pencil DXA, N = 25), underwent two consecutive DXA scans. A whole body phantom withknown values for fat mass, bone mineral content and fat-free soft tissue mass was also used to validate each DXA device.Additionally, the criterion phantom was scanned 20 times by each DXA to assess reliability. Test–retest reliability of DXAanthropometric measures were derived from repeated fan and pencil DXA scans. Fat-free soft tissue mass and bone mineralcontent from both DXA units showed strong correlations with, and trivial differences to, the criterion phantom values. Fatmass from both DXA showed moderate correlations with criterion measures (pencil: r = 0.64; fan: r = 0.67) and moderatedifferences with the criterion value. The limits of agreement were similar for both fan beam DXA and pencil beam DXA(fan: fat-free soft tissue mass = −1650 ± 179 g, fat mass = −357 ± 316 g, bone mineral content = 289 ± 122 g; pencil: fat-free soft tissue mass = −1701 ± 257 g, fat mass = −359 ± 326 g, bone mineral content = 177 ± 117 g). DXA also showedexcellent precision for bone mineral content (coefficient of variation (%CV) fan = 0.6%; pencil = 1.5%) and fat-free softtissue mass (%CV fan = 0.3%; pencil = 0.5%) and acceptable reliability for fat measures (%CV fan: fat mass = 2.5%,percent body fat = 2.5%; pencil: fat mass = 5.9%, percent body fat = 5.7%). Both DXA provide precise measures of fat-freesoft tissue mass and bone mineral content in lean Australian Football players. DXA-derived fat-free soft tissue mass andbone mineral content are suitable for assessing body composition in lean team sport athletes.

Keywords: fat-free soft tissue mass, DXA, anthropometry, team sports, precision

Introduction

In high-performance athletes, both fat-free soft tissuemass and fat mass are of particular interest as theyimpact physical performance (Behnke & Royce,1966; Stewart, 2001) and risk of injury and illness(Duthie, 2006; Hagmar, Berglund, Brismar, &Hirschberg, 2013). Indeed, it is now common thathigh-performance athletes regularly assess changesin body composition to determine the effectivenessof training and nutritional interventions (Duthie,2006). Furthermore, since small changes in bodycomposition may impact upon athletic performance,it is important that assessment methods have goodprecision (Santos et al., 2010).

Body composition is commonly assessed in profes-sional sport by practical methods such as skinfolds,bioelectrical impedance analysis (BIA), air displace-ment plesmography and dual-energy X-ray

absorptiometry (DXA). These methods are preferredin comparison to criterion methods such as four-com-partment techniques, due to their portability, practi-cality and low cost. Due to its speed of measurecompared to four-compartment models and goodlevel of accuracy, DXA is widely used to evaluatebody composition in athletes (Buehring et al.,2014). DXA provides a minimally invasive measure-ment of the three-compartment model of body com-position (i.e. fat mass, fat-free soft tissue mass andbone mineral content), each of which are of interest tosport and exercise scientists. While many studies haveassessed the measurement precision of DXA inhealthy population (Rothney et al., 2012), relativelyfew studies (Pineau, Filliard, & Bocquet, 2009;Santos et al., 2010) have compared the measurementprecision of DXA in trained athletes. Despite this,DXA is now widely accepted as a practical criterion

Correspondence: Johann Christopher Bilsborough, Sport and Exercise Discipline Group, University of Technology Sydney (UTS), Lindfield, Australia.E-mail: bcb1@iprimus.com.au

Journal of Sports Sciences, 2014http://dx.doi.org/10.1080/02640414.2014.926380

© 2014 Taylor & Francis

Dow

nloa

ded

by [

Mem

oria

l Uni

vers

ity o

f N

ewfo

undl

and]

at 1

8:38

17

July

201

4

method for assessing fat mass and fat-free soft tissuemass in athletes (Stewart & Hannan, 2000; Stewart &Sutton, 2012).

There have now been many validation studies,mostly in healthy (non-athletic) humans, showingthat the level of accuracy of these devices is greaterthan skinfold and BIA measures (Avlonitou,Georgiou, Douskas, & Louizi, 1997; Lohman,Harris, Teixeira, & Weiss, 2006). Despite this, sev-eral studies that compared DXA to a four-compart-ment model showed a mean underestimation ofpercent body fat (Toombs, Ducher, Shepherd, &De Souza, 2012; Wong et al., 2002), with under-estimation greater in leaner individuals (Van DerPloeg, Withers, & Laforgia, 2003). However, it haspreviously been shown that DXA estimation for fat-free soft tissue mass and fat mass can vary signifi-cantly among differing models (hardware and soft-ware) under different laboratory conditions (Nana,Slater, Hopkins, & Burke, 2012; Schoeller et al.,2005). At present, however, no studies have com-pared the measurement precision of the differentDXA devices in athletic populations.

The different technology amongst DXA devices(i.e. fan (wide and narrow fan) vs. pencil beam)may influence the measurement accuracy. With pen-cil DXA technology, X-ray beams pass through anarrow collimator and data are acquired via a recti-linear pattern separated by a few millimetres over thelongitudinal axis of the patient (Pludowski et al.,2010). In contrast, fan beam technology uses widerX-ray beams, allowing faster scanning and betterimage resolution but at the cost of higher radiationdosage and magnification errors (Ackland et al.,2012; Soriano et al., 2004). These magnificationerrors have been reduced via newer technologyintroduced, known as narrow angle fan beam.These DXA units scan in a rectilinear fashion witha fan beam that is slightly narrower than the olderpencil beams. Each pass of the beam over the parti-cipant overlaps the previous one, and the images arereconstructed to form a more accurate estimation ofthe depth of the bone than the previous fan beamtechnology (Toombs et al., 2012). Moreover, recentadvances in software-assisted automated analysishave improved precision of the newer fan beamunits by reducing the technicians’ manual handlingerrors (Fan et al., 2008; Libber, Binkley, & Krueger,2012). However, despite these recent improvementsin technology, there is currently a poor understand-ing of the influence that discrete DXA technologiesmay have on measurement precision in athletes thathave proportionally high fat-free soft tissue mass andlow fat mass.

There is also relatively poor information on themeasurement precision of DXA devices in trainedor elite athletes. We reviewed 98 studies between

1998 and 2012 that used DXA to measure compo-nents of body composition in athletes and found thatonly 15 (15.4%) reported validity measures and 57(58.7%) reported measurement reliability.Moreover, the studies that examined measurementvalidity used a range of criterion measures (i.e.known scale weights, hydrostatic weighing and var-ious multi-compartment models), making accuratecomparison between the studies difficult. Despitethese limitations, these studies revealed good overalltest–retest reliability of DXA for assessing fat-freesoft tissue mass, fat mass, bone mineral contentand bone mineral density (coefficient of variation(%CV): 0.8 ± 0.4%, 2.6 ± 1.2%, 1.1 ± 0.7% and1.0 ± 0.9%, respectively; all mean ± standarddeviation (s)). Several other factors such as athletepositioning, hydration, body morphology (skeletalproportions, body thickness and tissue composition),auto-analysis inaccuracies and different software andhardware may explain some of the measurementerror in these previous studies; however the sourcesof error have yet to be fully elucidated. This errormay be important when assessing changes in bodycomposition in highly trained athletes as smallchanges in anthropometry may greatly affect compe-titive performance.

To date, no studies have directly compared preci-sion of body composition measures in highly trainedteam sport athletes using different DXA technology(i.e. fan vs. pencil beam). Therefore, the aims of thisstudy were to (1) compare the measurement preci-sion of two DXA devices for assessing body compo-sition in highly trained Australian football players,(2) determine the test–retest reliability of these DXAdevices for assessing body composition in highlytrained Australian Football players and (3) validatealternate practical methods that are commonly usedto assess body composition in team sports.

Methods

Thirty-six male professional Australian Footballplayers (mean ± s; age 22.7 ± 3.0 years, mass84.4 ± 5.62 kg, stature 186 ± 5 cm) were recruitedfor the validity component of this study. A custom-made whole body phantom with known weights offat-free soft tissue mass, fat mass and bone mineralcontent served to validate each device in an addi-tional component of the validation study. From thepool of 36 players, two separate subgroups wereformed. One group consisted of 22 players (mean ± s;age 22.5 ± 2.8 years, mass 83.80 ± 6.09 kg, stature184 ± 5 cm) and another of 25 players (mean ± s;age 21.3 ± 2.1 years, mass 82.91 ± 8.40 kg, stature184 ± 7 cm) each of which completed the reliabilityaspect of the study. All players were contracted tothe same Australian football club, which participated

2 J.C. Bilsborough et al.

Dow

nloa

ded

by [

Mem

oria

l Uni

vers

ity o

f N

ewfo

undl

and]

at 1

8:38

17

July

201

4

in the Australian Football League. Only players thatdid not exceed the length of the scanning table(192 cm) were included in the study. The two sepa-rate groups underwent the scanning procedures tolimit radiation exposure. The methods for the studywere approved by a university ethics committee andby the Australian Football League club involved. Allparticipants provided written informed consent priorto commencing the study, and the UniversityHuman Ethics Review Panel approval was received.

Study overview

Procedures. Prior to scanning, each DXA unit (i.e. anarrow fan beam (Prodigy, Lunar Corp, Madison,WI, USA) and a pencil beam (DPX-IQ, LunarCorp.)) was calibrated with criterion phantomdevices according to manufacturer guidelines. Tominimise scanning errors, procedures were standar-dised according to the recommendations of theAustralian and New Zealand Bone and MineralDensity Society. The athletes were all placed withhands in a pronated flat position within the scanrange, and the leg position was standardised andsecured with straps to reduce bone overlap in lowerlimbs. The athletes removed metal objects or jewel-lery from their body and wore the same minimalclothing (underwear) for each scan. The athleteswere also instructed to follow standard protocols offood and fluid to ensure hydration was optimal priorto each scan. Moreover, all players urine specificgravity (Digital Refractometer, ATAGO, Tokyo,Japan) were assessed prior to scanning to ensurehydration status before each testing session. Thesame technician analysed all scans using a manualanalysis for the pencil beam DXA (SoftwareSmartScan Version 4.7e), and although automaticanalysis was performed with the fan beam DXA(Software version enCORE™ 2009, Version13.20.033), regions of interest were confirmed bythe technician.

Accuracy. To assess the accuracy of each DXA sys-tem, a whole body phantom (OrthimetrixIncorporated™, Naples, FL, USA) was used as thecriterion measure. This custom-made phantom hasknown weights of regional and total body fat-free softtissue mass, fat mass and bone mineral content andis composed of 11 discrete polymer resin pieces. Thedistinct difference to other manufacturer’s phantomsis that the Orthimetrix device is in the shape of anactual person. The 11 polymer resin pieces weremoulded to form 2 upper arms, 2 forearms (withhands), 2 upper legs, 2 lower legs, pelvic region,trunk and head. The phantom pieces consist ofknown quantities of soft tissues and alumina hydro-xide that are moulded in polyurethane. The replicate

bone mineral content, known values of calcium car-bonate were used to obtain a material with the sameX-ray absorption as bones. The whole phantomcould be modified by removing any of the 11 pieces(each consisted of combined known fat-free soft tis-sue mass, fat mass and bone mineral content) toprovide a range of measures of fat-free soft tissuemass, fat mass and bone mineral content for calibra-tion. The phantom was designed with low quantitiesof fat mass. The range of criterion measures takenwere 13.02–23.22 kg for fat-free soft tissue mass,0.63–1.41 kg for fat mass and 1.26–2.11 kg forbone mineral content.

To assess accuracy, a range of criterion measureswere used by adding/removing pieces (such as arms,legs or trunk) of the phantom, which were scannedby each DXA device to provide 10 different criterionmeasures. Each criterion phantom was scannedtwice in the same position. The position of the phan-tom was standardised for each scan, and the distancebetween the limbs was measured to be the same foreach scan.

Reliability. To assess test–retest reliability of the fanbeam DXA and the pencil beam DXA units, twogroups of players completed two separate scansunder standard conditions. Separate matched groupsof Australian football players were randomly allo-cated to complete the reliability scans on either thefan beam or pencil beam DXA device. Twenty-twoplayers were assessed twice via a fan beam DXA(Prodigy, Lunar Corp.), while 25 players wereassessed twice on a pencil beam DXA (DPX-IQ,Lunar Corp.). The same qualified technician reposi-tioned the players on the scanning bed between testsand completed the analysis of each scan.

Validity of field methods. All players also had their sumof seven skinfolds and estimated body fat from askinfold derived body fat formula and BIA (detailslater) compared against fat mass and percent bodyfat results from both DXA units to determine validityof these practical inexpensive methods for use inAustralian football players. A sample of 47 playerswas used for this aspect of the study as it was con-ducted in conjunction with the reliability studies ofeach DXA. All skinfold and body mass measureswere taken by the same Level II ISAK (TheInternational Society for the Advancement ofKinanthropometry)-qualified anthropometrist accord-ing to standard methods (Stewart, Marfell-Jones,Olds, & de Ridder, 2011). The skinfold sites weretriceps, subscapular, biceps, supraspinale, abdominal,thigh and medial calf. All skinfold measurements weretaken using Harpenden skinfold callipers (BritishIndicators, Hertfordshire, UK). Body mass was mea-sured with digital scales (British Indicators) to the

Body composition and team sports 3

Dow

nloa

ded

by [

Mem

oria

l Uni

vers

ity o

f N

ewfo

undl

and]

at 1

8:38

17

July

201

4

nearest 0.01 kg. The between test technical error forsum of seven skinfolds and body mass was 2.0 mm(2.8%) and 0.2%, respectively. Percent body fat wasthen calculated from a soccer-specific (skinfold-derived) body fat formula (Sutton, Scott, & Reilly,2008).

The players’ body composition was also estimatedusing a BIA system at a fixed signal frequency of50 kHz and 500 µA, using manufacturer guidelines(BF-662W, Tanita Corporation of America, Inc.,Arlington Heights, IL, USA). All participants stoodin minimal clothing (underwear only) on the metalfootplates of the device. The athletes were alsoinstructed to follow standard protocols of food andfluid to ensure hydration was optimal prior to eachscan. Moreover, urine specific gravity of all theplayers (Digital Refractometer, ATAGO, Japan)was assessed prior to scanning to ensure appropriatehydration status before each testing session. Leg-to-leg impedance of the lower extremities and bodyweight were measured simultaneously while the par-ticipant stood on the scale. Estimates of fat-free softtissue mass, percent body fat and fat mass wereprovided using proprietary algorithms. The %CVfor this model of BIA is 23.3%.

Statistical analyses

Data are presented as means with 90% confidencelimits (CL) and confidence intervals, respectively.To assess accuracy, differences for fat-free soft tissuemass, fat mass and bone mineral content measuresfrom each DXA device with the criterion values ofthe phantom were calculated. The limits of agree-ment (LOA), technical error of measurement, %CV,Pearson’s correlation coefficients and standardisedbias were calculated using customised spread sheets(www.sportsci.org). Standardised bias was calcu-lated as the mean bias divided by the standard devia-tion of the criterion and interpreted using a modified

Cohen scale: <0.20, trivial; 0.2–0.6, small; 0.6–1.2,moderate; 1.2–2.0, large; >2.0, very large. The test–retest reliability was determined using intra-classcorrelation coefficient (ICC), and typical error wasexpressed as a %CV. Significance was set at P < 0.05(Hopkins, Marshall, Batterham, & Hanin, 2009).

Results

Accuracy

Accuracy measures for the pencil and fan beam DXAunits for fat-free soft tissue mass, fat mass and bonemineral content are summarised in Table I. The fat-free soft tissue mass and bone mineral content valuesderived from both of the DXA units showed strongcorrelations with the criterion whole body phantomvalues. The fat mass measures from both DXA unitsshowed moderate correlations with criterion measures(pencil beam DXA r = 0.64; fan beam DXA r = 0.67).

Reliability

Table III details the reliability measures for bothtotal and regional composition measures from boththe fan and pencil beam DXA units. Both DXAunits showed excellent reliability for total bonemineral content and fat-free soft tissue mass andacceptable reliability for total fat mass and percentbody fat. The test–retest reliability for the BIA unitfor determining percent body fat showed %CV’s of25.9% (20.1–37.1) with an ICC of 0.54 (0.22–0.76).

Discussion

The primary aim of this study was to determine theaccuracy and precision of a fan and pencil beamDXA for use in the evaluation of body compositionin professional Australian Football players. In addi-tion, the study also examined the validity of alternatecommonly used methods to assess body composition

Table I. Accuracy (mean, 90% CL) of known measures of fat mass, fat-free soft tissue mass and bone mineral content (BMC) determinedvia fan and pencil beam DXA units with whole body phantom as criterion method (N = 20).

Mean ± s (g) Bias ± 90% LOA TEM (%) r CV (%) Standardised bias

Pencil beam FM (g) 726 ± 128 −1650 ± 257 30.66 0.64 (0.35–0.82) 17.9 (13.9–25.6) 0.79 (moderate)Fan beam FM (g) 729 ± 130 −1701 ± 257 30.32 0.67 (0.39–0.84) 17.2 (13.4–24.6) 0.76 (moderate)Pencil beam FFSTM (g) 17,435 ± 3078 −357 ± 316 6.60 1.00 (1.00–1.00) 0.51 (0.4–0.7) 0.03 (trivial)Fan beam FFSTM (g) 17,486 ± 3112 −359 ± 326 6.38 1.00 (1.00–1.00) 0.55 (0.4–0.8) 0.03 (trivial)Pencil beam BMC (g) 1892 ± 279 289 ± 122 7.25 0.99 (0.97–0.99) 2.32 (1.8–3.2) 0.16 (trivial)Fan beam BMC (g) 1911 ± 281 177 ± 117 7.91 0.99 (0.98–1.00) 2.2 (1.7–3.1) 0.15 (trivial)

Notes: There were poor correlations between the percent body fat estimated from a soccer-specific skinfold equation (r = 0.32 and 0.32),while moderate correlations with BIA estimated percent body fat (r = 0.64 and 0.66) and Σ7 skinfolds (r = 0.71 and 0.66) with criterionmeasures taken from the pencil beam DXA and fan beam DXA, respectively (Table II). The body mass derived from the pencil beam DXAwas better correlated with the scale weight of the participants (r = 0.98) than the fan beam DXA (r = 0.96); however, these were consideredtrivial and small for each devices (d = 0.13 and 0.20 for pencil and fan DXA, respectively). FM – fat mass, FFSTM – fat-free soft tissuemass, LOA – limits of agreement.

4 J.C. Bilsborough et al.

Dow

nloa

ded

by [

Mem

oria

l Uni

vers

ity o

f N

ewfo

undl

and]

at 1

8:38

17

July

201

4

in team sport athletes. The main findings from thestudy were that both fan beam DXA and pencilbeam DXA provide precise and accurate measuresof fat-free soft tissue mass and bone mineral content;however, poorer accuracy and precision wereobserved for fat mass. We also observed weakerrelationships between estimated fat content mea-sures determined through skinfold and bioelectricalimpedance and DXA-derived results.

There have been many studies that have examinedthe validity of DXA using different compartmentmethods in both athletic (Clark, Sullivan, Bartok, &Carrel, 2007; Egan, Wallace, Reilly, Chantler, &Lawlor, 2006; Fornetti, Pivarnik, Foley, & Fiechtner,1999; Santos et al., 2010; Stewart & Hannan, 2000)and non-athletic populations as well as dissection andashingmethods in animals (Clarys et al., 2010; Hunteret al., 2011). These studies have shown differencesbetween DXA and criterion measures in percentbody fat ranging from −5.3% to 2.9%, with the

majority showing an underestimation of fat with leanerpopulations (Prior et al., 2001; Toombs et al., 2012;Van Der Ploeg et al., 2003; Williams et al., 2006).However, the limitation associated with most of thesestudies has been the use of a “soft validation” methodwhere the total fat-free soft tissue mass, fat mass andbone mineral content are added together and com-pared against body mass measured on a calibratedscale. The present study is a precision study that com-pares fat-free soft tissue mass, fat mass and bonemineral content against known masses with two DXAunits in elite athletes. The present results showed thatboth fat-free soft tissue mass (%CV = 0.5–0.6%) andbone mineral content (%CV = 2.2–2.3%) correspondwell with known values provided in the criterion phan-tom with both the fan and pencil beam DXA. In con-trast, themeasurement error for fatmass for bothDXAunits showed a moderate typical error of measurement(pencil beam = 278 g; fan beam = 275 g and poor%CV (17.2–17.9%)). Moreover, the poor validity

Table III. Reliability of total body and regional composition determined with a fan and pencil beam DXA (mean, 90% CL).

Lunar prodigy (fan beam) Lunar DPX-IQ (pencil beam)

Total N = 22 %CV ICC N = 25 %CV ICCBody mass (kg) 0.2 (0.2–0.3) 1.00 (1.00–1.00) 0.3 (0.3–0.5) 1.00 (1.00–1.00)BMC (g) 0.6 (0.4–0.8) 1.00 (0.99–1.00) 1.5 (1.2–2.0) 1.00 (1.00–1.00)FFSTM (g) 0.3 (0.2–0.5) 1.00 (1.00–1.00) 0.5 (0.4–0.6) 1.00 (1.00–1.00)FM (g) 2.5 (1.9–3.6) 0.99 (0.98–1.00) 5.9 (4.7–7.8) 0.98 (0.96–0.99)%BF 2.5 (1.9–3.5) 0.99 (0.97–1.00) 5.7 (4.6–7.6) 0.98 (0.95–0.99)

RegionalArms FFSTM (g) 2.7 (2.1–3.9) 0.89 (0.76–0.95) 1.8 (1.5–2.4) 0.98 (0.97–0.99)Arms FM (g) 4.3 (3.3–6.1) 0.93 (0.83–0.97) 7.3 (5.8–9.9) 0.96 (0.91–0.98)Arms %BF 3.0 (2.3–4.4) 0.93 (0.83–0.97) 6.8 (5.5–9.0) 0.95 (0.91–0.98)Legs FFSTM (g) 1.3 (1.0–1.8) 0.98 (0.96–0.99) 1.6 (1.3–2.1) 0.97 (0.95–0.99)Legs FM (g) 2.3 (1.8–3.4) 0.99 (0.98–1.00) 6.4 (5.1–8.6) 0.97 (0.95–0.99)Legs %BF 2.2 (1.7–3.2) 0.99 (0.98–1.00) 6.4 (5.1–8.4) 0.97 (0.94–0.98)Trunk FFSTM (g) 1.4 (1.1–2.0) 0.97 (0.93–0.99) 1.7 (1.4–2.2) 0.98 (0.95–0.99)Trunk FM (g) 4.5 (3.4–6.4) 0.98 (0.96–0.99) 7.5 (6.0–1.0) 0.97 (0.94–0.99)Trunk %BF 4.1 (3.2–6.0) 0.98 (0.95–0.99) 6.7 (5.4–9.0) 0.97 (0.95–0.99)

Note: FM – fat mass, FFSTM – fat-free soft tissue mass, BMC – bone mineral content, %BF – percentage body fat, %CV – coefficient ofvariation, ICC – intra-class correlation coefficient, 90%CL – confidence limits.

Table II. Comparison of different methods to assess body composition in Australian football players with DXA-derived fat mass as criterionmethod (mean, 90% CL).

N Mean ± s TEM (%) r CV (%) Standardised bias

%BF formula vs. pencil beam %BF 47 9.1 ± 1.0 vs. 7.6 ± 1.6 17.25 0.32 (0.10–0.51) 22.4 (19.1–27.3) 0.96 (moderate)%BF formula vs. fan beam %BF 47 9.1 ±1.0 vs. 8.5 ± 1.8 12.72 0.32 (0.11–0.51) 22.8 (19.5–27.8) 0.96 (moderate)Σ7 skinfolds vs. pencil beam %BF 47 46.2 ± 5.3 vs. 7.6 ± 1.6 101.62 0.71 (0.57–0.81) 15.8 (13.5–19.2) 0.71 (moderate)Σ7 skinfolds vs. fan beam %BF 47 46.2 ± 5.3 vs. 8.5 ± 1.8 97.25 0.66 (0.50–0.77) 17.3 (14.8–21.1) 0.76 (moderate)BIA %BF vs. pencil beam %BF 47 10.7 ± 1.9 vs. 7.7 ± 1.7 25.44 0.64 (0.43–0.79) 18.6 (15.2–24.1) 0.78 (moderate)BIA %BF vs. fan beam %BF 47 10.7 ± 1.9 vs. 8.9 ± 1.8 17.02 0.66 (0.46–0.80) 16.8 (13.7–21.7) 0.76 (moderate)Calibrated scale BM vs. pencil

beam BM (g)47 83, 856 ± 5828 vs.

83, 308 ± 58060.85 0.99 (0.99–0.99) 0.9 (0.8–1.1) 0.13 (trivial)

Calibrated scale BM vs. fanbeam BM (g)

47 83, 857 ± 5828 vs.84, 584 ± 5873

1.19 0.98 (0.97–0.99) 1.4 (1.2–1.7) 0.20 (small)

Note: BIA – bioelectrical impedance analysis.

Body composition and team sports 5

Dow

nloa

ded

by [

Mem

oria

l Uni

vers

ity o

f N

ewfo

undl

and]

at 1

8:38

17

July

201

4

coefficients for fat mass in the present study (r = 0.64–0.67) are similar to a previous report that compared fatmass from DXA to a four-compartment model(r = 0.54–0.62) (Santos et al., 2010). The exceptional-ly low fat mass in the current phantom (0.63–1.41 kg)may also have affected the poorer accuracy results.Together, the present results show the measurementaccuracy of these DXA devices to be poor for assessingfat mass in lean athletes, and such error should beconsidered carefully when assessing body compositionin highly trained athletes.

While the fat-free soft tissue mass measures in thisstudy showed excellent accuracy, the DXA devicesshowed moderate differences between fat mass mea-sures determined from the phantom. Notably, whilethe coefficients of variation for fat mass were similarbetween DXA units, the moderate differences in fatmass may be attributed to the different analysis soft-ware and algorithms used to determine the compart-mental masses (Malouf et al., 2013; Soriano et al.,2004). Additionally, variability in DXA estimatescan arise from technical error generated by theunit, failure to standardise the positioning of theparticipant in the scanning area and the biologicalvariation, including hydration status and the effectsof diet, exercise, food and fluid in the hours prior tothe scan (Nana et al., 2012). These findings showthat fat mass measures taken from different DXAshould not be used interchangeably and that serialmeasures of body composition with athletes shouldbe completed with the same device and software.

A secondary aim of this study was to validatecommon practical methods used to assess body com-position in team sport athletes using DXA-derivedfat mass. In agreement with previous studies on leanathletic populations (De Lorenzo et al., 2000;Fornetti et al., 1999; Silva, Fields, Quitério, &Sardinha, 2009), we observed poor validation coeffi-cients between DXA-derived fat measures and fatmeasures determined through BIA, skinfolds or esti-mated from sport-specific prediction equations. Forexample, De Lorenzo et al. (2000) reported signifi-cantly lower fat mass and higher fat-free soft tissuemass estimated from both skinfolds and BIA in com-parison to DXA derived results. Likely explanationsfor the poor relationships between the DXA andpractical methods in this study are the nature of thehomogenous group investigated, lack of standardisa-tion in hydration and rest status, the measurementnoise from the BIA methods (Lukaski, 2013) andnon-linear relationship between subcutaneous fatand fat mass – especially in very lean or obese indi-viduals (Roche, 1987). Specifically, high levels ofmusculoskeletal development and elevated bonemineral density in this trained population may resultin differing proportions of water, mineral and pro-tein to the normal population (Prior et al., 2001).

This may, in turn, affect the precision of skinfoldsand BIA methods. It is possible if the relationshipswere assessed in a more heterogeneous group with agreater range of body fat content that the validitycoefficient could be improved. On the basis of thepresent findings, we suggest that sum of seven skin-folds may be the most valid practical portable meth-ods for estimating fat content; however, werecommend against comparing fat estimation usingthese alternate methods with those taken from DXA.

Taken collectively, the present results show thatDXA provides a precise method for assessing bodycomposition in lean professional Australian Footballplayers. However, since the DXA fat-free soft tissuemass showed better precision than fat mass mea-sures, these variables may be more suitable for mon-itoring athlete’s responses to training and dietaryinterventions. The present results also revealedpoorer validity of BIA, sum of seven skinfolds anda skinfold-derived soccer-specific formula. Takentogether, we recommend that scientists can confi-dently assess fat-free soft tissue mass in highlytrained team sport athletes through both fan andpencil beams.

In the present study, the reliability of two DXAdevices was compared using two separate, matchedgroups for each DXA device. Results demonstratedthat fat-free soft tissue mass and bone mineral con-tent showed excellent reliability but slightly poorerreliability for fat mass, which compares well withprevious results using DXA in athletes (Nana et al.,2012; Stewart & Hannan, 2000). For example, Nanaet al. (2012) recently reported very low test–retesterror in fat mass (cyclists: 1.9%, strength: 2.5%), fat-free soft tissue mass (cyclists: 0.8%; strength: 0.6%)and bone mineral content (cyclists: 0.8%, strength:1.0%) assessed from a narrowed fan beam DXA in agroup of trained cyclists and strength athletes. Theregional reliability of each scanner in the presentstudy also compares well to previous researchersdisplaying excellent reproducibility from each device(De Lorenzo et al., 2000). Similarly, the test–retestreliability was slightly poorer in regional measurescompared to the whole body measures regardless ofDXA unit. It has been suggested that DXA measure-ment accuracy for fat mass is poorer in leaner indi-viduals (Toombs et al., 2012; Van Der Ploeget al., 2003), but the precision is comparable tonon-exercising populations (Toombs et al., 2012).

We also observed slightly poorer reliability mea-sures taken in the pencil beam DXA, which might beexplained by the need for manual handling from thetechnologist (Toombs et al., 2012). Indeed, the fanbeam software used in this allowed for more auto-mated analysis and required less technician interven-tion in the analysis, which could have influenced thereliability. On the basis of these observations, we

6 J.C. Bilsborough et al.

Dow

nloa

ded

by [

Mem

oria

l Uni

vers

ity o

f N

ewfo

undl

and]

at 1

8:38

17

July

201

4

recommend that to increase measurement precision,scientists adopt methods that require less interven-tion from technicians. Despite the slightly poorerreliability in the pencil beam technique, both DXAunits provided good reliability, particularly for fat-free soft tissue mass and bone mineral content.Indeed, the reliability of both DXA devices providedmuch better measurement precision than theBIA-derived fat mass measures. These observationsagree with many previous studies that have alsoshown poor measurement reliability of various BIAmethods for estimating fat mass in lean athletes (DeLorenzo et al., 2000; Fornetti et al., 1999). Thepresent reliability data also allow us to use theapproach of Batterham and Atkinson (2005) whouse a nomogram and the test %CV to determineappropriate sample sizes for single group design stu-dies with a statistical power of 90%. On the basis ofthe present results, we suggest that a sample of ~20 isadequate to detect a 5% change in fat mass usingboth DXA, while a sample of >400 would berequired to detect a similar change if BIA wasused. However, much lower samples (i.e. <5)would be required to detect similar changes in fat-free soft tissue mass, which are more likely related tophysical performance in team sports such asAustralian Football.

There are a few limitations of research design inthe present study that needs to be acknowledged.Although each of the groups were matched for phy-sical characteristics in the reliability assessmentbetween the DXA devices, to limit radiation expo-sure we had different participants in each group.This may have impacted on some of the differencesin reliability errors and makes precise comparison ofthe reliability measures between the fan beam DXAand pencil beam DXA difficult. Nonetheless, manysteps were taken to standardise and limit othersources of error (i.e. the same technologist per-formed and analysed all scans, all players werehydrated). However, the professional AustralianFootball players in the current study were not all ina fasted and rested state – this may have influencedboth the BIA and the DXA results. Other research-ers have suggested that athletes be scanned in afasted and rested state to minimise test error; how-ever, the logistics of scanning large groups (i.e., >45players) from the same club in a fasted and restedstate within a busy training schedule is impractical.Therefore, while acknowledging that the presentapproach may increase measurement error, we sug-gest that our approach may also explain why thereliability measures in some variables in the currentstudy are slightly higher than recent similar reportsin strength athletes and trained cyclists (Nana et al.,2012).

Summary

In summary, the present results show that both fanbeam DXA and pencil beam DXA units provide avalid and reliable measure of fat-free soft tissue massand bone mineral content in highly trained and leanAustralian Football players. However, we alsoobserved that both DXA devices provide poorermeasurement precision when assessing fat mass inlean athletes but have greater validity and reliabilitythan similar estimates derived from BIA and skin-folds. We recommend that while less expensive andmore easily accessible, the practical methods used inthis study may not be appropriate for monitoringchanges in body composition in highly trained teamsport athletes. Finally, on the basis of the presentfindings, we recommend that DXA-derived fat-freesoft tissue mass and bone mineral content measuresto be the most valid and reliable variables for assess-ment. Indeed, since small changes in fat-free softtissue mass has been shown to influence physicalperformance, it may be meaningful to regularlymonitor fat-free soft tissue mass via DXA to informthe training process in team sport athletes.

Funding

The Carlton Football Club supported this research.

References

Ackland, T. R., Lohman, T. G., Sundgot-Borgen, J., Maughan,R. J., Meyer, N. L., Stewart, A. D., & Müller, W. (2012).Current status of body composition assessment in sport:Review and position statement on behalf of the ad hoc researchworking group on body composition health and performance.Sports Medicine, 42(3), 227–249.

Avlonitou, E., Georgiou, E., Douskas, G., & Louizi, A. (1997).Estimation of body composition in competitive swimmers bymeans of three different techniques. International Journal ofSports Medicine, 18, 363–368.

Batterham, A., & Atkinson, G. (2005). How big does my sampleneed to be? A primer on the murky world of sample sizeestimation. Physical Therapy in Sport, 6, 153–163.

Behnke, A. R., & Royce, J. (1966). Body size, shape, and compo-sition of several types of athletes. Journal of Sports Medicine andPhysical Fitness, 6(2), 75–88.

Buehring, B., Krueger, D., Libber, J., Heiderscheit, B.,Sanfilippo, J., Johnson, B., & Binkley, N. (2014). Dual-energyx-ray absorptiometry measured regional body composition leastsignificant change: Effect of region of interest and gender inathletes. Journal of Clinical Densitometry, 17(1), 121–128.

Clark, R. R., Sullivan, J. C., Bartok, C. J., & Carrel, A. L. (2007).DXA provides a valid minimum weight in wrestlers. Medicineand Science in Sports and Exercise, 39(11), 2069–2075.

Clarys, J. P., Scafoglieri, A., Provyn, S., Louis, O., Wallace, J. A.,& De Mey, J. (2010). A macro-quality evaluation of DXAvariables using whole dissection, ashing, and computer tomo-graphy in pigs. Obesity, 18(8), 1477–1485.

De Lorenzo, A., Bertini, I., Iacopino, L., Pagliato, E., Testolin,C., & Testolin, G. (2000). Body composition measurement in

Body composition and team sports 7

Dow

nloa

ded

by [

Mem

oria

l Uni

vers

ity o

f N

ewfo

undl

and]

at 1

8:38

17

July

201

4

highly trained male athletes. A comparison of three methods.Journal of Sports Medicine and Physical Fitness, 40(2), 178–183.

Duthie, G. M. (2006). A framework for the physical developmentof elite rugby union players. International Journal of SportsPhysiology and Performance, 1(1), 2–13.

Egan, E., Wallace, J., Reilly, T., Chantler, P., & Lawlor, J. (2006).Body composition and bone mineral density changes during apremier league season as measured by dual-energy x-rayabsorptiometry. International Journal of Body CompositionResearch, 4(2), 61–66.

Fan, B., Lewiecki, E. M., Sherman, M., Lu, Y., Miller, P. D.,Genant, H. K., & Shepherd, J. A. (2008). Improved precisionwith Hologic Apex software. Osteoporosis International, 19(11),1597–1602.

Fornetti, W. C., Pivarnik, J. M., Foley, J. M., & Fiechtner, J. J.(1999). Reliability and validity of body composition measuresin female athletes. Journal of Applied Physiology, 87(3), 1114–1122.

Hagmar, M., Berglund, B., Brismar, K., & Hirschberg, A. L.(2013). Body composition and endocrine profile of maleOlympic athletes striving for leanness. Clinical Journal of SportMedicine, 23(3), 197–201.

Hopkins, W. G., Marshall, S. W., Batterham, A. M., & Hanin, J.(2009). Progressive statistics for studies in sports medicine andexercise science. Medicine and Science in Sports and Exercise, 41(1), 3–13.

Hunter, T. E., Suster, D., Dunshea, F. R., Cummins, L. J., Egan, A.R., & Leury, B. J. (2011). Dual energy x-ray absorptiometry(DXA) can be used to predict live animal and whole carcass com-position of sheep. Small Ruminant Research, 100(2–3), 143–152.

Libber, J., Binkley, N., & Krueger, D. (2012). Clinical observa-tions in total body DXA: Technical aspects of positioning andanalysis. Journal of Clinical Densitometry, 15(3), 282–289.

Lohman, T. G., Harris, M., Teixeira, P. J., & Weiss, L. (2006).Assessing body composition and changes in body composition:Another look at dual-energy x-ray absorptiometry. Annals of theNew York Academy of Sciences, 904(1), 45–54.

Lukaski, H. C. (2013). Evolution of bioimpedance: A circuitousjourney from estimation of physiological function to assessmentof body composition and a return to clinical research. EuropeanJournal of Clinical Nutrition, 67(Suppl. 1), SS2–SS9.

Malouf, J., DiGregorio, S., Del Rio, L., Torres, F., Marin, A. M.,Farrerons, J., … Domingo, P. (2013). Fat tissue measurementsby dual-energy x-ray absorptiometry: Cross-calibration of 3different fan-beam instruments. Journal of ClinicalDensitometry, 16(2), 212–222.

Nana, A., Slater, G. J., Hopkins, W. G., & Burke, L. M. (2012).Effects of daily activities on dual-energy x-ray absorptiometrymeasurements of body composition in active people. Medicineand Science in Sports and Exercise, 44(1), 180–189.

Pineau, J. C., Filliard, J. R., & Bocquet, M. (2009). Ultrasoundtechniques applied to body fat measurement in male andfemale athletes. Journal of Athletic Training, 44(2), 142–147.

Pludowski, P., Jaworski, M., Matusik, H., Kobylinska, M.,Klimek, P., & Lorenc, R. S. (2010). The evaluation of con-sistency between body composition assessments in pediatricpopulation using pencil beam and fan beam dual-energy x-ray absorptiometers. Journal of Clinical Densitometry, 13(1),84–95.

Prior, B. M., Modlesky, C. M., Evans, E. M., Sloniger, M. A.,Saunders, M. J., Lewis, R. D., & Cureton, K. J. (2001).

Muscularity and the density of the fat-free mass in athletes.Journal of Applied Physiology, 90(4), 1523–1531.

Roche, A. F. (1987). Some aspects of the criterion methods forthe measurement of body composition. Human Biology, 59(2),209–220.

Rothney, M. P., Martin, F. P., Xia, Y., Beaumont, M., Davis, C.,Ergun, D., … Rezzi, S. (2012). Precision of GE lunar idxa forthe measurement of total and regional body composition innonobese adults. Journal of Clinical Densitometry, 15(4), 399–404.

Santos, D. A., Silva, A. M., Matias, C. N., Fields, D. A.,Heymsfield, S. B., & Sardinha, L. B. (2010). Accuracy ofDXA in estimating body composition changes in elite athletesusing a four compartment model as the reference method.Nutrition & Metabolism, 7, 22.

Schoeller, D. A., Tylavsky, F. A., Baer, D. J., Chumlea, W. C.,Earthman, C. P., Fuerst, T., & Lohman, T. G. (2005). QDR4500A dual-energy x-ray absorptiometer underestimates fatmass in comparison with criterion methods in adults.American Journal of Clinical Nutrition, 81(5), 1018–1025.

Silva, A. M., Fields, D. A., Quitério, A. L., & Sardinha, L. B.(2009). Are skinfold-based models accurate and suitable forassessing changes in body composition in highly trained athletes?Journal of Strength and Conditioning Research, 23(6), 1688–1696.

Soriano, J. P., Ioannidou, E., Wang, J., Thornton, J. C., Horlick,M. N., Gallagher, D., … Pierson, R. N. (2004). Pencil-beamvs. fan-beam dual-energy x-ray absorptiometry comparisonsacross four systems: Body composition and bone mineral.Journal of Clinical Densitometry, 7(3), 281–289.

Stewart, A. D. (2001). Assessing body composition in athletes.Nutrition, 17(7–8), 694–695.

Stewart, A. D., & Hannan, W. J. (2000). Prediction of fat and fat-free mass in male athletes using dual x-ray absorptiometry as thereference method. Journal of Sports Sciences, 18(4), 263–274.

Stewart, A. D., Marfell-Jones, M., Olds, T., & de Ridder, H.(2011). International standards for anthropometric assessment.Lower Hutt: International Society for the Advancement ofKinanthropometry.

Stewart, A. D., & Sutton, L. (2012). Body composition in sport,exercise and health. Abingdon: Routledge.

Sutton, L., Scott, M., & Reilly, T. (2008, May 15–16). Validationof a new anthropometric equation for the prediction of body composi-tion in elite soccer players. Paper presented at the Proceedings ofthe 1st World Conference on Science and Soccer, Liverpool.

Toombs, R. J., Ducher, G., Shepherd, J. A., & De Souza, M. J.(2012). The impact of recent technological advances on thetrueness and precision of DXA to assess body composition.Obesity (Silver Spring), 20(1), 30–39.

Van Der Ploeg, G. E., Withers, R. T., & Laforgia, J. (2003).Percent body fat via DEXA: Comparison with a four-compart-ment model. Journal of Applied Physiology, 94(2), 499–506.

Williams, J. E., Wells, J. C., Wilson, C. M., Haroun, D., Lucas,A., & Fewtrell, M. S. (2006). Evaluation of lunar prodigy dual-energy x-ray absorptiometry for assessing body composition inhealthy persons and patients by comparison with the criterion4-component model. American Journal of Clinical Nutrition, 83(5), 1047–1054.

Wong, W. W., Hergenroeder, A. C., Stuff, J. E., Butte, N. F.,Smith, E. O. B., & Ellis, K. J. (2002). Evaluating body fat ingirls and female adolescents: Advantages and disadvantages ofdual-energy x-ray absorptiometry. The American Journal ofClinical Nutrition, 76(2), 384–389.

8 J.C. Bilsborough et al.

Dow

nloa

ded

by [

Mem

oria

l Uni

vers

ity o

f N

ewfo

undl

and]

at 1

8:38

17

July

201

4