Hockey is an extremely popular sport in Canada and many other countries. There are more than 570 000 registered minor hockey players in Canada.1 Currently, one of the most contentious issues within minor hockey is body checking. In 2002, the Canadian Hockey Association implemented a policy change in minor hockey that lowered the age groups for Pre-Novice, Novice, Atom, Peewee, Bantam, and Midget by a single year.2 This age change resulted in the introduction of body checking 1 year earlier than in previous seasons (ie, from 12 to 11 years). In some jurisdictions, this age change corresponded with introducing body checking at the Atom level (9 and 10 year olds).3
Many individuals involved in minor hockey believe that introducing body checking at an earlier age results in fewer and perhaps less severe injuries in older divisions because of a player's instinctive ability to protect himself/herself. However, this belief is largely anecdotal. The purpose of this study was to evaluate the effect of the Canadian Hockey Association's change in age categories in relation to hockey injuries treated in emergency departments (EDs) following 2 cohorts of minor hockey players from the Atom division through the Bantam division.
Injury data were obtained from the 7 EDs that serve Edmonton, Alberta, and the surrounding communities of St Albert and Leduc, through the Ambulatory Care Classification System (ACCS) administrative database. The ACCS is an administrative health database that contains clinical (eg, diagnoses, procedures) and demographic information (eg, sex, age) for persons who present to EDs. Data from ED charts, nursing notes, and consultation reports were entered into the ACCS by trained nosologists. Within this database, there is a coding system to specifically identify what sport and recreational activity an individual was engaged in at the time of injury. There are more than 120 sport and recreation codes pertaining to different activities.
Emergency department injury data for ice hockey injuries for males aged 9 to 15 years were obtained from September 1 to April 31 for each year from 1997 to 2010. Within these 13 hockey seasons, we identified 2 cohorts of players: 1 before and 1 after the age change (Table 1). The pre–age change cohort was identified as players 10 and 11 years of age during the 1997/98 season (Atom), who were then followed through the 1999/2000 season (Peewee: aged 12 and 13 years) and 2001/2002 season (Bantam: aged 14 and 15 years). The post–age change cohort was identified as those who were 9 and 10 years of age during the 2003/2004 season (Atom), who were then followed through the 2005/2006 season (Peewee: aged 11 and 12 years) and the 2007/2008 season (Bantam: aged 13 and 14 years). Injury records for ice hockey that occurred outside of the normal winter hockey season (May 1 to August 31) were excluded. Injuries were categorized into the division for which the individual's age matched. For example, an injury to a 12-year-old was categorized as an injury in the Peewee division.
All 10 diagnosis fields recorded in the ACCS were searched for International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) (before 1 April, 2002) and International Classification of Diseases, Tenth Revision, Canada (ICD-10-CA) (after 1 April 2002) pertaining to injuries. Where multiple injuries (eg, a concussion and a fracture) were recorded in the same ACCS record (single visit), each injury contributed to 1 occurrence in the numerator for the respective injury category. Injuries were classified as fractures (ICD-9-CM: 800-829; ICD-10-CA: S02, S12, S22, S32.0-S32.8, S42, S52, S62, S72, S82, S92, T02, T08.1, T10.1, T12.0, T14.2, T90.2, T91.1, T91.2, T92.1, T92.2, T93.1, T93.2), head and neck injuries (800-804, 830, 848.0-848.2, 850-854, 870-874, 900.0-900.3, 910, 918, 921, 940-941.59, 950, 807.5, 807.6, 920, 925.1, 925.2, 947.0, 953.0, 954.0, 957.0, 959.01, 959.09; S00-S19), and other injuries. Head and neck injuries were additionally classified as intracranial (800-804, 850-854; S06) and nonintracranial injuries.
Injury rates were calculated by summing the total number of hockey injuries treated in EDs for the respective 2-year age grouping and were divided by the total number of registered players in the corresponding division (ie, Atom, Peewee, Bantam). Risk ratios and 95% confidence intervals (CIs) were calculated using Stata version 10 (StataCorp LP, College Station, Texas). Results were considered significant where P < 0.05. Ethical approval was obtained from the University of Alberta's Health Research Ethics Board.
There were 8026 males, aged 9 to 15 years, registered with the Edmonton Minor Hockey Association (EMHA) for the 1997/98, 1999/00, 2001/02, 2003/04, 2005/06, and 2007/08 hockey seasons. The pre–age change cohort consisted of 4215 registered players, and the post–age change cohort consisted of 3811 registered players. The number of injuries treated for each age within each division, the number of registered players per division, and the injury rates for each division are displayed in Table 2. Overall, significantly higher injury rates for the pre–age change period were observed in the Peewee division, whereas injury rates in the Bantam division approached significance (P = 0.054). It was also observed that with the exception of the Atom division, the youngest players had a greater frequency of injury compared with the older players in each division, regardless of the 2 cohorts (Table 2).
The rates of injuries treated in EDs, by injury type, are shown in Table 3. Minor fluctuations between the pre–age change and post–age change cohorts were observed, but there were few statistically significant differences. No significant differences were observed for the Atom division. When stratified by injury type, injury rates for fractures and injuries to the head and neck (including fractures and intracranial injuries, such as concussions) were lower in the pre–age change cohort for the Atom and Peewee divisions but higher for the Bantam division; however, no significant differences were observed by injury type. Significant differences were only observed in the other injury category for the Peewee and Bantam divisions.
As player registration data did not include age, a sensitivity analysis was conducted to explore how injury rates would differ based on different age distributions within each division of minor hockey. Injury rates were recalculated by assuming (1) an equal distribution of players within each age (50% younger age vs 50% older age), (2) a 10% greater number of players in the older age (40% vs 60%), and (3) a 10% greater number of players in the youngest age (60% vs 40%). These sensitivity analyses provided parameters similar to the raw data for all 3 scenarios (Table 4). The following results pertain to the sensitivity with an equal age distribution. The overall rate of injury was significantly higher for only the Peewee players in the pre–age change cohort. When stratified by injury type, significant differences were only observed in the other injury category. In the Bantam division, no significant differences were observed for overall injuries or when stratified by injury type, or when comparing the youngest and oldest players in the pre–age change/post–age change cohorts. In the Atom division, injury risk ratios were only significantly lower for other injuries in the younger players (RR = 3.16; 95% CI, 1.05–9.46) after the age change took effect.
Injury risk ratios for fractures and other injuries, for the youngest players within the Peewee division, were lower than those in the oldest age group (fracture RR = 1.08 vs 1.31; other injury RR = 1.87 vs 3.00), but these differences were not significant. Injury risk ratios for head and neck injuries (both intracranial and nonintracranial) were also lower for the younger players within the Peewee division than those observed for the older players. In fact, the injury rates for the youngest players were higher after the age change (head and neck injuries RR = 0.72; 95% CI, 0.36–1.47), whereas the injury rates for the oldest players were lower after the age change took effect (RR = 1.53; 95% CI, 0.62–3.81), although neither was significant. In the Atom division, injury rates for the other injuries for the younger players were significantly lower in the post-age change period but were higher, although not significantly, for the oldest Atom players.
Using 2 cohorts of minor hockey players, this study investigated injury trends following 8026 registered hockey players through 3 divisions of minor hockey (Atom, Peewee, and Bantam). We observed that when stratified by injury type and division, there were significant differences in the rate of injuries that were neither fractures nor head and neck injuries for both the Peewee and Bantam divisions in the post–age change cohort. The rate of hockey-related injuries treated in EDs was not observed to significantly decrease or increase for any other injury type.
Two previous studies have also used ED data to investigate injury rates in minor hockey following this policy change4,5; however, the findings from these studies were inconsistent. Our results are similar to those reported by Kukaswadia et al,5 who compared injury rates using a similar data source in Ontario, when body checking was introduced at the Atom level (9 and 10 year olds). Their findings indicated that injury rates after the age change were similar to, and in some cases lower than, injury rates reported before the age change. Additionally, they reported that injury rates specifically because of body checking remained stable.5 However, the results of our study contrast findings reported by others. A study in the same health region by Hagel et al4 observed that injury rates for 10-year-olds and 12-year-olds were not significantly different after the age change, whereas 11-year-olds were 1.9 times more likely (95% CI, 1.5–2.4) to present to an ED with a hockey-related injury. Although we observed a significantly lower rate of overall injuries in the Peewee division and injuries other than fractures and head and neck injuries in the Peewee and Bantam divisions after the age change for the injuries, these differences may be because of a number of differences in methodology, including fewer injury-specific categories used in our analysis and the method by which we categorized injuries into the different divisions of minor hockey. Additionally, our sensitivity analysis only compared injury rates within divisions (youngest vs oldest age) and not between specific ages before and after the age change.
The debate regarding introducing body checking has 2 positions: (1) the introduction of body checking should be done with younger athletes, so they learn this skill in a safe manner to prevent injury in older divisions, and (2) body checking should be eliminated from young age groups where there is considerable size differentials or from minor hockey entirely. The purpose of this study, however, was to evaluate if the rate of hockey-related injuries treated in EDs changed after the change in age categories. This study was not intended to evaluate injury risk when introducing body checking in earlier divisions of minor hockey (ie, Atom).
It has been suggested that introducing body checking at an earlier age leads to decreasing injury rates over time, possibly due to a learning effect6; however, this conclusion has been disputed.7,8 Proponents of body checking at younger ages may suggest that the lower injury rates we observed in the post–age change seasons were due to an “instinctive” ability or increased skill to protect themselves due to earlier teaching of body checking. Two recent studies, which have investigated such claims, have reported that injury rates for players in Bantam are higher9 and lower,10 compared with other Bantam players, who were introduced to body checking in Peewee; however, neither study reported statistically significant differences. Several other studies have shown that injury rates increase when body checking is introduced4,6,11–13 and continue to increase significantly through the older divisions.12,14 For instance, Emery et al11 reported that Peewee hockey players in Alberta had injury rates 3.07 times greater (95% CI, 2.21–4.27) than Peewee players in Quebec, where body checking is not allowed in Peewee, whereas Darling et al13 reported that teams that were permitted to body check had injury rates that are 3.75 times higher (95% CI, 1.51–9.34) than teams who are not allowed to body check.
Body checking is commonly reported to account for a large proportion of injuries in minor hockey.5,9,11,12 In a review by Warsh et al,8 body checking was reported as the mechanism of injury for 2.9% to 91% of injuries; however, their findings indicated that the majority of research identified that body checking is the mechanism of injury in 40% or more of the reported injuries. Warsh et al8 also identified only 1 study,15 which has been questioned due to the denominator and self-reported injuries,4,8 that did not observe an increased risk of injury due to body checking. Additionally, another systematic review16 reported that the risk of injury is 2.45 times greater (95% CI, 1.7–3.6) when body checking is allowed. However, a recent study by Darling et al13 reported that 66% (95% CI, 60%–72%) of injuries in minor hockey are due to unintentional contact, rather than intentional acts such as body checking.
In the Peewee and Bantam divisions, we only observed significant decreases in injury rates that were neither fractures nor head and neck injuries. Injury rates were not observed to significantly increase either. It is unclear as to why the rate of injuries that were neither fractures nor head and neck injuries significantly decreased for the post–age change cohort in the Peewee and Bantam divisions. However, if a true potential benefit of introducing body checking 1 year earlier does exist, it seems to be negligible in later years because there were no significant differences between injury rates observed for fractures or head and neck injuries. Additionally, the results of the sensitivity analysis support that if there were an equal distribution of players in the younger and older ages of a division, the injury rates were not significantly lower in the post–age change cohorts for fractures or head and neck injuries.
How body checking is introduced, taught, and practiced is likely to be very important, but evidence shows that injury rates increase when body checking is introduced. As body checking is part of elite, semiprofessional, and professional men's hockey, it is unlikely that body checking will be removed from most levels of minor hockey, as advocated by some.17 A balance between hockey policy and injury prevention must be reached that allows children to develop skills, be safe, and most importantly have fun, while participating in this cherished sport. While parents should have input, as any policy decision will have implications to their children, policy makers at the national level must ultimately weigh the costs of increased injury rates for young children and the unproven belief that injuries will decrease due to a learning effect. However, given the percentage of players who reach the college or professional hockey ranks is quite small, following the precautionary principle would not support the introduction of body checking at earlier ages.
There are limitations to our study. First, the retrospective use of an administrative database limited our ability to identify all hockey-related injuries that would have occurred. For instance, injuries that were treated with first aid or where treatment was sought from a family physician, a sports medicine physician, a “walk-in” clinic, or elsewhere cannot be identified. Therefore, it is likely that minor injuries are underreported. Additionally, because of the retrospective design, the data obtained are limited to that entered into the ACCS by the trained nosologists, and we were unable to obtain information on the circumstances surrounding the injury. As such, we could not determine the mechanism of injury.
Second, the data obtained from the ACCS did not indicate the date of birth for the injured individual. Therefore, injuries were categorized into divisions based on the age at presentation to an ED. Categorizing injured players based on their age at presentation decreased the precision to which we were able to place injuries into the correct division (ie, a 12-year-old player participating in the Peewee division, who turns 13 during the second half of the season and then presents to an ED with an injury, would have been incorrectly classified as an injury in the Bantam division). However, this method of classification has been used previously.9
Third, our analysis was constrained by not having player registration data for each specific age, so we were unable to compare injury rates before and after the age change for specific ages. Therefore, our injury results by division may be confounded by age; however, our sensitivity analysis yielded similar results to the raw data.
Fourth, it is possible that variables related to the different periods may have affected the results (there was a 6-year gap between participation in the same division for the 2 cohorts). The average annual percentage change (AAPC) for hockey injuries that were neither fractures nor head and neck injuries treated in the local EDs decreased from the 1997/98 to the 2009/10 season (AAPC = −3.24%), whereas the AAPC for fractures (0.44%), head and neck injuries (0.88%), and intracranial injuries (1.43%) increased during the same period (Alberta Centre for Injury Control and Research, unpublished data, 2010).
Additionally, beginning in the 2007/08 season, the EMHA implemented an injury reduction strategy, which aimed to reduce head and neck injuries, reward fair play, and promote sportsmanship. This prevention strategy—which applies only to the Bantam post–age change cohort in this study—may have influenced the results by reducing the injuries in the 2007/08 season; however, during the first year of the injury prevention program, only nonintracranial head and neck injuries decreased (P > 0.05) in the Bantam division from the previous season (2006/07), whereas the rates for other injury types increased (P > 0.05). Therefore, we believe that it is unlikely that the data for the post–age change Bantam cohort were heavily influenced by this intervention during this single season.
Despite these limitations, our study does have strengths. First, by using 2-year intervals for the pre–age change/post–age change cohorts, we have essentially captured the same players in each cohort who continued to participate in minor hockey from the Atom division through the Bantam division. By defining our cohorts in this manner, the same children who were present in the pre–age change/post–age change cohorts in the Atom division should be present in the Peewee and Bantam divisions during the subsequent seasons, if they continued to participate in minor hockey in Edmonton. Therefore, we have defined cohorts that are more stable than if we had used consecutive years. Second, by using an administrative health database to calculate injury rates, we have likely captured a substantial proportion of serious injuries such as fractures and recognized intracranial injuries. By doing so, we have likely eliminated the variability and bias that occurs by using self-report, parents, volunteers, or team “trainers” to document and report injuries.
The age at which body checking is introduced to minor hockey players remains a contentious issue. Evidence suggests that introducing body checking increases the risk of injury.4,6,7,11–13,18 Although limited statistically significant differences have been reported, injury rates have been observed to be slightly lower in older age groups when body checking is introduced earlier,5,10 it has also been reported that injury rates due to body checking are slightly higher when body checking is introduced earlier.9
Based on our findings and the current literature, we do not believe that lowering the age at which body checking is introduced even further (ie, to Atom or Novice) would result in significantly lower injury rates in older divisions. Although we did observe slightly lower and mostly nonsignificant decreases in the rate of injuries after the age change in the 2002/2003 season, we suggest that there is limited reason to introduce body checking to children as young as (or younger than) 9 and 10 years, subjecting them to the 2-fold to 5-fold increased risk of injury,4,11 to achieve what seems to be limited and nonsignificant decreases in injury rates in older divisions. Because the number of players who participate in minor hockey decreases in oldest age groups, the efficacy of such potential benefit is ultimately reduced due to increased injuries in the youngest age groups. Additionally, as the debate surrounding when to introduce body checking is likely to continue, the development of a standard criteria for data collection and reporting would greatly aid further research in this field.