INTRODUCTION
The National Hockey League (NHL) is the professional league for ice hockey in North America.1 Ice hockey demonstrates a high risk of injury with overall incidence rates ranging from 5.93-15.6 per 1000 athlete exposures.1,2 This high injury rate is related to the level of physicality required for the sport. Players skate in excess of 32-48km per hour3,4 with contact occurring via body checking the most common mechanism of injury.1,5–7 These high rates of injury come at a cost to teams in the league. Fifty point nine percent of NHL players missed at least one game resulting in a total of $218 million dollars in lost salaries (based on individual player salary and number of days missed due to time loss injuries) per season between 2009-2010 to 2011-2012 seasons.8 Due to the injury and financial burden to players and teams, a comprehensive, transparent approach to injury surveillance is needed to attempt to mitigate injury risk. Further, the time trend analysis of injury patterns is an essential step in this process for both risk factor identification and evaluation of modulators, such as equipment changes9 or body checking.4
A systematic approach in injury prevention research has been suggested by van Mechelen et al. to determine injury incidence and burden to inform future studies on preventative measures.10 Data among NHL players for injury surveillance are often reported by a singular body region,11,12 or do not encompass the whole league making comparisons by body region and position challenging.13 Additionally, given the risk of repeated high impact contact, players are at risk for subsequent injuries, further impacting injury burden.5,12,14–16 Players who sustain an injury are at risk of sustaining a second or third injury, particularly to the shoulder,12 groin,15 or specific to a diagnosis of concussion.14 One study analyzed league wide injury rates across multiple body regions in the NHL from 2006-2007 to 2011-2012 seasons. Authors of this study demonstrated an incidence rate of 15.6 per 1000 athlete exposures.1 Since then, rule changes on illegal checks17 and equipment updates have occurred.18 During the 2010-2011 NHL season, lateral or blind side hits meant to target the head, known as an ‘illegal check to the head’ resulted in a two minute minor or a match penalty18; furthermore, visor wear was mandated for those entering the league in 2013. This warrants an updated epidemiological injury profile that reflects temporal trends to improve injury mitigation programs and assess efficacy of league rule changes on injury.
One way to identify injury incidence and temporal trends among NHL players is through publicly available data.8,19–21 Publicly available data improves transparency, and allows for collaboration among organizations to advance data robustness and distribution among stakeholders.22,23 This transparency and potential for collaboration is essential to promote open science initiatives.24,25 An open science approach allows researchers to assess, reproduce, and conduct studies for independent research with data that is easily accessible.24,25 Publicly available data have been utilized across professional leagues and demonstrated high reporting reliability.21,26 However most public data utilization in the NHL has been for cost analysis of injury8 and influence of schedule density on injury,19 not injury surveillance. Publicly available data are accessed via a computer iterative repeatable process which is an efficient method of data extraction.12,27 This process increases repeatability, and offers the potential for shared league wide injury risk identification and injury mitigation programs.28 This study sought to use publicly available data to determine injury and illness incidence by position and body region and determine initial versus subsequent injury risk. A secondary purpose of this study was to determine temporal trends of injury and illness incidence by position and body region.
METHODS
Study Design
This was a retrospective cohort study. Participants were NHL players 18 years or older who competed in at least one season between 2007-2008 to 2018-2019. Players were categorized by position. Two online resources were used to create a combined data set for this study, including 1) https://www.prosportstransactions.com, and 2) http://www.nhl.com/stats/.20,21 These data have been previously utilized in Major League Baseball, the National Football League, and the National Basketball Association.20,21 The data can be accessed through the Open Science Framework data repository https://osf.io/rx4jb/. NHL stakeholders were included to aid in research question development and clinical interpretability including team physicians, athletic trainers, data analysts, and performance specialists. Strengthening the Reporting of Observational Studies in Epidemiology for Sport Injury and Illness Surveillance guideline was used.29
Injury and Illness Classification
Injuries and illnesses that occurred from the first game of the regular season to the last game of the post season were included in this study. Injury was defined as tissue damage or derangement of normal physical function occurring during any training session or competition that resulted in at least one day time loss.21,29 Illness was defined as a complaint or disorder reported by a player and his team, not related to injury, resulting in at least one day time loss.21 The authors defined injury based on a specific joint or body segment as defined by the Orchard Sports Injury Classification System.30 Concussions were highlighted separately from head/neck injuries, and represented the only distinct diagnosis highlighted separate from body region (i.e. head/neck). Some data points are presented as a crude descriptor (i.e. “upper body”) and were reported in this study as general classifications (upper body, lower body, or other), if injury to a joint or segment was not discernable. Initial injury was defined as the first injury documented; subsequent injuries were defined as either multiple or recurrent injuries, or an exacerbation of a previous injury.31
Athlete Exposure
Athlete exposure was calculated based on game exposure (AGE) only, as determining practice exposure was not possible with this data set. For the 2015-2019 seasons, AGEs were calculated based on all 32 NHL teams and a 23 active man roster (8 defensemen, 8 wing players, 4 centers, 3 goalies) playing 82 regular season games between 2007-2008 to 2011-2012, and 2013-2014 to 2018-2019 seasons. The 2012-2013 season was representative of a player strike; this the AGEs were calculated based on 32 NHL teams and a 23 active man roster (8 defensemen, 8 wing players, 4 centers, 3 goalies) playing 48 regular season games. For all seasons, a postseason exposure adjustment was included to account for post season injuries based on the number of playoff games that occurred each season, with a reduction in the number of active players as teams were eliminated.1
Data Extraction, Data Reduction and External Validation
For a detailed description of data repository used refer to Supplemental File 1. Data extraction, data reduction, and external validation used have been previously described.20,21,27 External validation demonstrated high reliability for injury reporting (98%). Refer to Supplemental File 2 for detailed methods.
Statistical Analyses
Injury and illness count data was converted to seasonal incidence proportion (IP) and incidence rate (IR). IP was calculated by total number of injuries or illness divided by total number of players per season.16 IR was calculated by sum of injuries and divided by the sum of player-games, multiplied by 1000 x Athlete-Game Exposures (AGEs). Initial and subsequent injury and illness risk ratios were calculated by specific position incidence for initial or subsequent injury, divided by all other position incidence for initial or subsequent injury for combined 13 seasons. Subsequent injury risk ratios were further stratified by number of subsequent injuries by position to explore injury burden. 95% confidence intervals (CIs) were reported for all IP and IR calculations. Temporal trends were reported. Sensitivity analyses were performed to calculate injury incidence for aggregated four season intervals to evaluate influence of seasonal outliers. All analyses were performed in R version 4.02 (R Core Team, 2020) using the rvest, tm, and xm12 packages.
RESULTS
Over thirteen seasons, 10,549 athletes participated in the NHL and 9,734 injuries and illnesses were recorded. Centers had the highest combined injury and illness incidence rate (15.14 per 1000 AGEs 95% CI:15.12-15.15) (Table 1).
Initial and Subsequent Injuries by Position
Centers presented with the highest incidence of multiple injuries reported for two (IP: 22.1 95% CI: 21.2-23.0) and three (IP: 9.0, 95% CI: 8.6-9.4) total injuries per season. (Table 2).
Centers were 1.4 (95% CI: 1.3-1.6) times more likely to sustain a subsequent injury, with a higher likelihood of sustaining multiple injuries, or new body region (RR: 1.9, 95% CI:1.8-2.1) versus an injury exacerbation or recurrent injury (RR: 1.4, 95% CI: 1.2-1.6) (Table 3). Refer to Supplemental file 3-5 for further information on initial versus subsequent injury and illness IP.
Incidence by Body Region and Position
The groin/hip/thigh was the most commonly injured body region with an IR of 1.14 per 1000 AGEs (95% CI: 1.06-1.21), followed by the head/neck (0.72 per 1000 AGEs, 95% CI: 0.66-0.78), and knee (0.60 per 1000 AGEs, 95% CI: 0.55-0.66) (Table 4). By position, centers demonstrated the highest IR among the top three most injured body regions (groin/hip/thigh: 1.41 per 1000 AGEs, 95% CI: 1.23-1.59; Head/Neck: 0.86 per 1000 AGEs, 95% CI: 0.72-1.00; knee: 0.80 per 1000 AGEs, 95% CI: 0.66-0.93) (Table 4).
Temporal Trends
Combined injury and illness IR peaked in 2009-2010 season at 12.01 (95% CI: 11.22-12.79) (Supplemental File 6). Among the most commonly injured body regions, the groin/hip/thigh demonstrated peak incidence during the 2007-2008 season (Overall: 2.53, 95% CI: 2.17-2.90) (Supplemental File 7) along with the knee (Overall: 1.46, 95% CI: 1.19-1.74) (Supplemental File 8); head/neck demonstrated a peak incidence in 2010-2011 season (Overall: 1.03, 95% CI: 0.81-1.26) (Supplemental File 9). All three body regions demonstrated a variable, though declining incidence over the study time frame. Injuries reported as ‘Lower Body’ demonstrated an increase in incidence over time peaking in 2015-2016 (Defense: 2.68, 95% CI: 2.09-3.28; Goalie: 1.45, 95% CI: 0.80-2.11) or 2016-2017 seasons (LW+RW: 2.31, 95% CI: 1.75-2.86, Center: 2.57, 95% CI: 1.78-3.35) (Supplemental File 12). For further breakdown of temporal trends, please reference Supplemental Files 6-13.
Sensitivity Analyses
Centers reported the greatest mean consecutive four-season incidence proportion for the groin/hip/thigh (13.09) (Supplemental Files 14, 17), knee (7.40) (Supplemental Files 14, 18), concussion (5.66) (Supplemental Files 14, 19) and shoulder/arm/elbow (5.43) (Supplemental Files 14, 20). The mean difference was similar between season and four-season analyses across all body parts for overall, and most commonly injured body regions (Supplemental Files 14, 15). For most commonly injured body region by position, the Groin/Hip/Thigh mean difference comparing season and four-season analyses were: defense (Season: 5.48 vs 5 Season: 5.93), LW+RW (9.37 vs 10.17), centers (12.07 vs 13.09), and goalies (4.31 vs 4.69) (Supplemental Files 14, 15). Further information on sensitivity analyses can be found in Supplemental Files 14- 22.
DISCUSSION
This study sought to establish injury incidence, initial versus subsequent injury incidence and determine temporal trends of injury incidence among NHL players. Centers had the highest combined injury and illness IR and the greatest incidence of subsequent injury. The groin/hip/thigh, knee, and head/neck were the most commonly injured body regions for all positions except LW+RW, which demonstrated the highest incidence for the groin/hip/thigh, head/neck, followed by equal injury incidence for the Shoulder/Arm/Elbow, and concussion. This suggests some position specific variability in injury may occur. Temporal trends for combined injury and illness incidence demonstrated the highest incidence during the 2009-2010 season. Head/neck injuries peaked in 2010-2011 season, prior to implementation of rule changes identifying illegal checks to the head. Groin/hip/thigh and knee injuries demonstrated declines over the study time frame, although ‘lower body’ injuries demonstrated an increase over time. Similar results were found in the sensitivity analyses using four-year increments for all temporal trends by body region and position suggesting four year or seasonal variance in injures may still capture the overall trends in injury data.
Initial and Subsequent Injuries by Position
Centers demonstrated the highest overall injury and illness IR and were 1.4 times more likely to sustain a subsequent injury within the same season. Research on injury by position is limited among NHL players and distribution can be conflicting depending on the league and level of play.32–34 A previous one-year prospective study among Swiss professional ice hockey players demonstrated near equal distribution of injuries by position for in-game injuries (i.e., right forwards: 23%, left forwards: 20%, centers: 17%, right defenders: 21%, left defenders: 15%). However this study did not include all teams in the league, and ice surface area is larger compared to the NHL (Swiss: 60 x 30 meters, NHL: 61.0 x 25.9 meters) potentially impacting injury incidence or severity.33 Similar to our findings, collegiate male ice hockey players have demonstrated that forwards account for 48.3-61.1% of injuries, but did not account for specific offensive positions.32,34 The increased incidence of initial and subsequent injury demonstrated among centers may be attributed to the demands of the position. Centers are expected to cover the largest zone of ice, may have increased defensive responsibilities compared to wing (LW+RW) players, and handle the puck more than other positions, all of which may influence contact with other players in open ice areas, or contribute to recurrent injuries. However, variability in coaching strategies for how centers are used are likely, and may not fully explain the higher injury incidence among centers.
Incidence by Body Region and Position
The groin/hip/thigh, knee and head/neck were the most commonly injured body regions for centers, defense, and goalies. This finding is similar to previous studies among collegiate8 and professional players.33 Swiss professional players reported the most common injury to the hip/groin/thigh (23%), followed by head (17%)33; whereas a recent study among collegiate male players demonstrated that the head or face was the second most injured body region (15.2%) followed by the hip or groin (12.1%).35 Groin and hip injuries have previously been shown to be evenly distributed across all players,34 although intra-articular injuries are more common in goalies.36 Hip/groin/thigh injuries in hockey can be acute or chronic in nature36 and may present as intra-articular36 or extra-articular issues.6 Additionally, hip/groin/thigh injuries that present as chronic in nature may be related to overuse mechanisms, and may be underreported.34 Among knee injuries, medial collateral ligament tears followed by anterior cruciate ligament tears are the most common knee injuries resulting in time loss via contact.4 Head/neck injuries often include facial injuries (4.1 per 1000 player games),7 as well as cervical spine injuries.35 The NHL requires players to wear half shields on helmets, compared to collegiate ice hockey full face shield requirements. Previous literature has suggested a lower incidence of facial injuries when wearing a full facial shield among collegiate players.37 The current data are similar to or lower than previous collegiate ice hockey reports comparing head/neck injuries (0.72 per 1000 AGEs versus 0.34-4.71 per 1000 Athlete Exposures).38 The current study did not differentiate among players who wore full versus half shield protection. Therefore, the impact of face shield type on head/neck injuries cannot be determined from our data and warrants further investigation. The high incidence demonstrated among hip/groin/thigh, knee, and head/neck injuries for a majority of players informs clinicians on need to consider injury mitigation programs for the lower extremity and cervical region, keeping in mind positional needs (i.e., goalie’s positional stance versus skaters).
One position subcategory, LW+RW, demonstrated differences among highest injured body regions after hip/groin/thigh. These players demonstrated the second highest injury incidence for the trunk/back/buttock, followed by equal injury incidence for the shoulder/arm/elbow, and concussion. Similar to our results, collegiate and Swiss professional players demonstrated the trunk or abdomen/thorax injuries accounted for 9-9.04% of all injuries and the fourth or fifth most commonly injured body region2,33,34; however these studies did not report specific incidence by individual positions. The shoulder is the most commonly injured joint in the upper extremity, oftentimes presenting with acromioclavicular sprains,12 glenohumeral instability,13 as well as minor injuries such as contusions4 which likely reflects the majority of injuries among the shoulder/arm/elbow reported in the current study. In addition to centers, the finding that LW+RW players demonstrated a higher incidence of concussion compared to defense or goalies is consistent with previous NHL literature.11,39 Offensive players sustain hits from defensemen that are on average larger in stature.39 Furthermore, previous research has demonstrated that location of concussion event is more evenly distributed across all zones compared to defensemen and goalies. Forwards also incurred more concussions when ‘on the rush’ when the player is skating at a higher speed.39 These variations of position stature, location, and nature of play may explain the current study findings of higher incidence of concussion among offensive players.
Temporal Trends
Head/neck IP peaked in 2010-2011 season, followed by a drop beginning in 2013-2014 seasons. From 2010-2011 season to 2013-2014 season, modifications were made to rules defining illegal checks to the head and subsequent penalties17 following research on concussion incidence and screening protocols. This rule change likely contributed to the decreased incidence in head and neck injuries noted in our results.14 Furthermore, the 2013-201418 season implemented a mandate requiring all players to wear a protective visor; players who wore a visor demonstrated more than a four-fold decreased risk of orbital or eye injuries.9 Visor wear increased from 32% during the 2002-2003 season9 to 97% during the 2018-2019 season,18 likely contributing to the decrease in injuries in the body region category of ‘head/neck’ injuries seen after the 2013-2014 season in our findings. However, these studies were not based on individual visor wear and the impact of injury should be interpreted with caution. Specific to concussion, incidence decreased over time after the highest peak noted during the 2011-2012 season likely in part to improved concussion recognition and stricter protocols.17,39 However, a spike in suspensions for illegal checks were demonstrated in 2013-2014 and 2018-2019 seasons which may in part contribute to the slight increase in overall or position specific (centers and RW+LW players) incidence, compared to the 2014-2015 to 2017-2018 seasons.25 Notably, the slight increase in overall or position specific incidence of concussion during the 2013-2014 may have been impacted by the NHL player strike, resulting in a condensed season, and should be considered. Impact of regulation on body checks and association with lower head/neck and concussion rates have been established among youth ice hockey players.39 These results suggest that rule implementation preventing body checks to the head and neck may also have a positive impact on lowering the incidence of head/neck and concussion injuries when rules are adequately followed and enforced in the NHL.
Among the most injured body region, the groin/hip/thigh and knee injuries demonstrated declines over the study time frame, although ambiguous labels including ‘Lower Body’ injuries demonstrated an increase over time. Therefore, the current study may reflect an underrepresentation of injury incidence across multiple body regions including the lower extremity. The NHL is required to disclose to the public a player’s injury status39; however, teams are not required to disclose the specific nature of the player’s injury, such as pathology.39 Many teams have opted to use the terms ‘lower body’ injury which may explain the increase in incidence of injury labeled ‘lower body’ over time compared to hip/groin/thigh.39,40 This decreased transparency in injury reporting is in contrast to other leagues such as the National Basketball Association, who are required to disclose the specific nature of the player’s injury.39 Furthermore, decreased transparency of publicly available injury data and hesitancy to disclose epidemiological data to researchers3 is likely reflected in the substantially less injury surveillance research among NHL players compared to other professional leagues.41 Decreased transparency in data reporting and access impacts volume and quality of studies that can improve player injury outcomes and appropriate resource allocation.3,41 Concussion was the specific diagnosis that the current study was able to report beyond the Orchard Injury Classification System body region labels. This increased transparency of reporting specific to this diagnosis likely aided in informing rule changes, injury recognition, and implementation of stricter protocols which likely contributed to the decreased incidence rate over time.17,40 This application of transparency may be applicable to other diagnoses of upper and lower body injuries for improved quality of injury surveillance data across the NHL to ensure appropriate injury mitigation programs and effective resource allocation implementation.
Limitations
Only NHL players were assessed, decreasing generalizability of the results to other professional ice hockey leagues, amateur ice hockey players, or female ice hockey players. Further, the public data set does not allow for missing data to be quantified, which may impact the precision of these results. However, external data validation was performed with other publicly available data to increase the interpretability of these results. Due to ambiguous terms such as ‘lower body’ the injury incidence by specific body region may be underreported or misclassified.40 Furthermore, these ambiguous terms may impact the ability to differentiate by body region for subsequent injuries (i.e., initial injury reported as ‘knee injury’ if reinjured could be reported as a ‘lower body injury’). This data was not reported by a clinician or other medical practitioner, which may also influence misclassification. Additionally, an estimated athlete-game exposures (AGEs) value was used to calculate IR based on typical games played each season. Although this approach has been previously performed to estimate player-game exposures,1,20,21,42 athletes by position may have different exposure to sport, and residual confounding is possible, impacting the clinical interpretability of these findings. In future studies examining effects of policy or intervention effects, AGEs may be suitable to use as a denominator, though exposure that captures both practice and game exposures may be more applicable to discern individual treatment effects. Furthermore, research varies in how athlete exposure is measured (i.e. 1000 athlete exposures versus 10000 athlete game), impacting comparability of the results. Finally, some injuries were reported to the nearest anatomical body part or nonspecific labels such as ‘lower body’ with specific injury classification (e.g., knee injury versus lateral meniscal tear of the right knee) not possible, decreasing the clinical interpretability of these findings.
CONCLUSION
Centers demonstrated the highest overall combined injury and illness incidence, were more likely to report multiple injuries, and more likely to report a subsequent injury compared other positions. The groin/hip/thigh, knee, and head/neck were the most commonly injured body regions for centers, defense, and goalies similar to previous literature; however, LW+RW players demonstrated higher incidence for groin/hip/thigh, trunk/back/buttock, shoulder/arm/elbow, and concussion. Head/neck injuries peaked in the 2010-2011 season but demonstrated a substantial decrease in the following seasons. These findings inform clinicians on specific injury incidence that may reflect positional variability, which may inform injury mitigation programs that address these position specific variations. Further, the current findings provide initial insight into injury trends that occur during periods of rule changes for illegal checking and visor wear for those who entered the league after 2013. Future literature is needed to investigate effects of rule or equipment changes to further confirm their effectiveness. Clinicians should be aware that although injury incidence decreased for groin/hip/thigh and knee, ‘lower body’ injuries demonstrated an increase over time, indicating that these injuries are likely underreported, citing a need for greater transparency of reporting injury.
Conflicts of interest
The authors report no conflicts of interest
Funding
None to disclose.