Position specific peak impact and running demands of professional rugby union players during game play

Abstract

This study quantifies the maximum number of impacts and peak running demands during 1– to 10– minute rolling window periods in elite rugby union matches using a multi-team dataset (n = 2232 player-games). Maximum values for impacts (impacts·min⁻¹) and running (m·min⁻¹) were calculated for 161 athletes from four teams across the 2018 and 2019 Super Rugby seasons. The effect of window duration and playing position on peak impact and running demands were estimated using linear mixed effect models and prediction intervals. The peak impact and running demands for a 1-min period were 4.5 – 5.5 impacts·min⁻¹ and 150 – 180 m·min⁻¹, depending on playing position. While small variations in mean impact and running movements could be observed by position, the large prediction interval and individual player variation meant that there was no practically meaningful difference by position. As such, when prescribing training drills to replicate the peak demands in rugby union, impact and running movements of players can be similar, regardless of position. Using a prediction interval allows us to identify the range where the demands in a future game may fall, and are beneficial to use when also trying to prepare players for the demands of rugby union.

Keywords

Global positioning system (GPS),performance analysis,playing position,team sport

Introduction

Rugby union (RU) is an international professional team sport comprised of intermittently high and low intensity activities.¹ Locomotor demands have been widely reported with distance covered between 3124 and 7427 m per game, and the average relative distance between 61 m·min⁻¹ and 85 m·min⁻¹.^1–5 The occurrence and frequency of high force contacts during competition is one of the main differences between rugby codes (RU, rugby league, rugby sevens) and other team sports. Understanding the contact demands on players has been reported in rugby league, and less so in RU.⁶ Given the prescence of different collision based activities in RU that are not present in rugby league (e.g. rucks, mauls, and contested scrums), it is important to differentiate these sports regarding collision demands. As such, current research on collisions in rugby league⁶ are likely poorly transferable to RU. The average number of contact activities in RU game play (including tackles, carries into contact, and rucks involving collisions) from manual video analysis⁷ and accelerometer detection⁷,⁸ is 11 to 44, depending on position, with the highest occurance in back row forwards. Greater involvement in contact activities are known to increase neuromuscular fatigue and energy expenditure,⁹ as well as muscle damage, and perception of effort¹⁰,¹¹ in players. These by-products of contact activities may affect other aspects of performance; during RU and rugby league training sessions, running outputs have been shown to decrease as the number of structured contact activities in training increase.¹⁰,¹² Given the importance of collision events in RU, and the contribution they have on subsequent muscle damage and fatigue, it is essential to quantify collisions in game play in order to appropriately prepare players for the demands of the sport.

Understanding the average running and contact game demands of RU is useful for the workload management and training prescription of players. However, given the intermittent nature of RU, average running and contact demands do little to inform training practices as they fail to quantify the most intense passages of play. Recent studies have attempted to quantify the peak periods of RU game play by separating the game into shorter time periods. Using pre-defined 10- minute time epochs show similar running demands to whole-game averages,³,⁴ while greater movement demands are observed when games are only analysed when the ball is in play (0.3–1.2 contacts·min⁻¹, 100–122 m·min⁻¹).¹³ When using shorter durations (<10 minutes) with a rolling average to quantify peak RU demands, even greater maximum running demands are achieved, ranging from 154 ± 21 m·min⁻¹ for front row forwards to 184 ± 28 m·min⁻¹ for inside backs.¹⁴ Within these peak periods of play (up to 10 min in duration) the movement profile of players are unlikely to be constant, and will likely still include numerous acceleration and deceleration phases, change of direction, and contact activities, all of which would contribute to the overall load and energy cost on players. While the acceleration demands have been quantified using this 1- to 10- min rolling average method,¹⁴ no research is currently available that reports peak contact activities during these time periods in RU.

Given the importance of contact activities in RU, and the significant physical toll they place on athletes, it is pivotal to quantify the peak contact frequencies in order to improve training prescription practices when trying to replicate the demands of competition. As such, the aim of the study was to quantify and describe the maximum and mean number of contact activities and running movements that occur during 1- to 10- minute rolling window periods in elite RU game play.

Methods

This was a retrospective cohort study using professional male RU players contracted to Australian franchise clubs (n = 4) during the 2018 and 2019 Super Rugby seasons. This dataset was provided by Rugby Australia (RA) the governing body for RU in Australia. Players (n = 161) were grouped by position; front row (n = 45), lock (n = 18), back row (n = 27), inside back (n = 45), outside back (n = 26). All procedures were approved by the La Trobe University Human Research Ethics Committee (HEC19375), where individual player consent was waived due to using a provided dataset where no contact was made with any players.

The physical demands of Super Rugby games (n = 95 games; 2232 player-games; front row = 604 files, lock = 293 files, back row = 382 files, inside back = 582 files, outside back = 371 files) were assessed. The micro technology devices (EVO, GPSports, Canberra, ACT) worn by players in each game held a 10 Hz global positioning system (GPS) chip to measure position and speed, and a 100 Hz triaxial accelerometer to measure linear acceleration. Devices were switched on outside approximately 60 minutes prior to the start of the game and were fitted inside a tight pocket sitting between the scapulae on the upper back of the playing jersey. Players were assigned to specific units to minimise inter-device variability. Raw GPS and accelerometer data were exported to excel in order to calculate variables for relative distance and impacts.

Registered software (GPSports Console v1.71, GPSports, Canberra, ACT) was used to download data files following each game. The quality of the data was confirmed via the horizontal dilution of position (HDOP, mean ± SD, 1.0 ± 0.2), and the number of satellites (9.5 ± 4.7) each device was connected to during each game. Any files with an average HDOP of >1.5 were removed from the analysis (equating to 1.8% of files removed). Data were exported into RStudio (Version 1.3.1073, RStudio, Boston, MA) in raw format so that moving averages for each time period could be calculated for each dependent variable (relative distance and impacts). Moving averages were calculated for various epochs (ranging from 1 to 10 min, at 1-min increments) with mean and maximum values for each epoch being recorded for each dependant variable, as described previously.¹⁵

Given several GPS devices that report collisions from an algorithm are yet to be validated, impact frequencies were quantified directly from the raw accelerometer data. As such, from here onwards we will use the term ‘impacts’ rather than ‘collisions’ to identify this variable based on the methods used in this study. An internal validation process was undertaken to confirm suitability of a g-force threshold for impact detection to determine contact-based events in this RU dataset. A 0.05 s rolling average was used to smooth the acceleration data before accelerometer files were matched with corresponding time stamped video footage (OPTA Sports, Stats Perform, London, UK). Different g-force thresholds were tested, starting with a resultant acceleration value of 2 g, and increasing by 0.5 g until there was agreement between the identified impacts and the time stamped video footage. In line with previous research,¹⁶ the internal validation process where raw accelerometer files were manually compared to video files, found that an impact threshold of 5 g was most appropriate to accurately measure contact events such as tackles (made and by opposition) and in-play contact with other players (rucks and breakdowns). Therefore, impacts less than 5 g were excluded from the analysis as these likely represent foot contacts from walking, running, or changes in direction which were not relevant for this study. There was no upper threshold for impacts events included in this study.

Linear mixed effect models¹⁷ were used to describe the relationships between time period, playing position, and player identity on maximum and mean relative impacts, and maximum and mean relative running distance. Position (front row, lock, back row, inside back, and outside back) and time period (1- to 10- min) were included as interacting fixed effects, while player was included as a random intercept. The relationship between the outcome variables and time period was anticipated to be non-linear¹⁴ so time period was included in the models as a restricted cubic spline term with three degrees of freedom to allow for non-linearities without assumptions about their functional form. The distributions of maximum and mean relative impacts are positively skewed and non-negative so were log transformed before modelling. Fixed effect relationships are presented as unconditional positional means with 95% prediction intervals (PI). Prediction intervals were chosen because they show the range of values the outcome variables could be expected to fall in a future game. The combined effect of player identity (random intercept) and position is visualised using conditional estimates (and 95%PI) of each outcome variable for a 5-minute time window. All analyses were conducted using the R statistical programming language.¹⁸

Results

The effects of position and time period on maximum relative impacts and running distance are presented in Figures 1 and 2, respectively. These figures show that shorter time periods have greater peak intensities and the shape of both maximum impacts and distance curves do not vary noticeably by position. From these positional graphs, the average maximum impact demands are as high as 4.5 to 5.5 impacts·min⁻¹ and the average maximum running demands between 150 to 180 m·min⁻¹ in the peak 1-min period. The prediction intervals for most positions show considerable overlap for both maximal impacts and running movements, indicating that in a future game the average player demands from any positional group could not be reliably predicted to be greater or lower than any other position. As an example, the average for an inside back is comparable to the upper prediction interval for front row players, showing that these positions may have different maximal demands, but in any individual game you cannot reliably predict this positional difference. Conditional estimates for a 5-minute window (Figures 1 and 2) show that the between player variation in maximal impacts and running is comparable in magnitude to the effects of playing position (i.e. the location of the dots varies a similar amount within groups as it does between groups). As such, the maximal running and impact demands of elite RU players are uncertain even if player identity and position are known.

Figure 1.

Maximum relative impacts performed during professional rugby union games during rolling intensity periods (1–10 min). The left figure presents the unconditional positional estimate (solid line) and 95%PI (shaded region). The dashed line represents the curve for inside backs as a reference point for comparison. The right figure shows the conditional estimate (and 95% PI) of demands in a 5- minute period for individual players within each positional role. As the mid-point of all windows, the 5- minute period was selected for representation, and the same outcomes is observed for all period windows.

Figure 2.

Maximum relative distances covered during professional rugby union games during rolling intensity periods (1–10 min). The left figures solid line shows the unconditional positional estimate and shaded region shows the 95%PI. Dashed line shows the curve for inside backs as a reference point for comparisons. The right figures show the conditional estimate (and 95% PI) of demands in a 5- minute period for individual players within each positional role. The 5- minute period was selected as it is the mid-point of the analysis and all period windows reflect the same outcome.

While the intensity is greater during shorter time periods for mean impacts (Figure 3) and mean running (Figure 4), there is little variation by position. The average mean impact demands are as high as 1.5 to 2.0 impacts·min⁻¹ between 95 to 115 m·min⁻¹ during a 1-min period (Figure 3 and 4). The mean conditional estimates for a 5-minute period show similar effects for the between player variation and the effects of playing position. Given the prediction intervals for all positions show considerable overlap for both mean impact and running demands, the future game demands on elite RU players are also uncertain even if player identity and position are known.

Figure 3.

Mean relative impacts performed during professional rugby union games during rolling intensity period (1-10 min). The figure on the lefts solid line shows the unconditional positional estimate and shaded region shows the 95%PI. Dashed line shows the curve for inside backs as a reference point for comparisons. The right figures show the conditional estimate (and 95% PI) of demands in a 5- minute period for individual players within each positional role. The 5- minute period was selected as it’s the mid-point of the analysis and all period windows reflect the same outcome.

Figure 4.

Mean relative distance covered during professional rugby union games during rolling intensity period (1-10 min). The figure on the lefts solid line shows the unconditional positional estimate and shaded region shows the 95%PI. Dashed line shows the curve for inside backs as a reference point for comparisons. The right figures show the conditional estimate (and 95% PI) of demands in a 5- minute period for individual players within each positional role. The 5- minute period was selected as it’s the mid-point of the analysis and all period windows reflect the same outcome.

Discussion

This study incorporated the rolling window method to identify peak and mean periods of impacts and running within elite level RU over various time durations. While showing the maximum impact and running demands, this study presented data from multiple Australian RU teams across two seasons. Window duration appears to influence the relative impact and running movements of players. However, considerable uncertainty around impact and running demands were observed when viewed at the individual game level, where individual differences were often greater than the observed positional effect. When prescribing drills to replicate the ‘peak periods’ that may be experienced in a game of RU, it is important to consider the impact demands on players, where up to 5 impacts·min⁻¹ may be observed. Given the random effect for player often being greater than the positional effect, establishing individual player movement demands for both impacts and running may be better for the optimal prescription of training to meet the peak demands of game play in RU.

The average maximum (4.5–5.5 impacts·min⁻¹) and mean (1.5–2.0 impacts·min⁻¹) relative impact demands within a peak 1 min period that RU players may be expected to complete in a game is greater than previously reported (0.3–1.2 contacts·min⁻¹).²,¹³ This study also shows that the peak and mean impacts over a rolling duration appear similar between positions, despite previous research indicating forwards are much more heavily involved in contact events than backs due to their involvement in scrums, rucks, and mauls.¹⁹ While other studies report positional differences in the absolute⁷ and relative¹³ impact demands between forwards and backs in RU, this study highlights all positions may experience similar peak impact frequencies at some point within a game. As such, given the overlap between positional and individual influences, it casts doubt over the practical importance of tailoring training to meet the peak period demands specific to playing position in RU. The contact conditioning requirements of all positions may therefore be similar when targeting peak period demands in training.

The rolling window analysis utilised in this study demonstrates the large variance of peak impact and running values performed by RU players over time, in comparison to whole match information which often underestimates the maximal impact and running intensities that occur within a game. This study also utilised a predictive interval, which provides an indication of where subsequent game data may fall in future games. Graphically, we can see that the moving averages of 3- to 10- minutes in duration yielded similar average maximum (∼2 impacts·min⁻¹) and mean (0.5 to 1.0 impacts·min⁻¹) impact intensities, which are still greater than previously reported impact intensities during ball in play analysis.¹³ The predictive interval also gets smaller with longer period windows, while the predictive range during peak 1- and 2- min windows are considerably large. As such, while the average peak demands may be 4–5 impacts·min⁻¹, in a subsequent game, players may be expected to complete up to 10–12 impacts·min⁻¹ in a peak 1-min period. This highlights the large variance that needs to be accounted for when targeting peak period intensities for training prescription. It appears that for time period durations of less than 3- minutes, up to 12 impacts·min⁻¹ may need to incorporated into drills, while drills greater than 3- min in duration, up to 2.5 impacts·min⁻¹ could be reasonably expected to occur in any subsequent games, and hence should be included in training prescription methods to suitably prepare athletes for the peak demands of elite RU.

Peak running periods over 1- to 10-minute durations were similar to previous research in elite RU.¹⁴,¹⁵,²⁰ Given the multi-team dataset used in this study, these results support previous literature being representative of elite RU game demands, despite the use of single-team datasets.¹⁴,²¹ While the teams involved in this study are from the Australian conference of Super Rugby, teams from other countries or conferences may differ in their peak game running demands. Inside and outside back positional groups experienced greater running demands in each period in comparison to their forward counterparts (back row, front row, and lock). Considering a back’s role is to perform a large amount of support running, defence coverage, and kick chase movements,¹⁹ it is of no surprise that backs cover greater relative distances. However, the positional differences are not always greater than the individual variance, and so for peak running demands, there may not be as great of positional difference as previously thought. The current study also looked at the mean running demands across a 1- to 10- minute period and found that there were small variations in the average intensity across the whole 1- to 10- minute duration. Indeed, mean running within the 5- to 10- minute rolling windows were similar to mean whole game data previously published in RU,²,²²,²³ and highlight that typical mean peak periods within RU game play maintain their intensity across 5- to 10- minute durations and are similar to whole game outcomes. This is likely related to typical ball in play durations,¹³ where phase durations greater than 5-min would include substantial ball-out-of-play time, and as such, are less relevant for peak period training prescription.

Understanding the mean and maximal relative impact and running demands of game play is useful for practitioners in order to guide training prescription in professional RU. Drills of varying durations can be compared to club-specific data or the data presented in this study to determine how closely training drills reflect peak competition demands. While this study provides peak impact and running demands players may be exposed to during games, it is currently not known what proportion of game time is spent at or near these peak values. Future research that quantifies this distribution of running and impact loads in relation to peak demands would further assist the utility of this information for training prescription purposes. While this study shows large variations, and hence cannot confirm substantial positional effects for peak period impact and running demands, other aspects may contribute to differences in positions not observed in this study (e.g. team playing style or tactics) and may need to be accounted for in future research. An interaction between technical elements (e.g. collisions) and running has been shown to occur in rugby league where, as collision frequency increased, relative running distance for both backs and forwards are reduced.⁶ This has similarly been shown in Australian football, where increasing technical involvements coincided with decreased average speed across shorter period durations (up to 5 min), but increases in average speed over longer periods (7 and 10 min).²⁴ These recent findings from other sports highlight the importance for understanding the context of observed peak periods and how technical involvements (through various collision events) may influence the running demands and vice versa in RU, specific to position. Understanding the preceding physical and technical demands, as well as the recovery phase following peak demands may also be important to further understand the relationship between running and collision activities and how to optimise training prescription to prepare for these peak demands.

Conclusion

This study reports the maximum and mean relative impact and distances covered during 1- to 10- min periods in elite Australian RU game play. Findings from this study allow practitioners to prescribe training drills that are representative of both the peak impact and running demands that players may be exposed to during game play. Average maximum impact demands were found to be as high as 5 impacts.min⁻¹, with a predictive interval as high as 12 impacts·min⁻¹ during a peak 1- minute period. The individual variation was considerably high though, and in some cases larger than the positional effects. As the time period duration increased, the relative number of impacts and distance covered decreased. Although this study highlights the relative impact demands alongside the running demands across a rolling window time period, future research should investigate the interaction between running and impact demands in elite RU match play and how each may influence the other.

Footnotes

Acknowledgements

The authors would like to thank Rugby Australia for their assistance with data availability for this project.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The authors would like to acknowledge the financial assistance of Melbourne Rebels Rugby Union and La Trobe University through the form of an industry-funded PhD.

References

Cunniffe

Proctor

Baker

, et al. An evaluation of the physiological demands of elite rugby union using global positioning system tracking software. J Strength Condition Res 2009; 23: 1195–1203.

Austin

Gabbett

Jenkins

The physical demands of super 14 rugby union. J Sci Med Sport 2011; 14: 259–263.

Coughlan

Green

Pook

, et al. Physical game demands in elite rugby union: a global positioning system analysis and possible implications for rehabilitation. J Orthopaed Sports Phys Ther 2011; 41: 600.

Jones

West

Crewther

, et al. Quantifying positional and temporal movement patterns in professional rugby union using global positioning system. Eur J Sport Sci 2015; 15: 488–489.

McLaren

Weston

Smith

, et al. Variability of physical performance and player match loads in professional rugby union. J Sci Med Sport 2016; 19: 493–497.

Johnston

Weaving

Hulin

, et al. Peak movement and collision demands of professional rugby league competition. J Sports Sci 2019; 37: 2144–2151.

Reardon

Tobin

Tierney

, et al. Collision count in rugby union: a comparison of micro-technology and video analysis methods. J Sports Sci 2017; 35: 2028–2034.

Roe

Halkier

Beggs

, et al. The use of accelerometers to quantify collisions and running demands of rugby union match-play. Int J Perform Anal Sport 2016; 16: 590–601.

Johnston

Gabbett

Seibold

, et al. Influence of physical contact on neuromuscular fatigue and markers of muscle damage following small-sided games. J Sci Med Sport 2014; 17: 535–540.

10.

Johnston

Gabbett

Jenkins

DG.

Influence of number of contact efforts on running performance during game-based activities. Int J Sports Physiol Perform 2015; 10: 740–745.

11.

Johnston

Gabbett

Seibold

, et al. Influence of physical contact on pacing strategies during game-based activities. Int J Sports Physiol Perform 2014; 9: 811.

12.

Roe

Darrall-Jones

Till

, et al. The effect of physical contact on changes in fatigue markers following rugby union field-based training. Eur J Sport Sci 2017; 17: 647–655.

13.

Pollard

Turner

Eager

, et al. The ball in play demands of international rugby union. J Sci Med Sport 2018; 21: 1090–1094.

14.

Delaney

Thornton

Pryor

, et al. Peak running intensity of international rugby: implications for training prescription. Int J Sports Physiol Perform 2017; 12: 1039.

15.

Delaney

Scott

Thornton

, et al. Establishing duration-specific running intensities from match-play analysis in rugby league. Int J Sports Physiol Perform 2015; 10: 725–731.

16.

Oxendale

Twist

Daniels

, et al. The relationship between match-play characteristics of elite rugby league and indirect markers of muscle damage. Int J Sports Physiol Perform 2016; 11: 515–521.

17.

Douglas

Martin

Ben

, et al. Fitting linear mixed-effects models using lme4. J Stat Softw 2015; 67: 1–48.

18.

R Development Core Team. R: a language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing, 2019.

19.

Duthie

Pyne

Hooper

Applied physiology and game analysis of rugby union. Sports Med 2003; 33: 973–991.

20.

Whitehead

Till

Weaving

, et al. The use of microtechnology to quantify the peak match demands of the football codes: a systematic review. Sports Med 2018; 48: 2549–2575.

21.

Reardon

Tobin

Tierney

, et al. The worst case scenario: locomotor and collision demands of the longest periods of gameplay in professional rugby union. PLoS One 2017; 12: e0177072.

22.

Quarrie

Hopkins

Anthony

, et al. Positional demands of international rugby union: evaluation of player actions and movements. J Sci Med Sport 2013; 16: 353–359.

23.

Roberts

Trewartha

Higgitt

, et al. The physical demands of elite English rugby union. J Sports Sci 2008; 26: 825–833.

24.

Johnston

Murray

Austin

, et al. Peak movement and technical demands of professional Australian football competition. J Strength Condition Res. Epub July 2019; Publish Ahead of Print 1 DOI: 10.1519/JSC.0000000000003241