In 1999, the National Institutes of Health Consensus Development Panel declared the incidence of mild traumatic brain injury (mTBI) had reached epidemic proportions and concluded that reducing incidence, severity, and postinjury symptomology should be a national priority (37). Four years later, in response to the increasing number of diagnosed cases, the young age of the at-risk population, and the possibility of long-term disability from repetitive injury, the National Center for Injury Prevention and Control declared mTBI occurring in sports, which is commonly diagnosed as concussion, an important public health problem that required an increase in research, treatment, and prevention efforts (8). Since that time, sports-related concussion has become a prominently discussed topic in academic, public, and government forums because of an ever-growing body of evidence that concussion history may lead to a higher likelihood of developing mild cognitive impairment, clinical depression, and early onset of Alzheimer disease (23,24). It has even been hypothesized that the repetitive subconcussive head impacts sustained in contact sports (e.g., football, hockey, and lacrosse) (3,10–12), even in the absence of diagnosed concussion, may potentially lead to the deleterious effects of chronic traumatic encephalopathy, a degenerative brain disease with clinical presentation similar to amyotrophic lateral sclerosis and Alzheimer disease (18).
Developing strategies for preventing concussion has been challenging primarily because of the difficulty in determining the causal relationship between head kinematics and injury. Traditionally, laboratory reconstruction techniques using surrogates (i.e., cadaveric specimens, animal models, and anthropomorphic test devices) have been used to replicate the human response to impact (21,27,38). Three primary limitations exist when trying to relate measures of head kinematics obtained from laboratory impacts to those experienced by athletes who are diagnosed with concussion in sports: 1) sports-related concussion is typically diagnosed by signs of neurological or neuropsychological dysfunction and self-reported symptomology, which cannot be easily deduced from surrogates; 2) surrogate tests do not account for the complex system of intrinsic and extrinsic variables (e.g., contact force and direction, player physiology at time of impact, equipment condition, and player anticipation) that influence kinematic response to impact; and 3) single-impact events created in the laboratory may be an insufficient injury model considering impact and/or injury history may modulate an athlete’s tolerance to impact. These variables, which may vary widely for a representative cohort, are impractical to obtain and difficult to replicate using traditional laboratory reconstruction.
To overcome these limitations, sporting fields, particularly those for contact sports, have been identified as living laboratories to explore the human response to impact because athletes routinely sustain head contact during play (35). One of the most impact-rich sporting environments is American-style football. In this sport, concussion is the third most common game injury with nearly 5% of all players diagnosed with concussion per season and 15% of those injured athletes diagnosed with multiple concussions in the same season (13,26). To leverage the large sample size, high frequency of impacts, and high potential for injury, Head Impact Telemetry (HIT) technology (Simbex, Lebanon, NH) was developed to record head impact exposure (frequency, location, and kinematics of head impact) sustained during play (9,15). Using this technology, Duma et al. (15) first reported the aggregate head impact exposure of 38 collegiate football players over a single season, one of whom sustained a concussion. Since that time, the same general methodology has been used in numerous studies to quantify the kinematics of head impact across several athletic populations (5,10–12,33,41).
Although the pathomechanics of head impact exposure leading to concussion diagnosis still remains unclear, several studies have provided preliminary insight into the relationship between on-field measures of head impact and diagnosed injury. Guskiewicz et al. first postulated also (25), and recently concluded also (17), that it may be difficult to identify a threshold for concussion after observing wide variation in head acceleration recorded before 13 cases of diagnosed concussion in collegiate football players. Greenwald et al. (20) observed similar variation in head acceleration when examining 17 impacts associated with diagnosed concussion; however, they demonstrated that injury predictions based on measures of head impact exposure could be improved by combining impact location, peak head acceleration, and impact duration into a single, independent metric through the use of principal component analysis. Similarly, Broglio et al. (5) found that measures of head acceleration before diagnosed concussion were similar between high school and collegiate football players, and a combination of linear acceleration, rotational acceleration, and impact location best differentiated 13 impacts associated with injury from head impacts not associated with injury. Alternatively, Rowson et al. (39,40) suggested that the lack of specificity of these measures to injury was due, in part, to underreporting and was able to develop injury risk curves using estimates of injury prevalence and both linear and rotational acceleration distributions for impacts associated with and without injury. They went on to show that risk curves based on simple biomechanical parameters have a high level of power for predicting concussions when compared with epidemiological data (16). Although insightful, one significant limitation of these initial studies is the relatively small sample of recorded head impacts associated with diagnosed concussion used in their respective analyses.
To overcome sample size limitations, we have collectively pooled the injury cases recorded with helmets instrumented with HIT System technology across multiple institutions and studies (5,20,25,41). With this much larger retrospective data set, we aim to elucidate the biomechanical basis of mTBI in American-style football. The purpose of this study is to compare two components of head impact exposure (frequency and kinematic response) on days with and without diagnosed concussion and to identify the sensitivity and specificity of single-impact severity measures to diagnosed injury. Specifically, we tested the hypotheses that individual players will sustain more impacts and impacts of higher severity on days of diagnosed concussion than on days without diagnosed concussion. Through this analysis, we begin the process of identifying components of head impact exposure that best correlate with diagnosed injury, which provides quantitative data from which protective equipment and safety standards can be developed and could potentially lead to both rule changes to mitigate at-risk behavior and new methods for identifying impaired athletes who, using current gold-standard methods, go undiagnosed.
PARTICIPANTS AND METHODS
For a 6-yr period (2005–2010), 1208 players from eight collegiate football teams and six high school football teams wore instrumented helmets (HIT System, Simbex) to measure head impacts during practices, games, and scrimmages—designated as team sessions. Yearly participation by each organization, as well as inclusion for each individual within the organization, was voluntary, with no consideration given to a player’s previous history of concussion or playing position. Because instrumentation was only available for Riddell helmets (Riddell, Inc., Chicago, IL), participation was limited to subjects already wearing VSR-4 (24%), Revolution (72%), or Speed (4%) helmet models. A total of 673 players were instrumented during multiple years, providing a yearly subject pool of 230 players from six teams in 2005, 330 and 550 players from 11 teams in 2006 and 2007, and 422, 426, and 352 players from eight teams in years 2008–2010. At all institutions participating in the research, approval for data collection and reduction was received by an institutional review board and informed consent was obtained, including parental consent in the case of minors.
Instrumented helmets were used to continuously monitor the head during all competitive activity and record head acceleration in real time after impact. The HIT System is composed of an in-helmet data acquisition system, a sideline transceiver, and a laptop computer (10,15). The in-helmet unit positions six single-axis accelerometers (Analog Devices, Norwood, MA) against a player’s head, providing isolated head acceleration measures (28). Before use, all helmet model and size (M, L, and XL) combinations were tested to meet on-field use requirements, which included meeting standards set by the National Operating Committee for Standards of Athletic Equipment. During play, when any accelerometer exceeded a 14.4g threshold, 40 ms of data were recorded (8 pretrigger and 32 posttrigger, 10 bit, 1000 Hz per channel), time stamped, and transmitted wirelessly (903–927 MHz) to a transceiver and laptop computer positioned on the sideline. Communication range typically exceeds 200 yd; however, in the event of poor communication, each unit was capable of storing up to 100 impacts in nonvolatile memory to minimize the potential for data loss. To verify the accuracy of on-field data collection, processing, and reduction, a multiphase validation process was conducted, which included laboratory testing (2,9,15,28), video correlation of on-field events (7,15,34), and multisite field trials (5,10–12,15,33,41).
During the period of study, concussion was generally defined as an alteration in mental status, as reported or observed by the player or team’s medical staff, resulting from a blow to the head, which may or may not have involved loss of consciousness. For all cases of injury, a certified athletic trainer (ATC) or team physician at each respective institution diagnosed and treated the injury at their professional discretion. After symptom resolution, the medical staff provided the date of injury, the suspected time of injury, the approximate time of diagnosis, the day of symptom resolution, and the player anthropometrics (age, height, and weight). In addition, anecdotal descriptions of the events surrounding injury (e.g., description of the impact, method of identifying the injury, and on-field observations regarding clinical presentation) were provided by each team when available.
Impact location and linear and rotational acceleration of the head center of gravity were computed for each impact from acceleration data collected with the instrumented helmets (9,40). Events recorded outside of an organized team session (practice, scrimmage, or game) or with peak linear acceleration below 10 g were removed before analysis, because these were considered to be outside of the measurement range of the device (10,33,36). Other identified non–head impact-related events, such as throwing a helmet, were also removed from the data set.
From the processed acceleration data, measures of impact kinematics available for analysis included the peak magnitude of linear and rotational acceleration and three additional metrics calculated from the linear acceleration time series data: Gadd severity index (GSI) (17), head injury criteria (HIC15) (43), and change in head velocity (Δv). In addition, the total number of daily head impacts sustained and the number of head impacts above the 50th and 95th percentile values of peak linear and rotational acceleration for all players were calculated. The percentile cutoff values for peak linear acceleration (50th = 20.5g, 95th = 62.7g) and peak angular acceleration (50th = 981 rad·s−2, 95th = 2975 rad·s−2) were previously reported by Crisco et al. (11) and Rowson et al. (40), respectively.
Only impacts sustained by athletes diagnosed with concussion during the period of study were considered for the purposes of this analysis (Fig. 1). For each kinematic measure, the median (50th percentile) and 95th percentile levels for these individual players on days with and without diagnosed concussion were calculated. Results were expressed as median values and 25%–75% interquartile range. A Wilcoxon signed-rank test for matched pairs was used to test the significance between the 50th and 95th individual player percentiles on days with and without diagnosed injury because the study variables were not normally distributed (Lilliefors test, P < 0.001). Similarly, distributions of impact frequency were skewed toward lower occurrence (Lilliefors test, P < 0.001), so the same method was used to test the hypothesis that a greater number of impacts per player occur on days of diagnosed concussion than those without diagnosis. This analysis was performed using all impacts as well as only with impacts greater than the 50th and 95th percentile acceleration levels to determine whether differences exist even when only considering the highest magnitude impacts.
Receiver operating characteristic (ROC) curves were generated to evaluate the sensitivity and specificity of single-impact severity measures to diagnosis of concussion. Impacts recorded immediately before a player was removed from participation and diagnosed with concussion were defined as immediate diagnosis impacts and used as positive ROC input. Negative ROC input cases were designated as all impacts for the concussed players occurring on days without diagnosis of concussion. For each ROC curve, the null hypothesis of the true area under the curve (AUC) equaling 0.5 (same as guessing) was tested, and an asymptotic significance value (P value) is reported. Hanley’s method for comparing area under ROC curves was used to test if any of single-impact severity measures were more sensitive to diagnosed concussion than peak linear acceleration.
Binary logistic regression was conducted to determine the odds ratios for concussion risk relative to incremental increases of each impact severity metric. This method determines how much the potential for diagnosed injury increases based on the measured severity and the presence or absence of clinically defined injury after impact. Again, immediate diagnosis impacts were used as positive input into the analysis, and all impacts sustained by concussed athletes on days without diagnosis was used as negative input. Results of this analysis include the regression coefficients (α, β), SE of the regression coefficient, the Wald statistic used to test the significance of each regression coefficient, the odds ratio, and the 95% confidence interval of the odds ratio.
All statistical analyses described above were performed with custom MATLAB scripts (version 7.11; The MathWorks Inc., Natick, MA) in combination with built-in statistical toolbox functions. A significance level of α = 0.05 was set a priori for each of the statistical tests.
A total of 161,732 head impacts were recorded over 10,972 player days from 95 athletes clinically diagnosed with mTBI (Fig. 1). Eight of the subjects sustained two diagnosed concussions and one had three, yielding 105 identified cases of injury. The median reported age, height, and weight of all concussed athletes was 19.2 ± 2.2 yr (15–23 yr), 183.5 ± 6.7 cm (165.1 ± 198.1 cm), and 94.6 ± 16.3 kg (63.5–138.8 kg), respectively. Collegiate athletes accounted for 68 of the diagnosed injuries with the remaining 37 sustained by high school players. Seventy of the cases (66.6%) occurred during games or scrimmages with the remainder occurring during practices. The time of symptom resolution was reported for 89 of the 105 cases, and of these, symptoms resolved in a mean of 5.9 ± 7.4 d (range, 15 min to 59 d) from the reported time of injury.
Kinematic measures for head impacts sustained on days with diagnosed concussion were higher than on days without diagnosed concussion (Table 1). Statistical significance was observed for both the 50th and 95th levels for all kinematic measures except 50th percentile rotational acceleration (P = 0.08, Table 1). On days when injury occurred, athletes also sustained a greater number of head impacts than on days when no injury was diagnosed (Fig. 2). The difference was found to be significantly different when considering all impacts as well as those with peak linear acceleration greater than the 50th and 95th percentile of all impacts (P < 0.001, Table 2).
In 45 of the injury cases (43%), the player did not continue playing after an impact that directly preceded diagnosis of concussion. In the other 60 cases, the player was not immediately removed from play, and the diagnosis did not occur until either later that day or in the following days because either the signs and symptoms of injury were not immediately recognizable or the player did not self-report. These 60 cases of delayed diagnosis were excluded from both ROC and logistic regression analysis because of the potentially confounding factor of sustaining additional head impacts after onset of symptoms.
Impacts sustained before immediately diagnosed concussions had mean severity of 112.1g ± 35.4g of peak linear acceleration, 4253 ± 2287 rad·s−2 of peak rotational acceleration, 321.5 ± 239.4 HIC15, 439.3 ± 315.2 GSI, and 4.29 ± 1.71 m·s−1 of change in velocity. The area under the ROC curves generated for each severity measure (0.921–0.983) were statistically higher than 0.5 (P < 0.001), indicating that all measures of severity are better than guessing the outcome of diagnosed concussion (Fig. 3). Peak linear acceleration and HIC15 were most sensitive to immediately diagnosed concussion (AUC = 0.983), but this was not significantly different than either GSI (AUC = 0.982) or change in head velocity (AUC = 0.980). The only severity metric significantly different from peak linear acceleration was peak rotational acceleration, which had a lower sensitivity to immediately diagnosed concussion (AUC = 0.921, P = 0.019).
The odds ratios and associated 95% confidence intervals provided in Table 3 indicate the increase in odds of sustaining a diagnosed concussion for a single unit measure increase of each severity metric. For example, a 1g increase in linear acceleration corresponds to a 1.052 greater odds of sustaining an immediately diagnosed concussion, or more practically, a player has 10.3 times greater odds of sustaining an immediately diagnosed concussion after a mean 95th percentile impact (84.9g) than a mean 50th percentile impact (38.9g) because the increase in odds equals the odds ratio raised to the power of change in a single unit measure (e.g., odds increase = odds ratio(top 95th − top 50th)).
On days of injury, 95 athletes with one or more diagnosed concussions sustained impacts with higher associated kinematic response than on noninjury days. Because many individual factors (e.g., style of play, playing position, and team tendencies) could influence susceptibility to injury, it is interesting to note that on days without injury, these players sustained head impacts typical for all football players. The 50th and 95th percentile peak linear (20.7g and 63.5g) and rotational (848 rad·s−2 and 2761 rad·s−2) accelerations recorded on noninjury days were nearly identical with those reported by Crisco et al. (20.5g and 62.7g) and Rowson et al. (981 and 2975 rad·s−2) who used similar methods to quantify the head impact exposure of collegiate football players who were not diagnosed with concussion from three collegiate football teams for 3 yr. This comparison is especially compelling considering the injured athletes evaluated in this study came from a larger range of seasons and a higher number of teams and is inclusive of both high school and collegiate players. Given the relatively large sample of injury cases presented in this analysis, it is clear that a significant distinction exists between kinematics sustained on days with concussion and other days of play. These in vivo measures of head acceleration represent a foundation of quantitative data that can be used to develop future protective equipment and test standards for that equipment. In addition, because head impact kinematics on days without concussion appear to be similar for all athletes (both those who were never diagnosed and those who were), the differences identified suggest that implementing a procedure to screen athletes for injury based on daily head impact exposure could lead to increased injury detection.
Similarly, and maybe less intuitive because injured athletes are commonly removed from competition, athletes sustained more head impacts on days with diagnosed injury than on days without diagnosis. Athletes also sustained more impacts above the 50th and 95th percentile levels of peak linear and rotational acceleration on days of diagnosed concussion, with the median number of highest severity impacts ranging between 2.0 and 2.85 times higher on days of injury. These data indicate that players not only experienced more head impacts on days of diagnosed concussion but also sustained more high-severity head impacts on these days. Previously, pilot studies have suggested a link between the number of head impacts sustained and in-season cognition (1,42); however, there has been little evidence suggesting that impact frequency is predictive of concussion. Although it is still unclear if multiple impacts predispose an athlete to injury (i.e., the high number of impacts lowers a player’s threshold of injury) or if the athlete simply has a higher risk of injury from a single event because of the higher number of impacts sustained, it is clear that the number of impacts a player sustains is a key measure to consider when evaluating the link between head impact and injury. This finding, combined with several previous studies showing impact frequency is related to several factors including team, playing position, skill level, and session type (6,10,11,33,41), suggests that injury mitigation strategies, such as rule changes to limit head contact, can be developed to reduce the occurrence of concussion in sports.
In a 2007 report, Schnebel et al. (41) provided a detailed description of two diagnosed cases of concussion and highlighted the difficulty of associating an injury diagnosis with a single impact. Issues that confound this association include multiple impacts occurring within a short period, symptoms that either resolve quickly or only become pronounced over time, and, most importantly, the reliance on a player’s self-report to initiate the medical evaluation. More than half of the diagnosed concussions reported in this study were not immediately identified by the team’s medical staff and, in the majority of cases, went undiagnosed until after play had ended. To mitigate uncertainty, we limited our analysis of injury risk to only those impacts sustained immediately before diagnosis of concussion as positive cases and all impacts for those players sustained on days without diagnosis of concussion as negative cases. By focusing solely on impact events with clearly discernible outcome, risk estimates presented within this study most likely underestimate risk of sustaining any concussion due to the exclusion of cases with delayed diagnosis. In addition, it has been estimated that up to 50% of all head injuries in football go undiagnosed (31), and preliminary findings have been presented that indicate a subset of athletes exists who experience in-season cognitive decline without experiencing abnormal symptomology (1). Because of this, the risk estimates presented here are defined as the risk for sustaining an immediately diagnosed concussion rather than the risk of sustaining any concussion or signs and symptoms of a concussion. Differences in head impact exposure between cases of immediate and delayed injury diagnosis were not evaluated directly; however, this will be the subject of future communications.
Although the risk estimates presented in this study are limited to those of immediately diagnosed concussion, it is still valuable to compare these results to historical studies of traumatic brain injury to evaluate commonly accepted theories. In the 1950–1960s, Gurdjian et al. (22) first observed a relationship between an impact event and clinical indicators of TBI through a series of tests conducted on anesthetized canines and human cadavers. From these experiments and supplementary data, a brain injury tolerance curve was created, known as the Wayne State Tolerance Curve (WSTC) that relates brain injury to linear head accelerationand duration (21,27). The WSTC is the basis for both GSI and HIC15, which are still used today in the development and standardization of head injury protective devices for both the automotive and helmeted sports industries. Although it has long been accepted that both peak acceleration and duration play a role in brain injury, it is interesting to note that, in our study, no statistical difference was found between peak linear acceleration and GSI, HIC15, or Δv when assessed as a predictor of those immediately diagnosed. One probable reason for this finding is that head impacts in football typically have very similar temporal characteristics (duration of 8.99 ± 3.01 ms) (6), and the associated injuries are less severe, making the data set described in this study more homogenous than the one used to develop the WSTC, which included injuries ranging between skull fracture and loss of consciousness after (linear) acceleration durations ranging between <0.001 and 0.60 ms (27). Although this does not discredit the role of impact duration to concussion in general, it does appear that the magnitude of linear acceleration without the inclusion of a temporal component is sufficient for differentiating impacts associated with concussion from those that are not when considering only head impacts sustained within a single helmeted sport.
Although many studies have shown that rotational acceleration is the most likely cause of diffuse axonal injury, historical literature on this association is primarily derived from animal surrogates undergoing pure rotational acceleration (i.e., whiplash events) (19,29). These data have been supported for humans by simulating brain tissue deformation after impact with finite element brain models and associating resulting measures of strain with the input kinematics (44). The question still remains, however, if these surrogate data apply to the impact scenarios occurring in a sports environment. Although the mechanism of injury cannot be assumed, studies focusing on impacts sustained during football have shown that combining rotational acceleration with other impact measures such as linear acceleration and impact location increases the specificity of injury prediction (5,20,44); however, for this analysis, we chose to treat each impact measure independently for the purpose of developing single-measure ROC curves. Although the predictive capabilities of peak linear acceleration and measures derived from the linear acceleration resultant were not found to be statistically significant, peak rotational acceleration was found to be the least sensitive of all evaluated severity measures to immediately diagnosed concussion. At this time, it is unclear why a discrepancy exists between the evidence that rotational acceleration is the cause of brain injury, and yet, it is the least predictive measure of immediately diagnosed concussion. These results could indicate a discrepancy between the pathomechanics of injury previously explored in the laboratory and the injury being defined as a concussion in helmeted sports, (14) or that the association between individual kinematic parameters and injury could simply be masked by the correlated relationship of these parameters in a football head impact (40).
There are several potential limitations of this study. First, all concussions were diagnosed by a trained medical professional using their clinical judgment and best practice guidelines at the time of injury; however, it has been well established that concussion symptoms frequently go unreported (8,31). It has also been shown that some players may experience cognitive change without any perceived symptomology (1,30,42). Although we can be reasonably assured that athletes diagnosed with concussion sustained an injury, the converse cannot be assumed. By limiting our analysis to head impacts sustained by players with at least one sustained concussion, we limit the potential for underestimating concussion while maintaining a large “control” sample of impacts not associated with injury, thus providing what we believe to be a robust estimate for risk of diagnosed concussion. To overcome this limitation in the future, additional methods for screening athletes on a consistent basis throughout the season could be implemented to evaluate in-season clinical presentation (i.e., presence/absence of signs and symptoms typically associated with concussion) rather than focusing solely on clinical diagnosis. Second, data were not analyzed separately by subject demographic information (e.g., playing position, high school vs. college, and helmet type). Although it has been shown that demographic-specific trends for head impact exposure exist between and within athlete populations (3,6,15,33,41), it is important to note that frequency, location, and the kinematic response to head impact is highly dependent on an individual player (10–12). For example, collegiate football players tend to sustain more impacts over the course of a season and impacts resulting in higher head acceleration more frequently than high school players; however, the range of head impact exposure athletes experience is quite large, so it is quite common for individual high school players to sustain impacts at a frequency and acceleration level that is on par with collegians (4). Because sports-related concussion is a highly individualized and complex pathophysiological process (32), our initial focus was to determine the head impact exposure measures most associated with injury for individual players. This within-subject design provides a better understanding of the biomechanical variables most related to diagnosed concussion, independent of these extrinsic factors that may have contributed to the level of exposure each athlete experienced. Moving forward, the results from this study can be combined with typical head impact exposure profiles already established for nondiagnosed athletes to determine whether different conditions of participation place an athlete more at risk for injury. Through this approach, strategies for injury mitigation can be developed. Finally, the study design used was not epidemiological in nature and only tracked cases of diagnosed concussion for athletes while wearing instrumented helmets. Because of this, readers should be careful not to estimate concussion rates from the data presented or interpret concussion risk based on the occurrence of injury by specific demographic information alone.
To the authors’ knowledge, this work presents the largest collection of in vivo biomechanical head impact data associated with diagnosed concussion to date. The key findings of this initial communication indicate that players sustain both a greater number of impacts and impacts of higher severity on days of diagnosed concussion than on days without diagnosed concussion. In addition, kinematic measures associated with peak linear acceleration are similar predictors of immediately diagnosed concussion, whereas predictive capability of rotational acceleration is significantly lower. Although further analysis is required, the data introduced in this study provides a foundation for identifying the biomechanical basis of head injury from which future communications will build upon.
This manuscript is the first in a series of communications within Medicine & Science in Sports & Exercise by the collaborating authors investigating the biomechanical basis of mild traumatic brain injury through the use of in-vivo biomechanical data obtained from on-field head impact monitoring in sports.
We appreciate and acknowledge the researchers and institutions from which the data were collected, including Mike Goforth, ATC, Virginia Tech Sports Medicine; Dave Dieter, Edward Via Virginia College of Osteopathic Medicine; Russell Fiore, ATC, Brown University Sports Medicine; Bethany Wilcox, Brown University; Ron Gatlin, ATC, Casady HS Oklahoma City, OK; Jeff Frechette, ATC, and Scott Roy, ATC, Dartmouth College Sports Medicine; Dean Kleinschmidt, ATC, and Brian Lund, University of Indiana Sports Medicine; Jesse Townsend, ATC, Greensburg Salam HS, Greensburg, PA; Jeff Cienick, ATC, Blackhawk HS, Beaver Falls, PA; John Burnett, ATC, Karns City HS, Karns City, PA; Chris Ashton, MS, ATC, University of Minnesota Sports Medicine; Scott Hamilton, Unity HS, Tolono, IL; Scott Oliaro, Scott Trulock, and Doug Halverson; and UNC-Chapel Hill Sports Medicine.
In addition, we would like to especially thank Ann-Christine Duhaime, MD, Massachusetts General Hospital, and Arthur Maerlender, PhD, Dartmouth Medical School, for reviewing the manuscript; Lindley Brainard and Wendy Chamberlin, Simbex, for coordination of data collection from Dartmouth College, Brown University, and Virginia Tech; and Rema Raman, PhD, and Sonia Jain, PhD, University of California San Diego, for review of the statistical analysis.
This work was supported in part by award R01HD048638 and R01NS055020 from the National Institute of Health, R01CE001254 and 5R49CE000196 from the Centers for Disease Control and Prevention, and NOCSAE (07-04, 14-19). HIT System technology was developed in part under NIH R44HD40473 and research and development support from Riddell, Inc. (Chicago, IL).
Joseph J. Crisco, Richard M. Greenwald, Jeffrey J. Chu, Jonathan G. Beckwith, and Simbex have a financial interest in the instruments (HIT System, Sideline Response System [Riddell, Inc.]) that were used to collect the data reported in this study. The remaining authors have no financial interests associated with this study.
The authors acknowledge that publication of the results of the present study do not constitute endorsement by the American College of Sports Medicine.
1. Beckwith JG, Chu JJ, McAllister TW, et al.. Neurocognitive function and the severity of head impacts sustained in athletic competition. In: Eighth World Congress on Brain Injury
. Washington (DC): Brain Injury; 2010. p. 446.
2. Beckwith JG, Greenwald RM, Chu JJ. Measuring head kinematics in football: correlation between the head impact telemetry system and Hybrid III headform. Ann Biomed Eng
. 2012; 40 (1): 237–48.
3. Brainard LL, Beckwith JG, Chu JJ, et al.. Gender differences in head impacts sustained by collegiate ice hockey players. Med Sci Sports Exerc
. 2012; 44 (2): 297–304.
4. Broglio S, Surma T, Ashton-Miller J. High school and collegiate football athlete concussions: a biomechanical review. Ann Biomed Eng
. 2012; 40 (1): 37–46.
5. Broglio SP, Schnebel B, Sosnoff JJ, et al.. Biomechanical properties of concussions in high school football. Med Sci Sports Exerc
. 2010; 42 (11): 2064–71.
6. Broglio SP, Sosnoff JJ, Shin S, He X, Alcaraz C, Zimmerman J. Head impacts during high school football: a biomechanical assessment. J Athl Train
. 2009; 44 (4): 342–9.
7. Brolinson PG, Manoogian S, McNeely D, Goforth M, Greenwald RM, Duma SM. Analysis of linear head accelerations from collegiate football impacts. Curr Sports Med Rep
. 2006; 5 (1): 23–8.
8. Centers for Disease Control and Prevention (CDC). Report to Congress on Mild Traumatic Brain Injury in the United States: Steps to Prevent a Serious Public Health Problem
. Atlanta (GA): National Center for injury Prevention and Control; 2003. pp. 1–45.
9. Crisco JJ, Chu JJ, Greenwald RM. An algorithm for estimating acceleration magnitude and impact location using multiple nonorthogonal single-axis accelerometers. J Biomech Eng
. 2004; 126 (6): 849–54.
10. Crisco JJ, Fiore R, Beckwith JG, et al.. Frequency and location of head impact exposures in individual collegiate football players. J Athl Train
. 2010; 45 (6): 549–59.
11. Crisco JJ, Wilcox BJ, Beckwith JG, et al.. Head impact exposure in collegiate football players. J Biomech
. 2011; 44 (15): 2673–8.
12. Crisco JJ, Wilcox BJ, Machan JT, et al.. Magnitude of head impact exposures in individual collegiate football players. J Appl Biomech
. 2012; 28 (2): 174–83.
13. Dick R, Ferrara MS, Agel J, et al.. Descriptive epidemiology of collegiate men’s football injuries: National Collegiate Athletic Association Injury Surveillance System, 1988–1989 through 2003–2004. J Athl Train
. 2007; 42 (2): 221–33.
14. Duhaime A-C, Beckwith JG, Maerlender AC, et al.. Spectrum of acute clinical characteristics of diagnosed concussions in college athletes wearing instrumented helmets. J Neurosurg
. 2012; 117 (6): 1092–9.
15. Duma SM, Manoogian SJ, Bussone WR, et al.. Analysis of real-time head accelerations in collegiate football players. Clin J Sport Med
. 2005; 15 (1): 3–8.
16. Funk J, Rowson S, Daniel R, Duma S. Validation of concussion risk curves for collegiate football players derived from HITS data. Ann Biomed Eng
. 2012; 40 (1): 79–89.
17. Guskiewicz KM, Mihalik JP. Biomechanics of sport concussion: quest for the elusive injury threshold. Exerc Sport Sci Rev
. 2011; 39 (1): 4–11.
18. Gavett BE, Stern RA, McKee AC. Chronic traumatic encephalopathy: a potential late effect of sport-related concussive and subconcussive head trauma. Clin Sports Med
. 2011; 30 (1): 179–88, xi.
19. Gennarelli TA, Thibault LE, Adams JH, Graham DI, Thompson CJ, Marcincin RP. Diffuse axonal injury and traumatic coma in the primate. Ann Neurol
. 1982; 12 (6): 564–74.
20. Greenwald RM, Gwin JT, Chu JJ, Crisco JJ. Head impact severity measures for evaluating mild traumatic brain injury risk exposure. Neurosurgery
. 2008; 62 (4): 789–98.
21. Gurdjian ES, Roberts VL, Thomas LM. Tolerance curves of acceleration and intracranial pressure and protective index in experimental head injury. J Trauma
. 1966; 6 (5): 600–4.
22. Gurdjian ES, Webster JE, Lissner HR. Observations on the mechanism of brain concussion, contusion, and laceration. Surg Gynecol Obstet
. 1955; 101 (6): 680–90.
23. Guskiewicz KM, Marshall SW, Bailes J, et al.. Association between recurrent concussion and late-life cognitive impairment in retired professional football players. Neurosurgery
. 2005; 57 (4): 719–26.
24. Guskiewicz KM, Marshall SW, Bailes J, et al.. Recurrent concussion and risk of depression in retired professional football players. Med Sci Sports Exerc
. 2007; 39 (6): 903–9.
25. Guskiewicz KM, Mihalik JP, Shankar V, et al.. Measurement of head impacts in collegiate football players: relationship between head impact biomechanics and acute clinical outcome after concussion. Neurosurgery
. 2007; 61 (6): 1244–52.
26. Guskiewicz KM, Weaver NL, Padua DA, Garrett WEJ. Epidemiology of concussion in collegiate and high school football players. Am J Sports Med
. 2000; 28 (5): 643–50.
27. Hodgson VR, Thomas LM, Prasad P. Testing the validity and limitations of the severity index. SAE Technical Paper
. 1970: 700901.
28. Manoogian S, McNeely D, Duma S, Brolinson G, Greenwald R. Head acceleration is less than 10 percent of helmet acceleration in football impacts. Biomed Sci Instrum
. 2006; 42: 383–8.
29. Margulies SS, Thibault LE. A proposed tolerance criterion for diffuse axonal injury in man. J Biomech
. 1992; 25 (8): 917–23.
30. McAllister TW, Flashman LA, Maerlender A, et al.. Cognitive effects of one season of head impacts in a cohort of collegiate contact sport athletes. Neurology
. 2012; 78 (22): 1777–84.
31. McCrea M, Hammeke T, Olsen G, Leo P, Guskiewicz K. Unreported concussion in high school football players: implications for prevention. Clin J Sport Med
. 2004; 14 (1): 13–7.
32. McCrory PR, Meeuwisse W, Johnston K, et al.. Consensus statement on concussion in sport. The 3rd International Conference on Concussion in Sport held in Zurich, November 2008. J Sci Med Sport
. 2009; 12 (3): 340–51.
33. Mihalik JP, Bell DRM, Marshall SW, Guskiewicz KM. Measurement of head impacts in collegiate football players: an investigation of positional and event-type differences. Neurosurgery
. 2007; 61 (6): 1229–35.
34. Mihalik JP, Guskiewicz KM, Jeffries JA, Greenwald RM, Marshall SW. Characteristics of head impacts sustained by youth ice hockey players. Proc Inst Mech Engrs, Part P: J Sports Eng Technol
. 2008; 222 (1): 45–52.
35. Moon DW, Beedle CW, Kovacic CR. Peak head acceleration of athletes during competition–football. Med Sci Sports Exerc
. 1971; 3 (1): 44–50.
36. Ng TP, Bussone WR, Duma SM. The effect of gender and body size on linear accelerations of the head observed during daily activities. Biomed Sci Instrum
. 2006; 42: 25–30.
37. NIH Consensus Development Panel on Rehabilitation of Persons With Traumatic Brain Injury. Rehabilitation of persons with traumatic brain injury. J Am Med Assoc
. 1999; 282 (10): 974–83.
38. Pellman EJ, Viano DC, Tucker AM, Casson IR, Waeckerle JF. Concussion in professional football: reconstruction of game impacts and injuries. Neurosurgery
. 2003; 53 (4): 799–814.
39. Rowson S, Duma SM. Development of the STAR evaluation system for football helmets: integrating player head impact exposure and risk of concussion. Ann Biomed Eng
. 2011; 39 (8): 2130–40.
40. Rowson S, Duma SM, Beckwith JG, et al.. Rotational head kinematics in football impacts: an injury risk function for concussion. Ann Biomed Eng
. 2012; 40 (1): 1–13.
41. Schnebel B, Gwin JT, Anderson S, Gatlin R. In vivo study of head impacts in football: a comparison of National Collegiate Athletic Association Division I versus high school impacts. Neurosurgery
. 2007; 60 (3): 490–6.
42. Talavage TM, Nauman EA, Breedlove EL, et al.. Functionally-detected cognitive impairment in high school football players without clinically-diagnosed concussion. J Neurotrauma
. 2010 Oct 1. [Epub ahead of print].
43. Versace J. A review of the severity index. In: Stapp Car Crash Conference, 15th Annual
. Warrendale: Society of Automotive Engineers; 1971. pp. 771–96.
44. Zhang L, Yang KH, King AI. A proposed injury threshold for mild traumatic brain injury. J Biomech Eng
. 2004; 126 (2): 226–36.