Training & Testing

Gaudino P et al. Monitoring Training in Elite �… Int J Sports Med

accepted after revision February 11 , 2013

BibliographyDOI http://dx.doi.org/10.1055/s-0033-1337943Published online: 2013Int J Sports Med © Georg Thieme Verlag KG Stuttgart · New YorkISSN 0172-4622

Correspondence Paolo Gaudino, Master Degree in Sport Science Department of Biomedical Sciences for Health Università degli Studi di Milano Via Kramer 4 Milano Italy 20129 paolo.gaudino@unimi.it

Key words GPS acceleration deceleration energy cost high-intensity position

Monitoring Training in Elite Soccer Players: Systematic Bias between Running Speed and Metabolic Power Data

tunity to derive valid and reliable estimates of the speed attained and distance covered during a range of activities [ 29 �– 31 , 34 ] . Consequently an increasing body of literature is beginning to emerge which serves to quantify the external load placed upon athletes during multi-direc-tional sports such as soccer [ 8 , 11 , 19 , 20 ] . To date, assessment of the external load during soccer-related activities using GPS technology has frequently centred around evaluating the dis-tance covered or time spent at speci c velocities with particularly attention focused upon the volume of high-speed activity given its reported importance to match-play performance [ 4 , 14 , 21 , 32 ] . This representation of the external load, however, does not account for the addi-tional distance covered or energy demands asso-ciated with accelerations and decelerations that can also be derived through GPS. As a conse-quence, since accelerations and decelerations further increase the energy demands placed on the athlete even when running within low speed thresholds, the traditional approach will under-estimate the total energy cost associated with

Introduction Training is a process of adaptation where enhancements in performance are achieved through progressive manipulation of the training load [ 23 , 26 ] . As a consequence, accurate assess-ment of an individual�’s training load represents an essential component of e ective training pre-scription. Evaluating the physical demands of training requires accurate assessment of both the internal and external load. This is particularly important in team sports such as soccer since di erences in individual responses to the same external work-load arise [ 23 ] . A number of approaches are fre-quently used to quantify the internal training load [ 1 , 5 , 10 , 16 ] . However, the multi-directional basis of sports such as soccer has previously made quanti cation of the external training load di cult to achieve. Consequently, traditional approaches have often focused solely upon the duration and frequency of the training stimuli [ 7 ] . More recently, the evolution of global posi-tioning systems (GPS) have provided the oppor-

Authors P. Gaudino 1 , 2 , F. M. Iaia 2 , G. Alberti 2 , A. J. Strudwick 1 , G. Atkinson 3 , W. Gregson 1 , 4

A liations 1 Research Institute for Sport and Exercise Sciences, Liverpool John Moores University, Liverpool, United Kingdom 2 Department of Biomedical Sciences for Health, Università degli Studi di Milano, Milano, Italy 3 Health and Social Care Institute, Teesside University, Middlesbrough, United Kingdom 4 ASPIRE, Academy for Sports Excellence, Doha, Qatar

Abstract We compared measurements of high-intensity activity during eld-based training sessions in elite soccer players of di erent playing posi-tions. Agreement was appraised between meas-urements of running speed alone and predicted metabolic power derived from a combination of running speed and acceleration. Data was collected during a 10-week phase of the com-petitive season from 26 English Premier League out eld players using global positioning sys-tem technology. High-intensity activity was estimated using the total distance covered at speeds > 14.4 km · h 1 (TS) and the equivalent metabolic power threshold of > 20 W · kg 1 (TP), respectively. We selected 0.2 as the minimally

important standardised di erence between methods. Mean training session TS was 478 ± 300 m vs. 727 ± 338 m for TP (p < 0.001). This di erence was greater for central defenders (~ 85 %) vs. wide defenders and attackers (~ 60 %) (p < 0.05). The di erence between methods also decreased as the proportion of high-intensity distance within a training session increased (R 2 = 0.43; p < 0.001). We conclude that the high-intensity demands of soccer training are underestimated by traditional measurements of running speed alone, especially in training ses-sions or playing positions associated with less high-intensity activity. Estimations of metabolic power better inform the coach as to the true demands of a training session.

Dow

nloa

ded

by: L

iver

pool

Joh

n M

oore

s U

nive

rsity

. Cop

yrig

hted

mat

eria

l.

Training & Testing

Gaudino P et al. Monitoring Training in Elite �… Int J Sports Med

soccer related activity [ 9 , 12 , 27 ] . However, limited attention to date has focused on acceleration and deceleration activity in elite soccer players [ 14 ] and thus the contribution of these activ-ities to estimates of the external load incurred by soccer players. In line with such observations, di Prampero et al. [ 12 ] recently introduced a new approach to estimating the energy cost of accelerated and decelerated running. When combined with tra-ditional estimates of running speed, this method permits a more comprehensive assessment of the overall energy cost of the activity in any given moment based on metabolic power values [ 12 ] . Using this approach, Osgnach et al. [ 27 ] recently reported that the energy cost associated with high-intensity activity dur-ing match-play was 2�–3 to three times larger than estimates based solely on running speed. Despite these observations, no study to date has compared the energy costs of eld-based training in soccer using the 2 di er-ent approaches to determine the extent to which traditional approaches may underestimate the true demand. Such informa-tion is important since accurate determination of the training load placed upon athletes is critical in attempting to maximise performance enhancement and injury prevention strategies. Therefore, the aim of the current investigation was to compare the energy cost of training when derived from the approach recently introduced by di Prampero et al. [ 12 ] as opposed to the traditional approach of distance covered at speci c running speeds. Furthermore, this investigation also served to evaluate whether the degree to which the 2 approaches di er was dependent upon playing position and the type of training ses-sion undertaken.

Methods Players and training observations Training data was collected from 26 soccer players competing in the English Premier League (age = 26 ± 5 years; height = 182 ± 7 cm; body mass = 79 ± 5 kg) over a 10-week period during the 2011�–2012 in-season competition period. A total of 628 individual training observations were undertaken on out eld players (goal-keepers were excluded) with a median of 24 training sessions per players (range = 3�–36). Players were assigned to one of 5 positional groups: central defender (training observations = 92), wide defender (training observations = 110), central mid elder (training observations = 103), wide mid elder (training observa-tions = 145) and attacker (training observations = 178). Only data derived from the team eld-based training sessions was ana-lysed, and no individual rehabilitation or individual tness ses-sions were included for analysis. The warm-up period prior to each training session was not included for analysis. All players were noti ed of the aim of the study, research procedures, requirements, bene ts and risks before giving written informed consent. The Ethics Committee of the University of Milan approved the study, which was performed in accordance with the ethical standards of the IJSM [ 18 ] .

Data collection The players�’ physical activity during each training sessions was monitored using a portable global positioning system (GPS) technology (GPSports SPI Pro X, Canberra, Australia). This ver-sion of the SPI Pro provides raw position, velocity and distance data at 15 Hz (15 samples per second). For the purpose of this study, every 3 raw data points was averaged to provide a sam-

pling frequency of 5 Hz. Previous models of this GPS have been shown to provide valid and reliable estimates of distance and velocity during linear, multidirectional and soccer-speci c activities [ 29 �– 31 , 34 ] . All devices were always activated 15-min before the data collection to allow acquisition of satellite signals [ 15 , 34 ] . The minimum acceptable number of available satellite signals was 8 (range 8�–11) [ 31 , 34 ] . Data was eliminated on days when the satellite signal was below this value. In addition, play-ers wore the same GPS device for each training sessions in order to avoid inter-unit error [ 15 , 22 ] .

Energy cost and metabolic power In order to estimate the energy cost (EC) and metabolic power (P met ) at any given moment within a training session, the equa-tion proposed by di Prampero et al. [ 12 ] based on previously studies of Minetti et al. [ 25 ] and then modi ed by Osgnach et al. [ 27 ] to evaluate soccer players sprinting on grass was adopted (eq. 1).

EC = (155.4 · ES 5 �–30.4 · ES 4 �–43.3 · ES 3 + 46.3 · ES 2 + 19.5 · ES + 3.6) · EM · KT (1)

Where EC is the energy cost of accelerated running on grass (in J · kg 1 · m 1 ), ES is the equivalent slope: ES = tan (90-arctan g/a f ); g = Earth�’s acceleration of gravity; a f = forward acceleration; EM is the equivalent body mass: EM = (a f 2 /g 2 + 1) 0.5 ; and KT is a con-stant (KT = 1.29). Consequently, P met in W · kg 1 (eq. 2) was calculated by multiply-ing EC (in J · kg 1 · m 1 ) by running speed (v; in m · s 1 ) at any given moment (i. e., every 0.2 s):

P met = EC · v (2)

Physical performance The physical demands of each training session for each player were evaluated through the assessment of speed and P met . The following 3 high-speed categories were used: high speed (HS; from 14.4 to 19.8 km · h 1 ), very high speed (VHS; from 19.8 to 25.2 km · h 1 ) and maximal speed (MS; > 25.2 km · h 1 ) [ 14 , 17 ] . Metabolic power categories were de ned as: high power (HP; from 20 to 35 W · kg 1 ), elevated power (EP; from 35 to 55 W · kg 1 ) and maximum power (MP; > 55 W · kg 1 ) [ 27 ] . To compare the high-intensity energy costs of training when based on speed compared to P met , the total distance covered at a speed > 14.4 km · h 1 (TS) and the equivalent P met ( > 20 W · kg 1 ; total high-metabolic power; TP) were estimated. This threshold was set since 20 W · kg 1 is the P met when running at a constant speed of approximately 14.4 km · h 1 on grass [ 27 ] . Training duration, total distance covered and distance covered in the di erent speed categories were calculated using a custom Excel spreadsheet from instantaneous raw data of time, speed and distance available from the SPI Pro X software Team AMS (GPSports SPI Pro X, Canberra, Australia). In the same program instantaneous acceleration values were calculated by dividing change in velocity by the change in time. Finally, equations 1 and 2 were also integrated into the custom spreadsheet in order to calculate total energy expenditure, average metabolic power and distance covered in di erent metabolic power categories.

Statistical analysis Data is presented as mean ± SD. It has been advised that method agreement-type studies should involve at least 40 participants

Dow

nloa

ded

by: L

iver

pool

Joh

n M

oore

s U

nive

rsity

. Cop

yrig

hted

mat

eria

l.

Training & Testing

Gaudino P et al. Monitoring Training in Elite �… Int J Sports Med

for adequate precision of error estimates [ 3 ] . Our sampling frame was a typical squad size (n = 26) for an elite soccer team. However, up to 24 repeated measurements were made on indi-vidual players, which increases statistical precision [ 2 , 35 ] . Data was analysed with a Generalized Estimating Equation model, which included the within-subjects factors of method type, player position and repeated training session. Signi cant main e ects and interactions between factors were followed up with least signi cant di erence comparisons [ 28 ] . The level for evaluation of statistical signi cance was set at p < 0.05. Simple e ect size, estimated from the ratio of the mean di erence to the pooled standard deviation, was also calculated. E ect size values of 0.2, 0.5 and 0.8 were considered to represent small, moderate and large di erences, respectively [ 33 ] .

Results Training duration and distance covered in each high speed category Table 1 shows the mean duration of training and distance cov-ered in each speed category across the di erent playing posi-tions. Overall session duration was 56.7 ± 18.0 min (mean ± SD) during which the players covered 3 772 ± 1 276 m. Mean HS, VHS and MS distance completed were 357 ± 218 m, 102 ± 94 m and 19 ± 34 m, respectively (mean ± SD). Training duration was simi-lar between playing positions ( Table 1 ). With the exception of attackers, the total distance covered by central mid elders was signi cantly greater compared to all other positions (central defenders, e ect size = 0.5; wide defenders, e ect size = 0.3; wide mid elders, e ect size = 0.4; p < 0.05; Table 1 ) with a similar distance observed between remaining positions. The highest and lowest HS distance was covered by central mid eld-ers and central defenders, respectively (p < 0.001; Table 1 ). The amount of VHS undertaken by wide defenders (p = 0.026, e ect size = 0.6) and attackers (p = 0.002, e ect size = 0.5) was greater than central defenders, respectively ( Table 1 ). No other di er-ences were observed between positions including MS, which was similar between all positions ( Table 1 ).

Energy cost and metabolic power Table 2 outlines the EC and P met of training in relation to play-ing position. EC and P met per training session was 24.7 ± 8.8 kJ · kg 1 and 7.7 ± 1.1 W · kg 1 , respectively (mean ± SD). Within each P met category mean distances of 425 ± 202 m, 147 ± 67 m and 155 ± 83 m were observed for HP, EP and MP, respectively. With the exception of attackers, EC during training was greater in cen-tral mid elders compared to all other playing positions (central defenders, e ect size = 0.5; wide defenders, e ect size = 0.4; wide mid elders, e ect size = 0.5; p < 0.05; Table 2 ). Similarly, P met was greater in central mid elders compared to all other positions (central defenders, e ect size = 0.5; wide defenders, e ect size = 0.8; wide mid elders, e ect size = 0.7; attackers, e ect size = 0.6; p < 0.01; Table 2 ). With respect to the di erent P met categories, central mid elders completed a greater HP dis-tance compared to all other positions (central defenders, e ect size = 0.5; wide defenders, e ect size = 0.8; wide mid elders, e ect size = 0.7; attackers, e ect size = 0.6; p < 0.01) and a greater EP distance compared to central defenders (p = 0.001, e ect size = 0.6; Table 2 ). Distance covered in the MP category was greater in central mid elders compared to central defenders (p = 0.009, e ect size = 0.5) and wide mid elders (p < 0.001, e ect size = 0.6). No other di erences were observed between playing positions ( Table 2 ).

Training demands: TS vs. TP Table 3 compares the high-intensity activity distance covered during training when expressed as total high-speed run-ning ( > 14.4 km · h 1 ; TS) and total high-metabolic power ( > 20 W · kg 1 ; TP). The TP (727 ± 338 m; mean ± SD) was signi -cantly greater than TS (478 ± 300 m; (p < 0.001; e ect size = 0.8). The magnitude of this di erence ( % change) was also dependent upon playing position ( Table 3 ) with central defenders dis-playing a greater % change relative to both wide defenders (p = 0.01, e ect size = 0.4) and attackers (p = 0.02, e ect size = 0.4). No other di erences were observed between the remaining playing positions. Further analysis of the di erence in high-intensity activity when derived by the 2 methods indicated that the magnitude of the di erence increased as the percentage of

Table 1 Training session duration and distance covered at di erent speed in relation to playing position (mean ± SD).

Central defender

(n = 92)

Wide defender

(n = 110)

Central mid elder

(n = 103)

Wide mid elder

(n = 145)

Attacker

(n = 178)

Follow-up Tests

Duration (min) 53.4 ± 18.4 55.9 ± 18.1 58.1 ± 19.7 55.2 ± 16.6 59.5 ± 17.4 CD = WD = CM = WM = A Total Distance (m) 3 498 ± 1 204 3 647 ± 1 302 4 133 ± 1 538 3 618 ± 1 138 3 906 ± 1 183 (CM > CD = WM = WD) = A* HS (m) 285 ± 128 370 ± 218 442 ± 332 347 ± 183 344 ± 180 CM > WM = A = (WD > CD)* VHS (m) 72 ± 57 112 ± 89 108 ± 105 91 ± 77 116 ± 112 CM = WM = (WD = A > CD)* MS (m) 16 ± 31 20 ± 33 21 ± 37 17 ± 28 21 ± 38 CD = WD = CM = WM = A HS = High speed (14.4�–19.8 km · h 1 ); VHS = Very high speed (19.8�–25.2 km · h 1 ); MS = Maximal speed ( > 25.2 km · h 1 ). *Signi cant di erence between playing positions (p < 0.05)

Table 2 Energy cost and metabolic power in relation to playing position (mean ± SD).

Central Defender

(n = 92)

Wide Defender

(n = 110)

Central Mid elder

(n = 103)

Wide Mid elder

(n = 145)

Attacker

(n = 178)

Follow-up Tests

EC (kJ · kg 1 ) 23.3 ± 8.3 23.8 ± 8.7 27.6 ± 10.7 23.0 ± 7.7 25.6 ± 8.3 (CM > CD = WD = WM) = A* P met (W · kg 1 ) 7.8 ± 1.1 7.5 ± 1.0 8.4 ± 1.2 7.6 ± 1.0 7.7 ± 1.2 CM > (CD = WD = WM = A)* HP (m) 356 ± 136 429 ± 208 513 ± 276 418 ± 170 411 ± 183 CM > (CD = WD = WM = A)* EP (m) 130 ± 53 143 ± 69 168 ± 79 145 ± 60 146 ± 68 (CM > CD) = WD = WM = A* MP (m) 143 ± 68 153 ± 73 182 ± 93 136 ± 73 164 ± 92 (CM > CD = WM) = WD = A* EC = Energy cost; P met = Mean metabolic power; HP = High P met (20�–35 W · kg 1 ); EP = Elevated P met (35�–55 W · kg 1 ); MP = Maximal P met ( > 55 W · kg 1 ). *Signi cant di erence between playing positions (p < 0.05)

Dow

nloa

ded

by: L

iver

pool

Joh

n M

oore

s U

nive

rsity

. Cop

yrig

hted

mat

eria

l.

Training & Testing

Gaudino P et al. Monitoring Training in Elite �… Int J Sports Med

high intensity distance covered per session (average between %TS and %TP) decreased (R 2 = 0.43; p < 0.001; Fig. 1 ).

Discussion Attempts to evaluate the physical demands of soccer-speci c activity have routinely centred upon the distance covered or time spent at di erent speeds. Recent observation from match-play, however, suggest that such approaches underestimate the total energy cost since they fail to take into account the energy demands associated with accelerations and decelerations [ 27 ] . Using the approach of Osgnach and colleagues [ 27 ] , based on metabolic power values, the present ndings demonstrate that previous approaches also underestimate the high-intensity demands of soccer training in elite players to a similar extent to those observed in match-play. Furthermore, this underestima-tion is greater in training sessions or playing positions associ-ated with less high-intensity activity. To the authors knowledge the present investigation represents the rst attempt to quantify the overall external load typically associated with daily eld-based training activity in elite soccer players. The mean total distance covered (3 772 m; range = 831�–9 502 m) and the mean TS distance covered (478 m; range = 0�–2 272 m) per session equated to ~40 and 20% of the distance typically covered during match-play, respectively [ 13 , 27 , 32 ] . In line with match-play observations [ 13 , 14 , 17 , 32 ] , there were signi cant di erences between playing positions re ecting position-speci c training methodologies that are rou-

tinely adopted in order to prepare the players for the physical and technical demands of match-play. For example, the total dis-tance was generally highest in central mid eld players with the lowest values in central defenders. Similarly, the highest and lowest TS distance was observed in central mid elders and cen-tral defenders, respectively, with attackers covering the greatest VHS distance. Interestingly, the distance covered at MS did not show any di erences between playing positions. This may partly re ect the fact that players may not frequently reach maximal speeds during soccer-speci c training activities [ 24 ] . Indeed, the average distance covered at MS during training sessions repre-sented only the 0.5 % of total distance covered. To date, assessment of the external training load in soccer using GPS technology has typically centred around the distance cov-ered or time spent undertaking movements at determined speed categories [ 4 , 20 ] . This approach, however, underestimates the total energy cost since it fails to incorporate the energy cost associated with accelerations and decelerations which fre-quently arise during soccer-speci c activities [ 6 , 27 ] . The latter represent more energetically demanding movements than con-stant-velocity movements to such an extent that high metabolic demands also arise at low running speeds in the presence of high accelerations or decelerations [ 9 , 12 , 27 ] . In the present investigation the mean energy cost associated with training was ~25 kJ · kg 1 (range = 5�–67 kJ · kg 1 ) compared to ~60 kJ · kg 1 observed during match-play [ 27 ] . When expressed as average metabolic power this equated to ~7.5 W · kg 1 (range = 4.6�–12.8 W · kg 1 ). Interestingly, when comparing the total energy cost of training sessions between playing positions mean energy cost was similar in central mid elders and attackers. In contrast, when expressed as metabolic power, higher values were observed in central mid elders (~8.5 W · kg 1 vs. ~7.7 W · kg 1 ). This likely re ects the higher volume of high-intensity activity undertaken by central mid elders compared to attackers and suggests that assessments of the physical demands of training using metabolic power data is more precise since it take into account both speed and acceleration values. Further support for the application of metabolic power is pro-vided when examining the high-intensity component of training which typically represents the most physically demanding ele-ments. Since the metabolic power when running at constant speed on grass at 14.4 km · h 1 is approximately 20 W · kg 1 [ 27 ] , the extent to which the use of speed per se underestimates the true energy cost of activity can be further explored by comparing the distance covered at a speed > 14.4 km · h 1 (TS) with the dis-tance at a metabolic power > 20 W · kg 1 (TP). In the present study 13 % of the total distance was covered at TS compared to 19 % at TP indicating that traditional approaches may underestimate the high-intensity demands of training by ~6 %. These estimations compare favourably with underestimation of ~8 % (18 vs. 26 %) recently reported by Osgnach et al. [ 27 ] during match-play.

Table 3 Total high-intensity training distance covered estimated from high-speed running ( > 14.4 km · h 1 ; TS) and high-metabolic power ( > 20 W · kg 1 ; TP) relative to playing position (mean ± SD).

Central defender

(n = 92)

Wide defender

(n = 110)

Central mid elder

(n = 103)

Wide mid elder

(n = 145)

Attacker

(n = 178)

Follow-up Tests

TS (m) 373 ± 179 502 ± 306 570 ± 403 455 ± 259 482 ± 291 (CM > WM) = WD = A > CD* TP (m) 628 ± 250 + 725 ± 342 + 863 ± 436 + 699 ± 291 + 722 ± 324 + CM > (CD = WD = WM = A)* % change 84 ± 59 62 ± 49 70 ± 48 72 ± 53 63 ± 38 CM = WM = (CD > WD = A)* TS = Total high-speed ( > 14.4 km · h 1 ); TP = Total high-metabolic power ( > 20 W · kg 1 ). + Signi cant di erence from TS (p < 0.05). *Signi cant di erence between playing posi-tions (p < 0.05)

Fig. 1 Bland-Altman plot comparing the di erence in high-intensity training distance covered ( % of total training session distance) estimated from high-speed running ( > 14.4 km · h 1 ; % TS) and high-metabolic power ( > 20 W · kg 1 ; % TP) relative to the mean (p < 0.001).

16 y = –0.3861x + 12.519R2 = 0.4269

14

12

10

% T

P - %

TS

8

6

4

2

00 5 10

Average % TP and % TS15 20 25 30 35

Dow

nloa

ded

by: L

iver

pool

Joh

n M

oore

s U

nive

rsity

. Cop

yrig

hted

mat

eria

l.

Training & Testing

Gaudino P et al. Monitoring Training in Elite �… Int J Sports Med

The degree to which the high-intensity demands of training are underestimated when based upon speed categories only may also be in uenced by playing position. For example, central defenders displayed a greater di erence (represented as % change) between TS and TP compared to both wide defenders and attackers. It would seem likely that these di erences are directly related to the nature of their involvement in match-play and training [ 14 ] . The reactive nature of the work undertaken by central defenders as they attempt to counter the movements of the opposition may require a high number of brief explosive accelerations and decelerations [ 14 ] . In contrast, wide defend-ers, particularly more attacking-orientated defenders and attackers may produce less explosive acceleration as a function of the freedom they have to dictate their own activity pro le as a consequence of the need to initiate movement patterns to cre-ate attacking opportunities [ 14 ] . Alternatively, the high TS val-ues reported in wide defenders may indicate simply that a large proportion of their high-intensity activity occurs at �“constant high�” speeds. As accelerations and decelerations are physical demanding tasks [ 9 , 12 , 27 ] , and underestimation of the stress imposed by these activities on players during training could in uence the degree to which training in uences adaptation and thus performance as well as the incidence of injury. Limited attention to date has focused upon acceleration in elite soccer players [ 14 ] . Consequently, further work is required in order to characterize the acceleration and deceleration pro les encoun-tered by di erent playing position during both match-play and training in order to inform the training process. Alongside examining the in uence of playing position, we also sought to determine whether the type of training session, spe-ci cally the amount of high-intensity activity undertaken within a training session, in uenced the degree to which the 2 methods of estimating training load di ered. We observed a trend for the magnitude of the di erence between methods to decline as the amount of high-intensity activity within a train-ing session increases. This suggests that during training sessions which incorporate a large percentage of high-intensity activity, underestimation of the true external load using traditional monitoring approach will be minimised. Conversely, training sessions with limited high-intensity activity in the presence of accelerations and decelerations will magnify the di erence between the 2 methods. Given that di erent approaches to training are employed by di erent coaches (e. g. di erent types of small-sided games), the present data may suggest that certain training strategies may have greater implications for deriving true estimates of the external load placed upon players using traditional approaches. It should be noted, however, that no dif-ferentiation between types of training session was undertaken within the present investigation. Consequently, further work is needed to provide a more detailed comparison of the two esti-mates across di erent types of training. In conclusion, the high-intensity demands of soccer training in elite players are underestimated by traditional measurements of running speed alone, especially in training sessions or playing positions associated with less high-intensity activity. Estima-tions of metabolic power better inform the coach as to the true demands of a training session. Consequently, the use of this monitoring approach may contribute to the development of training programmes which serve to further enhance perform-ance and reduce the incidence of injury.

Acknowledgements The authors want to thank Dr. Richard Hawkins and Robin Thorpe for their assistance with data collection. The authors have no con ict of interest that is directly relevant to the content of this research. No nancial support was provided for this investigation.

References1 Alexiou H , Coutts A J . A comparison of methods used for quantifying

internal training load in women soccer players . Int J Sports Physiol Perform 2008 ; 3 : 320 �– 330

2 Atkinson G , Batterham A M . The use of ratios and percentage changes in sports medicine: time for a rethink? . Int J Sports Med 2012 ; 33 : 505 �– 506

3 Atkinson G , Davison R C , Nevill A M . Performance characteristics of gas analysis systems: what we know and what we need to know . Int J Sports Med 2005 ; 26 (Suppl 1 ): S2 �– S10

4 Aughey R J . Applications of GPS technologies to eld sports . Int J Sports Physiol Perform 2011 ; 6 : 295 �– 310

5 Borresen J , Lambert M I . The quanti cation of training load, the train-ing response and the e ect on performance . Sports Med 2009 ; 39 : 779 �– 795

6 Bradley P S , Di Mascio M , Peart D , Olsen P , Sheldon B . High-intensity activity pro les of elite soccer players at di erent performance levels . J Strength Cond Res 2009 ; 24 : 2343 �– 2351

7 Brink M S , Nederhof E , Visscher C , Schmikli S L , Lemmink K A . Monitor-ing load, recovery, and performance in young elite soccer players . J Strength Cond Res 2010 ; 24 : 597 �– 603

8 Casamichana D , Castellano J . Time-motion, heart rate, perceptual and motor behaviour demands in small-sides soccer games: e ects of pitch size . J Sports Sci 2010 ; 28 : 1615 �– 1623

9 Cavagna G A , Komarek L , Mazzoleni S . The mechanics of sprint running . J Physiol 1971 ; 217 : 709 �– 721

10 Coutts A J , Rampinini E , Marcora S M , Castagna C , Impellizzeri F M . Heart rate and blood lactate correlates of perceived exertion during small-sided soccer games . J Sci Med Sport 2009 ; 12 : 79 �– 84

11 Dellal A , Hill-Haas S , Lago-Penas C , Chamari K . Small-Sided Games in Soccer: Amateur vs. Professional Players�’ Physiological Responses, Physical, and Technical Activities . J Strength Cond Res 2011 ; 25 : 2371 �– 2381

12 di Prampero P E , Fusi S , Sepulcri L , Morin J B , Belli A , Antonutto G . Sprint running: a new energetic approach . J Exp Biol 2005 ; 208 : 2809 �– 2816

13 Di Salvo V , Baron R , Tschan H , Calderon Montero F J , Bachl N , Pigozzi F . Performance characteristics according to playing position in elite soccer . Int J Sports Med 2007 ; 28 : 222 �– 227

14 Di Salvo V , Gregson W , Atkinson G , Tordo P , Drust B . Analysis of high intensity activity in Premier League soccer . Int J Sports Med 2009 ; 30 : 205 �– 212

15 Du eld R , Reid M , Baker J , Spratford W . Accuracy and reliability of GPS devices for measurement of movement patterns in con ned spaces for court-based sports . J Sci Med Sport 2010 ; 13 : 523 �– 525

16 Foster C , Florhaug J A , Franklin J , Gottschall L , Hrovatin L A , Parker S , Doleshal P , Dodge C . A new approach to monitoring exercise training . J Strength Cond Res 2001 ; 15 : 109 �– 115

17 Gregson W , Drust B , Atkinson G , Di Salvo V . Match-to-match variability of high-speed activities in premier league soccer . Int J Sports Med 2010 ; 31 : 237 �– 242

18 Harriss D J , Atkinson G . Update �– ethical standards in sport and exercise science research . Int J Sports Med 2011 ; 32 : 819 �– 821

19 Hill-Haas S V , Dawson B , Coutts A J , Rowsell G J . Physiological responses and time-motion characteristics of various small-sided soccer games in youth players . J Sports Sci 2009 ; 27 : 1 �– 8

20 Hill-Haas S V , Dawson B , Impellizzeri F M , Coutts A J . Physiology of small-sided games training in football: a systematic review . Sports Med 2011 ; 41 : 199 �– 220

21 Iaia F M , Rampinini E , Bangsbo J . High-intensity training in football . Int J Sports Physiol Perform 2009 ; 4 : 291 �– 306

22 Jenninngs D , Cormack S , Coutts A J , Boyd L J , Aughey R J . Variability of GPS units for measuring distance in team sport movements . Int J Sports Physiol Perform 2010 ; 5 : 565 �– 569

23 Manzi V , D'Ottavio S , Impellizzeri F M , Chaouachi A , Chamari K , Castagna C . Pro le of weekly training load in elite male professional basketball players . J Strength Cond Res 2010 ; 24 : 1399 �– 1406

Dow

nloa

ded

by: L

iver

pool

Joh

n M

oore

s U

nive

rsity

. Cop

yrig

hted

mat

eria

l.

Training & Testing

Gaudino P et al. Monitoring Training in Elite �… Int J Sports Med

24 Mendez-Villanueva A , Buchheit M , Simpson B , Peltola E . Does on- eld sprinting performance in young soccer players depend on how fast the can run or how fast they do run? J Strength Cond Res 2011 ; 25 : 2634 �– 2638

25 Minetti A E , Moia C , Roi G S , Susta D , Ferretti G . Energy cost of walking and running at extreme uphill and downhill slopes . J Appl Physiol 2002 ; 93 : 1039 �– 1046

26 Mujika I , Padilla S , Pyne D , Busso T . Physiological changes associated with the pre-event taper in athletes . Sports Med 2004 ; 34 : 891 �– 927

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

28 Perneger T V . What�’s wrong with Bonferroni adjustments . BMJ 1998 ; 316 : 1236 �– 1238

29 Portas M D , Harley J A , Barnes C A , Rush C J . The validity and reliability of 1-Hz and 5-Hz global positioning systems for linear, multidirectional, and soccer-speci c activities . Int J Sports Physiol Perform 2010 ; 5 : 448 �– 458

30 Randers M B , Mujika I , Hewitt A , Santisteban J , Bischo R , Solano R , Zubillaga A , Peltola E , Krustrup P , Mohr M . Application of four di er-ent football match analysis systems: a comparative study . J Sports Sci 2010 ; 28 : 171 �– 182

31 Varley M C , Fairweather I H , Aughey RJ . Validity and reliability of GPS for measuring instantaneous velocity during acceleration, deceleration, and constant motion . J Sports Sci 2012 ; 30 : 121 �– 127

32 Vigne G , Gaudino C , Rogowski I , Alloatti G , Hautier C . Activity pro le in elite Italian soccer team . Int J Sports Med 2010 ; 31 : 304 �– 310

33 Vincent W J . Statistics in Kinesiology . Champaign, IL. : Human Kinetics , 1999

34 Waldron M , Worsfold P , Twist C , Lamb K . Concurrent validity and test-retest reliability of a global positioning system (GPS) and timing gate to assess sprint performance variables . J Sports Sci 2011 ; 29 : 1613 �– 1619

35 < http://sportsci.org > , http://sportsci.org/resource/stats/precision.html

Dow

nloa

ded

by: L

iver

pool

Joh

n M

oore

s U

nive

rsity

. Cop

yrig

hted

mat

eria

l.