The epidemiology of kicking injuries in professional Rugby Union: A 15‐season prospective study

While kicking in Rugby Union can be influential to match outcome, the epidemiology of kicking injuries remains unknown. This study therefore aimed to investigate the epidemiology of injuries attributed to kicking in professional rugby, including playing position‐specific effects and differences in kicking volumes and kick types.


| INTRODUCTION
Kicking has the ability to influence the outcome of professional Rugby Union (hereafter rugby) matches both directly by adding points via attempts at goal, and indirectly through manipulating territory which can lead to try-scoring opportunities 1,2 or relieve defensive pressure. While kicking is not the sole determinant of a team's success, more frequent kicking has been consistently associated with winning across a range of rugby competitions. [2][3][4] Team success is also known to be associated with player injury, as injuries within a squad affect player availability and hence final league and tournament position. 5,6 It is possible that injuries sustained by players who frequently kick may therefore play a role in limiting the effectiveness of their team. The epidemiology of injuries sustained in rugby activities has been extensively investigated, [7][8][9][10][11][12][13][14][15][16][17][18][19] and although kicking as a potential mechanism of injury has been reported in some studies, [15][16][17][18] it is frequently grouped alongside a combination of "other" mechanisms. Given the importance of kicking for match outcome, 2-4 a more detailed consideration of the nature and causes of injuries sustained while kicking is clearly warranted. While position-specific injury profiles have previously been considered, 10,20 these are likely to be affected by multiple position-specific events and the epidemiology of injury directly associated with kicking currently remains extremely limited.
Injury frequency is typically quantified through incidence (injuries per unit of time) or propensity (injuries per given number of events) calculations. Traditional methods of calculating match injury incidence include the total number of players on the pitch and do not consider the number of players typically involved in a given match event. As backs perform considerably more kicks than forwards, and there are differences in the number of kicks completed by each member of the backs, 21 traditional calculations of incidence based on whole-team exposure may underestimate the true frequency of kicking-related injuries. Brown et al. 22 described the use of a modified incidence value based on players who were directly involved in a given injury event (eg, only forwards or front row players were included in calculations of scrummaging injury incidence), therefore including only backs in an incidence calculation of kicking-related injuries appears worthy of consideration. However, as kicking is unlikely to be evenly distributed between backs, 21,23,24 this may still underestimate the frequency of kicking injuries associated with specific playing positions among the backs. In a similar vein, the propensity for injury has also been used to quantify injury frequency for discrete events such as scrummaging and tackling, 14,22 as these calculations directly account for the number of times the given activity is completed. Establishing the frequency of kicking by playing position would therefore enable positional propensity for kicking injuries to be calculated if the playing positions of those players sustaining an injury during kicking were identified. Furthermore, consideration of the typical kicking profiles (ie, frequency of different types of kick performed) could provide additional valuable information to direct future investigations of the potential mechanisms associated with kicking injuries and identification of potential conditioning and rehabilitation practices for those players most at risk from kicking injuries. The aim of this study was therefore to investigate the epidemiology of injuries directly attributable to kicking in professional rugby, and to consider playing position-specific effects and differences in kicking volumes and kick types.

| Injury epidemiology-data collection
Participants were male, first-team, rugby players contracted to professional clubs competing in the English Premiership or at international representative level for England. Players provided written informed consent for collection and processing of their personal and injury data for research purposes in line with England Rugby's ethical guidelines. The study was approved by the Research Ethics Committee of the academic institution where the Professional Rugby Injury Surveillance Project (PRISP) was hosted for each season.
Injury and exposure data were recorded by medical staff at participating teams as part of the PRISP as outlined in previous studies. 7,8,12,13 The injury diagnoses entered into the PRISP were categorized under the Orchard Sports Injury Classification System (OSICS), 25 and injury definitions were consistent with the international consensus statement on injury surveillance in rugby. 26 Injury data covered 15 seasons from 2002/03 to 2017/18 (2004/05 data were unavailable). Data included all match (Premiership, National Cup, European competition, England internationals) and training injuries from the start of pre-season through to the end of the competitive season.
Data relating to acute-onset injuries which were directly attributed to the act of kicking were extracted. Gradual-onset injuries which may have been caused or exacerbated by the activity of kicking could not be identified from the available data. The data extracted for each kicking injury included: playing position at time of injury; event (match or training); time in match (where applicable); number of days subsequently absent from rugby activity; OSICS code and additional diagnosis information; side injured; dominant limb; and kicking limb. Where kicking limb was not explicitly stated, the dominant limb was assumed to be the kicking limb. Injury data entered prior to the 2013/14 season used three-character OSICS codes, which were converted to the equivalent four-character OSICS codes prior to analysis to ensure consistency with data collected in the subsequent seasons. 25 Injury distribution was initially established by identifying the global body region (eg, lower limb, upper limb, etc.) from the corresponding OSICS code. Injury locations in the pelvis and sacrum, and lower | 1741 limb, were then further subdivided according to the injury consensus statement. 26 Additional information was gleaned from freehand injury descriptors provided by medical staff when entering injuries into the database, if present. Muscle injuries were limited to the muscle group unless information regarding specific muscles had been identified.

| Match kicking characteristicsdata collection
All kicks performed during two full seasons of the English Premiership (2016/17 and 2017/18, n = 269 matches; one match unavailable) were coded as part of the formal match analysis undertaken for Premiership Rugby. For each kick, the playing position of the kicker and the type of kick performed (box kick; chip/kick pass; distance punt; drop goal; ground kick (fly hacks and grubbers); hang chase; place kick (from a tee); and restarts (22 m and halfway)) were identified (descriptions of each kick type are available in Table S1).

| Injury epidemiology-data analysis
Injury incidence calculations were consistent with the international consensus statement. 26 Modified incidence (ie, using injuries and player-hours for a specific playing position group rather than whole squad data) and injury propensity were calculated in line with previously published methods. 11,14,22,27 For both match and training data, overall incidence and modified incidence (for backs only) were presented as the number of kicking injuries sustained per 1000 player-match-hours or 1000 player-training-hours. All subsequent analyses were based on match data only due to the low incidence of training injuries (see Section 3). The overall and positional propensity values were determined from all 15 seasons' injury data and both seasons' kick characteristic data, and were calculated as the number of kicking injuries sustained per 1000 kicks. Injury severities were recorded as the total number of days lost, in line with the consensus statement, 26 and reported as median (interquartile range, IQR) values due to the nonnormal distribution as determined with Shapiro-Wilk tests (P < .05). Mean injury severity was also determined in order to calculate the burden (ie, the product of injury incidence and mean severity) of certain categories of kicking injury.

| Match kicking characteristicsdata analysis
Mean and standard deviation values were calculated for the number of kicks per match across all 269 matches in the kicking characteristic dataset. Kick type and kick type by playing position were calculated as percentages of this entire dataset and the position-specific subsets (where playing position was reported), respectively. The relative differences between proportions of kick types, and between proportions of kick types completed by each position, were established and described in a format consistent with previously reported procedures. 28 In brief, these were calculated by subtracting the proportion of kicks in a given kick type category from that of the distance punts (for kick type) or fly-half (for kick type by position) as the reference values, as these were the categories with the highest counts for each variable. Confidence intervals (CI; 95%) and proportion ratios were calculated for each comparison using the Wilson method. 29 Proportion ratios were calculated by dividing each given proportion by that of the distance punt (for kick type) or fly-half (for kick type by position), with a scale of trivial (1.0), small (1.1), moderate (1.4), large (2.0), very large (3.3), and extremely large (10.0) used to describe the differences. 28 Inverse values were used for negative differences, that is, 0.9 (small), 0.7 (moderate), 0.5 (large), 0.3 (very large), and 0.1 (extremely large).
For both the kicking injury and kicking volume datasets, if entries did not provide all information, they were only excluded from the specific analysis relating to that piece of information. Where data are absent from calculations, this is acknowledged in the results.

| Injury frequency
A total of 134 acute-onset injuries (match: 77 [see Table 1 for positional proportions]; training: 55; unspecified: 2) directly attributed to the act of kicking were recorded over the 15 analyzed seasons. Across all 15 seasons, total exposures were 116 720 player-match-hours and 1 900 654 playertraining-hours. The overall incidence of match kicking injuries was 0.7/1000 player-match-hours, and training kicking injuries was 0.03/1000 player-training-hours. Due to the low incidence of training kicking injuries, all subsequent analyses were limited to match data only. Modified match injury incidence for the backs (ie, 7/15 of the total match exposure and only match injuries sustained by backs) was 1.4/1000 player-match-hours. Over two Premiership seasons, 17 832 kicks were performed (mean ± SD kicks per match = 66 ± 13, range = 38-110). Based on a typical English Premiership season of 135 matches, the mean number of kicks during a typical season was calculated (8949). This was then projected over 15 seasons and applied to the total number of match kicking injuries sustained (77) to yield an overall propensity for match kicking injury of 0.57 injuries per 1000 kicks. Fly-halves (47%) and full backs (17%) sustained the greatest proportions of match kicking injuries which equates to position-specific injury incidences of 4.6 kicking injuries/1000 player-match-hours for fly-halves and 1.7/1000 player-match-hours for full backs. The positional proportions of all match kicking injuries and the differences between flyhalves and other positional groups are shown in Table 1 (1 out of 77 entries did not specify the playing position at the time of injury).

| Time of injury in match
A greater number of injuries (n = 14) were sustained during the warm-up period than either the first (n = 7) or second (n = 13) quarters of the game (4 out of 77 entries did not specify time in match). Entries for warm-up injuries did not specify whether players were originally designated as a starter or replacement. Sixty-six percent of all injuries which occurred during the playing time (ie, 0-80+ minutes) occurred during the second half, and the quarter with the highest proportion of injuries was the third quarter for both starters and replacements (Figure 1).

| Injury type and location
Of all match kicking injuries recorded, 92% were located in the pelvis or lower limb (Table 2). Muscle or tendon tissue injury occurred in 82% of these match kicking injuries. In just the pelvis and lower limbs, muscle or tendon injuries accounted for 89% of all match kicking injuries. Of the pelvis and lower limb injuries, 56% were located in the thigh alone. Thigh injuries led to a median severity of 10 days (IQR = 5.0-17.8 days; mean: 19.4 days; burden: 6.6 days/1000 playermatch-hours) lost from rugby-related activity. Of these, the quadriceps were the most frequently injured muscle group (n = 21, 53%), followed by the hamstrings (n = 13, 33%) and the adductors (n = 6, 15%). The rectus femoris was the most frequently injured individual muscle, sustaining 21% of all match kicking injuries, more than any other individual muscle. Muscle strains accounted for the majority (88%) of thigh injuries, with the remainder being classified as "muscle spasm/tightness/trigger points." Eighty-one percent of all pelvis/lower limb match kicking injuries were to the kicking leg side: for thigh injuries alone (ie, the most commonly injured area), 78% occurred in the kicking limb. Match kicking injuries sustained in the stance limb were more severe (median: 19.5 days, IQR = 4.8-44.5 days; mean: 57.7 days) than in the kicking limb (median: 9 days, IQR = 5.0-16.3 days; mean: 14.3 days). However, due to the greater number of kicking limb injuries sustained, the mean burden of kicking limb injury (6.3 days/1000 playermatch-hours) exceeded that of the stance limb (5.9 days/1000 player-match-hours). Nine of the 77 entries did not provide enough information to determine either the dominant side or kicking limb and were excluded from these analyses.

| Match kicking characteristicsall players
Of all kicks performed during the two seasons analyzed, distance punts were most common ( Figure 2). Distance punts were therefore used as the standard for all quantitative comparisons against other kick types (see Table S2). Moderate differences were found between the proportion of distance punts completed and box kicks and restarts. Large differences were found between the proportion of distance punts and place kicks, while very large differences were found when compared with the chip/kick pass, ground kicks, and hang chase. Extremely large differences were found between the distance punt and drop goal attempts.

| Match kicking volumes-positional differences
Backs performed 99% of all kicks. Five of the 17 832 kick entries did not specify any playing position information and were excluded from all positional group-specific analyses (ie, backs vs forwards), while 84 out of 17 832 kick entries only identified the player as a member of the backs and were subsequently excluded from individual playing position-specific analyses. Fly-halves performed the greatest proportion of total kicks (46%; Figure 3), followed by scrum-halves (27%), full backs (11%), centers (9%), and wingers (6%). The difference between the proportion of kicks completed by the flyhalf and scrum-half was moderate, the differences were very large in comparison with centers, wingers and the full back, and extremely large in comparison with forwards (mean differences in proportions and their 95% CIs are presented in Table S3). The position-specific proportions of each kick type with reference to those of the fly-half are presented in  Table S4).

| Positional propensity for injury
Based on the proportion of match kicks completed by each playing position (Figure 3) and proportion of match kicking injuries sustained, positional propensity for match kicking injury was calculated (Figure 4). Using the categories detailed by Fuller et al, 14

| DISCUSSION
This study investigated the epidemiology of injuries directly attributable to kicking in professional rugby, including the consideration of position-specific differences and the potential influence of kicking volumes and kick types. The overall incidence of kicking injuries during matches was 22 times greater than that during training and as such, all subsequent analyses focussed on match injuries. The incidence of match kicking injuries was lower than other injury mechanisms previously reported in rugby. A meta-analysis examining rugby injury epidemiology studies reported that line-outs and scrummaging had the lowest incidences of all reported mechanisms with 1 and 7 injuries per 1000 player-match-hours, respectively. 9 Focussing on kicking injuries is therefore not a priority when considering the overall incidence of injury across the whole squad, with tackle-related injuries clearly a higher priority. 9 However, the importance of considering the injury and task-specific issues associated with each playing position 10 is demonstrated by the greater incidence of kicking injury for the fly-half (4.6/1000 player-match-hours) and full back (1.7/1000 player-match hours). Given the known importance of kicking for match outcome, 2-4 the current results regarding kicking injury epidemiology provide valuable information for the specific physical preparation of players within the squad who frequently kick. These data also provide novel position-specific injury propensity and kick type information which can be used to inform technical, tactical and physical preparation profiles, as well as to inform decisions regarding the return on investment associated with position-specific interventions intended to reduce the risk of injury. 20 Modification of injury incidence to only include a specific positional group can be more informative than calculations based on an entire team. 22 As backs performed 99% of all kicks in the current study, a modified calculation for kicking injury incidence was limited to the match injuries sustained by backs and their associated exposure hours. However, this modified incidence (1.4/1000 player-match-hours) still assumes that all backs are equally exposed to a potentially injurious mechanism, which is clearly not the case (Figures 3  and 4). Propensity for injury provides a more appropriate method of reporting injury frequency, particularly when subdivided by playing positions. 14,22 The overall propensity for injury due to kicking (0.57/1000 kicks) was lower than that of other match events reported previously (1.1/1000 line-outs; 2.0/1000 rucks). 16 However, a greater number of players are exposed to a potential injury mechanism in each line-out or ruck than in a single kick. Furthermore, when considered by playing position groups (Figure 4), there is not a linear relationship between the proportion of match kicks performed and the number of kicking injuries sustained. Consistent with a previous study of international rugby, 21 the fly-half performs the highest volume of match kicking (46%). The fly-half also sustains the greatest proportion of kicking injuries (47%) and thus has an average propensity for kicking injury. 14 While scrum-halves perform almost a quarter of all kicks, their kicking injury propensity (0.17/1000 kicks) is very low, whereas the wingers, centers, full backs, and forwards all have a high propensity for kicking injury (Figure 4). The propensity for kicking injury therefore appears to be affected by other position-specific factors rather than simply exposure to kicking.
While specific physical preparation and exposure to kicking during training are potential factors which could influence position-specific propensity for match kicking injury, these would likely not explain the observed differences between fly-halves and scrum-halves. The type of kick undertaken may also be an important consideration because there are numerous types performed, each of which may place varying demands on the kicker. Given the apparent differences in kick types performed between scrum-halves (ie, box kicks constituting almost 70% of a scrum-half's total kicking load, Figure 3) and the other playing positions who all exhibit higher propensities for kicking injury, it is possible that the mechanics of a box kick are potentially less injurious than other kick types. However, caution must be applied because while the mechanics of goal kicking [30][31][32][33] and punt kicking [34][35][36] have previously been studied in both rugby and Australian Rules Football, direct comparisons between kick type mechanics have not been undertaken and there have been no mechanical descriptions of the other rugby kick types included in the current study. Furthermore, the studies examining goal or punt kicking have focussed on identifying variables which are associated with successful performance, rather than the potential role of kicking mechanics in injury. Given this lack of research, it is challenging to make inferences regarding the potential for injury based on specific kick types. Direct comparisons of the mechanics of different kick types may provide useful biomechanical information which can be used to investigate the demands placed on the musculo-skeletal system and shed light on the potential role of kick type as an injury risk factor.
The time in the match is also a factor of importance when considering rugby injuries. Two-thirds of all match kicking injuries which occurred during the playing time (ie, 0-80+ minutes) were in the second half, with the third quarter the most injurious for both starters and replacements. This is consistent across the pooled rugby literature 9 and may be related to decrements in technique associated with fatigue in starters, as seen in other rugby-specific tasks such as tackling. 37 However, those observed technique decrements 37 were found to be mitigated by physical characteristics such as increased strength, demonstrating that task-specific physical preparation may be an important consideration. Incomplete warm-ups prior to the second half may also contribute to the increased frequency of injuries seen in the third quarter, 38 which may affect both starters and replacements. The incidence of match kicking injuries also included warm-up injuries as this directly involved the match squad and preparation outside of training hours. However, as the total exposures of warm-ups were unknown, this may therefore overestimate the incidence of match injuries. It is worth noting that the number of warm-up injuries exceeded those sustained in either the first or second quarters of the match. While the reason for this cannot be determined from the current analysis, kickers anecdotally perform a high volume of kicks in a relatively short space of time during a warm-up. Medical and performance staff may therefore need to consider the approach to kicking used in the warm-up if players are sustaining injury during a controlled period which is intended to prepare them for match play.
Consistent with previous rugby-based epidemiology studies, 9 the majority of kicking injuries were located in the pelvis or lower limb, with 81% of these occurring in the kicking limb. The median severity of stance limb kicking injuries was higher than that of the kicking limb, but the greater frequency of kicking limb injuries meant that the overall burden of kicking and stance limb injuries was similar (6.3 and 5.9 days/1000 player-match-hours, respectively), even despite two stance limb injuries with severities in excess of 100 days being sustained. The thigh was the most injured segment, and 84% of injuries in this location were muscular strains. Rectus femoris sustained more kicking injuries than all other specific locations and accounted for 21% of all match kicking injuries. This was followed by muscular strains in the hamstrings and adductors. In soccer-specific studies, the rectus femoris has been implicated in up to 48% of hip flexor-specific MRI assessments. 39 Given the relative similarity in the prevalence of anterior thigh muscle injuries in a kicking-dominant sport such as soccer and the kicking-specific rugby injuries in the current study, this suggests that anterior thigh strains, in particular to the rectus femoris, should be a primary focus when working with rugby players who are expected to kick.
The epidemiological data presented in the current study were obtained prospectively over multiple seasons. As the analysis of kick counts and types was only available for the two most recent domestic Premiership seasons, these were used to estimate the propensity for kicking injury sustained during Premiership and England matches across all seasons in which injury epidemiological data were recorded. It is possible that total kick counts could change over seasons (eg, due to style of play or law changes) and between levels of play (ie, domestic vs international). It was not possible to assess this in the current study given that kick count and type data were only available over two seasons, and future prospective analyses could be undertaken to assess these potential changes in kicking patterns. Future work could also seek to document the type of kick performed (and other situational factors) at the time of each kicking injury, as these data were also not available in the current study and could provide a clearer picture of which kick types lead to the greatest onset of injury. It is also worth noting that the estimates of kicking injury frequency in the current study may be conservative. It was only possible to extract acute-onset kicking injuries from the dataset and, as such, gradual-onset injuries which were caused or exacerbated by repeated kicking are excluded. It is currently unknown how kicking volumes may contribute to accumulated match fatigue and injuries classified as occurring during another match event, and vice versa. As such, it may be beneficial to investigate all injuries sustained by kickers in order to truly understand the position-specific problems, although this is clearly not without its challenges, due to the multi-factorial nature of injury risk.

| CONCLUSION
This is the first study to investigate the epidemiology of kicking-related injuries in professional Rugby Union with this level of detail and the first to present a breakdown of position-specific kick types over multiple seasons. Kicking injuries are most commonly sustained in the pelvic region and lower limbs, with muscular strains being the most common injury diagnosis, particularly of the kicking limb rectus femoris. Traditional incidence of kicking-related injuries is low in comparison with other mechanisms of injury; however, clear position-specific factors have been identified. The fly-half sustains the most injuries from kicking with an average propensity, scrum-halves have a very low propensity for kicking injury, and forwards and other backs have a high propensity for kicking injury. Time in match affects the incidence of kicking injury with the majority of kicking injuries being sustained in the second half and during the warm-up.

| PERSPECTIVE
Practitioners are advised to consider the current findings when incorporating kicking into position-specific training and conditioning programs due to the differences demonstrated between playing positions in both propensity and match kicking characteristics, and these should also be considered in warm-up preparations prior to matches. It may be that specific preparation for kicking (eg, during training) and other factors such as the type of kick performed may influence the observed differences in propensity, however, further research is required to directly investigate this. Further research is also needed to better understand the potential mechanisms for kicking thigh muscle strain injuries, potentially across the different kick types, as well as to explore whether continued exposure to kicking may influence the occurrence of other gradual-onset injuries.