Quantifying Collision Frequency and Intensity in Rugby Union and Rugby Sevens: A Systematic Review

Background Collisions in rugby union and sevens have a high injury incidence and burden, and are also associated with player and team performance. Understanding the frequency and intensity of these collisions is therefore important for coaches and practitioners to adequately prepare players for competition. The aim of this review is to synthesise the current literature to provide a summary of the collision frequencies and intensities for rugby union and rugby sevens based on video-based analysis and microtechnology. Methods A systematic search using key words was done on four different databases from 1 January 1990 to 1 September 2021 (PubMed, Scopus, SPORTDiscus and Web of Science). Results Seventy-three studies were included in the final review, with fifty-eight studies focusing on rugby union, while fifteen studies explored rugby sevens. Of the included studies, four focused on training—three in rugby union and one in sevens, two focused on both training and match-play in rugby union and one in rugby sevens, while the remaining sixty-six studies explored collisions from match-play. The studies included, provincial, national, international, professional, experienced, novice and collegiate players. Most of the studies used video-based analysis (n = 37) to quantify collisions. In rugby union, on average a total of 22.0 (19.0–25.0) scrums, 116.2 (62.7–169.7) rucks, and 156.1 (121.2–191.0) tackles occur per match. In sevens, on average 1.8 (1.7–2.0) scrums, 4.8 (0–11.8) rucks and 14.1 (0–32.8) tackles occur per match. Conclusions This review showed more studies quantified collisions in matches compared to training. To ensure athletes are adequately prepared for match collision loads, training should be prescribed to meet the match demands. Per minute, rugby sevens players perform more tackles and ball carries into contact than rugby union players and forwards experienced more impacts and tackles than backs. Forwards also perform more very heavy impacts and severe impacts than backs in rugby union. To improve the relationship between matches and training, integrating both video-based analysis and microtechnology is recommended. The frequency and intensity of collisions in training and matches may lead to adaptations for a “collision-fit” player and lend itself to general training principles such as periodisation for optimum collision adaptation. Trial Registration PROSPERO registration number: CRD42020191112. Supplementary Information The online version contains supplementary material available at 10.1186/s40798-021-00398-4.

studies quantified collision frequencies and/or intensities in training, three focused on both training and match-play, while 66 studies quantified frequencies and/or intensities of collisions in matches. Further investigation is needed to improve and understand the relationship between training and matches. • Per minute, rugby sevens players perform more tackles and ball carries into contact than rugby union players and forwards experienced more impacts and tackles than backs. Forwards also perform more very heavy impacts and severe impacts than backs in rugby union. • Integrating video-based analysis and microtechnology is recommended, and the metrics and grouping variables between training and matches should be consistent. • The frequency and intensity of collisions in training and matches may lead to adaptations for a "collisionfit" player and lend itself to general training principles such as periodisation for optimum collision adaptation.

Background
Rugby union and rugby sevens (henceforth called sevens) are invasion team sports that are characterised by frequent high speed running and physical collisions [1,2]. Although the two rugby codes differ in match duration (sevens = 14 min; rugby union = 80 min) and player numbers (sevens = 7 players; rugby union = 15 players) [3][4][5][6], the type of collisions are similar (i.e., tackles, scrums, rucks and mauls) [6]. Winning these collisions is associated with overall team success and player performance [7][8][9]. For example, Ortega et al. (2009) identified that winning teams complete more tackles than losing teams [7]. These collisions are also physically and technically demanding for players with an associated high injury incidence and burden (injury incidence rate X mean severity) [10][11][12][13]. For instance, in senior professional male rugby union players, 29.0 injuries per 1000 player hours occur when being tackled, 19.0 injuries per 1000 player hours occur when tackling and 17.0 injuries per 1000 player hours occur in the ruck/maul [14]. In sevens, 40.4 injuries per 1000 player hours occur when tackling, with 1.2 injuries per 1000 player hours occurring in the mauls and scrums [15]. Given the high injury incidence and burden, and the positive performance outcomes associated with winning collisions in rugby union and sevens, it is important for coaches and practitioners to adequately prepare players for competition. To do this, they need to know the frequency and intensity of these collisions in both training and matches [16]. In matches and training, the frequency and intensity of collisions have been quantified primarily using two methods: video-based analysis and microtechnology. Quantifying the frequency and intensity of collisions using video-based analysis requires the systematic observation and interpretation of video from matches and/or training [17,18]. Analysing collisions can occur while the matches or training session(s) are underway, although most detailed analyses occur post-match [17]. Previously, video-based analysis was the main method used to quantify collisions in both rugby cohorts [17]. Quantifying collisions in this manner however, is based on human observation, and as such, it is labour intensive and requires reliability checking to reduce bias and subjectivity [16]. For these reasons, a shift to automated methods of collecting collision data through the use of microtechnology has occurred.
In sport, microtechnology typically incorporates global positioning systems (GPS) and micro-electrical mechanical systems (MEMs) that capture the external physical demands of competition and training [19]. Commercially available microtechnology devices for team sports are designed to be unobstructive, so players can wear them during competition and training. One of the first studies using microtechnology to determine physical demands in rugby union was published in 2009 [20], and since then, research using these devices has grown [19]. Initially, GPS was only used to provide information on distance and speed [21,22]. Since then, MEMs have been built into GPS devices which now house triaxial accelerometers, gyroscopes and magnetometers [22]. Triaxial accelerometers measure acceleration in three different axes (anterior-posterior, medial-lateral and vertical) [16,22], and the sum of the acceleration in these three axes provides a vector magnitude (g force). This vector magnitude can be used to quantify the intensity of the collision [19,22]. Each manufacturer has a different algorithm that is used to quantify collisions [23]. As a consequence, validating collision metrics for these devices has been challenging [23]. Although quantifying collisions using microtechnology may be more time efficient than video-based methods, the validity and reliability of microtechnology in rugby union and sevens requires further investigation [16,24] due to the ambiguity in the current results [25].
To benefit coaches and practitioners, and aid injury prevention and injury management strategies, a synthesis of the frequency and intensity of collisions in rugby union and sevens to date, both in training and matches, is required. For example, a coach who understands the positional match tackle frequencies and intensities can optimise tackle training sessions to meet those position specific match demands. Since one of the roles of coaches and practitioners is to ensure positive adaptations to training and reduce maladaptation, understanding the frequency and intensity of collisions may also aid optimising recovery between training and matches. Therefore, the aim of this systematic review to synthesise the collision frequencies and intensities for rugby union and rugby sevens based on video-based analysis and microtechnology.

Search Strategy
The search strategy was based on a similar systematic review in rugby league [16]. The current systematic review was carried out in accordance with the PRISMA guidelines [28]. The search was conducted from 1 January 1990 to 1 September 2021 on four different electronic databases (PubMed, Scopus, SPORTDiscus and Web of Science). The search used the following combined key terms for collisions ('tackl*' OR 'collision' OR 'impact*') AND ('dose' OR 'frequency' OR 'intensity' OR 'demands') AND rugby union ('rugby' OR 'rugby union' OR 'rugby sevens'). For example, in Pub-Med the search was (((tackl* OR collision OR impact* OR collisions)) AND (dose OR frequency OR intensity OR demands)) AND (rugby OR rugby union OR rugby sevens). The reference list of the final full-text articles (n = 73) was also examined.

Selection of Studies
After consolidating the studies from the different electronic databases, LP removed the duplicates and screened the titles and abstracts (Fig. 1) for eligibility before retrieving the full text [28]. The review was registered with PROSPERO (registration number: CRD42020191112). The full text articles were further screened for eligibility by LP and MN. Any discrepancies in the screening process were discussed until agreed upon. A third researcher was available if consensus on the inclusion of an article could not be reached; however this was not required. The inclusion criteria were (i) any publication that quantified collisions in terms of frequency or intensity in rugby union and/or sevens (ii) study participants within each study had to be over 18 years of age. When collisions were based on 'impact metrics' , only impacts > 8 g were included in the data to eliminate possible confusion with running demands (i.e., high intensity accelerations or decelerations) unless stated otherwise [25]. Publications from conferences and annual meetings were excluded. Only peer-reviewed publications were included. Any publication that could not be translated into English was excluded. Authors were contacted for detailed information if necessary. The final full-text articles went through the data extraction process.
Collisions were broadly defined as any physical contact made with another player (teammate or opposition), which resulted in an alteration to the player's momentum. This included collisions such as the tackle (tackling and being tackled), scrums, rucks and mauls [26,27]. For this review the studies did not need to have a definition to be included.

Data Extraction
Data relating to participant characteristics (i.e., number, age, height, weight, level of competition, sex, cohort), context (i.e., match play or training), method used to quantify the collisions (i.e., video or microtechnology), the model and specifics of the device (i.e., GPS device rate, inertial sensors, number of files, software), videobased analysis characteristics (i.e., camera system, number of cameras, location of the devices and software), and collision characteristics were extracted from the final 73 full-text articles. Collision characteristics included type of collision, number of matches or training sessions, year of competition, absolute frequency (number), collisions in relation to playing time (number of collisions per minute) and the intensity of each collision. Collision intensity was commonly classified as very heavy (8-10 g), severe (> 10 g) or another range that was specific to the device based on the nature of the collision [29].

Data Analysis
All data were reported in the tables as mean ± standard deviation (SD) unless stated otherwise. Where possible, a meta-analysis (OpenMeta[Analyst]) was completed to produce a pooled mean and 95% confidence intervals (CI). An analysis was only conducted if there were at least two studies with mean and standard deviations. The DerSimonian-Laird continuous random-effects analysis method was used for the meta-analysis, with I-squared (I^2) used to assess the heterogeneity of the data. I^2 of 0-40% was considered low heterogeneity, 40-75%: moderate heterogeneity and > 70% was considered high heterogeneity [16]. The forest plots (mean and 95% CI) presented the results of the meta-analysis.
Sixteen studies recorded the intensity of collisions by using microtechnology (22%) ( severe impacts per match [29,76,77] (Fig. 2). Three studies recorded the relative frequency of collisions by intensity [81][82][83]. On average, forwards experience 9.1 (7.5-10.8) impacts > 5 g per min [81,83] (Fig. 3). Backs experience on average 9.5 (8.1-10.1) impacts > 5 g per min [81,83]. Note, Tee et al. only included > 5 g impact since it included > 8 g impacts [83]. Players experienced the highest amount of contacts in the first 20-30 min of a match and the least amount of contacts between 60 and 70 min [82]. Forwards experience more very heavy contacts in the second half of the match in comparison to the first half of the match. Backs experience fewer impacts in the second half of the match in comparison to the first half of the match [29]. There was no    difference in impacts > 8 g per min for backs and forwards across the match [81]. Forwards experience more impacts > 5 g per min in 0-10 and 50-60 min and experienced the least amount in the 20-30 min, 40-50 min and 60-70 min intervals of the match. Backs experience more impacts > 5 g in the 0-10 min interval of the match and the 20-30 min interval of the match and the least in the 70-80 min interval [81].

Rugby Union Training
Two studies recorded collision frequency using microtechnology during training (3%) [32,80]. Bradley et al. (2015) recorded the contact number of weekly training sessions of forwards and backs. Note, match data were also included in this training week [32]. Takeda et al.

Sevens Match-Play
Eight studies (11%) reported collision frequency using microtechnology during match-play [35-38, 47, 51, 62, 78]. One study reported positional groupings (forwards and backs) [78], another study reported the level of play [36] and another study reported collision frequency by sex [37] (Table 2). Collision types included impacts, collisions, tackles, rucks and scrums. Only one study recorded the relative frequency of tackles, ball carries in contact and rucks [62] and another study recorded relative frequency of impacts for forwards and backs [51].

Sevens Training
Only one study reported tackle frequency during training (on average 17.8 ± 4.4 tackles per week) [47].

Video-Based Analysis Rugby Union Match-Play
Thirty-seven studies recorded the collision frequency using video-based analysis methods (51%) [ (Fig. 8).

Rugby Union Training
Only one study reported collision frequency during training [90]. Vaz et al. (2012) reported that novice players perform an average of 28.2 ± 3.3 tackles during small-sided games, while experienced players perform 48.7 ± 3.3 tackles on average [90].

Sevens Match Play
Eight studies recorded the collision frequency by using video-based analysis (11%) (  (Fig. 9). Finally, backs and forwards experience more contacts in the second half of the match compared to the first half [60].

Sevens Training
No video-based training studies were found for sevens.

Discussion
To our knowledge, this is the first systematic review on quantifying collision frequency and intensity in rugby union and rugby sevens. This review demonstrates that video-based analysis and microtechnology are the main methods used to quantify collisions in rugby union and sevens. Not surprisingly, the absolute collision frequency during sevens matches was lower than rugby union due to the shorter duration of the game and fewer players on the field. When comparing relative frequencies though, rugby union players seem to perform less tackles and ball carries into contact than sevens players, while rucks per minute were similar between the two rugby codes [55,70]. Expressing collision frequencies relative to playing time provides coaches and players with the 'collision density' [96], a metric that can potentially be used in training Fig. 3 Meta-analysis of studies reporting relative > 5 g impacts frequency per match (n min −1 ) from microtechnology in rugby union. The forest plot (mean and 95% confidence interval (CI)) presents the results of the meta-analysis of the pooled data estimates for the > 5 g impacts per min per match frequency for forwards. The squares and horizontal lines represent individual study mean and 95% CI and the diamond presents the pooled mean and 95% CI. The bigger the square the larger the sample size Fig. 4 Meta-analysis of studies reporting absolute > 10 g impacts per match (n) from microtechnology in sevens. The forest plot (mean and 95% confidence interval (CI)) presents the results of the meta-analysis of the pooled data estimates for the absolute > 10 g impacts frequency per match. The squares and horizontal lines represent individual study mean and 95% CI and the diamond presents the pooled mean and 95% CI. The bigger the square the larger the sample size          to better prepare players for the collision demands of matches. With that said, only two studies expressed collisions or contact events per minute in sevens [62,70], which highlights an area for further work. In rugby union match-play, forwards experience more tackles than backs (12.8 (7.5-18.1) tackles and 7.6 (4.3-10.9) tackles, respectively). Another key finding of this review is that forwards experience more very heavy impacts (52.  [35,36,62,67,77,94] and two studies on both sexes [37,38]. Overall, there was a lack of consistency on the definition of a collision. Also, grouping variables (i.e., how the positions were grouped) made it hard to make comparisons. It is recommended to integrate microtechnology and video-based analysis simultaneously to ensure maximal accuracy of metrics. Given the high injury incidence and burden of collision events, it is important that we adequately prepare athletes for collisions in training to meet the collision demands of matches.
To optimise training, researchers, trainers and sport practitioners typically study competition activities and demands, and attempt to replicate these demands in training [76,78,93,97]. Training is subsequently monitored to ensure athletes meet said competition activities and demands [34]. Monitoring training also ensures athletes are not exposed to any unnecessary injury risks, and are positively adapting to training [34]. Only four studies quantified collision frequencies and/or intensities in training-three in rugby union [32,80,90] and one in sevens [47], while 66 studies quantified frequencies and/or intensities of collisions in matches. Three studies related the frequency and intensity of collisions during training to matches-two in rugby union [34,42] and  Meta-analysis of studies reporting absolute total scrums, rucks, and tackles per match (n) from video-based analysis in rugby union. The forest plot (mean and 95% confidence interval (CI)) presents the results of the meta-analysis of the pooled data estimates for the total a scrums, b rucks and c tackles per match. The squares and horizontal lines represent individual study mean and 95% CI and the diamond presents the pooled mean and 95% CI. The bigger the square the larger the sample size one in sevens [51]. In both studies, collision frequencies and intensities were lower in training, suggesting that players may not be adequately preparing for matches [34,51]. Indeed, the adaptations for a "collision-fit" player are likely to respond to general training principles including the concept of periodization [98]. Using general training concepts, such as periodisation, and collision demands data from match-play, coaches and practitioners can develop training programmes to enhance players' adaptability and capacity to repeatably engage in physical-technical contests without increasing their risk of injury; in other words, building a 'collision-fit' player. Recently, this has been suggested for skill training and Hendricks et al.
(2018) described such a periodised plan for the rugby tackle [99]. Understanding the adaptations for a "collision-fit" player will also allow for safer return to play protocols for collision sport athletes and reduce the risk of re-injury. To inform collision preparation practice, more work on collision training and its relationship to match demands, player development, performance and/or (re) injury risk is required. Collision training studies of this nature should also ideally be collected over more than one season and from multiple teams. Collision frequency and intensities have been quantified in studies using video-based analysis (n = 37), microtechnology (n = 24) or both methods (n = 12). Each method has its advantages and disadvantages. For example, video-based analysis is laborious and reliant on human observation, while it may capture more contextual detail of the collision event [16]. Conversely, microtechnology may be more efficient and objective, but its reliability and validity for quantifying collision demands is inconclusive at this stage [16,24,25]. Also, customised algorithms detect collisions, making study comparisons difficult [100]. With that said, studies are emerging to support collision metrics when used in conjunction with video-based analysis [23,25]. Although some literature supports the use of microtechnology for collision monitoring, there is still a lack of validity regarding other metrics and therefore more investigation is needed [23]. As such, a superior approach to quantifying collision demands from a research and practitioner perspective may be to integrate video and microtechnology [18,19]. Using both video and microtechnology, coaches, practitioners and researchers are able to cross check the microtechnology data with video, determine its accuracy and distinguish between collision events [18,24,25].
If the goal is to ensure players are well-prepared for matches by providing the optimal collision frequency and intensity dose, the metrics (i.e., collisions, contacts, scrums, tackles, rucks and mauls) and grouping variables (i.e., specific positions, forwards and backs) between training and matches need to be consistent and more accurate. In other words, how collision demands are reported for matches should be useful to the coach and practitioner, and transferable to a training setting. Therefore, metrics and grouping variables between the two settings need to be consistent to ensure this transfer. Strong engagement with the coach and practitioner when developing reporting metrics is therefore recommended [101]. Recently, a consensus document for the video-based analysis of contact events was published to improve the consistency and quality of video-based analysis work in rugby union and sevens [18]. A similar consensus-based Meta-analysis of studies reporting absolute tackles per match (n) from video-based analysis in rugby union. The forest plot (mean and 95% confidence interval (CI)) presents the results of the meta-analysis of the pooled data estimates for the absolute tackle frequency for a forwards and b backs. The squares and horizontal lines represent individual study mean and 95% CI and the diamond presents the pooled mean and 95% CI. The bigger the square the larger the sample size approach may be required for microtechnology collision metrics [16,22]. As mentioned, many studies report collisions differently, making study comparisons difficult between groups, methods used and between rugby cohorts. As a result, this limited the current synthesis.
Collision intensity metrics in particular were inconsistent between studies. The lack of consistency between studies is a key factor limiting our understanding of collision loads [16]. Additionally, the intensity of collisions is difficult to compare longitudinally, given that technology Fig. 7 Meta-analysis of studies reporting absolute tackles per match (n) from video-based analysis in rugby union. The forest plot (mean and 95% confidence interval (CI)) presents the results of the meta-analysis of the pooled data estimates for the absolute tackle frequency for a front row forwards, b back row forwards, c inside backs and d outside backs. The squares and horizontal lines represent individual study mean and 95% CI and the diamond presents the pooled mean and 95% CI. The bigger the square the larger the sample size Paul et al. Sports Medicine -Open (2022) 8:12 Fig. 8 Meta-analysis of studies reporting absolute tackles per match (n) from video-based analysis in rugby union. The forest plot (mean and 95% confidence interval (CI)) presents the results of the meta-analysis of the pooled data estimates for the absolute tackle frequency for a props, b locks, c hooker and d scrumhalf. The squares and horizontal lines represent individual study mean and 95% CI and the diamond presents the pooled mean and 95% CI. The bigger the square the larger the sample size is constantly evolving. More recent technology is likely more accurate as algorithms are improved over time ensuring MEMs have a high specificity and sensitivity, and are more likely to detect a collision when it occurs [23], although limited studies can confirm this [25]. The purpose of this review was to synthesise the frequency and intensity of collisions during training and matches in rugby union and sevens. In both rugby cohorts, future studies should investigate training in comparison to match-play. Additionally, future studies should explore women's rugby. Many of these groups were understudied and are very important in our rugby community. A consensus-based approach for microtechnology is warranted since grouping variables and metrics were inconsistent throughout the studies. Beyond this, there are a number of other factors that can affect how players respond and adapt to different frequencies and intensities of contact. Collision events in rugby union and sevens are dynamic and have a major technical-skill component [102,103]. The opposing players' technical ability may also affect the perceived intensity of the collision event. The perceived physical and technical demands of collision events can also be captured using subjective ratings such as rating of perceived exertion (RPE) [104] and rating of perceived challenge (RPC) [98,104], respectively. These subjective ratings are useful when planning and monitoring training [104]. Also, collisions are interspersed between periods of high intensity running (sprinting, accelerations, decelerations) and low-intensity activities (walking, jogging). As such, advanced collision Meta-analysis of studies reporting absolute tackles, rucks, and scrums per match (n) from video-based analysis in sevens. The forest plot (mean and 95% confidence interval (CI)) presents the results of the meta-analysis of the pooled data estimates for the absolute frequency of a tackles, b rucks and c scrums per match. The squares and horizontal lines represent individual study mean and 95% CI and the diamond presents the pooled mean and 95% CI. The bigger the square the larger the sample size training should also include periods of high-intensity running to mimic complete match demands and fatigue conditions [97].

Conclusion
In conclusion, this review found a discrepancy in the number of studies quantifying collision demands in training compared to matches. While more work on quantifying the collision demands of training is required, studies should also compare training and matches if we are to improve our understanding of the relationship between training and matches. Another key finding is that the main method for quantifying collisions was video-based analysis. To improve the relationship between matches and training, integrating both video-based analysis and microtechnology is recommended, and the metrics and grouping variables between training and matches should be consistent. Per minute, rugby sevens players perform more tackles and ball carries into contact than rugby union players and forwards experienced more tackles than backs (12.8 (7.5-18.1) tackles and 7.6 (4.3-10.9) tackles, respectively). Another key finding in this review is that forwards experience more very heavy impacts (52.5 (29.8-75.2) vs. 41.7 (26.4-57.0) very heavy impacts) and severe impacts (10.8 (4.4-17.1) vs. 6.7 (5.1-8.4) severe impacts) than backs in rugby union. The frequency and intensity of collisions in training and matches may lead to adaptations for a "collision-fit" player and lend themselves to general training principles such as periodisation for optimum collision adaptation. Subjective measures such as RPE and RPC should be incorporated into the monitoring and management of the collision section of training to understand the internal load.