What Is Injury in Ice Hockey: An Integrative Literature Review on Injury Rates, Injury Definition, and Athlete Exposure in Men’s Elite Ice Hockey

Injuries in men’s elite ice hockey have been studied over the past 40 years, however, there is a lack of consensus on definitions of both injury and athlete exposure. These inconsistencies compromise the reliability and comparability of the research. While many individual studies report injury rates in ice hockey, we are not aware of any literature reviews that have evaluated the definitions of injury and athlete exposure in men’s elite ice hockey. The purpose of this integrative review was to investigate the literature on hockey musculoskeletal injury to determine injury rates and synthesize information about the definitions of injury and athlete exposure. Injury rates varied from 13.8/1000 game athlete exposures to 121/1000 athlete exposures as measured by player-game hours. The majority of variability between studies is explained by differences in the definitions of both injury and athlete exposure. We were unable to find a consensus injury definition in elite ice hockey. In addition, we were unable to observe a consistent athlete exposure metric. We recommend that a consistent injury definition be adopted to evaluate injury risk in elite ice hockey. We recommend that injuries should be defined by a strict list that includes facial lacerations, dental injuries, and fractures. We also recommend that athlete exposure should be quantified using player-game hours.


Introduction
Ice hockey is a high intensity sport where players can reach speeds of up to 48 kph [1]. These speeds, and the nature of collision sports lead to musculoskeletal injuries at all levels of ice hockey [1][2][3]. There is a need to accurately quantify injury rates in men's elite ice hockey both for assessing player risk [4] and the associated economic burden [5]. Injury rates in ice hockey have been investigated in order to assess injury trends, injury types, injury location, and underlying injury mechanisms [6]. Injury rates can also be used to quantify the effects of rule changes [7]. Accurate data is needed in order to better investigate areas of concern while objectifying the effects of rule changes and other preventative measures [8,9].
Differences in the definitions for injury and athlete exposure (AE) lead to inconsistencies between studies, and obscure the resulting injury rates. Consensus statements on injury definitions and data collection procedures have been developed for soccer [10] and rugby [11], but have not been developed for ice hockey. Consistent definitions and methods to evaluate ice hockey injuries are required [12] to improve the comparability of published data [8]. Our objective was to review global musculoskeletal injury rates in men's elite ice hockey, as well as definitions of injury and athlete exposure. We focused our review on males as females have different types and rates of injury than males [13]. We focused on elite players aged 16 years and older playing junior hockey (United States Hockey League, North American Hockey League, Canadian Hockey League), US and Canadian College Hockey (NCAA Div. 1 and Div. III, Canadian Inter-University Sport), international or minor professional and professional hockey (Finnish Elite League, Swedish Elite League, Japanese Elite League, International Ice Hockey and the National Hockey League) as this cohort has not been as extensively studied as other levels such as high school and youth hockey [14,15]. In addition, the economic burden of injury at this level is high. During two seasons in the National Hockey League (NHL), injuries represented a total salary cost of US $218 million per year. While salary losses represent a significant financial burden, it is hoped that improved injury surveillance will reduce these costs.

Materials and Methods
We conducted an integrative literature review [16] to evaluate musculoskeletal injury rates, injury definition and athlete exposure measurement in elite ice hockey. We formulated three research questions a priori to focus our review: What is the rate of musculoskeletal injuries in men's elite ice hockey? In elite ice hockey, what injury definition is best suited to enable direct comparisons among research studies? Finally, in elite ice hockey, what measure of athlete exposure is best suited to achieve consistent and comparable injury rates?

Literature Search
A PubMed search strategy was created with the assistance of a University research librarian. PubMed was chosen as a search engine as it is the optimal tool in life sciences and biomedicine [17]. The search strategy used the key words: hockey AND (injury OR injuries) AND (NHL OR national OR international OR world OR competitive OR professional OR elite OR high caliber OR high caliber OR collegiate OR university OR intercollegiate OR NCAA OR "National Collegiate Athletic Association"). In addition, the same search strategy was performed on SPORTDiscus. The PubMed and SPORTDiscus records of these references were pooled and screened based on established inclusion and exclusion criteria ( Table 1). Articles that were not relevant to our research questions were excluded. The references in the remaining papers were reviewed to identify additional relevant articles. All studies were reviewed by both authors for their relevance to the three research questions.
Original, peer-reviewed, English language research articles evaluating the injury rates in elite ice hockey were included. Articles were excluded if they were editorials, abstracts, books or excerpts from conference proceedings. Articles were excluded if they did not contain one of the following variables: injury definition, injury rate, athlete exposure, injury mechanism or injury location. Unpublished data was not reviewed. Outcomes other than injury definition, injury rate, and athlete exposure, mechanisms and anatomical location The primary outcomes of interest are injury definition, injury rates, mechanisms and anatomical location.

Results
The PubMed and SPORTDiscus search identified 2463 references. An additional 3 pertinent articles were identified from the references for these articles. A total of 2212 articles were vetted after 254 duplicate articles were removed. Two-thousand, one-hundred and eighty-four of these articles were excluded as they were not relevant to any of our three research questions. No relevant articles were published prior to 1975. Accordingly, a total of 28 articles were included ( Figure 1).

Figure 1.
Flowchart describing the process for selecting relevant studies. The top row represents the identification process. The second and third rows represent the screening process. The fourth row represents the eligibility of the articles assessed and the last row identifies the articles included.

Rate of Musculoskeletal Injuries in Men's Elite Ice Hockey (Question #1)
Injury rate data, and study design characteristics are presented for each of the 28 studies in Table  2. Injury rates in competitive ice hockey range from 13.8 to 121/1000 player-game hours, depending on factors such as the league of play and exposure estimate. Professional players in Europe and North America experience musculoskeletal injury rates between 49 to 80/1000 AE as measured in playergame hours [4,18] while the collegiate hockey players in Canada and the United States experience lower rates (13.8 to 19.95/1000 AE) as measured in player games [19,20]. The highest injury rates are experienced at the junior level (39.8 to 121/1000 player-game hours) [21,22]. The majority of these musculoskeletal injuries are attributed to collision with other players, the boards or the hockey puck [18,20,23,24].

Rate of Musculoskeletal Injuries in Men's Elite Ice Hockey (Question #1)
Injury rate data, and study design characteristics are presented for each of the 28 studies in Table 2. Injury rates in competitive ice hockey range from 13.8 to 121/1000 player-game hours, depending on factors such as the league of play and exposure estimate. Professional players in Europe and North America experience musculoskeletal injury rates between 49 to 80/1000 AE as measured in player-game hours [4,18] while the collegiate hockey players in Canada and the United States experience lower rates (13.8 to 19.95/1000 AE) as measured in player games [19,20]. The highest injury rates are experienced at the junior level (39.8 to 121/1000 player-game hours) [21,22]. The majority of these musculoskeletal injuries are attributed to collision with other players, the boards or the hockey puck [18,20,23,24].       The injury rates in practice are much lower than games. Practice rates range between 1.4/1000 player-practice hours for Swedish Elite hockey [30] to 3.9/1000 player-practice hours for junior hockey [21], versus game injury rates of 74.3/1000 player-game hours [38] and 121/1000 player-game hours [34], respectively. Although the injury rates are lower for practices, the number of hours spent in practices is several-fold greater than games, so the actual number of injuries is higher than indicated by the injury rate.
There was a large variance in injury rates between studies. This large variance is a function of variability in the definitions for both injury and athlete exposure. As noted in previous papers, establishing consistent definitions of injury and athlete exposure are important first steps for objectifying injury risks in high caliber ice hockey [10,45].

Injury Definition in Men's Elite Ice Hockey (Question #2)
Probably the most important methodological factor affecting injury rate calculations is the definition of what constitutes an injury [45]. A review investigating the methods of data collection on injury surveillance identified three categories of injury definitions [45]. Category 1 defines injuries as all complaints regardless of time loss. All injuries are recorded, regardless of the severity or amount of time lost from competition. Category 2 defines injuries as events that require medical attention by a member of the medical staff. Therefore, according to this definition, a member of the medical staff, typically a team therapist or team doctor, must diagnose the injury. Finally, category 3 defines injuries as events that have a time loss element. Accordingly, an injury is only recorded if the athlete misses a team-related practice or competition. Individual studies typically fit into one, or more of these categories.
Our review identified 28 studies evaluating injuries in elite ice hockey. Early research investigating injury rates in the Swedish Elite League, and the Swedish National team used the time loss definition of injury (Category 3). As shown in Table 2, the majority of ice hockey injuries studies use either a time loss (Category 3) or medical attention definition (Category 2). None of the articles evaluating injuries in elite ice hockey used the all complaints definition (Category 1).
Our review found inconsistent definitions of a reportable injury in ice hockey research based on the time loss definition. In addition, the list of injuries has expanded over time. Facial lacerations were considered reportable injuries in 1991 [24], while sutures, fractures, dislocations and subluxations were added in 1992 [23]. Concussions, dental and eye injuries were added in subsequent years [20,21], potentially increasing injury rates by expanding the list of injuries. In addition, illness may be counted as an injury, inflating the injury rates [23].
The definition of injury based on medical attention (Category 2) has also been used to quantify competitive ice hockey injury rates [31,34,38]. However, this metric is often combined with the time loss component to result in a broader interpretation of injuries [20][21][22][23]35,42]. For example, injuries such as concussions, dental injuries, lacerations and eye injuries are captured with medical attention by a team physician or athletic trainer, resulting in a more extensive list of ice hockey related injuries compared to definitions that did not include these injuries [45]. Of note, some studies have expanded their list to include illnesses and psychological complaints that are unrelated to injury [46]. The time loss definition (Category 3) is the easiest to use as it is easy to track time loss. However, it leads to the fewest reported incidents [45] as it fails to capture the athletes that continue to train and play while injured [47]. Depending on the time of year, some injuries may be under reported as injured players continue to play throughout key time periods, such as playoffs. The medical attention definition (Category 2), though broader and encompassing a greater number of conditions, also has limitations. The subjective interpretation of what constitutes medical attention may lead to systemic bias [48], and the types of injuries managed by the various practitioners may differ based on their qualifications and status [45].

Athlete Exposure Metric in Men's Elite Ice Hockey (Question #3)
Athlete exposure is the second component of injury rate. An athlete exposure is defined as one athlete participating in a practice or game in which there is a potential for athletic injury [49]. Injury rates are typically based on 1000 athlete exposures. These exposure rates can be quantified as injuries per 1000 game-hours (or injuries per 1000 games), injuries per 1000 practice-hours, or overall injuries per 1000 AEs (games and practices combined). Injury per 1000 player-game hours is based on a 60-min active game and is calculated as the number of injuries/number of players on the ice at the same time (6)/number of games × 1000. Many researchers use this method [18,21,24,30,31,33,34]. However, this exposure estimate is not used consistently among researchers. For example, several studies accounted for both teams when calculating athlete exposure (number of injuries/number of players on ice at the same time (two teams)/number of games × 1000 [22,42]. In contrast, another study used a 20 person roster, including the back-up goaltender, to calculate athlete exposure per 1000 player-game hours [38]. This larger number of players will lead to a smaller injury rate.
Our review identified different nomenclatures pertaining to the athlete exposure metric, such as player-games and player-game hours [42]. The number of athletes used to quantify these exposure rates vary between studies, and are not consistently defined. For example, one researcher [42] calculated player-game injury rates based on 22 players competing for each team in a game (i.e., 44 players) while another [30] calculated player-game hours injury rates based on 6 players. This was based on the number of players on the ice at a time, and whether goaltenders were included. Other researchers have used roster averages over a set period of time [36,37], or a tournament [22,42] to calculate player-game injury rates.
Injury per 1000 games is the average number of injuries that one player experiences per 1000 games (number of injuries/total number of players (roster)/number of games × 1000 [20,37]. Our review found different implementations of this approach as there was some research that counted both rosters when computing athlete exposure [42]. This has an effect on total estimated exposures and can lead to reduced injury rates. Finally, several articles did not fully describe whether they included both rosters or a single team roster when calculating athlete exposures [19,23], making it difficult to determine accurate injury rates.
In addition, we investigated the impact of calculating injury rate based on the actual time on ice (TOI) [4,50]. Using the actual time on ice, injury rate was calculated as the number of injury events/sum of individual AE time as found on the player statistics page (www.nhl.com/stats/player). The time on ice was calculated based on the number of minutes and seconds that each individual played per game over the season. The difference between estimated athlete exposure (number of injuries/number of teams (30)/number of players on roster each game (19)/number of games (82)) and the TOI metric was large. As much as three times the amount of exposure was identified by estimating exposure rates. However, when comparing the time on ice metric to the estimated player game-hour metric, the differences were minimal. The player game-hour exposure (based on one hour per game rather than the actual amount of time that players spent on ice, which changes due to overtime periods and penalties) is similar to the time on ice calculations (14,676.2 h calculated as the sum of players' time on ice versus 14,760 h calculated as 30 teams × 82 games × 6 players) [4].
Our review found that practice athlete exposure was calculated consistently in most studies. Injury per 1000 practice hours (number of injuries/number of practice hours/number of players on team × 1000) was the standard [21,30,33,34].

Discussion
Injury rates in men's elite ice hockey are higher in professional leagues such as the Swedish Elite League [31] and Finnish National League [33] than college hockey [19,20,23]. This may be due to the differing demands as professional players play more games in a season, and therefore may experience more overuse injuries. It may also be due to the athlete exposure estimation (player-game hours vs. player-games) used to calculate injury rate. Style of play and hockey rink dimensions are additional variables that may influence injury rate. Overall, we observed the trend that injury rates have increased over time in professional European leagues [35] and college hockey [39], while decreasing in men's international ice hockey [42].
We observed a wide range of injury definitions. This affects both the reliability and comparability of injury surveillance research. There is currently a consensus-based injury definition in sports such as soccer and rugby [10,11]; however, there is no consensus injury definition in ice hockey. We recommend that hockey forms a consensus injury definition as this will resolve an important issue that currently impedes hockey injury research. A consistent injury definition would create clarity as to which injury is considered a recordable event. We identified the International Ice Hockey Federation's (IIHF) definition of injury as the most appropriate as it only captures events that are sufficiently severe that they influence participation in practices or games. The IIHF's definition describes a reportable event as "any injury sustained in a practice or game that prevented the player from returning to the same practice or game; any injury sustained in a practice or game that caused the player to miss a subsequent practice or game; a laceration which required medical attention; all dental injuries; all concussions; all fractures" [42]. Although no single definition suits all needs, the time loss definition is the most common and easy to identify. It is considered reliable and allows for the comparison of data between teams, seasons and various leagues [45]. It is also used in other professional sports such as cricket and Australian football [51,52]. The choice of definition should reflect the aims and goals of surveillance. With its consistency, ease of use, and comparability of published data [8] among the most important variables, we feel the time-loss definition best meets the needs of injury surveillance in men's elite ice hockey. However, like all definitions there are limitations in choosing this metric. First, athletes often continue to compete in the presence of injury. Delaying treatment may lead to missed injuries. Finally, the threshold for time loss may depend on the time of season and how important the player is to team success [45]. Despite these drawbacks, we feel the strengths of the time-loss definition outweigh its limitations and that the IIHF's time-loss definition is warranted in elite men's ice hockey.
We also noted that athlete exposure estimations were inconsistent in the literature. The major confusion lies in how many participants are included in the injury rate calculation. Several researchers used player-game exposure based on the entire team, or average team roster (19 players) [20,36,37], while others used player-game hour exposures based on 6 players [18,21,24,30,34]. This leads to difficulty in interpreting injury rates and comparing research. It was proposed that the gold standard in athlete exposure during games is time on ice. As much as three times the amount of exposure was accounted for by estimating exposure rates using the player-game approximation compared to time on ice. However, when comparing the time on ice metric to the estimated player game-hour metric (based on one hour per game, rather than the actual amount of time that players spent on ice) it appears that this difference is small [4]. Therefore, the simplest and easiest way to calculate athlete exposure is to use six players on the ice (player-game hours) unless position specific injury rate information is warranted. Using a consistent athlete exposure metric will increase intra-and inter-league injury rate reliability.
The majority of studies reviewed found that collision with other players is the leading mechanism of injury as well as contact with the boards, opponent's hockey sticks and hockey pucks [22,35,36]. This leads to an injury paradox: the goal of the sports performance specialist is to build bigger, faster, stronger, leaner, more powerful, robust players. However, these types of players also travel faster, and hit harder, elevating the risk of injury. This situation emphasizes the need for accurate injury surveillance methods as these may help reinforce rules and/or govern the addition of new rules enforcing safety for active players.

Limitations
There are limitations to this study. There is a relative paucity of studies evaluating injury rates in men's elite ice hockey, and the definitions of injury and athlete exposures vary between studies. Accordingly, the reported injury rates differ between studies and are difficult to interpret. Two databases (PubMed and SPORTDiscus) were used to identify research papers that were relevant to injury definition, injury rates and athlete exposure in elite ice hockey. While these databases are an excellent source for research articles in sports, life sciences and biomedicine, supplemental databases may have identified additional research studies.

Conclusions
In summary, this project represents the first integrative literature review investigating injury rates, injury definition and AE in men's elite ice hockey. The greatest opportunities for continued improvement lie in both consistency and comparability to refine, improve and streamline calculations of injury rate.
At the current moment, a uniform definition of injury is the most important step to better objectify injury data in ice hockey. A universal definition is required by sport governing bodies and researchers. Though each approach has its limitations, in order to compare exposure rates in both the intra-and inter-league, a workable, consistent definition is required. Specific responsibility should be given in terms of who will diagnose the injury if the definition is a time loss definition, a medical attention definition, or a combination. In addition, a detailed injury list is needed to clarify the definition of injury and whether specific injuries such as dental, concussions, and facial lacerations, are included.
Finally, disparate AE estimations diminish injury rates, which compromises research findings. Attendance rate in both practice and games (player-game hours based on 6 players per game and the full roster during practices) is the preferred method for calculating athlete exposure.

Further Research
Investigating anatomical areas prone to injury is crucial for team performance staff such as athletic therapists, physical therapists and strength and conditioning specialists as it may guide rehabilitation initiatives, performance program design and athlete monitoring [53]. We observed that the lower extremities was the most common site of musculoskeletal injury.
Future research should clearly define injury rate measurements to provide doctors, therapists, and coaches with accurate information to streamline return to play initiatives. In this regard, our review has exposed gaps including the disparate definition of injury and the lack of a consistent athlete exposure metric. Acknowledgments: The authors thank David Lesauvage, Library Assistant, University of Western Ontario, Canada, for his contribution in refining a comprehensive search strategy for our review. The authors declare they have no competing interest. The study complied with the laws of the country of the authors' affiliation.

Conflicts of Interest:
The authors declare no conflict of interest.