Portal de Periódicos da CAPES

Mechanisms of Injury as a Diagnostic Predictor of Sport-Related Concussion Severity in Football, Basketball, and Soccer

2016; Lippincott Williams & Wilkins; Volume: 63; Issue: Supplement 1 Linguagem: Inglês

10.1227/neu.0000000000001280

1524-4040

Scott L. Zuckerman, Douglas J. Totten, Kolin E. Rubel, Andrew W. Kuhn, Aaron M. Yengo‐Kahn, Gary S. Solomon,

Trauma and Emergency Care Studies

Sport-related concussion (SRC) has become an increasingly common injury in the United States. Approximately 1.6 to 3.8 million SRCs are reported annually,1 and the rate of recreation-related concussion and SRC has steadily increased from 2001 to 2012.2 Efforts have been made to curb increasing rates of SRC by studying mechanisms of injury.3-27 Video analysis has emerged as a common method to identify SRC mechanisms in professional ice hockey,10-12 soccer,7,28 baseball,8 rugby,9 and cricket.17 However, studies using video footage for nonprofessional sports are relatively sparse.5,6,22 Mechanisms of SRC in collegiate and high school athletics have relied primarily on self-report or team report and clinician documentation.3,4,22,29,30 If the exact mechanism of injury can be identified, athletes, coaches, and sports medicine professionals may be better equipped to target safety interventions to decrease the risk of SRC. Currently, there is no gold-standard diagnostic test to determine the severity of SRC,31 which presents an inherent challenge to those managing athletes with suspected concussive injury. Risk factors, including symptom severity, history of concussion, preinjury mood disorders, and family history of mood disorders, have been shown to predict a prolonged recovery after SRC.32-34 Should mechanism of injury affect concussion severity and outcome, these data may help sports medicine professionals to better individualize treatment plans for athletes after an SRC. Through a single-institution, regional sports concussion registry of middle school, high school, and collegiate athletes, we sought to further examine mechanisms of SRC and to elucidate potential applications of these data. The objectives of the present study were 2-fold: to provide a descriptive analysis of mechanisms of SRC in football, basketball, and soccer and to determine whether mechanism of injury was associated with symptom duration. For the second objective, because of the limited published empirical data available,22 we retained the null hypothesis that mechanism of injury would not affect SRC severity. METHODS A retrospective cohort study was conducted through in-depth patient interviews from the Vanderbilt Sports Concussion Center (VSCC) outcomes registry database. The VSCC outcomes registry database was launched in 2013, and efforts remain ongoing. Included in this database are athletes across greater middle Tennessee who presented to the VSCC for treatment. This study was approved by our Institutional Review Board (No. 130897). Inclusion and Exclusion Criteria All patients seen at the VSCC from January 2012 to May 2015 were compiled. During this time period, a total of 941 patients were seen by 5 interdisciplinary providers from the fields of adult and pediatric neurosurgery, neuropsychology, and primary care sports medicine. As noted in Figure 1, of the 941 patients treated, 699 sustained SRC and were eligible to be contacted. Complete data were collected for 295 patients (42%). The population studied included middle school, high school, and collegiate athletes playing sports with the highest reported incidence of SRC at the high school and collegiate level: football, soccer, and basketball.3,4 Athletes were subsequently excluded if they were not playing at these levels of competition, did not play these sports, or had a history of brain surgery, seizures, or meningitis. After these specific study exclusions were applied to the 295 patients with complete data, our final cohort included 180 individual athletes.FIGURE 1: Flowchart of the final cohort.Data Collection In-depth phone interviews were conducted to retrospectively collect patient and concussion information. Each interview was standardized and followed the same structured format. All patients and families were contacted at a minimum of 3 months after the injury to allow for symptom resolution. In addition to the interview with the concussed athlete, a parent or guardian was interviewed for each patient < 18 years of age or when available to provide additional information because most patients were adolescents or teenagers, and not all SRC information may have been readily known as a result of loss of consciousness or amnesia surrounding the event. Depending on availability, multiple contacts were required for some patients to obtain complete data. If the patient was still symptomatic during the interview, additional contacts were made until symptoms had resolved or 1 year had transpired after injury. Video analysis was not available for any SRC event. Several data points were collected for each patient, including demographics, medical history, family history, acute injury characteristics, and symptom duration. The institutional electronic medical record was used to cross-validate already gathered data. Any discrepancies between patient interviews and the electronic medical record were resolved with additional patient contacts. All data were maintained in a deidentified database. For the present study, extracted demographic, family, and past medical information included age, sex, race, body mass index, number of prior concussions, and history of multiple comorbidities such as attention deficit disorder/learning disorders, depression or anxiety, psychiatric disorders, or any positive history of such in any parent, sibling, or child. Concussion-related event data included mechanisms of injury (described below). Clinical outcome data included duration of symptoms and achievement of asymptomatic status. Mechanism of Injury Description Mechanisms were defined based on a review of the prior literature and our own clinical experience at the VSCC.3-27 It was found that prior literature described concussion mechanisms in 2 ways: what initiated contact with the concussed player6,7,9,10,22 and the sporting activity the player was engaged in during the time of injury.3,5,6,28,35 More often, the sporting activity was less descriptive than the source of contact. For instance, some studies mentioned "general play" instead of a specific sporting action, and data were often missing for this category.3 Some studies also evaluated the awareness of an oncoming collision or the "defenselessness" of a player using video analysis and a validated scale.6,15 Using these prior studies to shape our study design, we decided to stratify SRC mechanism on the basis of 3 components: a contact mechanism, a player mechanism, and an awareness mechanism. For purposes of our search, we operationally defined contact mechanism as what initiated contact with the athlete's head, or in other words, the source of the collision. The player mechanism explained the sport-specific action the concussed athlete was engaged in during the concussion, and the awareness mechanism signified whether the athlete was aware of the oncoming collision. Of note, a minority of studies described what body part of the concussed player incurred the impact,22,26 which was most often the head.6,22,26 Thus, for this study, it was assumed the concussed player's head was the principal point of contact, and this was not included as a mechanism. We conducted a pilot phase in which mechanism data were collected for 20 patients. On the basis of an analysis of the pilot data, we further refined our categories for specificity and relevance, as has been described in similar works.26 Patients were recontacted as needed to collect any necessary additional information. Statistical Analyses For all descriptive analyses, continuous variables were reported as mean (± SD) and binary count variables with number (percentage). For each sport, the occurrence of mechanisms within each contact, player, and awareness domain was compared proportionally via 1-way χ2 analyses for goodness of fit. Symptom duration data were represented with Kaplan-Meier survival plots.36 Starting time 0 was defined as time of injury. The event of interest was symptom resolution. Analysis time was thus defined as days to symptom resolution. The study period for each patient was 365 days. Patients who did not achieve asymptomatic status by 365 days were censored by design. These censored patients were included in the analysis but did not achieve the outcome of becoming asymptomatic. Because symptom duration was the time-variable outcome, these data were not assumed to be normally distributed and were presented with median and ranges. In addition to Kaplan-Meier plots, Cox proportional hazards models were run to estimate the relationship between our exposure (concussion mechanism) and the time-variable outcome (days of symptom duration). The hazard function was defined as the instantaneous risk of becoming asymptomatic at time (t) in the subset still with symptoms. As a result of the number of covariates associated with duration of symptoms after SRC, a propensity score analysis was used to avoid an unstable model.37 The propensity score was defined as the probability that the individual would report a certain mechanism on the basis of the following preinjury covariates: age, sex, race, body mass index, number of prior concussions, and history of multiple comorbidities, including attention deficit disorder/learning disorders, depression or anxiety, psychiatric disorders, or a family history of any of the listed comorbidities. Rather than running a single Cox regression model for each group of mechanisms, we ran a separate model comparing each mechanism with the remaining group. For example, for any of the categorical mechanism variables, instead of a single model with a reference group and indicator variables comparing each mechanism with the reference, we chose to run separate models to compare each mechanism with the collective group. For each Cox regression model, all hazard ratios with 95% confidence intervals not containing 1.0 were considered statistically significant. All statistical analysis was performed in STATA version 14 (StataCorp LP, College Station, Texas). RESULTS Demographics Demographic data are summarized in Table 1. The majority of athletes were male (70%), and football was the most represented sport (51%). Football players had the highest body mass index (25.3 ± 4.6 kg/m2) compared with basketball (21.7 ± 3.6 kg/m2) and soccer (21.6 ± 2.9 kg/m2) players. The majority of athletes had not sustained a previous concussion (69%), and migraine was the most prevalent comorbidity (18%). On average, the athletes were 15.9 ± 2.0 years old. High school athletes made up nearly three-quarters of the cohort (74%), followed by middle school athletes (17%) and collegiate athletes (9%). One-third of SRCs occurred during practice rather than an organized game. The median duration of SRC symptoms in all 3 sports was 21 days (3 weeks) with a wide range. A total of 4 patients (2%) did not reach asymptomatic status by 365 days.TABLE 1: DemographicsDescriptive A descriptive analysis of concussion mechanisms can be found in Table 2. Within the contact mechanism domain, contact with the ground or surrounding equipment was the most prevalent mechanism in basketball (40%) and was less commonly reported in football (17%) and soccer (22%). The most common contact mechanisms for each sport were helmet-to-helmet contact in football (74%), ground/equipment contact in basketball (40%), and player (nonhead)-to-head contact in soccer (33%). Head contact with the ball in soccer caused an SRC in 25% of players, and 9 of these 11 (82%) were female. The only sport in which differences in contact mechanisms reached statistical significance was football, in which helmet-to-helmet contact was significantly overrepresented (χ2 = 69.267, df = 2, P < .001).TABLE 2: MechanismsWithin the player mechanism domain, football mechanisms were relatively equal, with tackling (33%) and blocking (32%) the most common causes of SRC. In basketball, rebounding (45%) and defense/loose ball (33%) made up the majority of SRCs. In soccer, challenging a player and fighting for a loose ball (44%) and the act of heading (not necessarily with ball-to-head contact; 40%) were more commonly reported as the cause of SRC than offensive ball carrying (16%). The observed proportional difference for player mechanisms in soccer was statistically significant (χ2 = 6.533, df = 2, P = .04). Within the awareness mechanism domain, football players were almost evenly distributed, with 51% reporting they were aware of the oncoming collision. However, in basketball (χ2 = 8.805, df = 1, P = .003) and soccer (χ2 = 11.255, df = 1, P < .001), athletes were significantly more likely to be unaware of the oncoming collision leading to SRC. Regression To further explore each mechanism, several hazard regression models were calculated (Table 3-5). Furthermore, Kaplan-Meier survival curves show the time to asymptomatic status across all 3 mechanism domains in Figures 2 through 4. On the basis of the Cox regression models and controlling for several covariates with propensity scoring mentioned above, neither the contact mechanism nor the player mechanism revealed any significant associations with symptom duration. With respect to the awareness mechanism, it appeared that for soccer, being unaware of the oncoming collision was associated with a 2.54-times increased risk of not achieving asymptomatic status. Of note, the 12 soccer players aware of the oncoming collision were all female.TABLE 3: Cox Regression Models and Hazard Ratios for Contact MechanismTABLE 4: Cox Regression Models and Hazard Ratios for Player MechanismTABLE 5: Cox Regression Models and Hazard Ratios for Awareness MechanismFIGURE 2: Contact mechanism Kaplan-Meier graphs for (A) football, (B) basketball, and (C) soccer players.FIGURE 3: Player mechanism Kaplan-Meier graphs for (A) football, (B) basketball, and (C) soccer players.FIGURE 4: Awareness mechanism Kaplan-Meier graphs for (A) football, (B) basketball, and (C) soccer players.DISCUSSION There are many potential modifiers of SRC outcomes, including age, sex, and concussion history. One relatively understudied modifier is the mechanism of how these injuries occur. The purpose of this study was to describe the mechanisms of SRC among middle school, high school, and collegiate athletes across 3 major sports: football, basketball, and soccer. From the descriptive analysis, several predominant mechanisms emerged. Football concussions most commonly involved helmet-to-helmet contact with evenly distributed player mechanism among ball carrying/catching, blocking, tackling, and kickoffs/punts. Basketball did not have a specific predilection for contact or player mechanism, although elbow-to-head was a frequent contact mechanism. Among soccer players, player-to-player contact while challenging or attempting to gain possession of a loose ball and during the act of heading (not necessarily with ball-to-head contact) resulted in concussion most often. No specific mechanism predicted symptom duration except in soccer, in which lack of awareness of an oncoming collision resulted in significantly longer symptom duration. A majority of prior mechanism studies have used video analysis to describe mechanisms, mostly at the professional level7,9-14,16-18,28 and less commonly in collegiate and high school athletics.5,6,22 Although the study of professional sports is important for elite athletes, results may not be generalizable to collegiate, high school, and youth athletes. Furthermore, studies using video analysis are restricted to game concussions and do not include injuries sustained during practice. The absence of practice concussions is noteworthy in that a recent, large-scale epidemiological study of collegiate athletes across 25 sports reported that 47% of concussions occurred during practice.3 In our study, one-third of all concussions occurred during practice. Overall, our results highlight several worthwhile trends with regard to SRC mechanisms and safety. Descriptive Results Football Contact Among contact mechanisms, the majority of football SRCs involved helmet-to-helmet collisions, a statistically significant finding (P < .001). Intuitively, the helmet is the hardest object worn by a player and can cause severe injury.38 Contact with the ground was next most common, followed by contact with another body part of the opposing player. Our data support the findings of Kontos et al,39 who examined 468 youth football players and found that across 11 338 athletic exposures, head-to-head contact accounted for 45% of all SRCs sustained. Although professional football findings may not be generalizable to younger athletes, a video analysis study of National Football League concussions from 1996 to 2001 reported that 107 of 174 athletes collided with another player's head (61%), followed by player nonhead (22%) and the ground (16%).16 Player With respect to player mechanism, concussions were evenly distributed between the offensive plays of catching and running, tackling, blocking, and kickoffs/punts. In a large surveillance study of 396 high school and collegiate concussions from 2005 to 2006, the majority of concussions occurred during running plays (55%) followed by passing plays (16%) and kickoffs/punts (8%).4 A larger study of only collegiate athletes from 2009 to 2010 to 2013 to 2014 reported that more concussions occurred in athletes while initiating a block (20%) or tackle (20%) rather than being tackled (14%) or being blocked (12%). To improve concussion safety in football, authors have suggested that education is needed to instill proper "heads-up" tackling techniques to reduce the risk of SRC.40 Some have debated whether tackling in youth football should even be delayed.41 Although helmet-to-helmet contacts seemed to be the most prominent contact mechanism, several different player mechanisms commonly resulted in SRC. Basketball Contact Neither contact nor player mechanisms yielded statistically significant results among basketball players, although most players were unaware of the ensuing collision (P = .003). In our cohort, contact with the ground or surrounding equipment was most responsible for an SRC. Basketball is often played on a hard surface within a confined, walled gymnasium. The metal supporting structure of the hoop represents another possible site of impact. In addition, elbows were a prominent point of contact, a finding not observed in most other sports. We were unable to find studies mentioning contact mechanisms in basketball, specifically the incidence of elbow-to-head concussions. Player Rebounding (45%) and defense/loose ball (33%) were the most common player mechanisms observed, followed by offensive shooting/driving (23%). Our results support previous findings among high school and collegiate basketball athletes from 2005 to 2006 in which rebounding was the most common player mechanism in men's basketball (31%) compared with defending (22%) in women's basketball.4 The offensive moves of shooting (13% in men's basketball, 16% in women's basketball) and ball handling (10% men's basketball, 19% women's basketball) were second to defense and rebounding but more prominent in women's basketball. In a more recent collegiate study over 4 seasons, rebounding and defending were most common in both men's and women's basketball, followed by general play without mention of other mechanisms.3 Additional studies are needed, but it appears that defense and rebounding more commonly lead to SRC than offensive play, although sex differences may exist. Soccer Contact With respect to contact mechanism, sources of contact were evenly distributed among ground/equipment, head, player nonhead, and the ball, similar to an earlier single-institution study of male and female soccer players.22 In a large surveillance study of US high school soccer players, contact with another player was responsible for 78% and 62% of SRCs in men's and women's soccer, respectively.30 After player contact, the second most common contact mechanism in our study was ball contact (25%), and 9 of 11 ball contact events occurred in female athletes. In a recent study of collegiate athletes, 11% of women's concussions occurred as a result of ball contact as opposed to 0% in men's soccer.3 Although not directly generalizable to youth or high school soccer, a study at the professional level (20 Fédération Internationale de Football Association tournaments) also showed that women's concussions occurred more often as a result of ball contact than men's.28 Of note, the study included Under-17 and Under-20 World Championship tournaments, which may be applicable to high school and collegiate athletes. Overall, although player contact is the most common source of SRC, it appears that ball contact may be more of a source of SRC among women than men, indicating that future study of sex differences is needed. Player Within player mechanisms, challenging or attempting to gain possession of a loose ball was most commonly implicated (44%), followed closely by the act of heading (40%). In contrast, most studies have demonstrated heading as the most common activity involved in concussion at the high school,4,30 collegiate,3 and professional levels.7 In a study of high school and collegiate athletes, 41% of men's and 35% of women's concussions occurred during the act of heading.4 In addition, a retrospective analysis of surveillance data analyzing US high school soccer between 2005 to 2006 and 2013 to 2014 demonstrated that heading was the most common soccer-specific activity in men's (31%) and women's (25%) concussions.30 Heading has generated significant interest and debate. Amid litigation, the US Soccer Federation recently banned the act of heading for all players ≤ 10 years of age.42 Regression Results A subset of athletes can experience prolonged symptoms after SRC, referred to as postconcussion syndrome.43,44 Previously published risk factors for postconcussion syndrome in athletes include initial symptom score,33,34 loss of consciousness,33 prior concussion,32 and a personal or family history of migraine, learning disability, or psychiatric illness.32 SRC mechanisms represent a potential predictor of symptom duration after SRC. If coaches and sports medicine professionals know that a player is at increased risk of an SRC or protracted recovery if the injury occurs during a specific action, it may be foreseeable that safety interventions could be designed to reduce these specific occurrences. A notable example occurred in 2011, when National Football League owners voted 26 to 6 to move up the kickoff line from the 30- to the 35-yard line.45 In the following season, the number of kickoff concussions decreased from 35 to 20.45 On the basis of prior literature,22 the hypothesis retained in the present analysis was that mechanisms would not be associated with symptom duration. The only study to directly address this question analyzed SRC at a large Canadian university in 226 athletes across 3 sports: football (105), hockey (51), and soccer (70).22 The authors prospectively collected mechanism data through video analysis in addition to interviews with the injured players, teammates, and sideline medical personnel. Concussion severity was based on days to return to sport. Although the majority of the analysis focused on sex and position differences, no differences were found in the amount of time lost depending on sport played, position, where on the head or body the concussed athlete was hit, or whether loss of consciousness occurred. No statistical analysis, confidence intervals, or P values were reported for outcomes based on these mechanisms, only equivocal results. With a robust statistical analysis, our study garnered similar findings. When stratified by sport and contact or player mechanism, neither the player nor the contact mechanism was associated with symptom duration. However, the awareness mechanism yielded significant results. In male and female soccer players, being unaware of the oncoming collision was associated with a longer duration of symptoms. Although symptom duration was not assessed as an outcome, Lincoln and coauthors6 used videotape to analyze high school men's lacrosse games and found that 60% of concussed players did not anticipate contact and were rendered defenseless. The findings from the present study may suggest that those who are aware of an ensuing hit can brace their cervical musculature to absorb the impact, thus minimizing the force transmitted to the brain and skull. An earlier study found that among men's and women's high school soccer, basketball, and lacrosse players, the risk of concussion was reduced by 5% for every pound of increased neck strength.46 Further study is needed to empirically demonstrate a relationship between activation of the cervical musculature and the severity of concussion. Limitations The present study is not without limitation. Our results are subject to the biases of a retrospective analysis. Recall bias is certainly present in the ability of athletes and their guardians to recall specific SRC information months and sometimes years after their injury, which is why collateral information was obtained for nearly every patient. The capture rate of 42% was relatively low. Another limitation was the absence of video analysis, as has been used in prior studies of concussion mechanisms. Furthermore, sports were unable to be divided between sex or level of competition (eg, middle school, high school, collegiate) because of limited numbers. Although a propensity score analysis was used to minimize the number of covariates in each Cox regression model, our study may have been underpowered to detect differences in outcome. Furthermore, multiple models were used to compare each mechanism with the remaining group rather than with a reference group. Our study also represents trends in 1 region of the country, and sporting activities may not reflect regional trends found elsewhere. CONCLUSION The present study analyzed mechanisms of SRC via a regional SRC outcomes registry. In middle school, high school, and collegiate athletes, (1) a helmet-to-helmet collision was proportionally the most common contact mechanism of SRC in football; (2) ground and surrounding equipment, in addition to player elbows, caused the majority of basketball concussions, although this was not significant; (3) challenging a player, fighting for a loose ball, and heading (not necessarily with ball contact) were proportionally the most common mechanisms of SRC in soccer; and (4) awareness of an oncoming collision in soccer was the only mechanism that decreased the risk of prolonged symptom recovery. Understanding the mechanisms of SRC unique to specific sports may help league officials, coaches, and sports medicine personnel introduce new rule changes and educate athletes on how to mitigate potential risk factors and habitually learn proper sporting techniques. Disclosures Dr Solomon receives consulting fees from the Tennessee Titans (National Football League), the Nashville Predators (National Hockey League), and the athletic departments of several universities. In addition, he is a member of the ImPACT Scientific Advisory Board and receives reimbursement for expenses to board meetings, and he receives royalties from book sales. The authors have no personal, financial, or institutional interest in any of the drugs, materials, or devices described in this article.