Background The fast, random nature and characteristics of ice hockey make injury prevention a challenge as high-velocity impacts with players, sticks and boards occur and may result in a variety of injuries, including concussion.
Methods Five online databases (January 1970 and May 2012) were systematically searched followed by a manual search of retrieved papers.
Results Seventeen studies met the inclusion criteria. The heterogeneous diagnostic procedures and criteria for concussion prevented a pooling of data. When comparing the injury data of European and North American or Canadian leagues, the latter show a higher percentage of concussions in relation to the overall number of injuries (2–7% compared with 5.3–18.6%). The incidence ranged from 0.2/1000 to 6.5/1000 game-hours, 0.72/1000 to 1.81/1000 athlete-exposures and was estimated at 0.1/1000 practice-hours.
Discussion and conclusions The included studies indicate a high incidence of concussion in professional and collegiate ice hockey. Despite all efforts there is no conclusive evidence that rule changes or other measures lead to a decrease in the actual incidence of concussions over the last few decades. This review supports the need for standardisation of the diagnostic criteria and reporting protocols for concussion to allow interstudy comparisons in the future.
- Ice hockey
- Contact sports
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
Ice hockey is a fast collision sport rooted in a heavily reinforced culture of aggressive play1 that is played predominantly in North America, Europe and countries of the former Soviet Union. Players may reach speeds of up to 50 km/h and pucks being accelerated to up to 160 km/h.2 The fast, random nature and characteristics of this sport make injury prevention a challenge as high-velocity impacts with players, sticks and boards occur and may result in concussion.
In addition to contact with a stick or collisions, head or body checks are a major cause of concussion.3–6 Head checking can be referred to as a player being struck above the shoulders with either body, shoulder, elbow, knee or stick of an opposing player.7
The primary cause of concussion has shown to be associated with inertial, or acceleration, loading at the moment of contact and the resulting transient increase in pressure. Brain tissue is resistant to changing shape when subjected to slow or transient pressures. However, it has been shown to deform more readily in response to shear forces. As rapid head rotations, for example, as a result of a check, generate such shear forces through the brain,8 the resulting tissue damage may cause transient neurological impairments observed in concussion.9
While an elimination of concussion in contact sports like ice hockey is impossible, new initiatives such as educational and communicative efforts have been recommended in addition to further improvements in protective equipment and rule enforcement10 in the light of potential long-term neurological deficits originating from head trauma.11 ,12
The aims of this systematic literature review were (1) to determine the magnitude and incidence of concussions in international professional and collegiate ice hockey, (2) the possible influence of changes in rules or equipment on concussion rates, (3) to investigate whether differences between leagues or countries exist and (4) if so, whether any differences can be explained.
A comprehensive search strategy was developed by identifying all potentially relevant search terms, categorising these terms into specific search phases and subsequently combining them by using Boolean terms. This search strategy was applied to five different electronic databases: PubMed, MEDLINE, EMBASE, Web of Science, ScienceDirect.
Occasionally, mild traumatic brain injury (mTBI) and concussion are used interchangeably in the literature although these two terms refer to different injury constructs.9 In order not to miss potentially relevant studies, the terms “mild traumatic brain injury” and “mTBI” were included in the literature search. The detailed search strategy for PubMed is shown as figure 1.
All databases were searched using the search strategy described above. Appropriate minor modifications to the basic search template were made to optimise the strategy in each of the five databases. Papers were limited to human studies published between January 1970 and May 2012.
Searching other resources
The hand search included analysing references cited in studies selected from the original online search. Citation searches of relevant studies were conducted using the PubMed, MEDLINE and ScienceDirect databases.
Data extraction and management
For the purpose of this review AR acted as the principal reviewer. AG and WK were involved independently in the process of identifying relevant studies and participated in further analysis of the finally included papers.
The third reviewer was used for a majority decision in case discrepancies between two reviewers were not reconciled by discussion. To standardise the procedure between the reviewers, the principal reviewer developed a detailed protocol sheet for critical appraisal by which general information on objectives, design, participant's demographics and outcomes were extracted. Each reviewer retrieved the information independently. A test was conducted with two articles similar but unrelated to the review question and the procedures discussed.
Papers were limited to peer-reviewed journals with language restrictions to English and German. The inclusion criteria were: studies that were fully or partially concerned with the incidence of concussion in male adult professional and collegiate ice hockey players.
With regard to the research question, the data extraction consisted of eight main areas: (1) study design, (2) source of data, (3) period of data collection, (4) competitive level, (5) number of participants, (6) definition/grading of concussions, (7) concussions as a percentage of the overall number of injuries and (8) the calculated incidence of concussions. Where data were missing, efforts were made to contact the corresponding author by email.
It was intended to pool incidence data if identical outcome measures were identified. Simple descriptive statics were used; all data were exported to PASW Statistics 18 (SPSS Inc, 2009) for statistical analysis.
Methodological quality assessment
A qualitative assessment was not conducted as for this type of epidemiological studies the quality primarily hinges on the correct diagnosis of concussion. Until recently, no standardised assessment tools for such examinations were recommended and available, which renders a balanced grading particularly of older problematic studies.
Literature search results
Initially, the online search strategy identified 123 studies of which abstracts were screened individually by the reviewers. The application of inclusion/exclusion criteria and consensus by the reviewers on the titles and abstracts eliminated a further 84 papers. The most common reason for rejection was not meeting the selection criteria.
From the titles and abstracts of papers selected (n=39) and, where possible, full papers were reviewed and the reviewers applied the inclusion criteria to the full text. Eighteen studies met the inclusion criteria and were presented in this review. Four of these 18 were added after the hand search of reference lists of included papers (figure 2).
Overall, the 18 studies meeting the inclusion criteria covered a period from the 1988–1989 to 2007–2008 seasons and collected data over 113–17–164 seasons. The majority of studies were designed prospectively (11/18, 61%) and the diagnosis of concussion was based on the clinical examination of the team physician (10/18, 56%). In most cases, the incidence of concussions was investigated in professional hockey (12/18, 67%). Teams from seven nations were enrolled, mostly from the USA (7/18, 39%).
Unfortunately, less than half of the studies gave definitions of the diagnostic criteria for concussion (8/18, 44%), three of the authors kindly provided this and other information after being contacted.18–20 Only one study reported on the distribution of severity and also indicated recurrent concussions.21 Where data were collected over several seasons, only two studies presented the results separately per season to allow the appreciation of trends.22 ,23
The incidence of concussion was reported either as a percentage out of the total number of injuries or in terms of injuries per 1000 athlete-exposures (AE),15 ,18 ,23 practice-hours (PH),4 ,19 game-hours (GH),4 ,5 ,19 ,20 ,22 ,24 per 100 players22 or per 1000 games.25
The incidence varied between 2% and 22% of all injuries sustained for periods ranging from a single season to a whole career. When comparing the injury data of European and North American or Canadian leagues, the latter show a higher percentage of concussions (2–7% compared with 5.3–18.6%).
With regard to GH, similar incidences of 0.2/100020–1.8/100022 were reported between 1996 and 2004 seasons in both Europe and North America. In contrast, Tegner and Lorentzon5 reported a comparably high value of 6.5/1000 GH.
The reported incidence of concussion in relation to AE showed more variation between the studies, ranging from 0.72/100018 to 1.81/100023 depending on the season during which it was investigated. The two studies not using newspaper releases as a data source report an incidence of 0.72/100018–1.55/1000 AE.18
Wennberg et al23 ,25 took the unusual approach of reporting the incidence of concussions based on news releases from “The Hockey News” and “Sporting News Hockey Register”. The two papers present conflicting results with regard to the total number of concussions for the seasons 1997–2002 with differences ranging from 7% (2001–2002) to 30% (1997–1998).
While the data suggest that there is a high incidence of concussion in professional and collegiate ice hockey, it is difficult to come to comprehensive conclusions. Protective equipment such as full or half shields may reduce the severity of concussions,26 but there is no conclusive evidence that improvements of protective equipment or rule changes over the years actually reduced the incidence of concussions itself.
Among other things, this may be due to statistical reasons. Joint efforts to reduce the incidence of concussion as well as an increased awareness of the long-term effects of recurrent head trauma may have improved diagnostic procedures and consequently lead to a higher number of detected and reported cases. The real incidence on the other hand may have actually decreased over time, or the number of such injuries remained unaltered despite all efforts to the contrary.
Also, all but one of the included studies failed to indicate whether the concussions were first events.21 As it has been shown that there is an association between the reported number of previous concussions and likelihood of incident concussion,27 this may have affected the reported incidence.
Sources of bias in epidemiological studies on concussion
While in most instances the diagnosis for concussion originated from the team physician, athletic trainer or athletic therapist, Wennberg et al23 ,25 chose newspaper releases as a data source. This entails a high risk of reporting bias which renders the interpretation of this problematic data as seen in the discrepancy of up to 30% for the overall number of concussions between the two newspapers. Although this difference decreased to about 7% in recent years, it remains questionable whether such medical news releases have a value for research purposes. The quite similar values between the sources for the seasons 2000–2002 may indicate a higher awareness of the concussion issue in the media as this period coincides with a marked increase in publications on sports-related concussions.28
There was also generally selection bias throughout the studies as the data were collected from samples of convenience rather than randomisation and it remains unclear whether this affected the overall results.
Definition and grading of concussion
There is no doubt that concussion is one of the most complicated injuries faced by a team physician or athletic trainer as there are typically no visible signs and the diagnosis initially relies on the keen eye of medical personnel to detect subtle differences in the athlete.
So far, there have been no generally accepted guidelines for the diagnosis of concussion. Consequently, studies where different teams were followed up over several seasons could not necessarily rely on specific diagnostic criteria due to the heterogeneous methods employed by the medical personnel. This also renders any interstudy comparison and interpretation of results very difficult.
For example, as loss of consciousness is not frequently encountered in sports-related concussions,29 such a definition of concussion is likely to cause a gross underestimation of the career incidence as reported by Tegner and Lorentzon.5 Nevertheless, this definition appears reasonable as the athlete may be able to accurately recall instances with loss of consciousness rather than injuries associated with less-serious signs and symptoms.
Recently, guidelines for the definition, classification and evaluation of concussions have been released for ice hockey and the SCAT-2 form was recommended.10 If generally employed, this will simplify the interstudy comparison and interpretation of future research.
There is also reason for concern that concussion is under-reported in ice hockey.22 It is possible that team physicians employ (too) high thresholds for diagnosing concussions for the same rationale that athletes may fail to report a concussion. Reasons may include the feeling the injury is not serious, a personal desire and feeling pressured to continue play, a failure to recognise the symptoms of concussion by athlete and/or medical staff, or the fear of jeopardising future career and/or financial benefits in case of prolonged return-to-play periods.30
However, it has been suggested in recent years that in the light of growing concerns the players become increasingly communicative about concussion31 which may lead to higher reporting rates and earlier detection.
Period of data collection
As the studies often express the results of long data collection periods of up to 16 seasons4 as a mean incidence, such data does not allow for any interpretation of effects associated with changes in diagnostic procedures, rules, equipment or other factors such as modification in training over time. In addition, other variables such as changes in team composition, for example, in terms of previous concussions of individual players, cannot be traced and evaluated.
Calculating and expressing incidence rates
When looking at the reported incidences of concussions, the inclusion criteria with regard to the diagnosis of a concussion have to be considered. The fact that nearly half the studies did not report any diagnostic criteria severely limits the interpretation of results. If, for example, such criteria were limited to the presence of amnesia or unconsciousness, then an underestimation of the true incidence may be expected. All this renders any comparison and interpretation of percentage-based results problematic.
As with reporting concussions as a percentage of overall injuries, there was also heterogeneity when reporting the incidences per 1000 AE, PH or GH. AE refers to an athlete participating in one game or practice, in which he is exposed to the possibility of sports-related injury while game and PH represent the time spent in game or practice only.19 Such mean values do not allow for individual player evaluation since ice time is quite inconsistent. While concussions rarely occur during practice,4 ,19 a pooling of AE and GH results is not possible.
In addition, some authors2 ,13 ,16 assumed that six players were on the ice during games at any one time and based their calculations on this number. Consequently, the game injury rate may become inflated as a result. Others based the calculated game exposures on 125 ,20 or all players who participated in some form each game,19 for example, including the back-up goalkeepers as they were also at risk for injury. It may be argued that the latter two concepts allow more accurate incidence calculations.
However, the incidence rates for concussions in relation to AE were generally similar between the studies over time and regardless of nation or competitive level. While it is possible that the incidence may have remained at similar levels regardless of rule changes or improvements in equipment over the last few decades, the number of confounding factors does not allow conclusions on this matter.
With regard to GH, the incidence was also generally similar ranging from 1.4/100024 to 1.8/100022 between 1996 and 2004 seasons in both Europe and North America. The rather high incidence of 6.5/1000 reported by Tegner and Lorentzon5 is an exception. One explanation may be that all team physicians in the studied league reached a consensus on the diagnostic criteria for concussion prior to the study. The first author later reported a similar incidence using the same concussion definition over the course of 17 seasons, indicating that this high incidence may reflect a closer estimation of the true incidence.32
Changes in rules and equipment
It would be interesting to follow the incidence of concussion over the course of the last decades and look for correlations with advances in protective equipment and change in rules. Helmets are mandatory since 1979 in the National Hockey League (NHL)7 after being introduced earlier in European countries such as Sweden.24 Although helmets reduce the risks of focal head injury, there is a limited ability to limit the effect of forces that are delivered through jaw and face or to reduce head rotation.33
There is no difference in the risk of sustaining concussions regardless of facial protection,34 although the use of full facial protection may lead to a decrease in concussion severity and allow for a faster return to play.35 However, it appeared that there had been an increase in head-checking particularly in ice hockey over the years.7 Consequently, in 2002 the IIHF introduced rule changes (Rule 540) which aimed to eliminate any head-checks by increasing the penalty for any contact to the head. Unfortunately, there are no studies allowing a comparison of the concussion incidence before and after the introduction of Rule 540 in European hockey.
Beginning with season 2005–2006, the NHL rules were changed to prevent “obstruction” in all forms. As the number of minor penalties and resulting power-plays increased, the number of concussions appeared to decrease significantly,23 probably as a result of the less-physical play during man-disadvantaged situations. Up to this time, the incidence of concussions showed little variation.22 This conclusion was based solely on news releases with all potential limitations associated and the collection period of other studies did not include these seasons.
Surprisingly, no study covered injury rates in countries of the former Soviet Union. Various external and confounding factors render any direct comparison of North American and European hockey very difficult. The more physical ice hockey style in North America,17 or higher player aggression36 in combination with the smaller rink size37 may contribute to the apparently lower incidence of concussions in European ice hockey (see online supplementary table S1). However, such differences may also be associated with methodological factors and warrant further investigation.
The initiatives outlined at the recent concussion summits9 ,10 ,28 show that the need to react to the alarmingly high incidence of concussion is well understood at least by the medical staff and the proposed initiatives give valuable directions. Furthermore, the introduction of assessment tools such as the SCAT-2 form will aid to establish a standardisation of diagnostic procedures and return-to-play guidelines.
The included studies indicate a high incidence of concussion in professional and collegiate ice hockey. Improvements in protective equipment may reduce the severity of concussions and allow a faster return-to-play. However, due to the number of potential confounding factors there is no conclusive evidence that rule changes or other measures lead to a decrease in the actual incidence of concussions over time. Where reported, the heterogeneous diagnostic procedures and criteria prevented a pooling of data. This review supports the need for standardisation of the diagnostic criteria and reporting protocols for concussion to allow interstudy comparisons in the future.
There are numerous limitations regarding the interpretation of concussion-related epidemiological studies, most of which have been discussed already.
Also, very few studies collected individual equipment exposure information or other potential confounding variables such as physical characteristics, experience of players and medical personnel as well as the coaching or game strategy. This also appears to be a relevant limitation when pooling of results across different ice hockey teams.
What this study adds
Despite all efforts to the contrary, the incidence of concussion remains alarmingly high in both collegiate and professional ice hockey.
There is no conclusive evidence that the incidence of concussion reduced over the years.
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Files in this Data Supplement:
- Data supplement 1 - Online table
Contributors AR drafted the manuscript and performed the statistical analysis. AG and WK helped with the design of this review and drafting the manuscript. All authors read and approved the final manuscript.
Competing interests None.
Provenance and peer review Not commissioned; externally peer reviewed.
▸ References to this paper are available online at http://bjsm.bmj.com