The Relationship Between Subconcussive Impacts and Concussion History on Clinical Measures of Neurologic Function in Collegiate Football Players
- First Online:
- Cite this article as:
- Gysland, S.M., Mihalik, J.P., Register-Mihalik, J.K. et al. Ann Biomed Eng (2012) 40: 14. doi:10.1007/s10439-011-0421-3
- 1.1k Views
Concussions sustained during college and professional football careers have been associated with both acute and chronic neurologic impairment. The contribution of subconcussive impacts to this impairment has not been adequately studied. Therefore, we investigated the relationship between subconcussive impacts and concussion history on clinical measures of neurologic function. Forty-six collegiate football players completed five clinical measures of neurologic function commonly employed in the evaluation of concussion before and after a single season. These tests included the Automated Neuropsychological Assessment Metrics, Sensory Organization Test, Standardized Assessment of Concussion, Balance Error Scoring System, and Graded Symptom Checklist. The Head Impact Telemetry (HIT) System recorded head impact data including the frequency, magnitude, and location of impacts. College football players sustain approximately 1,000 subconcussive impacts to the head over the course of a season, but for the most part, do not demonstrate any clinically meaningful changes from preseason to postseason on measures of neurologic function. Changes in performance were mostly independent of prior concussion history, and the total number, magnitude and location of sustained impacts over one season as observed R2 values ranged between 0.30 and 0.35. Repetitive subconcussive head impacts over a single season do not appear to result in short-term neurologic impairment, but these relationships should be further investigated for a potential dose–response over a player’s career.
KeywordsConcussion historyCumulative exposureSubconcussive impactsMild traumatic brain injury
There is growing concern that head impacts (both concussive and subconcussive) sustained throughout a football player’s career may be a contributor to long-term brain dysfunction. Reports in the lay media of former professional football players’ struggles with depression and early onset dementia, combined with recent published case reports of chronic traumatic encephalopathy (CTE),21 beg the question about these potential cumulative effects. Concussions and their resulting impairments in brain function have been largely studied; however, little is known about the effects of the subconcussive impacts. Many studies demonstrate a dose–response relationship between the number of previous concussions and risk of subsequent concussions suggesting a lowered injury threshold exists in players who sustain multiple concussions.5,12,15,28 Division I football players reporting a history of three or more previous concussions were three times more likely to sustain subsequent concussions compared to players with no history of concussion.12 Concussion incidence is relatively high across all age levels of football participation.11,14,16,24,31 While little is known about subconcussive impacts, the topic has garnered a lot of interest of late. A recent case series investigation of 11 high school players reported negative changes in visual working memory and functional Magnetic Resonance Imaging (fMRI) in a subset of four players without concussion, but with a high frequency of impacts to the top of the head throughout the season.32 Thus, studying the potentially negative cumulative effect of repeated head impacts is an area of research that warrants investigation.
Previous research has explored the potential for mild traumatic brain injury resulting from the effect of repetitive soccer heading over time.9,20 Other research suggests that increased exposure to sports involving blows to the head, as well as duration of participation and increasing level of competition, appear to increase one’s risk of chronic neurologic impairment.28 One recent study indicated that former athletes with a history of concussion performed worse than controls on a recognition memory task 30 years postinjury.8 Another study investigating the potential acute impairments over the course of one football season suggests the cumulative effect of subconcussive head impacts has little effect on clinical measures of neurologic impairment.23 Players were evaluated preseason, midseason, and postseason, and no differences in neurocognitive measures as determined by the Standardized Assessment of Concussion (SAC) and Immediate Postconcussion Assessment and Cognitive Test (ImPACT) were observed. This work only studied the change in mental status and neurocognitive test scores, and did not include measures of balance and symptom reporting commonly used by sports medicine clinicians. Further, they were unable to objectively quantify characteristics of head impacts sustained by players in their sample.
No study has previously explored the relationship between repetitive subconcussive head impacts and clinical measures of neurologic impairment while monitoring sustained impacts throughout the course of a single collegiate football season. The purpose of our study was to explore the relationship between subconcussive head impacts and previous history of concussion on clinical measures of neurologic impairment over the course of a Division I Football Bowl Subdivision team’s season. Potential contributors of neurologic deficits according to frequency, magnitude and location of impacts, as well as the number of previous concussions and years of football participation at the college level, were all considered in the analyses.
Materials and Methods
Forty-six male collegiate football players were enrolled in this study (age = 19.65 ± 1.16 years, height = 189.43 ± 7.06 cm, mass = 112.72 ± 20.75 kg). The sample consisted of a variety of player positions including offensive and defensive linemen, linebackers, defensive backs, wide receivers, running backs, tight ends, and quarterbacks recruited from a single Division I Football Bowl Subdivision (formerly known as Division I-A) program. Participants were excluded if they had a concussion within 3 months prior to the scheduled baseline testing (over the course of the summer conditioning session), as well as any participants with a current vestibular, visual or balance deficit as determined by their medical history, or lower extremity injury that could disrupt balance performance. Any athlete that sustained a concussion during the course of the season was also excluded from postseason testing.
The Automated Neuropsychological Assessment Metrics test modules, cognitive domains evaluated, and brief overview of the test modules
Brief overview of test module
Simple reaction timea
Athlete quickly clicks mouse to respond to stimulus on screen
Mental processing speed
Athlete depresses correct mouse button in response to basic arithmetic problem
Match to sample
Athlete indicates which of two 4 × 4 checkerboard matrices exactly matches the original matrix previously presented for 2 s
Procedural reaction time
Numbers 2 through 5 randomly appear on screen. Athlete left-clicks if ‘2’ or ‘3’ appear; or right-clicks if ‘4’ or ‘5’ appear
Random symbol-digit pairings appear on the bottom of the screen. Athlete responds when random symbol-digit pairing matches any of the 9 symbol-digit combinations at the top of the screen.
Athletes respond to individual presentation of random letters on the screen to six letters presented to them at the start of the battery
Subjects completed the Standardized Assessment of Concussion (SAC) in a quiet and controlled environment. The SAC assesses domains commonly affected by mild traumatic brain injury such as orientation, immediate and delayed memory, and concentration; and additionally allows for a neurologic exam and clinical evaluation of symptoms with exertion.34 The SAC has been found to be both valid and reliable in high school and college athletes;3,17,19,34 however, practice effects have been previously identified. Two forms of the SAC were used across the two sessions to minimize practice effects.34 Additionally, the time lapse between preseason and postseason testing was sufficient to eliminate any practice effect from the initial testing session.17,19,33
The Sensory Organization Test (NeuroCom International Inc.; Clackamas, OR) was used to assess participants’ balance performance during baseline testing administered prior to preseason, as well as following the conclusion of the season. The SOT involves a force plate system capable of objectively evaluating a person’s ability to use sensory information from the visual, vestibular and somatosensory systems. Each participant underwent three, 20-s trials under six different sensory conditions: normal vision and normal support surface, eyes closed with normal support surface, sway-referenced visual input with normal support surface, normal vision and sway-referenced support surface, eyes closed and sway-referenced support surface, and sway-referenced visual and support surface. In the sway-referenced conditions, the conflicting sensory input required participants to suppress the inaccurate sensory information and rely on the remaining sensory systems to maintain their balance. The force platform recorded the center of pressure relative to a neutral starting position and computed a sway ratio based on anterior and posterior deviations.26 While inconclusive, some research has demonstrated a learning effect across trials when ordered serially4. Therefore, to eliminate the subjects’ ability to anticipate the condition (sway referencing, etc.) and minimize any learning effects, they performed their trials in randomized order.
The Balance Error Scoring System (BESS) consists of three stances (double, single, and tandem) completed once on a firm surface and repeated on a medium density foam pad (Airex Balance Pad 81000, Airex Power Systems, Knoxville, TN) for a total of six, 20-s trials. Participants were instructed to stand as motionless as possible with hands on iliac crests and eyes closed for all trials. Subjects were instructed to maintain control of the contralateral limb in 20° of hip flexion and 45° of knee flexion during the single leg balance tasks. If participants lost their balance at any point during the trial they were told to return to their original testing position as quickly as possible and resume the trial. Scoring for the BESS involved the summation of the total number of errors committed throughout the duration of all six of the trials. The six possible errors included: hands lifted off iliac crest, opening eyes, step, stumble or fall, moving hip into more than 30° of flexion or abduction, lifting forefoot or heel, and remaining out of testing position for more than 5 s.29 The total number of errors committed throughout the six trials resulted in their total BESS score. Each subject was videotaped and the lead author (SMG) later scored these videos to ensure accurate data collection. The BESS has previously been shown to have high intratester reliability (0.87–0.98).29 Only the lead author evaluated the BESS errors.
Participants were asked to fill out the Graded Symptom Checklist (GSC), which allowed them to self-report symptoms using a seven-point Likert scale ranging from asymptomatic (0) to mild (1) to severe (6). During both the baseline and postseason testing, subjects were asked to rate the severity of each symptom that they reported feeling “on a regular basis” (i.e., at least three or more times per week), on average, during the period of time leading into the testing session when they were not playing football.
The Head Impact Telemetry (HIT) System consists of six spring-mounted single axis accelerometers embedded within Riddell VSR-4 and Revolution football helmets used to track the frequency, location, and magnitude of impacts to the players’ helmets. A signal transducer transmits the data to the Sideline Response System on a laptop computer via radio wave transmission (903–927 mHz). In the event the accelerometer unit is unable to communicate directly with the sideline data collection system, the information from head impacts is stored on a nonvolatile onboard memory system. Data collection was initiated at the start of each practice or game throughout the season and information was synchronized and transferred to the database system following each practice and game to record all data. Weekly maintenance of accelerometers included status checks and battery replacements. The HIT System has been previously validated by laboratory testing.10
All subjects signed an informed consent form approved by the University of North Carolina at Chapel Hill Office of Human Research Ethics prior to testing. Baseline testing began in June and continued until the beginning of preseason camp in August. Postseason testing began in December, no sooner than 3 days following the last regular season game (average following last day of participation = 10.4 ± 3.6 days). All clinical testing took place in a quiet and controlled research laboratory.
Subjects completed a quasi-randomized testing order of each of the five testing measures (ANAM, SAC, SOT, BESS, and GSC). In order to facilitate testing, subjects were assigned to one of two testing orders during baseline testing. One group began with the computerized ANAM testing, while the other group began with the remainder of the clinical test battery beginning with the SOT and followed by the BESS. Postseason testing was kept in the same order for each participant. This eliminated the possibility of different testing orders affecting the outcome of our results from preseason to postseason. All athletes were asked to refrain from any physical activity for 2 h leading up to the test session (preseason or postseason). All postseason testing was completed within 3 weeks of each subjects’ last day of participation.
Raw head impact data were exported and reduced to include only impacts registering greater than 10 g. This cutoff value is consistent with previous literature in this subject area.10,13,22,25 Information on the frequency, magnitude, and location of impacts were recorded. The cumulative magnitude of impacts represented the summation of the recorded linear g’s of acceleration for each impact greater than 10 g sustained by a player over the course of the season. Our dependent variables were measured at preseason and postseason. The ANAM scores from each domain were analyzed individually. The total SAC score was used as our dependent variable for mental status (maximum possible score was 30 points), and consisted of the total of the four subtests as follows: orientation, immediate memory, concentration, and delayed memory. For the SOT, we used the composite equilibrium score computed using data collected during the 18 individual trials. This measure is indicative of the subject’s overall balance across all of the trials. Higher scores indicate better balance performance. For the BESS, we summed the number of errors committed by the subjects across each of the six conditions into a single total BESS score. Finally, individual symptom scores were recorded and summed for a total symptom severity score on the Graded Symptom Checklist (GSC). The total number of symptoms reported by each individual was also recorded, representing the total number of symptoms endorsed. After all of the baseline data and postseason data had been collected, change scores were computed for each of our dependent measures by subtracting our baseline measures from those recorded during postseason testing. Clinical data were then reviewed individually and an analysis of outlier data revealed four instances of depressed preseason baseline scores representing a very small fraction (<1%) of the overall data collected. Since these data deviated from the mean by more than 2.5 standard deviations and believed to be based on inaccurate data, these four data points were removed from our dataset prior to any further data analysis.
Statistical analyses included 12 separate multiple linear regression analyses (enter method)—one for each dependent variable—on change scores (postseason minus preseason) computed for each of our dependent variables, including all seven individual ANAM modules, the total SAC score, the SOT composite equilibrium score, the total number of BESS errors, the total symptom severity score, and total number of symptoms endorsed on the GSC. The six independent variables included for each multiple linear regression analysis consisted of: (1) the total number of impacts, (2) the total number of impacts greater than 90 g, (3) the total cumulative magnitude of impacts, (4) the total number of impacts to the top of the head sustained throughout the duration of the season, (5) the number of self-reported concussions sustained within the previous 5 years, and (6) the number of years of collegiate football participation. Prior research has reported that impacts to the top of the head result in the highest average magnitude and the highest likelihood of resulting in concussion.13,22 Additionally, significant results were examined more closely by looking at beta coefficients and p values to determine which of our independent variables contributed the most to the significant change in score from preseason to postseason. An alpha level of .05 was set a priori. Data were analyzed using SPSS for Windows version 16.0 (SPSS Inc.).
On average, the college football players in our sample sustained 1177.3 ± 772.9 head impacts during the football season, with an average of 12.0 ± 11.1 head impacts registering above our a priori 90 g “high-magnitude” impact threshold. The number of all head impacts above and below our threshold for high-magnitude impacts differed across playing position (F6,39 = 10.74; p < 0.001), with defensive and offensive linemen experiencing the most head impacts. There were no differences in the number of high-magnitude head impacts across playing position (F6,39 = 1.10; p = 0.38).
Neurocognitive and Mental Status Measures
Descriptive and statistical findings for clinical measures of neurologic function from preseason to postseason
Automated Neuropyschological Assessment Metrics
Other clinical measures of neurologic function
GSC total severitya
GSC total numbera
Sensory Organization Test
Predictor variables for significant regression models identified in Table 2
# Impacts >90 g
Cumulative magnitude of impacts (g)
# Impacts to top of head
# Previous concussions
# Years played
Δ SOT composite
Δ BESS total score
Δ Total number Sx endorsed
Balance Error Scoring System
On average, we observed an improvement in scores from preseason to postseason on the BESS (−5.84 ± 4.90 errors; lower scores indicate an improvement in postural stability). Using the same independent variables from our previous regression models, we observed a significant finding (p < 0.05; R2 = 0.27). The number of impacts (p = 0.01), cumulative magnitude of head impacts (p = 0.01), and the number of previous concussions sustained in the previous 5 years (p = 0.04) were significant predictor variables of the change in BESS total error score from preseason to postseason (Table 3). A higher number of head impacts and higher number of prior concussions was associated with an improved BESS score at postseason; however, a higher cumulative magnitude of head impacts was predictive of a worsened BESS score.
Graded Symptom Checklist
The independent variables in our regression model did not predict the changes we observed between preseason and postseason total symptom severity scores (p > 0.05; R2 = 0.22) (Table 2). The total number of symptoms endorsed was also recorded during both the preseason and postseason test sessions. A significant regression equation was found (p < 0.01; R2 = 0.35). Upon further evaluation, we observed that a higher number of severe head impacts (>90×g) (p = 0.03), a higher number of impacts to the top of the head (p = 0.04) sustained over the course of the entire season, and a higher number of years of collegiate football participation (p = 0.04), were all significant predictors of an increase in the total number of symptoms reported from preseason to postseason (Table 3).
Our findings suggest that football players sustain a substantial number of subconcussive impacts to the head over the course of a season, more than 1,000 on average, which is slightly higher but within the extrapolated range suggested in previous reports.7,10,30 For the most, we found that these impacts do not represent clinically meaningful changes from preseason to postseason on concussion tests often used to identify neurologic impairment. Most of the observed deficits (pre–postseason change scores), of which there were few, were subtle and appear to be independent of the total number of impacts sustained, total number of severe impacts (>90 g) sustained, total cumulative magnitude of impacts sustained, and total number of impacts sustained to the top of the head. Interestingly, we observed that the amount of college football exposure appears to be associated with preseason to postseason deficits on measures of symptoms and balance (SOT composite equilibrium). While our findings over the course of a single season do not support the notion that football exposure may contribute to depression, early onset dementia and CTE, the methodologies presented could help to establish a foundation for a more longitudinal study aimed at answering these questions.
Our results are consistent with previous work investigating uninjured football players throughout the duration of one season by testing them on neurocognitive tests at the beginning of the season, at midseason and at postseason.23 While significant increases in performance attributed to a practice effect were reported, the authors concluded that there were no meaningful changes observed across the three testing sessions.
Overall, we observed slight improvements in change scores from preseason to postseason on ANAM, but none of our predictor variables explained a significant amount of variance in these change scores in each of the domain modules. Previous studies have identified practice effects with serial administration of the ANAM test.6 Our procedures involved administering the concussion measures at only two different time intervals, nearly 6 months apart. Thus, we would expect any potential practice effects associated with the ANAM test battery to be washed out after a 6-month period between testing sessions.6 Another study investigating the validity of concussion measures found that the SAC did not show practice effects from repeated administration of the test.33 We observed a subtle decrease in the total SAC score from preseason to postseason, which was not related to the predictor variables.
Noting that there was only a small non-significant decrease in SAC scores from preseason to postseason, we re-examined our predictor variables. Based on our hypothesis, an increased total cumulative magnitude, for example, would have an associated decrease in SAC scores from preseason to postseason. One potential issue with this variable is that a player with a high total cumulative magnitude of impacts could be someone who has sustained a high frequency of impacts of very low magnitudes. While speculative at this point, it is plausible that many 10 g impacts may not have an effect over a season but by increasing the number of impacts exceeding 40 g, as an example, may result in a cumulative effect on concussion measures over the course of a season. This could explain why we did not observe a meaningful decrease in SAC score from preseason to postseason, and may have the same implications on our other measures of neurologic impairment. Only impacts registering a linear acceleration greater than 10 g were included for the purposes of our analyses as impacts below this threshold make it difficult to distinguish between head impacts and voluntary head movements. It would be interesting to explore the possible effects of subconcussive impacts occurring above a particular threshold on neurological impairment over time. For the BESS, there was an average decrease (i.e., improvement) of −5.84 ± 4.90 errors from preseason to postseason. This agrees with previous research findings that have found improvements in BESS scores after multiple testing sessions.34 As the total cumulative magnitude increased, the BESS score tended to result in an increased number of errors from preseason to postseason. Paradoxically, as the number of impacts increased, the BESS score resulted in a decreased number of errors from preseason to postseason. This relationship, although small, partially agrees with our hypothesis in that as total cumulative magnitude increases, the BESS deficits become more pronounced (i.e., increase in total BESS errors) from preseason to postseason. There is no reasonable explanation for why an increase in total head impacts would be associated with improved BESS scores, but future studies should attempt to clarify this finding. Changes observed for the SOT equilibrium scores do not represent meaningful postural deficits. As this is among the first study of its kind, we feel our findings relating to these individual balance evaluation methods may have implications to how clinicians interpret these separate balance outcomes in patients who sustain repeated sub injurious head impacts. Likewise, those players with a higher prior concussion history tended to perform slightly better on the BESS at post-season compared to preseason. Thus, the normal practice affect does not appear to be affected by prior concussions.
Finally, we observed an average increase of 0.90 ± 1.45 symptoms from preseason to postseason on the GSC. According to our model, the total number of impacts >90 g, total number of impacts to the top of the head, and number of years of collegiate football participation predicted symptom change score. As the total number of impacts >90 g increased, the total number of symptoms reported increased from preseason to postseason. Similarly, as the total number of impacts sustained to the top of the head increased, the total number of symptoms reported also increased from preseason to postseason. Additionally, as the number of years of collegiate football participation increased, the total number of symptoms reported also increased.
As a subjective self-reported measure, an average change score of less than one point is not alarming, as the symptoms could be brought on by other factors. However, it would be interesting to study this trend over multiple seasons to see if there may be a dose–response to repeated subconcussive head impacts. Our model accounts for 35% for the variability in these results, leaving approximately 65% of the variability accounted for by extraneous factors. Other possible reasons for the increased number of symptoms reported at postseason testing, such as drowsiness and fatigue, could be explained by the effect of participating in an intense and physically demanding sport for 6 months, while also balancing their academic responsibilities.
Although our regression model yielded significant variables that contributed to a pre–postseason change in select concussion measures, we acknowledge that these differences have limited clinical significance. A difference of one point on the SOT or GSC does not represent meaningful postural deficits or an alarming increased presence or severity of symptoms. These clinical measures of neurologic function have all been tested and found to be reliable measures and usually within a small degree of error, which most of our observed change scores likely fall within.17,19,27,33
While some of our outcomes may suggest a relationship exists between subconcussive impacts and history of previous concussion on our independent factors, it is also important to recognize that most of these change score values had observed R2 values in the range of 0.30–0.35. This would indicate that only 30–35% of the variance in these clinical scores could be explained by the independent variables of head impacts, football exposure, and concussion history. The measures we used are typically used for gross estimates of cognitive function following concussion and while widely used in the sports medicine setting, are most sensitive for detecting acute cognitive deficits (1–5 days postinjury).2,18 There may also be additional variables that we have not accounted for that might help explain changes in our resulting outcome measures. These potential contributing factors, which should be studied in future studies, include motivation or effort of subjects when performing these tests, presence of attention deficit disorder or learning disabilities, and sleep deprivation. These are all possible sources that may influence neurocognitive domains and presence of concussive-like symptoms.
There were a few individuals that sustained substantial time-loss injuries during the season and thus had limited impact exposure. Based on our data, the independent variables we hypothesized would explain any observed changes in clinical measures did not provide a complete explanation of this question. Our statistically significant regression models resulted in relatively low R2 values, suggesting a number of the above factors we did not study could potentially explain any noted changes in performance on these measures. The inclusion of a control group in future studies may permit us to elucidate more clearly the true effect of subconcussive head impacts on changes in clinical measures of neurologic function. We had to eliminate some individual data points due to extreme variations from group means or from the subject’s own previous score. We attributed these few outliers to a lack of motivation, which has previously been shown to adversely affect performance on neurocognitive testing.1 The underreporting of concussions by the athletes enrolled in our study could also attribute to the outliers we observed. Our sports medicine team employs a systematic concussion management protocol. When the injury is sufficiently mild to not manifest in any overt presence of signs that can be readily observed, it is sometimes very difficult to accurately render a concussion diagnose in an athlete unwilling to report their symptoms to the sports medicine team. Furthermore, there was no way to determine the relative contributions better muscle conditioning or lower extremity injuries may have had on our balance performance outcomes based on our methodology.
It is important to note that all of the change scores from our outcome measures did not amount to many clinically relevant deficits in performance over the course of a single season of football. Previous literature has investigated the cumulative effect of repeated concussions as well as the concept of the neurometabolic cascade effect. Currently, the relationship between the neurometabolic cascade effect and performance on clinical measures of neurologic function, as well as symptom presentation, is still largely unknown. However, if there was a relationship between altered brain metabolism and a measurable decreased performance on clinical tests, our study was unable to make a link between a season-long accumulation of head impacts and deficits on measures of neurologic function. However, future research should be conducted over multiple seasons looking at any potential changes on clinical measures of neurologic impairment while relating it to head impact data and overall football exposure. Given the recent attention that trauma-induced depression, dementia, and CTE have garnered, it would be beneficial to learn of potential long-term effects of subconcussive impacts over a football career. Our findings suggest that the number of previous years of playing exposure at any level should be considered in any investigations studying the link between neurologic function and cumulative head impacts. Furthermore, neuroimaging studies should be included as part of these future studies to identify potential correlates such as structural abnormalities, decreased white matter connectivity (Diffusion Tensor Imaging), and altered brain activity (functional Magnetic Resonance Imaging) from preseason to postseason, as well as over multiple seasons.
This study was supported in part by the Centers for Disease Control and Prevention and the National Operating Committee on Standards for Athletic Equipment.