Paul Sweeney Final year research project

A Comparison of Two Bioelectrical

Impedance Analysis Modes with DXA

for Estimating the Body Composition of

Elite Inter-County GAA Athletes

Name: Paul Sweeney

Student I.D: 11111321

A thesis submitted to the University of Limerick in fulfilment of the requirements

for the Degree of Bachelor of Science in Sport and Exercise Sciences

Department of Physical Education & Sport Sciences

Head of Department: Dr. Ann McPhail

Supervisor: Professor Phil Jakeman

Submitted: April 2015

.

i

Authors Declaration

I hereby declare that the work contained within this thesis is my own work, and was

completed without collaboration or assistance from others apart from the counsel

received from my supervisors, Name and department. This work has also not been

submitted to any other University of higher education institution, or for any other

academic award within this University.

Name

Date

Name

Date

ii

Acknowledgements

I would like to thank my parents Pat and Mary, for their great support throughout over

the years. They did well to put up with me during stressful times.

I would like to thank Linda, Anne-Marie, David, and Callum for their support and help

throughout what has been a demanding final year.

I would like to thank my supervisor Phil Jakeman for his advice and helpful knowledge

during this project.

I would like to thank Will McCormack and Katie Hughes for their advice during this

project. Thank your for responding to my many emails and always being available to

chat when needed.

Thanks to the fourth year Sport and Exercise Science class for a great four years. Best

of luck in the future.

To Murray, Tinny, Norris, O’Hare, and Leacy. The amount of craic that has been had

over the last few years will never be topped and for that I thank ye.

A special thanks must go to Tinny for the grinds in excel over the past year.

iii

Table of CoFntents Authors Declaration ........................................................................................................ i

Acknowledgements ....................................................................................................... ii

List of Figures ................................................................................................................ v

List of Tables ............................................................................................................... vii

Abbreviations ............................................................................................................. viii

Abstract ........................................................................................................................ ix

Chapter 1 - INTRODUCTION ....................................................................................... 1

Chapter 2 – LITERATURE REVIEW ............................................................................ 3

2.1 What is Body Composition? ................................................................................................ 3

2.2 Why Measure the Body Composition of Athletes? ............................................................. 3

2.2.1 Physical Demands of GAA ............................................................................................ 3

2.2.2 Body Composition and Athletic Performance .............................................................. 4

2.2.3 Seasonal Variations in Body Composition .................................................................... 6

2.3 Methods of Body Composition Assessment ....................................................................... 7

2.3.1 DXA ............................................................................................................................... 7

2.3.2 BIA ................................................................................................................................ 9

2.3.3 Problems Associated with BIA.................................................................................... 10

2.3.4 BIA vs. DXA ................................................................................................................. 10

2.4 Conclusion ......................................................................................................................... 12

Chapter 3 - METHODS ...............................................................................................13

3.1 Participants ....................................................................................................................... 13

3.2 Preparation ....................................................................................................................... 13

3.3 Procedures ........................................................................................................................ 13

3.3.1 Anthropometric Measurements ................................................................................ 13

3.3.2 Bioelectrical Impedance Analysis (BIA) ...................................................................... 13

3.3.3 Dual Energy X-ray Absorptiometry (DXA) .................................................................. 14

3.4 Statistical Analysis ............................................................................................................. 14

Chapter 4 – RESULTS ................................................................................................16

4.1 Descriptive Statistics ......................................................................................................... 16

4.2 BIA Athlete and Normal mode vs. DXA Analysis ............................................................... 16

4.2.1 Overview .................................................................................................................... 16

iv

4.2.2 BFM ............................................................................................................................ 16

4.2.3 LTM ............................................................................................................................. 16

4.2.4 BMC ............................................................................................................................ 17

4.2.5 FFM ............................................................................................................................. 17

4.2.6 BF% ............................................................................................................................. 17

4.2.7 Athlete vs. Normal mode ........................................................................................... 17

Chapter 5 – DISCUSSION ..........................................................................................23

5.1 Background and Purpose .................................................................................................. 23

5.2 Findings ............................................................................................................................. 23

5.3 Systematic Errors .............................................................................................................. 24

5.4 BIA Athlete vs. Normal mode ............................................................................................ 26

5.5 Limitations ......................................................................................................................... 26

Chapter 6 – CONCLUSION .........................................................................................27

6.1 Summary and Future work................................................................................................ 27

References ..................................................................................................................28

Appendices .................................................................................................................. A

Appendix A1 .............................................................................................................................. A

Appendix A2 .............................................................................................................................. A

Appendix A3 .............................................................................................................................. A

Appendix A4 .............................................................................................................................. A

Appendix A5 .............................................................................................................................. A

Appendix A6 .............................................................................................................................. A

v

List of Figures

Figure 4.1 - Bland Altman plots of Body Fat Mass (kg) with mean difference (dotted

lines) and 95% limits of agreement (dots) comparing Athlete and Normal modes to

DXA………………………………………………………………………………………….....18

Figure 4.2 - Bland Altman plots of Lean Tissue Mass (kg) with mean difference (dotted

lines) and 95% limits of agreement (dots) comparing Athlete and Normal modes to

DXA……………………………………………………………………………………..….…..19

Figure 4.3 - Bland Altman plots of Fat Free Mass (kg) with mean difference (dotted

lines) and 95% limits of agreement (dots) comparing Athlete and Normal modes to

DXA……………………………………………………………………………..………….…..20

Figure 4.4 - Bland Altman plots of Lean Tissue Mass (kg) with mean difference (dotted

lines) and 95% limits of agreement (dots) comparing Athlete and Normal modes to

DXA…………………………………………………………………………………..….……..21

Figure A1 - Lunar iDXA scanner (GE Healthcare, Chalfont St Giles, Bucks.,

UK)…………………………………………………………………………………………..…A2

Figure A2 - Fundamental principle of

DXA…………………………………………………………………………………………….A2

Figure A3 - Tanita MC-180MA Body composition Analyser (Tanita UK

Ltd.)…………………………………………………………………………………….………A3

Figure A4 - Bland Altman plots of Bone Mineral Content (kg) with mean difference

(dotted lines) and 95% limits of agreement (dots) comparing Athlete and Normal modes

to

DXA…………………………………………………………………………………………….A4

Figure A5 - Bland Altman plots of body fat mass (kg) with mean difference (dotted

lines) and 95% limits of agreement (dots) (Athlete vs.

Normal)……………………………………………………………………….………………..A5

Figure A6 - Bland Altman plots of lean tissue mass (kg) with mean difference (dotted

lines) and 95% limits of agreement (dots) (Athlete vs.

Normal)………………………………………………………………………………….……..A5

Figure A7 - Bland Altman plots of fat free mass (kg) with mean difference (dotted lines)

and 95% limits of agreement (dots) (Athlete vs.

Normal)……………………………………………………….………………………………..A6

vi

Figure A8 - Bland Altman plots of fat free mass (kg) with mean difference (dotted lines)

and 95% limits of agreement (dots) (Athlete vs.

Normal)……………………………………………………………………….………………..A6

vii

List of Tables

Table 4.1 - Anthropometrics for the 157 GAA players included in this study. Data are

reported as mean (standard deviation (SD)), median (interquartile range (IQR)) and

range (max – min); n = 157)

………………………………………….……………………..……………………..………15

Table 4.2 DXA and BIA athlete (a) and normal (n) measured body fat mass (BFM),

lean tissue mass (LTM), bone mineral content (BMC), fat free mass (FFM), and body

fat % (BF%), for all subjects (n = 157)

………………………………………………………………………………………..……..17

viii

Abbreviations

Air Displacement Plethysmography - ADP

Bioelectrical Impedance Analysis - BIA

Bioelectrical Impedance Spectroscopy - BIS

Body Fat Mass - BFM

Body Fat Percentage - BF%

Bone Mineral Content - BMC

Coefficient of Variance - CV

Correlation Coefficient - CC

Standard Error of Estimate - SEE

Limits of Agreement - LoA

Standard Deviation - SD

Dual Energy X-ray Absorptiometry - DXA

Extracellular Water - ECW

Fat Free Mass - FFM

Four Compartment model - 4-C model

Gaelic Athletic Association - GAA

Hydrostatic Weighing - HW

Intracellular Water - ICW

Lean Tissue Mass - LTM

Multi Frequency Bioelectrical Impedance Analysis - MF-BIA

National Collegiate Athletic Association - NCAA

Single Frequency Bioelectrical Impedance Analysis - SF-BIA

Total Body Water – TBW

ix

Abstract

A Comparison of Two Bioelectrical Impedance Analysis Modes with DXA for

Estimating the Body Composition of Elite Inter-County GAA Athletes

Student name Paul Sweeney Supervisor Professor Phil Jakeman

Background: The Tanita MC-180MA body compositional analyser is a multi frequency bioelectrical impedance analysis (BIA) device used within clinical settings. The equations provided by the manufacturer utilise gender, height, body mass, age, body type category (Athlete or Normal) and measured impedance to obtain body composition estimations. Purpose: The purpose of this study was to investigate which mode (Athlete or Normal) was demonstrated better agreement in estimating the body composition in elite inter-county GAA (football and hurling) players, compared to dual energy x-ray absorptiometry (DXA) as the reference method. Methods: One hundred and fifty seven inter-county GAA players aged 19-40 were recruited from five county teams in Ireland. For each subject estimates of body fat mass (BFM), lean tissue mass (LTM), bone mineral content (BMC), fat free mass (FFM), and body fat percentage (BF%) were taken by BIA in Athlete and Normal mode and DXA. Results: Both BIA modes demonstrated good relative agreement with DXA for all body composition measures. In absolute terms, there were significant differences observed between Athlete mode and DXA for all body composition variables, underestimating BFM, BMC, and BF% while overestimating LTM and FFM with large biases and wide limits of agreement found. No significant differences were observed between Normal mode and DXA for BFM and BF%, however underestimations were shown for LTM, BMC and FFM. Bias was smaller and limits of agreement were narrower in Normal mode compared to DXA. Conclusion: Compared with DXA, Normal mode displayed better accuracy than Athlete mode in estimating the body composition of elite GAA players. In absolute terms, Athlete mode provided large biases and wide limits of agreement for all body composition measures. Normal mode showed smaller biases, indicating that it may be used interchangeably with DXA for group measurements, however wide limits of agreement suggest that results of body composition assessments on individuals should be analysed with caution.

1

Chapter 1 - INTRODUCTION

Body composition is an important component of health and physical fitness that can

influence the performance of athletes (Rodriguez, 2009). Accurate assessments of

body composition are necessary in order to monitor training and nutritional status of

athletes (Moon, 2013). Strength and conditioning coaches can use body composition

measurements to evaluate the effectiveness of specific training programmes (Moon

2013). Sports nutritionists can also utilise body composition results to establish

personalised dietary interventions for their athletes (Esco et al., 2014; Segal, 1996).

Furthermore, body composition values can assist medical personnel in monitoring an

athlete’s physical and mental health, as radical changes in body composition can

indicate underlying health concerns (Fornetti et al., 1999). The Gaelic Athletic

Association (GAA) is the largest sporting organisation in Ireland and is comprised of

five distinctive sports; Gaelic football, hurling, camogie, handball and rounder’s.

Football and hurling are the most popular of these sports with an estimated 15% of

adult males participating in both codes in Ireland (Delaney and Fahey, 2005). Both

football and hurling are physically demanding contact sports in which high levels of

muscular strength, power and speed are advantageous (McIntyre et al., 2005; Reilly

and Doran, 2001). Therefore, the development of lean tissue mass (LTM) is desirable

as it is seen to enhance strength, power, and speed (Rodriguez, 2009). In contrast,

higher amounts of body fat mass (BFM) are detrimental to athletic performance,

increasing energy expenditure and reducing power to weight ratio, speed and

acceleration (Duthie et al., 2006; Sventesson et al., 2008; Harley et al., 2011).

There are many methods available for assessing and monitoring the body composition

of athletes. Laboratory methods include Dual Energy X-ray Absorptiometry (DXA),

Hydrostatic Weighing (HW) and Air Displacement Plethysmography (ADP), whereas

Bioelectrical Impedance Analysis (BIA) and Anthropometry are commonly used in field

settings (Ackland et al., 2012). DXA is frequently used method for body composition

analysis in clinical and sport settings and is considered to be a reliable and valid

method of assessing body composition in athletes (Buehring et al., 2014; Bilsborough

et al., 2014). DXA allows for a minimally invasive measurement of the three-

compartment model of body composition, consisting body fat mass (BFM), and two

components of fat free mass (FFM), i.e. LTM and bone mineral content (BMC).

Although DXA is an accepted technique for body composition measurement, it is

expensive and impractical, as each scan must be conducted by trained personnel.

Most inter-county GAA teams would not have time for, or access to DXA, therefore field

2

methods are often preferred within this population due their low cost and high

practicality.

BIA has emerged as a popular field alternative to DXA for use within athletic

populations (Moon, 2013), as the method is user friendly, inexpensive, and requires no

specialised facilities or expertise to take the measurements (Sillanpaa et al., 2014). BIA

estimates body composition by applying an electric current through the human body

measuring resistance and reactance (Kushner, 1992). The resistance measured as

well as gender, height, and body mass are then integrated into a regression equation

from which BFM, FFM and TBW can be estimated. The regression equations are

usually specific to the population for which they were established therefore the choice

of equation is important (Swartz et al. 2002). Elite athletes engage in rigorous training

and tend have leaner physiques (different body types) than non-athletes (Prior et al.,

2001) thus equations derived from sedentary populations may not be suitable (Swartz

et al., 2002). Recommended prediction equations have been developed for athletes in

the past (Yannakoulia et al., 2000; Oppliger et al., 1992), however their validity in

athletic populations is still unknown.

In a bid to improve the accuracy of body composition measurements in individuals with

different body types, some BIA systems have incorporated two programmed

algorithms, one for athletes (Athlete mode) and one for the non-athletes (Normal

mode). The choice of mode is based on the volume of exercise performed by the

subject per week. Although limited research has been conducted investigating the

accuracy of these settings on BIA devices in athletes, a study by Swartz et al. (2002)

examined whether the choice of BIA algorithm altered body composition estimates

compared to HW in a cohort of highly active, moderately active, and inactive young

men. “Normal adult” mode was found to overestimate BF% (4.5 - 5%) and

underestimate FFM (3.5 - 4 kg) in subjects who participated in greater than 2.5 hours of

exercise per week. Conversely, the “athlete” mode underestimated BF% (4.5%) and

overestimated FFM (3.7 kg) in individuals who participated in less than 2.5 hours of

exercise. The results of the above study emphasise the need for population specific

BIA equations to be created in order to accurately assess body composition of athletes.

The purpose of this study was to investigate which mode (Athlete or Normal) on the

Tanita MC-180MA Multi-Frequency BIA displays the most agreement in measuring

body composition components in elite inter-county GAA players compared to DXA as

the reference method. A secondary purpose was determine whether choosing BIA in

Normal vs. Athlete mode significantly changes the output of the components of body

3

composition. It was hypothesized that Athlete mode would display better agreement

than Normal mode for estimating the body composition of inter-county GAA players

compared to DXA as the criterion method

Chapter 2 – LITERATURE REVIEW

2.1 What is Body Composition?

Body composition has been described as “the chemical or physical components that

collectively make up an organisms mass, defined in a systematic way” (Stewart, 2010).

Body composition can be divided into a model consisting of 5 complex levels, (i)

Atomic; (ii) Molecular; (iii) Cellular; (iv) Tissue System; (v) Whole-Body. The majority of

research conducted on athletic populations has focused on investigating how the

quantity and distribution of molecular components including BFM and FFM (LTM, BMC,

and total body water (TBW)) can influence athletic performance (Malina et al., 2007)

2.2 Why Measure the Body Composition of Athletes?

Body composition plays an important role in the health and performance of an athlete

(Ackland et al., 2012). The measurement of body composition and changes in body

composition over time has many important applications to athletes and sport and

exercise science practitioners. Measurements can be analysed for many purposes

including, monitoring the success of training programs (Moon, 2013), establishing

individualised dietary interventions from estimating energy expenditure (Segal, 1996),

and evaluating the physical and mental well-being of the athletes (Fornetti et al., 1999).

Consequently, it is important for strength and conditioning coaches, nutrition experts

and health care practitioners working with athletes, that reliable, accurate and relatively

inexpensive methods for body compositional analysis are available.

2.2.1 Physical Demands of GAA

In order to play Gaelic football and hurling at a high level (i.e. inter-county), one must

display high levels of physical ability. Both codes are physically demanding contact

sports, characterised by intermittent changes of pace with anaerobic bouts overlapping

on moderate aerobic activity (Reeves and Collins, 2005). In a recent study on elite

adolescent GAA athletes, Cullen et al. (2013) stated that successful performance

required athletes to display several fitness components including high levels of

muscular strength, power, and speed, while also relying on the anaerobic and aerobic

systems. While literature investigating the physiological demands of hurling is limited,

research has shown that elite Gaelic footballers exhibit similar fitness profiles to

4

professional soccer and Australian football athletes. (Cullen et al., 2013; Reilly and

Collins, 2008; Reeves and Collins, 2005; McIntrye, 2005). Therefore, due to the highly

physical and intense nature of the sport it is reasonable to assume that body

composition influences performance.

2.2.2 Body Composition and Athletic Performance

Numerous studies have investigated the relationship between body composition and

physical performance in lacrosse, soccer, and ice hockey; yet, there are no studies

examining this relationship within the GAA. The majority of research has investigated

the effects of BFM on physical performance with a vast amount of literature supporting

the notion that excess body fat negatively impact’s athletic performance (Ackland et al.,

2012; Rodriguez, 2009; Duthie et al., 2006).

Excess BFM is suggested to be particularly detrimental to athletes participating in

sports which involve activities requiring the maximal displacement of one’s body

through space (i.e. sprinting, running, and jumping). This is because the BFM acts as

extra weight that must be propelled against gravity, negatively effecting an athlete’s

acceleration, speed, and power to weight ratio, while also leading to an increase in

energy expenditure (Harley et al., 2008; Svantesson et al., 2008; Malina and Geithner,

2007). A study by Matilla et al. (2007) found increased BFM to be strong predictor of

aerobic performance in a sample of 140 Finish conscripts. The authors documented

inverse relationship between adipose tissue and aerobic capacity, in that for every 1%

increase in BF%, there was a 19.3 meter reduction in coopers test running distance.

Similar findings were reported in a study conducted by Collins et al. (2014)

investigating the relationship between body composition and performance tests in 54

collegiate level lacrosse athletes. The subjects underwent body composition

assessments by air-displacement plethysmography (ADP) before participating in a

battery of tests, measuring maximum power production (one repetition maximum power

clean), upper-body muscular endurance (body weight bench press and dips to failure),

and both aerobic (one mile run) and anaerobic capacity (300 yard shuttle). Moderate

correlations were found between increased BF% and number of bench press and dips

repetitions (upper body muscular endurance) (r = -0.36) and one mile run times

(aerobic capacity) (r = 0.44), while a strong relationship existed between increased

BF% and 300 yard shuttle test time (anaerobic capacity) (r = 0.69).

It can therefore be suggested that in biomechanical terms, increased BFM acts as

ballast, that can lead to negative aerobic and anaerobic performance outcomes

(Ackland et al., 2012). The findings of Collins et al. (2014) and Matilla et al. (2007) are

5

in agreement with research by Potteiger et al. (2010) who investigated predictors of ice

skating performance in 21 elite ice hockey players. Body composition was assessed by

means of air displacement plethysmography (ADP). On-ice skating performance was

measured during 6 timed 89 m sprints with the results showing BF% to be moderately

correlated to average skating time (r = 0.57), such that greater relative fat levels were

associated to slower skating speed. Based on the findings from the above studies it is

clear BFM has a negative effect on many key fitness components related to hurling and

football including speed, and both aerobic and anaerobic capacities. This further

emphasises the importance of accurate body composition assessment methods for

practitioners within the sports.

Although the measurement of BFM and its effect on athletic performance has been the

main focus within the literature, many studies have also investigated the influence of

FFM components. LTM is of interest for athletes and sports practitioners, because like

BFM, its quantity and distribution is said to effect the performance of physical tasks.

This is because skeletal muscle is the tissue responsible for force production within the

body with a direct relationship existing between skeletal muscle cross-sectional area

and force generation (Ackland et al., 2012). Therefore, a phenotype displaying high

proportions of LTM is desirable for elite inter-county hurlers and footballers, as this is

directly related to higher power and strength-to-weight ratios, as well as enhanced

acceleration, speed, power, strength, and endurance (Svantesson et al., 2008; Duthie

et al., 2006). Numerous studies have been conducted on athletic populations

investigating the association between LTM and physical performance. Matilia et al.

(2007) investigated the relationship between fitness and physical performance in 140

conscripts and found lower body LTM to predict lower body explosive power in the form

of a standing broad jump test (r2 = 0.231). However, no relationship existed between

LTM and muscular strength or aerobic capacity. This was thought to be because

muscular strength was assessed using tests that involved using the subjects own body

mass as the external load (sit-ups, push-ups, back extensions, and pull-ups). The

authors concluded that they may have obtained different results it the administered

strength tests consisted of exercises involving pushing resistance away from the body

(e.g. 1 repetition maximum bench press). This may be due to the fact that absolute

FFM levels play an important role in the performance of tasks involving the projection

of objects or the movement of another individual (breaking tackles in hurling and

football) (Malina and Geitner, 2011). In contrast, a study recently conducted by

Hogstrom et al. (2012) on 48 male and female adolescent cross-country skiers (aged

15-17), showed LTM to influence aerobic capacity, reporting a positive moderate

6

association between LTM%, weight adjusted maximal oxygen uptake thresholds (VO2

max), and the onset of blood lactate accumulation (r = 0.47–0.67, p < 0.05).

Furthermore, when assessing differences in physical fitness amongst female wrestlers

Pallares et al. (2012) found elite wrestlers had higher levels of FFM than amateurs,

thus further supporting the suggestion that FFM influences athletic performance. As the

majority of studies have investigated the effect of BFM on physical ability, more

research is warranted relating to FFM components.

2.2.3 Seasonal Variations in Body Composition

It is clear that body composition is an important determinant of performance therefore it

is important to assess body composition changes that occur throughout a season and

how these variations effect athletic performance. Studies on soccer, rugby, and rugby

league have investigated the body composition of athletes at different points of the

season and have noted intra-seasonal changes caused by a number of factors

including injury, illness, and dietary practices (Harley et al., 2011; Carling and Orhant,

2010; Silvestre et al., 2006; Duthie et al., 2005). Therefore, monitoring body

composition at different phases of a season may help players avoid any adverse

variations while also providing target values to achieved by the player through training

and nutritional intervention after a period of injury (Harley et al., 2011). Research also

suggests that a GAA player’s body composition tends to change throughout the course

of a season due to the physical effects of training and the stage of competition reached

(Reilly and Keane, 2001; Reilly and Doran, 2001). Much of the literature investigating

soccer has noted beneficial changes in body composition over the course of a

competitive season. Osteojic and Zivanic (2003) assessed body composition

alterations using skinfold measurements in thirty male professional soccer players and

while there was no changes noted in FFM, the results displayed a reduction in BFM

expressed as BF%, between the beginning and the end of the season. This was in

conjunction with the findings from a similar study by Casajus et al. (2001) who found a

significant decrease in BFM expressed as BF% (8.6 ± 0.91 %FM to 8.2 ± 0.91 %FM)

derived from skinfold measurements in 15 elite soccer players between the beginning

(September), and middle (February) of the season. Furthermore, Silvestre et al. (2006)

investigated seasonal body composition changes on 25 male elite level collegiate

soccer players using DXA at the beginning (pre) and end (post) of a National Collegiate

Athletic Association (NCAA) season. The authors observed a significant increase in

whole body (0.9 ± 0.2 kg), and regional LTM in the legs (0.4 ± 0.0 kg) and trunk (0.3 ±

0.1 kg) from pre to post season phases. In contrast to the above findings, Harley et al.

(2011) reported detrimental body composition changes in elite rugby league players

7

during the competitive season. The authors performed measurements using DXA at

three intervals throughout the competitive season: PRE (end of pre-season), MID

(Middle of competitive season), and POST (a week after the conclusion of the season),

and found among other variables, a significant reduction in absolute LTM (-1.17 ± 1.33

kg) and BFM (0.90 ± 1.14 kg) from PRE to POST. The literature discussed in the above

paragraphs further demonstrates changes that occur throughout the course of a

sporting season once again highlighting the need for a quick, simple and reliable

method of body comp assessment in the GAA. No studies of this kind have been

conducted on a population of GAA athletes therefore, further research into seasonal

variations is required.

2.3 Methods of Body Composition Assessment

Numerous techniques to have been developed to estimate body composition and can

be divided into reference, laboratory, or field categories (Ackland et al., 2012).

Reference methods are the most accurate techniques to which all other methods are

compared. The four compartment (4-C) body composition model is considered the true

reference method for body composition assessment as it can provide estimates of FFM

without making assumptions relating to the density and hydration of individuals

(Toomey et al., 2015). The model is determined using a combination of techniques,

separating body mass into BFM, TBW, bone, and protein. BFM is measured by

hydrodensitometry, bone mass by DXA, TBW by isotope dilution, and protein from the

residual (Toomey et al., 2015). However, issues relating to time, cost, and accessibility,

limit its application within sporting populations (Santos et al. 2010). Laboratory methods

for body composition assessment include DXA, HW and ADP, while BIA and skinfolds

are commonly used field methods (Ackland et al., 2012). HW and ADP are based on

the two-compartment model of body composition measuring BFM and FFM by

estimating whole body density and relating it to BF%. For HW, the subject is

submerged in water and body density is calculated by dividing their body mass by the

volume. ADP is similar to HW however, density is estimated using a highly pressurised

air capsule and not underwater (Ackland et al., 2012). DXA and BIA were the two

techniques used within the current study therefore they will be discussed in greater

detail in the preceding paragraphs.

2.3.1 DXA

DXA is a minimally invasive three-compartment model of body compositional

assessment that estimates whole body and segmental BFM, as well as two FFM

components LTM and BMC. DXA offers many advantages to other laboratory

techniques for athletes as measurements are relatively quick (5-8 minutes), precise,

8

and use low levels of radiation (Ackland et al., 2012). DXA operates by transmitting x-

rays through the body at two separate photon energies, one low and one high (i.e. 40

and 70 KeV). The x-ray beams travel through tissues within the body and are

attenuated depending on the physical make up (density and thickness) of the tissues

they pass through (Toombs et al., 2012). Soft tissues (fat, muscle, water) are lower in

density than hard tissues (bone) and therefore allow more photons to travel through,

decreasing attenuation. DXA distinguishes between BF and FFM by calculating the

ratio of low-to-high photon energy attenuation in the soft tissue (Pietrobelli et al., 1996)

(Fig. 1). (Appendix 1)

A vast amount of literature has investigated the validity of DXA against a four-

compartment (4-C) model for assessing body composition within athletic populations

(Santos et al., 2010; Withers et al., 2004; Prior et al., 1997). Many studies have found

mean differences between the methods in BF% ranging from -3.5% to 2.9%, with the

majority showing larger underestimations of relative and absolute BFM in leaner

individuals (Toombs et al., 2012). The accuracy and precision of DXA is said to vary

depending on the tissue measured, with values for lean mass demonstrating better

accuracy and precision than BFM (Toombs et al., 2012). Santos et al. (2010)

investigated the accuracy of fan DXA compared to a 4-C model in 27 elite male judo

athletes. The results showed that on a group level, DXA provided better estimations of

FFM (r > ~0.95, Standard error of estimate (SEE) <1.98, Limits of Agreement (LoA);

0.6 kg to -7.0 kg) than BFM (r > ~0.78, SEE <2.6, LoA; 6.2 kg to -1.1 kg) and BF% (r >

~0.72, SEE <2.65, LoA; 8.8 to -2.4). This was indicated by the higher correlation

coefficient (CC), lower SEE and tighter LoA. Additionally, on an individual level large

differences were found between DXA and 4-C for all variables. Similarly, Bilsborough et

al. (2012) conducted a study on thirty-six elite Australian football players and found a

fan DXA to provide greater accuracy and precision for estimates of LTM and BMC,

than BFM. DXA measures were compared to a whole body phantom to assess

accuracy, whereas the athletes completed two separate scans under the same

conditions to determine DXA precision. DXA showed better accuracy for estimating

LTM and BMC (r = 0.98-1.00) than BFM (r = 0.39-0.84) showing stronger correlations

with the reference method. Furthermore, precision was higher for LTM and BMC (%CV

0.3%-0.6%) than BFM (%CV = 2.5%). The authors suggested that these findings could

have stemmed from physiological variations within participants, as the conditions were

not adequately controlled. The results from aforementioned studies demonstrate that

measures obtained from DXA should be analysed with caution especially for BFM in

athletes at both a group and individual level. Although there remains some uncertainty

9

about the accuracy of DXA obtained body composition measures, it has been found to

demonstrate similar results compared to other laboratory techniques (i.e. densitometry)

(Kohrt, 1998). Therefore, until the creation of a practical 4-C model for athletes, DXA is

the preferred method, due to its speed, practicality and precision (Ackland et al., 2012)

2.3.2 BIA

BIA is a popular field method of assessing body composition that has been widely used

in athletic populations due to its cost, accessibility, and practicality. BIA is based on a

three-component model providing estimations of FFM, BFM, and TBW (Moon, 2013).

BIA devices transmit harmless electrical currents through the body in order to calculate

impedance, (i.e. resistance and reactance of the current) (Kyle et al., 2004; Bolanowski

and Nilsson, 2001; Kushner, 1992). BIA operates on the principle that electrical

currents flow through body tissues at different velocities depending on their

composition, with the most resistance provided by BFM and the least by FFM as it is

rich in water and electrolytes (Kushner, 1992). Assuming TBW makes up a fixed

percentage of FFM (73%), body composition values can be estimated from specific

regression equations based on gender, height, body mass, and calculated resistance

(Pateyjohns et al., 2006; Kyle et al., 2004). Many choices of BIA systems are now

commercially available. Early BIA methods utilised a single frequency current (SF-BIA)

of typically 50kHz, travelling between surface electrodes placed on the hand and foot of

the subject to estimate the body composition of an individual (Kyle et al. 2004).

However, research investigating the accuracy of SF-BIA systems in athletic populations

has shown conflicting results, with some authors reporting good accuracy (Yannkoulia

et al., 2000; Fornetti et al., 1999) and others reporting poor accuracy (Esco et al.,

2011). These contradictory findings may have occurred due to the fact that single low

frequency currents (<100KHz) cannot fully penetrate through cell membranes and

therefore are unable to predict the concentration of intracellular water (ICW), and in

turn total body water (TBW) (Shafer et al., 2009). Advancements in technology have

led to BIA devices being developed using multiple frequency currents (5 to 500kHz)

(Silanpaa et al., 2014; Scharfetter et al., 2001). These Multi-Frequency BIA devices

(i.e. MF-BIA, Bioelectrical Impedance Spectroscopy (BIS)) are deemed to be more

accurate in determining distribution of ICW and extracellular water (ECW) and

therefore may be preferred to SF devices in the estimation of FFM (Matthie, 2008).

However, it has not yet been determined which method provides the most accuracy

(Kyle et al., 2004), therefore this topic requires more research.

10

2.3.3 Problems Associated with BIA

A problem associated with the use of BIA within athletic populations is its high

sensitivity to variations in hydration status. BIA estimates body composition by

assuming that FFM comprises of 73% TBW, thus, changes in hydration levels can lead

to prediction errors in body composition parameters. Elite GAA athletes perform

multiple bouts of acute exercise each week, and are therefore more susceptible to

greater hydration variation than the normal individual (Segal, 1996). These individual

acute exercise bouts may lead to loses in TBW through sweating causing increased

BIA measured resistance, and in turn falsely underestimates FFM and overestimates

BFM (Segal, 1996). Saunders et al. (1998) investigated the effects of altered hydration

on BIA in 15 endurance athletes aged 19 to 56 years. The results showed that hypo-

hydration induced by exercise, was incorrectly interpreted as changes in the athletes

BFM. This was supported by Frisard et al. (2005) who suggested that BIA

overestimates FFM and underestimates BFM in overly hydrated individuals, and

underestimates FFM and overestimates BFM in those who are dehydrated. The above

findings emphasise the importance of strict adherence to pre-test guidelines (i.e.

fasting, exercise avoidance) in order for accurate body composition values to be

obtained in athletes (Moon, 2013).

BIA regression equations are usually population specific, therefore choosing the correct

equation is of fundamental importance. To date, no generalised equation with a valid

estimation of TBW exist for use on athletic populations, which could cause inaccurate

FFM values due to the variability of FFM hydration in athletes (Moon, 2013). This is a

major limitation of BIA use in athletic populations and so some BIA devices have

developed two pre-programmed algorithms in their devices, one for athletes, and one

for non-athletes. It has not yet been established which mode provides more accurate

results and hence the current study was conducted.

2.3.4 BIA vs. DXA

The majority of research comparing BIA to DXA for body composition assessment has

been conducted on non-athletic populations with only one study within the literature

comparing the Tanita MC-180MA MF-BIA to DXA. This was conducted by Leahy et al.

(2012) on a large cohort (n = 403) of healthy men and woman aged 18-29 years. The

authors found BIA to underestimate median BFM (1.3 kg) and BF% (2.1%), while

overestimating FFM (1.5 kg) (p < 0.05) compared to DXA in all subjects. When the

biases were investigated further, the authors observed that the underestimations

became more apparent as absolute and relative fat tissue levels increased, whereas,

the overestimation of FFM remained constant across the spectrum of values. Similarly,

11

Sillanpaa et al. (2014) studied 882 mixed gender adults aged 18-88 years and found

MF-BIA to underestimate BFM (2.9 kg and 1.6 kg) and overestimate LTM (3.1 kg and

2.6 kg) in men and women respectively. BIA was found to overestimate LTM in leaner

subjects and underestimate BFM in those who were obese. The biases found between

BIA and DXA in the above studies could have stemmed from the algorithms used in

both BIA devices and the body geometry of the participants.

Conflicting results have been reported in studies comparing bioelectrical impedance to

DXA in athletes participating in numerous sports across several age ranges (Moon,

2013). Fornetti et al. (1999) compared a SF-BIA device to DXA for measuring BF% and

FFM in a cohort of 132 female athletes from a range of sports, and found BIA to

provide good relative agreement, as demonstrated by high correlations (r = 0.969-

0.983) and low a prediction error (1.1 kg) between the two methods. Yannakoulia et al.

(2000) created two athlete specific BIA equations derived from DXA, and found them to

provide accurate measures of body composition in a cohort of female dancers when

validated against DXA. The authors noted that cross-validation would be necessary in

order to adequately assess the accuracy of both equations in an athletic population.

The above findings are in contrast to more recent literature where BIA has been

reported to display inaccurate measures compared to DXA. In a sample of 43 highly

active male judo, karate, and water polo athletes, De Lorenzo et al. (2000) reported

that between SF-BIA and DXA, BIA underestimated BF% by 2.5% and overestimated

FFM by 2.4 kg. Similarly, Sventesson et al. (2008) studied elite male soccer (n= 17)

and ice-hockey athletes (n= 16) aged (18+) and found a a bioelectrical impedance

spectroscopy (BIS) device underestimated BF% by 4.6% in ice hockey players and

1.1% in soccer players with large differences also being reported at an individual level.

Esco et al. (2011) found further support for the above studies in a similar investigation

on 40 collegiate level female athletes using a hand-to-hand SF-BIA device. While good

relative agreement existed between both methods for BF% (r = 0.74, R2 = 0.55, SEE =

3.60, and p < 0.01), and FFM (r = 0.84, R2 = 0.71, SEE = 2.45, p < 0.01), poor absolute

agreement was found. This was indicated by the large biases and wide limits of

agreement showing BIA to underestimate BF% by 5.1% and overestimating FFM 3.4%,

with the differences becoming greater at higher levels of fat and lean mass. The above

findings were concurrent with more recent research by Esco et al. (2014) on 45 female

collegiate level athletes, who demonstrated a MF-BIA device to underestimate BF% by

3.3% and overestimate FFM by 2.1 kg compared to DXA. Based on the above findings,

it appears that BIA methods underestimate both absolute and relative body fat and

overestimate FFM and LTM compared to DXA in athletic populations. However, the

12

majority of studies were conducted on female athletes using SF-BIA devices; therefore,

further research is warranted into the use of MF-BIA devices on male team sport

athletes.

2.4 Conclusion

It can be concluded from the literature review that body composition can influence

athletic performance. Therefore, an easy to use, quick and accurate method for

measuring body composition of GAA players as this can provide valuable information

to athletes, coaches, and various other professionals within the field of sport and

exercise sciences. DXA and BIA are commonly utilised methods for body composition

assessment in athletic populations. While the literature suggests that BIA lacks

accuracy compared to DXA in athletic groups the majority of studies have been

conducted on female athletes using SF-BIA devices; therefore, further research is

warranted examining the accuracy of MF-BIA devices on male team sport athletes.

More research is also needed in order to evaluate the accuracy of population specific

regression equations as there is a gap in the research relating to BIA equations created

for athletes from a multi-compartment model with a valid estimate of TBW. The vast

amount of literature suggesting that BIA is inaccurate at extreme BF levels which

further strengthens the need for athlete specific equations to be developed.

13

Chapter 3 - METHODS

3.1 Participants

Following written, informed consent, 157 senior inter-county GAA athletes (put in

number of hurlers and footballers here) were recruited from five inter-county teams

across Ireland. Before the study commenced, all athletes completed a pre-test health

screening questionnaire.

3.2 Preparation

Data collection took place over a four year period (2009-2013) within the Physical

Education and Sport Sciences department of the University of Limerick. The

participants were instructed to avoid any form of organised training or exercise session

of 20 minutes or longer for a period of 12 hours before testing, refrain from ingesting

food for three hours before testing, drink 500 ml of water one hour before testing, and

empty their bladder or defecate immediately before testing if required.

3.3 Procedures

3.3.1 Anthropometric Measurements

Body mass was measured in minimal clothing to the nearest 0.1 kg using a Tanita MC-

180MA Body Composition Analyzer, (Tanita UK Ltd). Height was measured to the

nearest 0.1 cm using a stadiometer (Seca, Birmingham, UK). Subjects were instructed

to remove all jewellery and metal objects prior to testing to ensure accuracy of the BIA

and DXA measurements (Sun et al. 2005).

3.3.2 Bioelectrical Impedance Analysis (BIA)

BIA measurements were carried out before DXA for all participants to determine fat

free mass (FFM), body fat mass (BFM) and lean tissue mass (LTM). Whole and

segmental body composition was assessed using an eight-contact electrode multi

frequency bioelectrical impedance analyser (Tanita MC-180MA Body Composition

Analyzer, Tanita UK Ltd). Body composition of all participants was assessed in both

Normal and Athlete mode. According to the instructions of the manufacturer, Normal

mode was designed for individuals who participated in less than 12 hours exercise per

week. Athlete mode was designed for active individuals who were over the age of 18

and participated in 12 or more hours of training (exercise) per week. In accordance with

the manufacturer’s instructions participants stood barefoot on the stainless steel metal

panel of the Tanita MC-180 with their feet parallel and soles in contact with the four

heel and toe metallic electrodes, and body mass was recorded. Gender, height, body

mass, age and physical activity mode (“normal” or “athlete”) were manually entered into

14

the BIA keypad interface. Participants grasped the handgrips, with their thumbs,

fingers, and palms in contact with the four anterior and posterior placed metallic

electrodes, and with arms hanging naturally by their sides fully extended, and abducted

laterally to approximately 20 degrees to ensure contact between the arms and torso

was avoided (Hogan et al. 2011). The device obtained impedance measures from 5

different regions within the body (whole body, right leg, left leg, right arm, left arm), by

passing an electric current (less than or equal to 90uA) from the 8 polar electrodes,

through the body at various frequencies (5, 50, 250, and 500 kHz). Body composition

parameters were then estimated from specific equations using height, body mass,

physical activity and impedance values (Kyle et al. 2004). The impedance measure

had a Coefficient of variance of 0.4% (Leahy et al. 2012).

3.3.3 Dual Energy X-ray Absorptiometry (DXA)

Measurements of body fat mass (BFM), , Lean tissue mass (LTM), bone mineral

content (BMC) and fat free mass (FFM) were undertaken by a Lunar iDXA scanner (GE

Healthcare, Chalfont St Giles, Bucks., UK) with encore 2007 v.11 software.. Calibration

was performed daily according to the manufacturer’s instructions using a proprietary

phantom consisting of bone, lean, and fat tissue. Participants wore minimal clothing

and removed all jewellery prior to the scan. Measurements were performed and

supervised by trained technicians within the University of Limerick Physical Education

and Sport Sciences Department. Participants were instructed to lay supine and

motionless on the measurement table with their arms by their sides and hands in the

mid-prone position, making sure there was no contact between the arm and trunk

segments. The DXA scanner used within this study was capable of providing

segmental body analysis, splitting the body into three anatomical regions of interest

(arms, legs, and trunk). Leahy et al. (2012) defined the aforementioned regions by the

following body landmarks.The arm segment was defined as the area of tissue bisecting

the centre of the glenohumeral joint to the phalanges. The leg segment was the area of

tissue perpendicular to the axis of the neck of the femur, angled with the pelvic brim to

the phalanges. The trunk segment consisted of all remaining distal tissue from the

bottom of the skull excluding leg and arm segments. All composition data was

calculated by enCore software from DXA derived estimates of body mass. According to

Huizenga et al. (2007) the coefficient of variation for the iDXA measurement of body

composition is <1%.

3.4 Statistical Analysis

Statistical analysis were performed using PASW Statistics 18.0 for Windows (SPSS,

Inc., Chicago, IL. A Kolgomorov Smirnov test was conducted to determine whether

15

data were normally or non-normally distributed. Paired t tests and Wilcoxon signed

ranks tests were used in order to compare measures from each BIA mode to DXA (i.e.

Athlete vs. DXA and Normal vs. DXA) for whole body analysis, and also to compare

both modes to one another (i.e. Athlete vs. Normal). Pearson’s and Spearman’s

correlation were used to assess the relative agreement between the methods. Bland

Altman plots (Bland and Altman 1986) were used to assess the absolute agreement

and bias between both modes and DXA and between both modes independently.

Limits of agreement were determined as the mean of the difference between each

method +/- 1.96 x SD of the difference. All tests were two-tailed and with the

significance level set at 0.01 for correlation analysis and 0.05 for all other analysis.

16

Chapter 4 – RESULTS

4.1 Descriptive Statistics

Anthropometrics for the GAA players are reported in Table 4.1. Not all the data were

normally distributed; the mean, standard deviation and range are reported, as well as

the median and interquartile range (IQR).

Table 4.1 Anthropometrics for the 157 GAA players included in this study. Data are

reported as mean (standard deviation (SD)), median (interquartile range (IQR)) and

range (max – min); n = 157)

Mean (SD) Median (IQR) Range

Age (y) 25.5 (4.1) 25.3 (6.7) 19-40

Height (cm) 183.3 (4.59) 184.0 (8.0) 173-195

Mass (kg) 85.4 (7.1) 84.8 (9.4) 70-111

BMI (kg/m2) 25.3 (1.5) 25.1 (1.8) 21-31

LTMI (kg/m2) 20.1 (1.0) 19.9 (1.0) 18-23

ALTMI (kg/m2) 9.9 (0.5) 9.9 (1.0) 9-12

4.2 BIA Athlete and Normal mode vs. DXA Analysis

4.2.1 Overview

Comparisons of BIA Athlete and Normal mode to DXA for all body composition

variables are displayed in table 4.2. Strong positive correlations were found between

both modes and DXA for BFM, LTM, FFM, and BF% (r > 0.6; p = 0.000), however,

moderate correlations were found for BMC (r = 0.57-0.58; p = 0.000). There were

significant differences found between Athlete mode and DXA for all body composition

variables and between Normal and DXA for LTM, BMC, and FFM (p < 0.05; Table 4.2).

4.2.2 BFM

Athlete mode underestimated median BFM by 2.8 kg (-21.3%) (p = 0.000), (LoA; -1.6

kg to 7.3 kg). Normal mode overestimated median BFM by 0.2 kg (1.5%) (p = 0.183),

(LoA; -4.6 kg to +4.1 kg). Both modes underestimated BFM in individuals with greater

than 20 kg of BFM (Figure 4.1 a & b.).

4.2.3 LTM

Compared to DXA, Athlete mode overestimated mean LTM by 2.7 kg (3.9%) (p =

0.000), (LoA; -7.0 kg to 1.6 kg) (Figure 4.2 (a)). Normal mode underestimated mean

LTM by 0.2 kg (-0.2%) (p = 0.000), (LoA; -4.0 kg to +4.4 kg) (Figure. 4.2 (b)). There

was no clear trend in the difference between DXA and BIA over the range of LTM

values (Figure 4.2 a & b.).

17

4.2.4 BMC

For BMC, A Wilcoxon signed ranks test showed that there was significant differences

(p = 0.000) obtained by both modes and DXA, with Athlete mode and Normal mode

underestimating the median value by 0.2 kg (5.2%) and 0.3 kg (7.8%) respectively.

4.2.5 FFM

Athlete mode was found to overestimate mean FFM by 2.4 kg (3.3%) (p = 0.000), (LoA;

-6.9 kg to 2.1 kg) (Figure. 4.4 (a)). Normal mode underestimated FFM by 0.6 kg (0.8%)

(p = 0.000), (LoA; -3.8 kg to +5.1 kg) (Figure. 4.4 (b)). Similar to LTM there was no

obvious trend in the difference between DXA and either BIA mode with increasing FFM

(Figure 4.3 a & b.).

4.2.6 BF%

The mean difference between DXA and BIA in both Athlete and Normal modes for

BF% are illustrated in Figure 4.5. BIA Athlete mode underestimated mean BF% by

3.3% (p = 0.000), (LoA ranging from approximately -1.9% to +8.4%) (Figure. 4.5 (a)).

Normal mode overestimated mean BF% by 0.4%, (LoA -5.4% to +4.6% (Figure. 4.5

(b)). There was a trend towards BIA underestimating in individuals with greater than

20% BF (Figure 4.4 a & b.).

4.2.7 Athlete vs. Normal mode

Significant differences were noted between both modes for all body composition

estimates (p < 0.05). Athlete mode underestimated BFM (3 kg) and BF% (3.6%) and

overestimated FFM (3.1 kg) and LTM (2.9 kg) values compared to the Normal setting.

Bland-Altman plots revealed that underestimations became more evident at lower

levels of BFM, while the difference between the two modes was less at higher body fat

levels.

18

Table 4.2 DXA and BIA athlete (a) and normal (n) measured body fat mass (BFM), lean tissue mass (LTM), bone mineral content (BMC), fat

free mass (FFM), and body fat % (BF%), for all subjects (n = 157)

Variable Method Mean SD Median Range IQR r-value p-value

BFM DXA

BIA (a)

BIA (n)

13.9

11.1*

14.1*

4.4

4.0

3.9

13.1

10.3b,c

13.3b

7-31

4-27

5-30

5.0

5.0

5.0

0.74

0.77

0.000

0.183

LTM DXA

BIA (a)

BIA (n)

67.5a,*

70.2a,b,c,*

67.3a,b,c,*

4.7

4.8

4.6

67.3

70.0

67.2

56-79

61-81

58-77

7.0

7.0

7.0

0.89

0.89

0.000

0.000

BMC DXA

BIA (a)

BIA (n)

3.9

3.6*

3.5*

0.3

0.2

0.2

3.8

3.6b,c

3.5b,c

3-5

3-4

3-4

1.0

0.0

0.0

0.58

0.57

0.000

0.000

FFM

DXA

BIA (a)

BIA (n)

71.4a

73.8a,b,c,*

70.8a,b,c*

4.9

5.0

4.8

71.2

73.6

70.7

59-84

64-84

61-81

7.0

7.0

7.0

0.99

0.89

0.000

0.000

BF% DXA

BIA (a)

BIA (n)

16.1

12.8a,*

16.5a,*

4.0

3.7

3.5

15.3

12.6b,c

16.2b

8.7-27.9

5.1-24.7

6.1-26.8

5.3

5.4

4.7

0.77

0.69

0.000

0.077

(a = normal distribution; b = significant correlation between BIA and DXA measurement (p < 0.01); c = significant difference between BIA and

DXA measurements; r value = correlation coefficient; p = statistically significant at (p < 0.05); * = significant difference between BIA Athlete and

Normal mode measurements)

19

-0.2

4.1

-4.6

-8.0

-4.0

0.0

4.0

8.0

12.0

5.0 10.0 15.0 20.0 25.0 30.0 35.0

BFM

dif

fere

nce

be

twe

en

me

tho

ds

(DX

A -

BIA

No

rmal

(kg

)

BFM (mean of methods) (kg)

BFM Bland Altman (DXA vs BIA Normal) (b)

Figure 4.1 a & b Bland Altman plots of Body Fat Mass (kg) with mean difference

(dotted lines) and 95% limits of agreement (dots) comparing Athlete and Normal modes

to DXA

2.8

7.3

-1.6

-8.0

-4.0

0.0

4.0

8.0

12.0

5.0 10.0 15.0 20.0 25.0 30.0 35.0

BFM

dif

fere

nce

be

twe

en

me

tho

ds

(D

XA

- B

IA A

thle

te (

kg)

BFM (mean of methods) (kg)

BFM Bland Altman (DXA vs BIA Athlete) (a)

20

0.2

4.4

-4.0

-12.0

-8.0

-4.0

0.0

4.0

8.0

55.0 60.0 65.0 70.0 75.0 80.0

LTM

dif

fere

nce

be

twe

en

me

tho

ds

(D

XA

- B

IA N

orm

al (

kg)

LTM (mean of methods) (kg)

LTM Bland Altman (DXA vs BIA Normal) (b)

Figure 4.2 a & b Bland Altman plots of Lean Tissue Mass (kg) with mean difference

(dotted lines) and 95% limits of agreement (dots) comparing Athlete and Normal modes

to DXA

-2.7

1.6

-7.0

-12.0

-8.0

-4.0

0.0

4.0

8.0

55.0 60.0 65.0 70.0 75.0 80.0

LTM

dif

fere

nce

be

twe

en

me

tho

ds

(DX

A -

BIA

Ath

lete

(kg

)

LTM (mean of methods) (kg)

LTM Bland Altman (DXA vs BIA Athlete) (a)

21

0.6

5.1

-3.8

-12.0

-8.0

-4.0

0.0

4.0

8.0

55.0 60.0 65.0 70.0 75.0 80.0 85.0

FFM

dif

fere

nce

be

twe

en

me

tho

ds

(D

XA

- B

IA N

orm

al (

kg)

FFM (mean of methods) (kg)

FFM Bland Altman (DXA vs BIA Normal) (b)

Figure 4.3 a & b Bland Altman plots of Fat Free Mass (kg) with mean difference

(dotted lines) and 95% limits of agreement (dots) comparing Athlete and Normal modes

to DXA

-2.4

2.1

-6.9

-12.0

-8.0

-4.0

0.0

4.0

8.0

55.0 60.0 65.0 70.0 75.0 80.0 85.0

FFM

dif

fere

nce

be

twe

en

me

tho

ds

(D

XA

- B

IA A

thle

te (

kg)

FFM (mean of methods (kg)

FFM Bland Altman (DXA vs BIA Athlete) (a)

22

-0.4%

4.6%

-5.4%

-8.0%

-4.0%

0.0%

4.0%

8.0%

12.0%

5.0% 10.0% 15.0% 20.0% 25.0% 30.0%

BF%

dif

fere

nce

be

twe

en

me

tho

ds

(D

XA

- B

IA N

orm

al)

BF% (mean of methods)

BF% Bland Altman (DXA vs BIA Normal) (b)

Figure 4.4 a & b Bland Altman plots of Body Fat % with mean difference (dotted lines)

and 95% limits of agreement (dots) comparing Athlete and Normal modes to DXA

3.3%

8.4%

-1.9%

-8.0%

-4.0%

0.0%

4.0%

8.0%

12.0%

5.0% 10.0% 15.0% 20.0% 25.0% 30.0%

BF%

dif

fere

nce

be

twe

en

me

tho

ds

(DX

A -

BIA

Ath

lete

)

BF% (mean of methods)

BF% Bland Altman (DXA vs BIA Athlete) (a)

23

Chapter 5 – DISCUSSION

5.1 Background and Purpose

Body composition has gained particular interest within the GAA because athletic

performance is influenced by and dependent on the quantity and proportion of BFM

and LTM (Pritchard et al., 1998). BFM negatively influences many fundamental fitness

components of Gaelic football and hurling, including speed, aerobic and anaerobic

capacity (Collins et al., 2014; Potteiger et al., 2010; Matilla et al., 2007). Conversely,

LTM has been positively associated with athletic performance (Hogstrom et al., 2012).

Therefore there is a growing need for convenient and accurate assessment methods

within the GAA. DXA is now accepted as a reference method for estimating LTM and

BFM in athletes (Bilsborough et al., 2014; Stewart and Hannon, 2000). However, the

majority of inter-county GAA teams do not have access to this method it is expensive

and most often found in clinical or laboratory settings. BIA methods on the other hand,

offer an attractive alternative for athletes, as they are cost and time effective, non-

invasive, and easy to use. The accuracy of BIA is limited in populations displaying

extreme levels of body fat, and has been found to overestimate and underestimate fat

values in lean and obese cohorts respectively (Segal et al., 1988). Athletes tend to be

leaner and more active than the normal population This has led many manufacturers to

incorporate body types into their equations. The primary purpose of this study was to

determine which mode (Athlete or Normal) on the Tanita MC-180MA MF-BIA had

better agreement in measuring body composition components in elite inter-county GAA

players compared to DXA. A secondary purpose was to investigate whether choosing

one BIA mode over the other significantly changed the output of the components of

body composition.

5.2 Findings

The principle finding of this investigation was that, relative to DXA, BIA Normal mode

provided more accurate measurements of the body composition components than

Athlete mode of elite GAA athletes. Values obtained in Athlete mode displayed larger

biases and wider limits of agreement for all body composition components. Both BIA

modes showed good relative agreement with DXA as demonstrated by strong

correlation coefficients for BFM, FFM, LTM, and BF%, as well as moderate correlations

for BMC. Although the high correlation coefficients indicate a strong relationship

between methods, this type of analysis does not imply that the two methods agree with

one another. In absolute terms, Normal mode demonstrated good agreement with DXA

for BFM and BF% as indicated by small non-significant biases between the methods.

24

These findings are comparable with numerous other studies carried out on athletes that

have shown close agreement between BIA and DXA for body composition measures

(Yannaoulia et al., 2000; Fornetti et al., 1999). Athlete mode significantly

underestimated median BFM 2.8 kg and mean BF% by 3.3% compared to DXA.

Although significant differences were found between both BIA modes and DXA for FFM

and LTM values, the biases exhibited in Normal mode were small, underestimating

FFM by 0.2 kg and LTM by 0.6 kg in contrast to Athlete mode which overestimated

FFM and LTM by 2.4 kg and 2.7 kg respectively, with the overestimations remaining

consistent over the range of FFM and LTM values. The measures obtained from

Athlete mode agree with previous research conducted on athletic populations finding

SF and MF-BIA devices to display good relative agreement with DXA, but provide

underestimations of BF% and overestimations of FFM respectively in collegiate level

female athletes (Esco et al., 2011; Esco et al., 2014). The results of the current study

demonstrate that Normal mode was superior to Athlete mode in assessing body

composition of elite GAA athletes. The non-significant biases and high correlation

coefficients between Normal mode and DXA for BFM and BF%, suggest it to be a

reliable method for group level body fat analysis in athletic cohorts. However, the limits

of agreement were wide for all measures, therefore limiting its applicability to

estimating body composition in individual athletes (Pateyjohns et al., 2006).

5.3 Systematic Errors

On closer interpretation of the Bland Altman plots, systematic errors existed between

both BIA modes and DXA, with underestimations for BFM and BF% becoming more

apparent as body fat levels increased (>20%; >20 kg respectively). These observations

are in agreement with findings reported in similar studies comparing MF- BIA to DXA in

normal healthy populations (Sillanpaa et al., 2014; Leahy et al., 2012; Sun et al., 2005).

Leahy et al. (2012) showed Tanita MC-180MA MF-BIA to significantly underestimate

BFM and BF% compared to DXA, and noted that the underestimations increased in

subjects with higher body fat levels. The underestimations of BFM and BF% in the

current investigation and the above studies may be due to hydration status at varying

levels of body fat (Pateyjohns et al., 2006; Frisard et al., 2005). Overweight individuals

have been found to exhibit greater TBW and Extracellular water (ECW) than their

leaner counterparts (Steijaert et al. 1997) Therefore, given the highly sensitive nature

of BIA to variations in hydration, higher levels of TBW could be wrongly interpreted as

greater FFM and lower BFM levels in subjects.

To the authors knowledge this is the first study to compare the accuracy of the BIA

modes on the Tanita MC-180MA MF-BIA to DXA, in a group of elite athletes, therefore

25

no direct comparisons could be made. However, in a similar study Swartz et al. (2002)

investigated the accuracy of Athlete and Normal modes against HW as they criterion

method in 57 middle-aged men of varying levels of physical activity. Subjects were

divided according to activity level with seventeen who participated in greater than 10

hours aerobic exercise per week categorised as highly active. All participants were

scanned in both modes and the Athlete equation was found to display greater accuracy

for estimates of BF% and FFM in the highly active subjects. While no significant

differences existed between Athlete and HW for BF% and FFM (p = 0.309), Normal

mode overestimated BF% by 5% and underestimated FFM by 4kg in highly active

individuals (p < 0.001). The findings of this investigation one again demonstrated the

importance of choosing the correct BIA equation in highly active populations. These

results were in contrast to the current study where Normal mode displayed better

accuracy than Athlete mode for all body composition estimates compared to DXA. The

inconsistent findings of the two studies could stem from many explanations. Firstly, the

current study used a MF-BIA device opposed to the SF-BIA utilised by Swartz et al.

(2002). Previous research has shown differences between SF and MF BIA devices for

body composition estimations (Thompson et al., 2007; Pateyjohns et al., 2006). For

instance, Thompson et al. (2007) found SF-BIA to display larger bias and wider LoA

than MF when compared to DXA as the criterion method. Secondly, the two studies

were carried out on different population groups. While similar age ranges were

reported, Swartz et al. (2002) studied highly active men while current investigation was

conducted on elite athletes. Although Swartz et al. (2002) suggested that those who

were highly active had comparable activity levels to athletes, elite GAA players perform

specialised regimens each week involving many types of training (i.e. aerobic,

resistance, anaerobic), which can modify their physical make up away from

morphological norms (Ackland et al., 2012). Furthermore, the multiple bouts of acute

exercise performed during training and competition, make athletes more susceptible to

variations in fluid and electrolyte balance (Ackland et al., 2012). As BIA assumes the

constant hydration of FFM (73%), variability of impedance within participants may be

greater in elite athletes. Despite the participants in the current investigation being given

clear pre-test guidelines regarding fluid consumption and exercise avoidance, time

since last exercise bout was not measured. As GAA is an amateur organisation, many

players would be reluctant to interrupt their training schedules for a body composition

assessment, which may be a further reason for the contradictory findings of the two

studies.

26

5.4 BIA Athlete vs. Normal mode

The secondary purpose of this study was to determine whether choosing one BIA

analysis mode over the other significantly altered body composition estimates. Results

showed that Athlete mode significantly underestimated BFM and overestimated LTM

compared to Normal mode. Swartz et al. (2002) similarly reported differences in the

outputs of the BIA physical activity settings showing Normal mode to significantly

overestimate BF% by 6.8% (p < 0.001) and underestimate FFM by 5.5 kg (p < 0.001)

compared to Athlete mode in highly active individuals. These conflicting results

between the two studies further emphasise the need for more research to be

conducted on BIA devices that incorporate equations based on body types and

physical activity levels. This is because inaccurate body composition estimations could

provide GAA coaches and practitioners with false data relating to the effectiveness of

training programmes and nutritional interventions.

5.5 Limitations

Several limitations of the investigation should be noted. Subjects were given strict pre-

test guidelines (e.g. refrain from any form of organised training or exercise session of

greater than 20 minutes for a period of 24 hours before testing, refrain from ingesting

food for three hours before testing, consume 500ml of water one hour before testing,

empty bladder or defecate immediately before testing if required). However, we could

not determine the hydration status of the participants. Variations in hydration levels can

lead to errors in predicting body composition components as BIA assumes that FFM

comprises of 73% water (Esco et al., 2014). As previously mentioned, elite inter-county

GAA players are predisposed to deviation in FFM hydration due to rigorous training

regimens, therefore this could have affected the accuracy of BIA estimates in the

current study. Another possible limitation of the current study was that DXA was used

as the criterion method instead other laboratory techniques such as hydrodensitometry

(HD) or a 4-C model. Although studies in the past have found DXA demonstrates

similar accuracy to HD, (Kohrt, 1998) research on athletic cohorts comparing DXA to 4-

C models have found mean biases in BF% ranging from -3.5% to 2.9%, with the most

studies reporting larger BFM underestimations in leaner individuals (Toombs et al.,

2012). However, many other studies on athletic cohorts have utilised DXA as the

criterion method (Esco et al., 2014; Esco et al., 2011; Sventesson et al., 2008), due to

its speed and precision therefore, until the development of a practical 4-C model for

assessing athletes, DXA is an adequate reference method (Stewart and Hannon,

2000).

27

Chapter 6 – CONCLUSION

6.1 Summary and Future work

Based on the available literature, it was hypothesized that Athlete mode would display

better agreement than Normal mode for estimating the body composition of inter-

county GAA players compared to DXA as the criterion method. Although both settings

provided acceptable relative agreement with DXA, Normal mode was found to be more

accurate for all measures, showing excellent absolute agreement with DXA for BFM

and BF%. This indicates that Normal mode may be used interchangeably with DXA for

group comparisons of body composition, however, the wide limits of agreement

suggest that results of individual body composition assessments should be analysed

with caution.

The results of this study may have practical implications to practitioners within the GAA

(Strength and Conditioning Coaches, dieticians/sports nutritionists). As DXA is

expensive and inconvenient for use in field settings, BIA Normal mode may serve as a

practical alternative for measuring body composition of groups. This could save inter-

county GAA teams time and money, while also allowing body composition to be

assessed frequently throughout the season in order to evaluate the effects of training

and nutritional interventions. As the current study did not assess the validity of BIA for

assessing body composition over a period of time, future research could assess the

suitability of BIA to measure changes in body composition over the course of the

training year. As hydration levels in athletes can fluctuate more than non-athletes, and

no generalised equation with a valid estimation of TBW exists for use on athletic

populations, future work should also focus on creating athlete specific BIA equations

from multi-compartment models that can accurately assess TBW.

28

References

Ackland, T. R., Lohman, T. G., Sundgot-Borgen, J., Maughan, R. J., Meyer, N.

L., Stewart, A. D. and Muller, W. (2012) 'Current status of body composition

assessment in sport: review and position statement on behalf of the ad hoc

research working group on body composition health and performance, under

the auspices of the IOC Medical Commission'.

Bilsborough, J. C., Greenway, K., Opar, D., Livingstone, S., Cordy, J. and

Coutts, A. J. (2014) 'The accuracy and precision of DXA for assessing body

composition in team sport athletes', J Sports Sci, 32(19), 1821-1828.

Bolanowski, M. and Nilsson, B. E. (2001) 'Assessment of human body

composition using dual-energy x-ray absorptiometry and bioelectrical

impedance analysis', Medical Science Monitor, 7(5), 1029-1033.

Buehring, B., Krueger, D., Libber, J., Heiderscheit, B., Sanfilippo, J., Johnson,

B., Haller, I. and Binkley, N. (2014) 'Dual-energy X-ray absorptiometry

measured regional body composition least significant change: effect of region of

interest and gender in athletes', Journal of Clinical Densitometry, 17(1), 121-

128.

Carling, C. and Orhant, E. (2010) 'Variation in Body Composition in Professional

Soccer Players: Interseasonal and Intraseasonal Changes and the Effects of

Exposure Time and Player Position', Journal of Strength & Conditioning

Research, 24(5), 1332-1339.

Casajús, J. A. (2001) 'Seasonal variation in fitness variables in professional

soccer players', J Sports Med Phys Fitness, (41), 463-9.

Collins, S. M., Silberlicht, M., Perzinski, C., Smith, S. P., & Davidson, P. W.

(2014) ‘The Relationship Between Body Composition and Preseason

Performance Tests of Collegiate Male Lacrosse Players’ Journal of Strength &

Conditioning Research, 28(9), 2673-2679.

Cullen, B. D., Cregg, C. J., Kelly, D. T., Hughes, S. M., Daly, P. G. and Moyna,

N. M. (2013) 'Fitness Profiling of Elite Level Adolescent Gaelic Football

Players', Journal of Strength & Conditioning Research, 27(8), 2096-2103.

De Lorenzo, A., Bertini, I., Iacopino, L., Pagliato, E., Testolin, C. and Testolin,

G. (2000) 'Body composition measurement in highly trained male athletes. A

comparison of three methods', J Sports Med Phys Fitness, 40(2), 178-83..

Duthie, G., Pyne, D., Hopkins, W., Livingstone, S. and Hooper, S. (2006)

29

'Anthropometry profiles of elite rugby players: quantifying changes in lean

mass', Br J Sports Med, 40(3), 202-207.

Esco, M. R., Olson, M. S., Williford, H. N., Lizana, S. N. and Russell, A. R.

(2011) 'The accuracy of hand-to-hand bioelectrical impedance analysis in

predicting body composition in college-age female athletes', Journal Of

Strength And Conditioning Research / National Strength & Conditioning

Association, 25(4), 1040-1045.

Esco, M. R., Snarr, R. L., Leatherwood, M. D., Chamberlain, N., Redding, M.,

Flatt, A. A., Moon, J. R. and Williford, H. N. (2014) 'Comparison of total and

segmental body composition using DXA and multi-frequency bioimpedance in

collegiate female athletes', Journal of Strength & Conditioning Research.

Fornetti, W. C., Pivarnik, J. M., Foley, J. M. and Fiechtner, J. J. (1999)

'Reliability and validity of body composition measures in female athletes',

Journal of Applied Physiology, 87(3), 1114-1122.

Frisard, M. I., Greenway, F. L. and DeLany, J. P. (2005) 'Comparison of

Methods to Assess Body Composition Changes during a Period of Weight

Loss', Obesity Research, 13(5), 845-854.

Harley, J. A., Hind, K. and O'Hara, J. P. (2011) 'Three-Compartment Body

Composition Changes in elite Rugby League Players During a Super League

Season, Measured by Dual-Energy X-ray Absorptiometry', Journal of Strength &

Conditioning Research, 25(4), 1024-1029.

Hogstrom, G. M., Pietila, T., Nordstrom, P. and Nordstrom, A. (2012) 'Body

Composition and Performance: Influence of Sport and Gender Among

Adolescents', Journal of Strength & Conditioning Research, 26(7), 1799-1804.

Kohrt, W. M. (1998) 'Preliminary evidence that DEXA provides an accurate

assessment of body composition', Journal of Applied Physiology, 84(1), 372-

377.

Kyle, U. G., Bosaeus, I., De Lorenzo, A. D., Deurenberg, P., Elia, M., Gómez, J.

M., Heitmann, B. L., Kent-Smith, L., Melchior, J.-C., Pirlich, M., Scharfetter, H.,

Schols, A. M. W. J. and Pichard, C. (2004) 'Bioelectrical impedance analysis—

part I: review of principles and methods', Clinical Nutrition, 23(5), 1226-1243.

Leahy, S., O’Neill, C., Sohun, R. and Jakeman, P. (2012) 'A comparison of dual

energy X-ray absorptiometry and bioelectrical impedance analysis to measure

total and segmental body composition in healthy young adults', European

Journal of Applied Physiology, 112(2), 589-595.

Malina, R. M. and Geithner, C. A. (2011) 'Body Composition of Young Athletes',

30

American Journal of Lifestyle Medicine, 5(3), 262-278.

Malina, R.M., (2007) ‘Body composition in athletes: assessment and estimated

fatness’. Clin Sports Med, 26(1): p. 37-68.

Martin Bland, J. and Altman, D. (1986) 'Statistical methods for assessing

agreement between two methods of clinical measurement', The lancet,

327(8476), 307-310.

Matthie, J. R. (2008) 'Bioimpedance measurements of human body

composition: critical analysis and outlook', Expert Review of Medical Devices,

(5), 239-61.

Mattila, V. M., Tallroth, K. A. J., Marttinen, M., & Pihlajamäki, H. (2007)

‘Physical fitness and performance. Body composition by DEXA and its

association with physical fitness in 140 conscripts’, Medicine and science in

sports and exercise, 39(12), 2242-2247.

McIntyre, M. C. (2005) 'A comparison of the physiological profiles of elite Gaelic

footballers, hurlers, and soccer players', Br J Sports Med, 39(7), 437-439.

Moon, J. R. (2013) 'Body composition in athletes and sports nutrition: an

examination of the bioimpedance analysis technique', Eur J Clin Nutr, 67 Suppl

1, S54-9.

Oppliger, R. A., Nielsen, D. H., Shetler, A. C., Crowley, E. T., & Albright, J. P.

(1992) ‘Body composition of collegiate football players: bioelectrical impedance

and skinfolds compared to hydrostatic weighing’, Journal of Orthopaedic &

Sports Physical Therapy, 15(4), 187-192.

Ostojic, S. M. (2003) 'Seasonal alterations in body composition and sprint

performance of elite soccer players', Journal of Exercise Physiology, 6(3), 11-

14.

Pallarés, J. G., López-Gullón, J. M., Torres-Bonete, M. D. and Izquierdo, M.

(2012) 'Physical fitness factors to predict female Olympic wrestling performance

and sex differences', Journal of Strength & Conditioning Research, 26(3), 794-

80

Pateyjohns, I. R., Brinkworth, G. D., Buckley, J. D., Noakes, M. and Clifton, P.

M. (2006) 'Comparison of three bioelectrical impedance methods with DXA in

overweight and obese men', Obesity, 14(11), 2064-2070.

Pichard, C., Kyle, U. G., Gremion, G., Gerbase, M. and Slosman, D. O. (1997)

'Body composition by x-ray absorptiometry and bioelectrical impedance in

female runners', Medicine & Science in Sports & Exercise, 29(11), 1527-1534.

Pietrobelli, A., Formica, C., Wang, Z. and Heymsfield, S. B. (1996) 'Dual-energy

31

X-ray absorptiometry body composition model: review of physical concepts',

American Journal of Physiology-Endocrinology and Metabolism, 271(6), E941-

E951.

Potteiger, J. A., Smith, D. L., Maier, M. L. and Foster, T. S. (2010) 'Relationship

between body composition, leg strength, anaerobic power, and on-ice skating

performance in division I men's hockey athletes', Journal of Strength &

Conditioning Research, 24(7), 1755-1762.

Prior, B. M., Cureton, K. J., Modlesky, C. M., Evans, E. M., Sloniger, M. A.,

Saunders, M. and Lewis, R. D. (1997) 'In vivo validation of whole body

composition estimates from dual-energy X-ray absorptiometry', Journal of

Applied Physiology, 83(2), 623-630.

Reeves, S. and Collins, K. (2003) 'The nutritional and anthropometric status of

Gaelic football players', Int J Sport Nutr Exerc Metab, 13(4), 539-48.

Reilly, T. and Collins, K. (2008) 'Science and the Gaelic sports: Gaelic football

and hurling', European Journal of Sport Science, 8(5), 231-240.

Reilly, T. and Doran, D. (2001) 'Science and Gaelic football: A review', J Sports

Sci, 19(3), 181-193.

Reilly, T. and Keane, S. (2013) ‘SEASONAL VARIATIONS IN THE FITNESS

OF ELITE GAELIC FOOTBALLERS', Science and football IV, 86.

Rodriguez, N. R., DiMarco, N. M., & Langley, S. (2009). Nutrition and athletic

performance. Medicine and science in sports and exercise, 41(3), 709-731.

Santos, D. A., Silva, A. M., Matias, C. N., Fields, D. A., Heymsfield, S. B. and

Sardinha, L. B. (2010) 'Accuracy of DXA in estimating body composition

changes in elite athletes using a four compartment model as the reference

method', Nutr Metab (Lond), 7(22), 7075-7.

Saunders, M. J., Blevins, J. E. and Broeder, C. E. (1998) 'Effects of hydration

changes on bioelectrical impedance in endurance trained individuals', Medicine

and science in sports and exercise, 30(6), 885-892.

Scharfetter, H., Schlager, T., Stollberger, R., Felsberger, R., Hutten, H. and

Hinghofer-Szalkay, H. (2001) 'Assessing abdominal fatness with local

bioimpedance analysis: basics and experimental findings', Int J Obes Relat

Metab Disord, 25(4), 502-11.

Segal, K. R. (1996) 'Use of bioelectrical impedance analysis measurements as

an evaluation for participating in sports', The American journal of clinical

nutrition, 64(3), 469S-471S.

Segal, K., Van Loan, M., Fitzgerald, P., Hodgdon, J. and Van Itallie, T. B.

32

(1988) 'Lean body mass estimation by bioelectrical impedance analysis: a four-

site cross-validation study', The American journal of clinical nutrition, 47(1), 7-

14.

Shafer, K. J., Siders, W. A., Johnson, L. K. and Lukaski, H. C. (2009) 'Validity of

segmental multiple-frequency bioelectrical impedance analysis to estimate body

composition of adults across a range of body mass indexes', Nutrition, 25(1),

25-32.

Sillanpää, E., Cheng, S., Häkkinen, K., Finni, T., Walker, S., Pesola, A., &

Sipilä, S. (2014). Body composition in 18‐to 88‐year‐old adults - comparison of

multifrequency bioimpedanceand dual‐energy X‐ray

absorptiometry. Obesity,22(1), 101-1

Silvestre, R., Kraemer, W. J., West, C., Judelson, D. A., Spiering, B. A.,

Vingren, J. L., Hatfield, D. L., Anderson, J. M. and Maresh, C. M. (2006) 'Body

composition and physical performance during a national collegiate athletic

association division 1 men’s soccer season’, Journal of Strength & Conditioning

Research, 20(4), 962-970.

Steijaert, M., Deurenberg, P., Van Gaal, L. and De Leeuw, I. (1997) 'The use of

multi-frequency impedance to determine total body water and extracellular

water in obese and lean female individuals', International journal of obesity,

21(10), 930-934.

Stewart, A. D. and Hannan, W. J. (2000) 'Prediction of fat and fat-free mass in

male athletes using dual X-ray absorptiometry as the reference method', J

Sports Sci, 18(4), 263-74.

Stewart, A.D., (2010) ‘Kinanthropometry and body composition: a natural home

for three-dimensional photonic scanning’, J Sports Sci, 28(5): p. 455-7.

Sun, G., French, C. R., Martin, G. R., Younghusband, B., Green, R. C., Xie, Y.-

g., Mathews, M., Barron, J. R., Fitzpatrick, D. G., Gulliver, W. and Zhang, H.

(2005) 'Comparison of multifrequency bioelectrical impedance analysis with

dual-energy X-ray absorptiometry for assessment of percentage body fat in a

large, healthy population', The American journal of clinical nutrition, 81(1), 74-

78.

Svantesson, U., Zander, M., Klingberg, S. and Slinde, F. (2008) 'Body

composition in male elite athletes, comparison of bioelectrical impedance

spectroscopy with dual energy X-ray absorptiometry', Journal of Negative

Results in Biomedicine, 7, 1-1.

Swartz, A. M., Evans, M. J., King, G. A. and Thompson, D. L. (2002) 'Evaluation

33

of a foot-to-foot bioelectrical impedance analyser in highly active, moderately

active and less active young men', British Journal of Nutrition, 88(02), 205-210.

Thomson, R., Brinkworth, G. D., Buckley, J. D., Noakes, M. and Clifton, P. M.

(2007) 'Good agreement between bioelectrical impedance and dual-energy X-

ray absorptiometry for estimating changes in body composition during weight

loss in overweight young women', Clinical Nutrition, 26(6), 771-777.

Toombs, R. J., Ducher, G., Shepherd, J. A. and De Souza, M. J. (2012) 'The

impact of recent technological advances on the trueness and precision of DXA

to assess body composition', Obesity (Silver Spring), 20(1), 30-9.

Toomey, C. M., Cremona, A., Hughes, K., Norton, C. and Jakeman, P. (2015)

'A Review of Body Composition Measurement in the Assessment of Health',

Topics in Clinical Nutrition, 30(1), 16-32.

Wattanapenpaiboon, N., Lukito, W., Strauss, B., Hsu-Hage, B., Wahlqvist, M.

and Stroud, D. (1998) 'Agreement of skinfold measurement and bioelectrical

impedance analysis (BIA) methods with dual energy X-ray absorptiometry

(DEXA) in estimating total body fat in Anglo-Celtic Australians', International

journal of obesity, 22(9), 854-860.

Withers, R., Smith, D., Chatterton, B., Schultz, C. and Gaffney, R. (1992) 'A

comparison of four methods of estimating the body composition of male

endurance athletes', Eur J Clin Nutr, 46(11), 773-784.

Yannakoulia, M., Keramopoulos, A., Tsakalakos, N. and Matalas, A. L (2000)

'Body composition in dancers: the bioelectrical impedance method', Medicine &

Science in Sports & Exercise, 32(1), 228.

A1

Appendices

Appendix A1

Subject Consent Form

A2

Appendix A2

Dual Energy X-ray Absorptiometry

Figure A1 Lunar iDXA scanner (GE Healthcare, Chalfont St Giles, Bucks., UK)

Figure A2 Fundamental principle of DXA

A3

Appendix A3

Bioelectrical Impedance Analysis

Figure A3 Tanita MC-180MA Body composition Analyser (Tanita UK Ltd.)

A4

0.4

1.0

-0.2

-0.90

-0.40

0.10

0.60

1.10

1.60

3.00 3.20 3.40 3.60 3.80 4.00 4.20 4.40 4.60

B

MC

dif

fere

nce

be

twe

en

me

tho

ds

(DX

A -

BIA

No

rmal

kg)

BMC (mean of methods) (kg)

BMC Bland Altman DXA vs BIA Normal (b)

Appendix A4

BMC Bland Altman plots DXA vs BIA Athlete and Normal modes

Figure A4 a & b Bland Altman plots of Bone Mineral Content (kg) with mean difference

(dotted lines) and 95% limits of agreement (dots) comparing “athlete” and “normal”

modes to DXA

0.3

0.9

-0.3

-0.90

-0.40

0.10

0.60

1.10

1.60

3.00 3.20 3.40 3.60 3.80 4.00 4.20 4.40 4.60

BM

C d

iffe

ren

ce b

etw

ee

n m

eto

ds

(D

XA

- B

IA A

thle

te (

kg)

BMC (mean of methods) (kg)

BMC Bland Altman DXA vs BIA Athlete (a)

A5

Appendix A5

Athlete mode vs. Normal mode Bland Altman plots

Figure A5 Bland Altman plots of body fat mass (kg) with mean difference (dotted lines)

and 95% limits of agreement (dots) (Athlete vs. Normal)

Figure A6 Bland Altman plots of lean tissue mass (kg) with mean difference (dotted

lines) and 95% limits of agreement (dots) (Athlete vs. Normal)

-3.1

-2.0

-4.1

-5.0

-4.0

-3.0

-2.0

-1.0

0.0

4.0 9.0 14.0 19.0 24.0 29.0 34.0

BFM

dif

fere

nce

be

twe

en

me

tho

ds

(Ath

lete

- N

orm

al)

BFM (mean ofmethods) (kg)

BFM Bland Altman (Athlete vs. Normal)

2.9

3.9

1.9

0.0

1.0

2.0

3.0

4.0

5.0

57.0 62.0 67.0 72.0 77.0 82.0

LTM

dif

fere

nce

be

twe

en

me

tho

ds

(Ath

lete

- N

orm

al)

LTM (mean of methods) (kg)

LTM Bland Altman (Athlete vs. Normal)

A5

Appendix A6

Figure A7 Bland Altman plots of fat free mass (kg) with mean difference (dotted lines)

and 95% limits of agreement (dots) (Athlete vs. Normal)

Figure A8 Bland Altman plots of body fat percentage with mean difference (dotted

lines) and 95% limits of agreement (dots) (Athlete vs. Normal)

3.1

4.1

2.0

0.0

1.0

2.0

3.0

4.0

5.0

6.0

60.0 65.0 70.0 75.0 80.0 85.0

FFM

dif

fere

nce

be

twe

en

me

tho

ds

(Ath

lete

- N

orm

al)

FFM (mean of methods) (kg)

FFM Bland Altman (Athlete vs. Normal)

-6.0%

-5.0%

-4.0%

-3.0%

-2.0%

-1.0%

0.0%

4.0% 9.0% 14.0% 19.0% 24.0% 29.0%BF%

dif

fere

nce

be

twe

en

me

tho

ds

(Ath

lete

- N

orm

al)

BF% (mean of methods)

BF% Bland Altman (Athlete vs. Normal)