Article Text
PDF
Abstract
Objective To identify risk factors for injury in amateur club rugby.
Design Prospective cohort design; with follow-up over the 2004 season.
Setting Amateur club rugby in New Zealand.
Participants Seven hundred and four male rugby players, aged 13 years and over.
Assessment of risk factors The study investigated the independent effect on injury incidence of age, ethnicity, rugby experience, height, weight, body mass index, physical activity, cigarette smoking, previous injury, playing while injured, grade, position, training, time of season, warm-up, foul play, weather conditions, ground conditions and protective equipment. Generalised Poisson regression was used to estimate the effect of each factor after adjusting for all other factors.
Main outcome measures Game injury, defined as ‘any event that resulted in an injury requiring medical attention or causing a player to miss at least one scheduled game or team practice’.
Results A total of 704 players, representing 6263 player-games, contributed information on injury and exposure. Evidence was obtained of the effect on injury incidence of increasing age, Pacific Island versus Maori ethnicity (injury rate ratio (IRR)=1.48, 1.03–2.13), ≥40 h strenuous physical activity per week (IRR=1.54, 1.11–2.15), playing while injured (IRR=1.46, 1.20–1.79), very hard ground condition (IRR=1.50, 1.13–2.00), foul-play (IRR=1.87, 1.54–2.27) and use of headgear (IRR=1.23, 1.00–1.50).
Conclusions Opportunities for injury prevention might include promoting injury-prevention measures more vigorously among players of Pacific Island ethnicity, ensuring injured players are fully rehabilitated before returning to play, reducing the effects of ground hardness through ground preparation and stricter enforcement of the laws relating to foul play.
Statistics from Altmetric.com
Introduction
The literature on injury in rugby union football is dominated by descriptive studies, often concerned with elite or professional players.1 These studies suggest that the risk of injury in rugby is high, relative to other football codes,1 2 although comparisons are frustrated by a lack of uniformity in study designs.3 While many putative risk factors are named in the literature, there have been few attempts to confirm these in analytical studies, with two notable exceptions undertaken in Scotland4,–,7 and New Zealand.8,–,10 The objective of this study was to investigate the independent effect on injury incidence of a number of commonly postulated risk factors (as listed in methods).5,–,9 11
Methods
We used data collected during the development of a national system for the routine surveillance of injury and safety behaviour in amateur rugby (Gulliver and Chalmers, unpublished data, 2010) which employed a prospective cohort design.12
Participants
All active players are required to register with the New Zealand Rugby Union. For this study, eligible players were males aged 13 years and over who were registered in 2003 and for whom complete contact details were available. A sample of 640 players was needed (allowing for 20% loss to follow-up) to give 80% power (p=0.05) to detect a 15% change in incidence rates (IRs). A random sample of 3159 players was drawn in February 2004 and invitations to participate were posted to them. Purpose-trained interviewers telephoned, assessed eligibility and 836 players who consented to participate and were intending to play during the 2004 season were enrolled in the cohort (figure 1).
Participants and questionnaires by phase of study.
Procedure
A ‘background’ questionnaire was administered (figure 1), with data collected on demographic factors (age, ethnicity)7 9; experience9 (years of participation); anthropometric (height, weight)9 and performance9 (hours per week during the previous 4 weeks spent in ‘strenuous physical activity’) characteristics; preseason training11 (any participation in organised training); previous injury6 9 (any injury event over the past 12 months); playing while injured9 (in the past 12 months); and cigarette smoking history9 (non-smoker, ex-smoker or current smoker). A ‘preseason’ questionnaire was administered at the beginning of the competitive season (figure 1), with data collected on physical conditioning undertaken during training.11 13
Weekly follow-up interviews throughout the competitive season collected data on grade9 (Senior A, Senior B/Senior reserve, Colts, Under 18/19, Under 14/15/16/17/schoolboy, Presidents/social, other), position9 11 (front row, lock, loose forward, inside back, midfield back, outside back, playing out of usual position), practice attendance6 (ie, ‘scheduled events organised by team captain or coach’), game participation (excluding games not part of competition), subjective assessment of environmental conditions5 11 (rain, wind, temperature, ground), foul play (being a victim at any time during a game) and behaviour with regard to selected injury-prevention measures (time spent on warming up;use of protective equipment8 (headgear, mouth guard, shoulder padding, taping/strapping, shin guards)). The outcome measure was ‘game injury’, defined as ‘any event that resulted in an injury requiring medical attention or causing a player to miss at least one scheduled game or team practice’. Only new injuries occurring in a given week were recorded. No attempt was made to determine if an injury was a recurrence of a previous injury (ie, one occurring after return to full participation).3 Players who withdrew from a game due to injury were considered to be exposed to the whole game. The study was approved by the Otago Ethics Committee of the Ministry of Health (Ref: OTA/01/10/079). Additional information on the questionnaires can be obtained from the authors.
Analyses
Continuous variables were categorised into groups, as appropriate. Ethnicity was categorised into ‘Maori’, ‘Pacific Island’ and ‘other’ groups.14 The majority of players in the ‘other’ group were of European ethnicity, but there were small numbers of players of other ethnicities. Body mass index (BMI) was calculated from individual weight and height measurements (kg/m2) and players were grouped according to WHO BMI categories.15 BMI could not be calculated for 32 players, representing 304 player-games (4.8%), because those players were unable to recall either their height or their weight at the time of interview. Accordingly, height and weight were also included in the analysis as separate variables. Players who participated in preseason physical conditioning were assigned to one of three impact levels: low, moderate or high,13 with priority given to the higher level when multiple levels involved.
Univariate and multivariate analyses were undertaken. The injury IR was calculated as the number of injury events per 100 player-games. In the univariate analysis, injury IRs, injury rate ratios (IRRs) and CIs were estimated for each level of each factor. In the multivariate analysis, only IRRs and CIs were estimated.
For the multivariate model selection, we included variables based on p values obtained in the univariate analysis. Starting position, use of protective gear, preseason training and preseason conditioning were included regardless of univariate p values because of reported associations with IR.6 8 9 11 Twenty-two variables were entered into the multivariate model selection process: age, ethnicity, rugby experience, BMI, strenuous physical activity, previous injury, playing while injured, grade, position, playing out of usual position, preseason training, preseason conditioning, time of season, wind, ground condition, time spent on warming up, foul play, and use of mouth guards, shin guards, headgear, taping/strapping, and shoulder padding. The model selection involved (a) using a backward elimination procedure to sequentially discard non-significant variables until the largest p value was no greater than 0.1 and (b) using deviances to compare the final model with the initial 22-variable model.16 Generalised Poisson regression was used to model the IRRs.17 18 Each participant was included as a ‘cluster’. Over-dispersion was controlled for using a variance inflation factor based on Pearson χ2 values; observed to be close to unity in all models fitted (range: 1.08–1.12). The model fit was assessed by analysing residuals in the regression diagnostics.19
Results
During the 28-week season, data were collected from 704 players on 6263 player-games; an average of 8.9 games per player. On average, 88% of the cohort was successfully contacted each week of the follow-up period. One hundred and thirty-two players contributed no follow-up data, while 98 contributed some data but were lost to follow-up during the season (figure 1). Data collected before loss to follow-up were retained in the analysis. Six hundred and fifteen game injuries were reported; giving an average of 0.87 injuries per player and an injury IR of 9.8 per 100 player-games (or about 93 injuries per 1000 player-hours (Chalmers, Stephenson and McNoe, unpublished data, 2005)). Of the 615 injuries, 216 (35%) were to the lower limb, 184 (30%) to the head, face or neck and 143 (23%) to the torso; 256 (42%) were sprains or strains and 140 (23%) were contusions (table 1).
Injuries by site and type
Univariate analysis
With the exception of the 18–20-year and 35 plus-year age groups, there was an increasing risk with age (table 2). Pacific Island players, more experienced players, and taller and heavier players were at higher risk. A BMI above 25 kg/m2 was associated with higher risk and a BMI below 20 kg/m2 was associated with lower risk, when compared with players in the normal range (ie, 20–24 kg/m2). Players reporting strenuous physical activity of ≥40 h per week were at higher risk. There was no evidence that IR was associated with cigarette smoking. Players with a history of greater than or equal to two injuries in the past 12 months and players who played while injured were at higher risk.
Intrinsic factors associated with injury incidence: univariate results
There was no evidence of an association between IR and position, participation in preseason training or conditioning, weather conditions (precipitation, wind, temperature), in-season practice attendance or time spent on warming up (table 3). Players in the Senior A and Presidents/social grades were at higher risk than the reference group (Under 18/19). Playing out of usual position appeared to be associated with higher risk. There was a downward trend in the risk of injury over the course of the season. Very hard ground and foul play were associated with higher risk.
Extrinsic factors associated with injury incidence: univariate results
The majority of players wore mouth guards, while few taped/strapped their joints or wore shin guards (table 4). Injury risk was lower for mouth guard and shin guard users when compared with non-users. The wearing of headgear and shoulder padding appeared to be associated with higher risk.
Protective equipment associated with injury incidence: univariate results
Multivariate analysis
Of the 22 variables included in the multivariate model, eight were retained: age, ethnicity, grade, strenuous physical activity, playing while injured, ground condition, foul play and use of headgear (table 5). Data from 690 players, representing 599 injuries from 6030 player-games, were included. Injury risk was higher for all age groups when compared with the 13–15-year age group. Players of Pacific Island ethnicity had a 48% higher risk compared with Maori players (reference group). The Presidents/social grade had a 92% higher risk compared with the Under 18/19 grade. Players who did ≥40 h per week of strenuous physical activity had a 54% higher risk of injury, compared with those doing less. Players who had played while injured in the past 12 months had a 46% higher risk compared with their counterparts. Very hard ground was associated with a 50% higher risk compared with firm ground. Foul play was associated with an 87% higher risk. Headgear users were 23% more likely to sustain injury compared with non-users.
Effects of individual factors on the injury incidence risk after controlling for the effects of all other factors in the model: multivariate results
Discussion
We found evidence of associations between IR and increasing age, Pacific Island ethnicity, strenuous physical activity, playing while injured, hard ground, foul play and use of headgear. We found no evidence of associations between IR and rugby experience, height, weight, BMI, cigarette smoking, previous injury, grade, position, warm-up, preseason training, preseason physical conditioning, time of season, weather conditions, use of mouth guards, shoulder pads or taping/strapping of joints.
A consensus statement on injury definitions and data collection procedures in studies of injuries in rugby union was published after this project had commenced.3 The cohort design, study population, follow-up period and definitions of match and training exposure were generally consistent with those prescribed. Our definition of injury was not as broad as that prescribed.3 We excluded ‘practice injury’ from our outcome measure because of presumed differences in exposure to risk between games and practices, and to permit comparisons with previous studies.4,–,9 11 Practice injuries accounted for only 17% of the injuries sustained by the cohort and the IR of 1.2 injuries per 100 player-practices (Chalmers, Stephenson and McNoe, unpublished data, 2005) was one-eighth of that of game injuries.
The necessity to sample from players registered in the previous season and the large number who could not be contacted, means that the cohort cannot be said to be truly representative of the 2004 playing population. With a median starting age for rugby of 7 years (Chalmers, Stephenson, Eyles and McNoe, unpublished data, 2004) it is unlikely that the number of new entrants in 2004 aged 13 years and over was large. The demographic characteristics of the cohort were similar to those of the source population (Chalmers, Stephenson, Eyles and McNoe unpublished data, 2004).
In contrast to the earlier cohort studies which measured putative risk factors directly,5,–,7 9 10 this was not practicable in this nationwide study and may have introduced measurement errors, such as the tendency to overestimate height and underestimate weight and BMI.20 Because we did not record the team membership, we were unable to take into account any clustering of the effects of environmental conditions by the team. We included data on environmental conditions in the analysis, however, and the standard errors were inflated to reflect clustering of player-games.
It is likely that some important factors were not measured by the surveillance system from which the data for this study were obtained, such as standard of play, and could not be controlled for in the multivariate model. Our outcome measure, that is, the injury IR, does not measure the impact of injuries, as it does not take into account the severity of those injuries (cf. Quarrie et al who used the proportion of season missed due to injury as a proxy for severity).9
Recent reviews have drawn attention to the lack of research on non-elite and younger players, and the infrequency of multivariate analyses.1 21 To our knowledge, this is the first cohort study to investigate a large number of putative risk factors in a nationwide sample of amateur club rugby players. The earlier cohort studies undertaken in Scotland and New Zealand were restricted to senior players and players in a single city, respectively.4 10 In making the following comparisons with the earlier studies, differences in definitions, outcome measures and methods of analysis are acknowledged.4,–,6 9 10
Intrinsic factors
Consistent with Quarrie et al,9 we found no evidence of associations between IR and years of participation, height, weight, BMI or cigarette smoking. While observing an association between age and risk, the strength of the association was inconsistent across age groups. Quarrie et al found no evidence of an association after adjusting for other factors,9 while Lee et al reported an almost twofold higher risk for players aged 25–29 years compared with players aged less than 16 years.6 The evidence relating to this factor remains equivocal.
While observing no difference in risk between ‘other’ ethnicities (predominantly European) and Maori, the risk for players of Pacific Island ethnicity was 48% higher than that for Maori. Quarrie et al found no evidence of ethnic differences but grouped together players of Maori and Pacific Island ethnicity, precluding direct comparison.
Players who reported undertaking ≥40 h of strenuous physical activity per week were shown to have a 54% higher risk of injury when compared with those less active. This finding is consistent with Quarrie et al, who reported evidence of a strong association between this factor and ‘proportion of season missed’.9
Previous injury has been identified as a risk factor in many investigations of sports injury22 but the evidence in rugby is inconsistent. For instance, while Quarrie et al observed no association between IR and injury in the previous season, Lee et al reported an association between IR and injury at the end of the previous season.6 9 Similarly, while Quarrie et al reported an association between IR and injury that affected player's ability to participate in preseason training, Lee et al reported no effect for those injured during preseason training.6 9 We found no evidence of an association between a history of injury in the past 12 months and the risk of in-season injury. We observed, however, that players with a history of playing while injured had a 46% higher risk of in-season injury, supporting earlier exhortations that injured players should be fully rehabilitated before returning to play.6 9
Extrinsic factors
In contrast to Quarrie et al,9 we found no evidence of an overall association between grade and IR, although the IR for the Presidents/social grade was 92% higher than for the Under 18/19 grade. The latter may relate to quality of training and standard of play. Consistent with previous reports,1 9 we found no evidence of an association between position and IR in our univariate analysis. Lee et al observed that there may be an increased risk of injury for players playing out of their usual position7 but consistent with Alsop et al we found no evidence of this.11
We found no evidence of an association between IR and participation in preseason training or in preseason physical conditioning.9 Lee et al reported increasing IR with increasing levels of participation in preseason training but argued that this factor could be confounded by other factors such as player confidence or game intensity.6 Our measure of conditioning was any preseason participation in a range of low-, medium- and high-impact activities,13 but we did not assess the extent of such conditioning. We found no evidence of an association between IR and in-season practice attendance.
Consistent with earlier studies,5 11 our univariate analysis showed a downward trend in IR over the course of the season, but this factor was eliminated from our multivariate model. Orchard has argued that an ‘early-season (injury) bias’ reported in rugby studies can be explained by changes in ‘ground hardness and/or shoe-surface traction’ over the course of a season.23 While associations between IR and ground conditions have been observed in some studies, these have been confounded by time of season.5 24 One study reported strong associations between IR and temperature and wind strength, and concluded that weather conditions were more likely to influence injury risk than ground conditions.5 Another, however, reported that IR was substantially higher on hard grounds (vs soft grounds), after controlling for time of season.11 We found no evidence of an association between IR and weather conditions but observed a 50% higher risk for very hard grounds (vs firm grounds). On balance, the evidence supports Orchard's view.23
Warming up before games is advocated as a means of reducing injury risk.13 We were unable to show an association between pregame warm-up and IR, probably because players warmed up before 98.6% of player-games. We observed that being the victim of foul play at any time during a game increased the injury risk. Foul play can take many forms, including dangerous play and misconduct.25 Without further information on the nature of the foul play, standard of play or game management, we do not wish to speculate on possible explanations for this finding.
Of the items of personal protective equipment investigated, we observed only that players wearing headgear were at a 23% increased risk of injury. We acknowledge that as most protective gear is intended to reduce injury to specific sites our ability to assess their effectiveness was limited using ‘all injury’ as our outcome measure. A further limitation was that for several items of equipment, one level of the factor was very small relative to the other (eg, mouth guards, for which wearing is compulsory). Headgear has been reported to reduce the risk of injury to the scalp and ears but not concussion,8 26 although one recent study has reported a protective effect for players who always wear headgear during games.27 Our finding is consistent with the suggestion that players who wear headgear may play a riskier game because of the perceived protection provided by wearing headgear.8 21
What this paper adds
▶ While many putative risk factors are named in the literature on injury in rugby there have been few attempts to confirm these in analytic studies, especially at the community (amateur) level of the sport.
▶ The authors found evidence of associations between the injury IR and increasing age, Pacific Island ethnicity, strenuous physical activity, playing while injured, hard ground, foul play and use of headgear.
▶ The authors found no evidence of associations between the injury IR and rugby experience, height, weight, BMI, cigarette smoking, previous injury, grade, position, warm-up, preseason training, preseason physical conditioning, time of season, weather conditions, use of mouth guards, shoulder pads or taping/strapping of joints.
Meaning of the study
The reason for investigating putative risk factors is to assist in identifying opportunities for intervention. From our findings and evidence from earlier studies, such opportunities might include promoting injury-prevention measures more vigorously among players of Pacific Island ethnicity, ensuring injured players are fully rehabilitated before returning to play,13 reducing the effects of ground hardness through ground preparation,23 and stricter enforcement of the laws relating to foul play.28
Unanswered questions and future study
Although we found no association between IR and preseason training or conditioning, we did not measure the nature or extent of participation in training. We were unable to confirm the effect of conditioning on the ability of players to withstand early season impacts.9 Further research is needed to understand the increased risk carried by players of Pacific Island ethnicity and the association between foul play and injury risk. Further analyses are required in which the outcome measures are limited to those injury types expected to be prevented by each item of personal protective equipment.
Acknowledgments
The authors gratefully acknowledge the support and assistance of the New Zealand Rugby Union and, in particular, that of Ken Quarrie, Injury Prevention Manager, for his assistance with the design of the questionnaires and his comments on an earlier version of this manuscript. The authors are grateful to Shaun Stephenson and Dan Russell for developing and maintaining the data entry software and database. Finally, the authors are grateful to the players for volunteering their time to participate in the study and the interviewers for their conscientious data gathering.
References
Footnotes
-
Funding Health Research Council of New Zealand.
-
Competing interests All authors have completed the Unified Competing Interest form at http://www.icmje.org/coi_disclosure.pdf (available on request from the corresponding author) and declare that (1) they have financial support for the submitted work from the Health Research Council of New Zealand and (2) they are employees of the University of Otago (DJC is a retired employee); and have no financial relationships with commercial entities that might have an interest in the submitted work. All authors declare that they have had financial support from grants awarded to the University of Otago by the Accident Compensation Corporation (ACC). All authors also declare that they have (3) no spouses, partners, or children with relationships with commercial entities that might have an interest in the submitted work and (4) no non-financial interests that may be relevant to the submitted work.
-
Ethics approval Otago Ethics Committee of the New Zealand Ministry of Health.
-
Provenance and peer review Not commissioned; externally peer reviewed.