Title: Energy expenditure derived from micro-technology is not suitable for assessing

internal load in collision-based activities

Submission type: Brief report

Authors: Jamie Highton, Thomas Mullen, Jonathan Norris, Chelsea Oxendale, Craig

Twist*

Affiliation

Department of Sport and Exercise SciencesUniversity of ChesterParkgate RoadChesterCH1 4BJ

*Author for correspondence: Craig Twist (c.twist@chester.ac.uk)

Preferred running head: Metabolic power in contact sports

Abstract word count: 154

Text word count: 1741

Number of figures: 2 Number of tables: 1

1

1

2

3

4

5

6

7

8

9

101112131415

16

17

18

19

20

21

22

23

24

Abstract

This aim of this study was to examine the validity of energy expenditure derived from

micro-technology when measured during a repeated effort rugby protocol. Sixteen

male rugby players completed a repeated effort protocol comprising 3 sets of 6

collisions during which movement activity and energy expenditure (EEGPS) were

measured using micro-technology. In addition, energy expenditure was also estimated

from open circuit spirometry (EEVO2). Whilst related (r = 0.63, 90%CI 0.08-0.89),

there was a systematic underestimation of energy expenditure during the protocol (-

5.94 ± 0.67 kcalmin-1) for EEGPS (7.2 ± 1.0 kcalmin-1) compared to EEVO2 (13.2 ±

2.3 kcalmin-1). High-speed running distance (r = 0.50, 95%CI -0.66-0.84) was

related to EEVO2, while Player Load was not (r = 0.37, 95%CI -0.81-0.68). Whilst

metabolic power might provide a different measure of external load than other

typically used micro-technology metrics (e.g. high-speed running, Player Load), it

underestimates energy expenditure during intermittent team sports that involve

collisions.

Key words: rugby, tackle, training load, GPS, accelerometry

2

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

Introduction

Global positioning systems (GPS) with in-built accelerometers are regularly used in

contact sport training1,2 and competition1,3 to quantify internal and external player

load. A metric recently introduced to these devices is metabolic power (i.e. energy

expenditure), which estimates the metabolic ‘internal’ cost of activities from players’

‘external’ movements.4 Metabolic power values of ~9-11 W·kg-1 have recently been

employed to estimate energy expenditures of ~24-45 kJkg-1 during elite rugby

league5 and Australian football match-play.6 In these studies, metabolic power was

proposed as a more appropriate method than traditional speed-dependant thresholds

on the basis that the latter might underestimate the true metabolic demands associated

with accelerated running. Some authors also suggest this metric might be useful for

informing players’ nutritional programming.1,5 However, the validity of metabolic

power to quantify the internal load of team sport activity has recently been questioned

due to a reported underestimation of energy expenditure owing to an inability to

detect non-ambulatory related activities.7 What remains unknown, however, is the

utility of this GPS-derived metric in collision-based sports, such as rugby. This is

particularly pertinent given that when physical contact is combined with intermittent

running, increases in heart rate, rating of perceived exertion and blood lactate

concentration suggest a greater metabolic response and anaerobic contribution when

compared to intermittent running alone. 8 Accordingly, this study sought to establish

the validity of energy expenditure derived from micro-technology when measured

during a repeated effort rugby protocol.

3

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

Methods

Participants and design

After ethics approval, sixteen university rugby players (age 23.8 ± 4.8 y; stature 1.80

± 0.05 m, body mass 84.5 ± 8.6 kg) volunteered to participate in this study.

Participants completed a single visit comprising a repeated effort protocol during

which movement activity and energy expenditure were measured.

The protocol comprised three sets of six ‘efforts’ with a strictly enforced 60 s standing

recovery between each set. One effort involved an 8 m run at 4 m·s -1 to make a

collision with a standard 30 kg tackle bag (Rhino Products, Leeds, UK). The

participant took the bag to the ground after collision, paused for 2 s, stood to re-

position the bag and then ran backwards at 2.5 m·s-1 to the start position. The protocol

was performed on a 3G artificial grass surface with running speeds controlled by an

audio signal emitted from a CD player. Environmental conditions were 22.8 ± 0.4ºC

and 37.3 ± 1.0% relative humidity.

Direct measures of internal load

Expired air was collected for each work (n = 3) and passive recovery (n = 3) period of

the protocol using a facemask connected to a Douglas bag (Hans Rudolph, UK). A

researcher ran alongside the participant carrying the Douglas bag, ensuring it

remained connected and did not disrupt running or tackling technique. Expired air

was analysed immediately for oxygen and carbon dioxide fractions using a previously

calibrated gas meter (Servomex 5200, Crowborough, UK) and gas volume using a dry

gas meter (Harvard apparatus, Kent, UK). These data were then used to calculate

oxygen uptake (VO2), RER and energy expenditure (EEVO2)10 for each work and rest

4

68

69

70

71

72

73

74

75

76

77

78

79

80

81

82

83

84

85

86

87

88

89

90

91

92

period. These values were then averaged over the whole protocol. The VO2 and

associated energy expenditure whilst standing was measured in all participants (2.9 ±

0.6 kcal) before exercise and subtracted from exercise VO2 to exclude resting

metabolic rate from calculation of internal load. Heart rate was measured using a

monitor (Polar Electro Oy, Kempele, Finland) strapped to the participant’s chest, with

data stored on the GPS unit (see below). Blood was collected from a fingertip

capillary sample 5 min after completing the protocol and analysed for lactate

concentration (Lactate Pro, Arkray, Japan).

GPS and accelerometer measures

A 10 Hz GPS device fitted with a 100 Hz tri-axial accelerometer, gyroscope and

magnetometer (Optimeye S5, Catapult Innovations, Melbourne, Australia) was

securely positioned between the participant’s scapulae using a custom-made vest. The

available satellites and HDOP were 16.3 ± 0.9 (range 15 - 18) and 0.7 ± 0.1 AU

(range 0.5 - 1.2), respectively. Data were later downloaded to a laptop and analysed

(Sprint, Version 5.1, Catapult Sports, VIC, Australia) to calculate distance (m), high-

speed running (>14 kmh-1) distance (m), Player Load (AU) and energy expenditure

(EEGPS; kcalmin-1) for each work and rest period.

Statistical analysis

All data are presented as mean ± SD. Data were checked for normality using the

Shapiro-Wilk statistic (P > 0.05). The difference between energy expenditure

determined via open circuit spirometry and GPS was assessed using a paired-samples

t-test. Agreement between measures of energy expenditure was assessed using 95%

limits of agreement (95%LoA), calculated as the mean difference between measures

5

93

94

95

96

97

98

99

100

101

102

103

104

105

106

107

108

109

110

111

112

113

114

115

116

117

(i.e. systematic bias) ± 1.96 x the SD of the differences between measures (i.e. the

‘random’ error in measurement). Pearson’s correlation coefficients with associated

95% confidence intervals (95%CI) were also performed to explore any relationship

between GPS-derived external load measurements (metabolic power, high-speed

running distance, Player Load) and the measurement of energy expenditure. To assess

the magnitude of the relationship, the following criteria were applied: < 0.1, trivial; >

0.1 – 0.3, small; > 0.3 – 0.5, moderate; > 0.5 – 0.7, large; > 0.7 – 0.9,  very large; and

> 0.9 – 1.0, extremely large.9 In all instances the alpha was set to P < 0.05.

Results

Physiological responses to the exercise protocol are reported in Table 1. Mean energy

expenditure over the protocol according to open circuit spirometry and GPS were 13.2

± 2.3 and 7.2 ± 1.0 kcalmin-1, respectively (Figure 1). This amounted to a total

energy expenditure of 82.8 ± 14.1 kcal using open circuit spirometry and 47.4 ± 6.9

kcal using GPS. The 95%LoA indicated a systematic underestimation of energy

expenditure in EEGPS (-5.94 ± 0.67 kcalmin-1), although the measures were

moderately related (r = 0.63, 95%CI 0.08-0.89; Figure 2). A moderate relationship

was also observed between high-speed running distance and EEVO2 (r = 0.50, 95%CI -

0.66-0.84, P < 0.05), but was low between Player Load and EEVO2 (r = 0.37, 95%CI -

0.81-0.68, P > 0.05).

**********Insert figures and table approximately here**********

6

118

119

120

121

122

123

124

125

126

127

128

129

130

131

132

133

134

135

136

137

138

139

140

141

142

143

Discussion

This study has demonstrated that, despite sharing a moderate relationship, energy

expenditure measured using micro-technology during a repeated effort contact drill in

rugby players is systematically lower than that measured using open-circuit

spirometry. The potential magnitude of underestimation during this specific form of

exercise, according to our 95%LoA, is between 5.27 and 6.61 kcalmin-1.

Extrapolated to, for example, a duration associated with rugby match-play (~40 min),3

then the absolute underestimation of energy expenditure could be 210 to 264 kcal.

The acceptability of this level of agreement is somewhat dependent on the purpose of

measurement. However, to provide some relevant context, the previously estimated

energy expenditure during ~40 min of rugby league match is approximately 908

kcal.11 In this context, future studies should investigate the magnitude of agreement in

energy expenditure using GPS and open-circuit spirometry associated with exercise

more closely simulating the exercise duration of contact team sports.

We reaffirm findings previously reported in team sport athletes1,7 that energy

expenditure is underestimated when calculated using metabolic power from micro-

technology. Our 45% lower energy expenditure using GPS derived metabolic power

is in agreement with Buchheit et al.,7 who employed similar procedures to report a

~51% difference between methods during a soccer-specific circuit. This is noteworthy

because, unlike Buchheit and co-workers,7 our measurements appropriately accounted

for the energetic cost of standing in the direct measure of energy expenditure.12 The

differences between methods in this study are much higher than the 11%

disagreement observed by Walker et al.1 However, rather than the direct measures

using open circuit spirometry as used here, Walker and colleagues1 used

7

144

145

146

147

148

149

150

151

152

153

154

155

156

157

158

159

160

161

162

163

164

165

166

167

168

accelerometer based predictions of energy expenditure that are known to

underestimate true values.13

The underestimation of energy expenditure in the present study might be because the

GPS is unable to detect increased energy transfer during periods of rest associated

with intermittent exercise (i.e. excess post-exercise oxygen consumption; EPOC).7

Indeed, Figure 1 shows that the difference between energy expenditure measures is

almost entirely accounted for by an increased VO2 during the rest periods. However,

the ‘real-time’ agreement between methods was not statistically analysed here,

because of a ‘lag’ in pulmonary VO2 kinetics relative to muscle VO2 (20-35 s).12 As

such, data were analysed over the whole protocol. This also served to incorporate the

inclusion of VO2 measured during recovery, which could in part account for the

replenishment of resources depleted using anaerobic metabolism.13 We acknowledge

that anaerobic metabolism would not be accounted for with measurements of VO2

whilst exercising, but which would presumably be incorporated in the GPS

measurement of metabolic power given it describes the energy required to

resynthesize ATP required for work performed at that time.12 Indeed, the relatively

high blood lactate concentrations observed here (10.5 ± 3.6 mmoll-1) provide an

indirect indication that there was a significant contribution of anaerobic metabolism to

ATP resynthesis during the exercise protocol.

Another potential explanation for the underestimation of energy expenditure by the

GPS is an inability to detect energy expenditure associated with non-locomotor

exertion. Metabolic power is derived from measures of acceleration and deceleration

and their corresponding energy cost predicted from steady state incline running.4 Thus

8

169

170

171

172

173

174

175

176

177

178

179

180

181

182

183

184

185

186

187

188

189

190

191

192

193

the energy cost predicted from acceleration/deceleration associated with contact, or

indeed measurement of energy expenditure with a static exertion, cannot be

determined from GPS. Given that contacts, such as those experienced during certain

team sports, increase the physiological exertion associated with intermittent running,8

it is likely that energy expenditure is increased by this type of activity, yet it is not

incorporated adequately into a calculation of metabolic power. While measures of

instantaneous velocity by the 10 Hz GPS device have acceptable validity and

reliability,15 criterion velocity during accelerations is still underestimated by these

devices.15 That these GPS measures contribute to the calculation of metabolic power

might also explain the observed underestimation.

We observed moderate associations between open-circuit spirometry derived energy

expenditure and metabolic power (r = 0.63) that was similar in magnitude but greater

than high-speed running (r = 0.50). Our data support previous assertions that

metabolic power more adequately reflects the physiological load associated with

intermittent running than traditional speed zones.5 Based on the low, non-significant

relationship (r = 0.37) between Player Load and energy expenditure, we suggest

Player Load does not reflect the metabolic load associated with intermittent exercise

employing collisions. Finally, we conclude that energy expenditure is underestimated

from micro-technology and therefore should be applied with caution when attempting

to quantify training load during intermittent team sports that involve collisions (e.g.

rugby, Australian rules football).

9

194

195

196

197

198

199

200

201

202

203

204

205

206

207

208

209

210

211

212

213

214

215

216

217

Acknowledgements: The authors wish to thank Richard Bott and Simon Cushman

for their technical assistance during this study.

References

1. Walker EJ, McAinch AJ, Sweeting A, Aughey RJ. Inertial sensors to estimate the energy expenditure of team-sport athletes. J Sci Med Sports. 2015;19:177-181.

2. Weaving D, Marshall P, Earle K, Nevill A, Abt G. A. combination of internal and external training load measures explains the greatest proportion of variance in certain training modes in professional rugby league. Int J Sports Physiol Perf. 2015;9:905-912.

3. Oxendale C, Twist C, Daniels M, Highton J. The relationship between match-play characteristics of elite rugby league and indirect markers of muscle damage. Int J Sports Physiol Perf. 2015.

4. Osgnach C, Poser S, Bernardini R, Rinaldo R, di Prampero P. Energy cost and metabolic power in elite soccer: A new match analysis approach. Med Sci Sports Exerc. 2010;42:170-178.

5. Kempton T, Sirotic AC, Rampinini E, Coutts AJ. Metabolic power demands of rugby league match-play. Int J Sports Physiol Perf. 2015;10:23-28.

6. Coutts AJ, Kempton T, Sullivan C, Bilsborough J, Cordy J, Rampinini E. Metabolic power and energetic costs of professional Australian football match-play. J Sci Med Sports. 2015;18:219-224.

7. Buchheit M, Manouvrier C, Cassirame J, Morin JB. Monitoring locomotor load in soccer: is metabolic power, powerful? Int J Sports Med. 2015;36:1149-1155.

8. Mullen T, Highton J, Twist C. The physiological and perceptual responses to a forward-specific rugby league simulation protocol with and without physical contact. Int J Sports Physiol Perf. 2015;10:746-753.

9. Hopkins WG. A scale of magnitudes for effect statistics. Retrieved from http://www.sportsci.org/resoruce/stats/index.html

10. Weir JB. New methods for calculating metabolic rate with special reference to protein metabolism. J. Physiol. 1949;109:1-9.

10

218

219

220

221

222

223

224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259

11. Coutts AJ, Reaburn P, Abt G. Heart rate, blood lactate concentration and estimated energy expenditure in a semi-professional rugby league team during a match: a case study. J Sports Sci. 2003;21:97-103.

12. Osgnach C, Paolini E, Roberti V, Vettor M, di Prampero PE. Metabolic power and oxygen consumption in team sports: A brief response to Buchheit et al. Int J Sports Med. 2016;37:77-81.

13. Tabata I, Irisawa K, Kouzaki M, Nishimura K, Ogita F, Miyachi M. Metabolic profile of high intensity intermittent exercises. Med Sci Sports Exerc. 1997;29:390-395.

14. Crouter SE, Churilla JR, Bassett DR. Estimating energy expenditure using accelerometers. Eur J Appl Physiol. 2006;98:601-612.

15. Varley MC, Fairweather IH, Aughey RJ. Validity and reliability of GPS formeasuring instantaneous velocity during acceleration, deceleration, and constant motion. J Sports Sci. 2012;30:121-127.

11

260261262263264265266267268269270271272273274275276277278

279

Table 1. Internal and external load measured during the exercise protocol

Mean ± SDInternal Load

Energy Expenditure (kcal∙min-1) 13.2 ± 2.3Heart rate (%max) 87.4 ± 4.1B[La] (mmol∙l-1) 10.5 ± 3.6

External LoadDistance (m) 391.3 ± 16.8High-speed running (m) 53.9 ± 31.2Metabolic power (kcal∙min-1) 7.2 ± 1.0Metabolic power >20 W∙kg-1 (s) 41 ± 9Player load (AU) 54.4 ± 5.5

12

280

281

282

283

284

Figure legends

Figure 1. Energy expenditure determined using global position system-derived

metabolic power (GPS) and open-circuit spirometry (VO2) during exercise and rest

bouts. White bars denote the metabolic rate associated with standing, whilst the

shaded bars represent the energy expenditure associated with exercise (i.e. delta

energy expenditure). * Significant (P < 0.05) difference between measures.

Figure 2. Relationship between energy expenditure predicted from open circuit

spirometry and GPS

13

285

286

287

288

289

290

291

292

293

294

295

296

297

Figure 1

14

298299

300

301

Figure 2

15

302303

304