head impacts during high school football: a biomechanical … › 2010 › 09 › ... ·...

Journal of A ihleíic Training 2009;44(4):342-349© by ihe National Athletic Trainers' Association, Incwww.nata.org/jat origina! research

Head Impacts During High School Football: ABiomechanical AssessmentSteven P. Broglio, PhD, ATC*; Jacob J. Sosnoff, PhD*; SungHoon Shin, MS*;Xuming He, PhD*; Christopher Alcaraz, MDf; Jerrad Zimmerman, MDt*University of Illinois at Urbana-Champaign, Champaign, IL; fCarle Foundation Hospital, Urbana, IL

Context: Little is known about the impact biomechanicssustained by players during interscholastic football.

Objective: To characterize the location and magnitude ofimpacts sustained by players during an interscholastic footballseason.

Design: Observational design.Setting: On the field.Patients or Other Participants: High school varsity football

team (n = 35; age - 16.85 ± 0.75 years, height = 183.49 ±5.31 cm, mass - 89.42 ± 12.88 kg).

Main Outcome Measure(s): Biomechanical variables (linearacceleration, rotational acceleration, jerk, force, impulse, andimpact duration) related to head impacts were categorized bysession type, player position, and helmet impact location.

Results: Differences in grouping variables were found foreach impact descriptor. Impacts occurred more frequently andwith greater intensity during games. Linear acceleration wasgreatest in defensive linemen and offensive skill players andwhen the impact occurred at the top of the helmet. The largest

rotational acceleration occurred in defensive linemen and withimpacts to the front of the helmet. Impacts with the highest-magnitude jerk, force, and impulse and shortest durationoccurred in the offensive skill, defensive line, offensive line,and defensive skill players, respectively. Top-of-the-helmetimpacts yielded the greatest magnitude for the same variables.

Conclusions: We are the first to provide a biomechanicalcharacterization of head impacts in an interscholastic footballteam across a season of play. The intensity of game playmanifested with more frequent and intense impacts. Thehighest-magnitude variables were distributed across all playergroups, but impacts to the top of the helmet yielded the highestvalues. These high school football athletes appeared to sustaingreater accelerations after impact than their older counterpartsdid. How this finding relates to concussion occurrence has yet tobe elucidated.

Key Words: concussions, mild traumatic brain injuries. HeadImpact Telemetry System, acceleration

Key Points

The mean linear acceleration resulting from impacts recorded during both high school games and practices exceeded thatreported at the collegiate level.Impacts to the top of the head yielded the greatest linear acceleration and impact force magnitude. Coaches must teachproper tackling techniques in order to reduce the risk of concussions and serious cervical spine injuries.High school offensive and defensive line players sustained the lowest-magnitude impacts but the highest number ofimpacts during games and practices.

Participation in sporting activities has been estimatedto result in 1.6 million to 3.8 fnillion brain injuriesannually.I Injuries occurring during collegiate and

professional football often receive the greatest attention,with an incidence rate of 4.8% to 6.3% of collegiateathletes23 and 7.7% of National Football League (NFL)athletes.'* On an annual basis, therefore, concussions occurto an estimated 3264 to 4284 of the 68 000 collegiate playersand 130 ofthe 1700 NFL athletes. Interscholastic athletes(ie. high school ). however, represent the single largestcohort of football players and account for the majority ofsport-related concussions. In a given year, 3.6% to 5.6% ofthe 1.2 niillion interscholastic football athletes sustainconcussions.-^'^ corresponding to an estimated 43 200 to67 200 injuries. The true injury incidence is likely muchhigher, because 53% of concussed high school athletes arcsuspected of not reporting their injuries to medicalpersonnel.'' Despite the fact that the greatest number of

concussions occur in high school football players,'' ourpresent understanding of the injury has focused oncollegiate and professional athletes.

Concussion assessment has improved over the years, andevaluative measures of concussion-related symptoms,neurocognitive functioning, and postural control havebeen proposed as assessment devices.** The strength ofthese tests lies in providing objective information forwithholding an injured athlete and in making return-to-play decisions. For example, the Standardized Assessmentof Concussion'' and the Balance Error Scoring System'"have been used effectively as sideline assessment tools thatare administered once an athlete with a suspected injuryhas been identified. Recognizing athletes with a suspectedinjury, however, remains problematic.'' The Head ImpactTelemetry System (HITS) (Simbex LLC. Lebanon. NH) isa novel wireless monitoring system with the potential torapidly identify athletes who have sustained an impaet to

342 Volume 44 • Number 4 • August 2009

the helmet capable of producing an injury. The HITSprovides real-time, postimpact datai' to the clinicianpositioned on the sideline, but the novelty of the systemand a dearth of information related to impact biomechan-ics presently limit utility of the HITS as a diagnostic tool.

Duma et aP- were the first to report on the HITS when38 athletes (up to 8 at a time) were equipped with thesystem for 10 games and 35 praetiees. The distribution ofimpact magnitudes was skewed positively, and the meanresultant linear aeeeleration was estimated at 32 ± 25g(l,tr = 9.8 m/s-). The investigators later expanded theirsample and reported similar Undings when up to 18 athletes(52 athletes total, according to the authors) were equippedwith the HITS during the 2003 and 2004 football seasons.'"^During the 67 praetiees and 22 games, a total of 11 604impacts were recorded. Aeross all players fitted with thesystem, an average of 135 impacts occurred during gamesand 129 impacts during practices, with a mean resultantlinear acceleration of 20.9 ± 18.7g. With identical methodsin a 2-year Investigation at a different institution, the meanlinear acceleration in collégiale athletes was 22.25 ± 1.79̂ ,'across all athletes and sessions. Further analyses showeddifferent impact magnitudes for session type, playerposition, and helmet location.'•+ Schnebel et al'*^ were thefirst to evaluate impacts occurring during interscholasticfoi>tball but failed to report the overall mean accelerationsfor the cohort.

The HITS teehnology has increased our biomechaniealunderstanding of impaets sustained during collegiate-levelfootball participation, but using both common and novelvariables associated with blows to the head to characterizeimpacts sustained at the interscholastic level is a newapproach. For instance, in addition to the often-re portedlinear and rotational aeeelerations that result from a headimpact, other investigators of brain trauma biomeehanicshave evaluated variables sueh as force, veloeity, impulse.and impact duration. i<' Many of these variables have yet tobe applied to a sport context or examined in relation tomore commonly reported values.

Although the incidence rate of concussion appears to besimilar across levels of play, the absolute number of injuriesoccurring during high sehool football where the disparityin medical coverage is the greatest, drives the need for abetter understanding of impact biomeehanics speeific to theyounger athlete. In particular, il is unclear whether thefrequency and magnitude of impacts mimic those of theolder and more mature players. Therefore, the purposes ofour ongoing investigation were to characterize impacts tothe head sustained by high school football athletes, using avariety of biomeehanical variables, and to compare theseoutcomes across session type, playing position, andlocation of head impact.

METHODS

All varsity-level interscholastie football athletes (n = 35)from a single central Illinois high sehool (Class 3A) wererecruited to participate in this study. Before data collectionbegan, all athletes and their parents completed institutionalreview board approved written informed assents andconsents, respectively. The institutional review board alsoapproved all study proeedures. Each athlete then provideddemographic information such as age, height, mass, year in

school, primary position, and helmet age. The athletes werefitted with either a new or a 1-year-oId Riddell Revolutionhelmet {Riddell/Ail American. Elyria. OH), and a HITSencoder was placed inside. Only 32 athletes wore a HITS-equipped helmet at any given time. Three athletes initiallyrecruited to participate were lost to orthopaedic injuriesbut were replaced by junior varsity athletes. Data werecollected aeross an entire season of football partieipation,including all preseason practices, all regular-season gamesand practices, and all postseason games and practices.

The HITS incorporales 2 components: an encoder unitlocated in the football helmet and a sideline computer. Theencoder consists of 6 sitigle-axis accelerometers. a telemetryunit, a data storage device, and an onboard battery pack.all encased in waterproof plastic and retrofitted within thepadding of the football helmet (Figure 1 ). Helmetsequipped with the HITS encoder look and functionidentically to other helmets and continue to meet NationalOperating Committee on Standards for Athletie Equip-ment (NOCSAE) standards for safety. Impact data arereeorded at 1000 Hz and transmitted to the sidelinecomputer for clinical use and data storage for lateranalysis. If the encoder-equipped helmet goes out of range(more than 137 m [150 yd]) of the eomputer, the onboarddata storage unit will record up to 100 impacts. Theseimpacts then download to the computer once ihe helmet isagain within range. The sideline computer is connected to aradio receiver, which is housed within an environmentallyresistant plastie case. Constant communication between thehelmet and sideline receiver is achieved through acontinually changing. Federal Communications Commis-

Figure 1. The Head Impact Tetemetry System encoder ptacedwithin the padding of a tootbaii heimet around the crown of thehead. (Photograph courtesy of Simbex LLC, Lebanon, NH.)

Journal of Athletic Training 343

sion-approved radio frequency range (903 to 927 MHz).Data received by the sideline computer are processedthrougii a novel algorithm to determine the location andmagnitude of each impact to the head." Accuracy of theHITS was established by comparing impact outcomes fromHITS-equipped football helmets with similarly protectedHybrid III crash dummies equipped with a 3-2-2-2accclerometer array. The correlation between the HITSand Hybrid III dummies was high {r = 0.98). with a 4%error rate when estimating both linear and angularaccelerations.'- Similarly, the location of impact wasnearly identical under the NOCSAE testing protocol,where the HITS accurately identified the location ofimpact within ±0.41 cm.i-

Data related to a head impact were recorded when asingle accelerometer exceeded the preset I5,if threshold.Following this event, the 8 milliseconds before impact andthe 32 milliseconds after impact were transmitted andstored. Because the data storage is triggered by a singleaccelerometer. the resultant linear acceleration might fallbelow 15,íí. To reduce the inclusion of errant impacts in thedata set, an investigator reviewed all impacts on a dailybasis. Athletes with high-magnitude impacts were ques-tioned about the nature ofthe impact. For example, if thehelmet was dropped, thrown, or swung against anotherhelmet, the impact was marked and later removed from thedata set and all analyses. The data available from thecomputer contained all pertinent impact data, includingpeak linear acceleration, rotational acceleration (derivedfrom the x-axis and y-axis angular accelerations), impactlocation, and date and time stamp.

Data Analysis

Data were grouped based on the session type (practiceand game), impact location, and athlete position. Impactswere categorized into front, back. side, and top. Thoseimpacts occurring above 60° of elevation over the head'scenter of mass were categorized as lop. Those below 60° ofelevation were divided into quadrants based on theazimuth location of impact (front, back, or side; Figure 2).Player positions were grouped into offensive line (center,guard, or offensive tackle; n = 8), offensive skill(quarterback, receiver, tight end, running back, or fullback; n = 15). defensive line (defensive end. nose tackle, ordefensive tackle; n = 7). and defensive skill (hnebacker.corner, or safety; n = 4). As is common to many highschool teams, some athletes played both offensive anddefensive positions (n = 8, 23%). For the purposes of thisinvestigation, players self-identified their primary position,which was used for analyses.

Automated features within the HITS software providethe resultant linear and angular acceleration and locationof each impact. Additionally, a custom MATLAB (version6.5.1; The MathWorks, Inc. Natick. MA) program waswritten to evaluate each impact and calculate the followingvariables: body mass index (BMI), estimated head mass.'^number o{' impacts per session per athlete, estimated peakhead jerk (m/s^), estimated impact force (N). impactimpulse (N/s), and impact duration (ms). A representativeimpact curve is shown in Figure 3. Impulse is the amountof force applied over a period of time, with greater impulseresulting in a greater change in momentum of the struck

-135" azimuth 135" azimuth

45° azimutti

-45" azimuth

0" azimuth

Figure 2. Distribution of impact locations across the helmet.Impacts to the right and left sides were grouped together.

object. Larger changes in momentum are speculated topresent a greater injury risk.if* Impulse was calculated bysumming the area under the acceleration curve using thetrapezoidal estimation method and multiplying that valueby impact duration. Impact duration was determined as thedifference in time from point A to point C. Jerk is ameasure of velocity, defined as the change in accelerationdivided by the change in time: m-s---s-i. Not all of theimpact measurements could be processed with our customprogram; fewer than 3% (n = 580) were excluded from theanalyses.

Before analysis, we visually inspected the dependentvariables (linear acceleration, rotational acceleration, ierk,force, impulse, and duration), revealing a positively skeweddistribution. To control for the violation of normality inthe subsequent analyses, these data were converted with anatural log function, and the statistical analyses wereconducted on the converted data. All dala presented beloware in the original metric. Further inspection of the dataindicated a violation to the homogeneity of variance for alldependent variables when grouped by player posilion andhelmet impact location (P < .01). No homogeneityviolations were noted when the data were grouped bysession type (P > .01).

We first performed a x~ goodness-of-fit assessment forimpact frequency based on session type, player position,and helmet location. This technique adjusts the number ofexpected events by accounting for the numbers of gameand practice sessions and athletes per position group. Asecond set of analyses included assessments of thedependent variables by session type, player position, andhelmet location. Analyses of variance were used to evaluatedifferences in the demographic variables and dependentvariables based on session type. When main effects werefound, post hoc comparisons were corrected using theBonferroni method. To control for the homogeneity ofvariance violation when the dependent variables weregrouped by player position and hclmcl location, weimplemented the Brown-Forsythe technique and the


35

Figure 3. Representative impact curve. A number of variableswere calculated from the impact data and participant demograph-ics. Jerk = (B^ KV\B A time. Force = B^ * head mass. Impuise =[Area under curve from Aa lo C;,] * time. Impact duration = lime fromA, to C^ a indicates acceleration values; t, time values.

(iames-Howell post hoc analysis. Fiiiiilly. Spearmancorrelations were calculated to better understand therelationship between the commonly used variables of linearand rotational acceleration and other measures of theimpacts. All analyses were completed with SPSS (version15.0; SPSS Inc. Chicago, IL) with statistical significance setat P < .01.

RESULTS

During the 2007 football season, we collected data across 68sessions, which included 35 practice days and 13 games. Atotal of 19224 Impacts were included in the analyses. Someimpacts (n = 82) were excluded from the data set for eventsnot directly t elated to sport participation, such as dropping orthrowing a helmet. No differences were noted atnong theplayers" ages, heights, and helmet ages (P > .01; Table 1).However, differences were seen for player mass (F3.30 =19.757, f* < .01 ). BMI (Fi.,,, = 13.20, P < .01 ). and head massiFy_M) = 16.44, P< .0\). For these variables, the offensive litieplayers were larger than the offensive skill (P < .01) anddefensive skill {P< .01) players, and the offensive skill playerswere smaller than the defensive linemen (P < .01).

The mean number of impacts incurred by all playersduring all sessions was 15.87 (SD = 17.87). The frequeneiesof impacts by player position for both session types arepresented in Figure 4. More impacts occurred duringgutnes {mean = 24.54 ± 22.41) than during practicesessions (mean = 9.16 ± 8.64) for all athletes (x-i =7452.88, P < .01). Further analyses indicated differences(xh = 1405.46, P < .01) among the number of impacts byplayer position for all sessions. The defensive line players

30

25

è 20

I"¡,0

„ . • • • • • • • ^ ^

, . . 1

-^Offensive line-•-Onensive skill••*•• Defensive line

' • " Defensive skill

—

Practices Games

Figure 4. Mean number of impacts sustained during practices andgames by player positions.

sustained the greatest number of impacts per session, withimpacts occurring 1.04. 1.69, atid 1.98 times motefrequently than those experienced by the offensive linemen,offensive skill players, and defensive skill players, respec-tively. Finally, differences in impact frequency by helmetlocation were ptesent {y}y = 4815.02. P < .01), withimpacts to the front ofthe helmet 1.79. 2.88, and 3.32 timesmore frequent than those to the back. side, or top oi' thehelmet, respectively.

Linear head accelerations alter impact by player positionand helmet location are presented in Figure 5. Differencesby session type were noted (F1J9222 = 50.23, P< .01). withgame impacts (mean = 24.76 ± 15.72 .t,') resulting in higherlinear accelerations than practice impacts (mean = 23.26 ±14.48^). Main effects for resultant hnear accelerations wereseen for player position (F3.10149.7n ^ 9.53, P < .01) andimpact location (fi,j2fí?9.5fi = 243.17, P < .01). Post hoeanalyses revealed that the defensive linemen and offensiveskill players sustained similar-magnitude linear accelera-tions, but only the defensive line players had greater linearaccelerations than the defensive skill and offensive lineplayers (/' < .01). For impact location, hits to the top ofthe helmet produced linear accelerations that were greaterthan those at all other locations {P < .01). In descendingorder, the magnitude of resultant linear accelerations wastop, front, back, and side.

Rotational head accelerations alter impact by playerposition and helmet location are presented in Figure 6.Analyses ofthe resultant rotational acceleration also showeda dilTerence by session type (Fij.,220 = H9.39, P < .01), withgame impacts (mean = 1669.79 ± 1249.41 rad/s-) generatingmore rotational acceleration than practice impacts (mean =1468.58 ± 1055.00 rad/s^). Additional analyses showed maineffects Ibr player position iFyM_)\f-9.w = 23.00, P < .01) andimpact location iF^wm^.si = 421.43, P < .01). Follow-uptesting revealed that the offensive and defensive line playerssustained similar magnitudes of rotational accelerations.

Table 1. Demographic Information for the High School Football Athletes (Mean ± SD)

Players Age, y Height, cm Mass, kg Body Mass Index Head Mass,

Offensive lineOffensive skillDefensive lineDefensive skillAll athletes (n

(n(n

(n(n

= 8)= 15)= 7)= 4)

34)

16.97 ± 0.5416.96 ± 0.9516.78 ± 0.4616.09 ± 0.8016.85 ± 0.75

184.85 ± 4.74181.17 ± 5.52185.16 ± 3.77182.78 ± 6.90183 49 ± 5.31

100.12 ± 10.1476.81 ± 6.2695.91 ±7.1781.64 ± 4.5089.42 ± 12.88

29.3723.4427.9724.4726.54

± 3 40• 2.14± 1.90± 1.34± 3.59

5.65 t 0.185.20 .±0.155.58 ±0.155.31 ± 0.165.45 ± 0.26

Head mass was estimated from the participants' heights and weights, based on recommendations by Zatsiorsky.'?


40

3.9 302aiÖ; 20uu<S 10 f

oit

0 fE

oCOm

n5

nCß

ve Li

1 11Q.O

ne _ F

ront

f t

p

1m

>ns

i

veS

•11

Top o

lt

De t

p.

m

îns

E

i

(U

ve L

i l

T1Q.O

ne

olt

De f

Ü

CD

3ns

=

(U

veS

11

—

Top

Fro

nt

^ B

ack

—'1

ï-

•acô

ayers

h

Top

Figure 5. Resultant linear acceleration by player position and impact location.

which were greater than those of ihe offensive and defensiveskill players (P < .01 ). The post hoc impact location analysesdemonstrated that impacts to the front of the helmet weregreater than to al! other helmet areas {P < .01). Indescending order, the greatest rotational acceleration result-ed from impacts to the front, back. side, and top.

The fnial analyses included evaluations of the maximumhead jerk, impact force, impact impulse, and duration ofimpact variables (Table 2). Compared with practices.larger magnitudes of all variables {P < .01) resulted duringgame play. We found a main effect for ail variables byplayer position {P < .01). with follow-up analysisindicating that the offensive line players had less head jerkthan the offensive skill and defensive line players {P < .01)but did not differ from the defensive skill players (F > .01).Forces of impact did not differ between the offensive anddefensive skill players (P > .01). but they were both lessthan those of the offensive and defensive linemen (P < .01).The evaluation of impact impulse revealed equivalentvalues for the offensive and defensive lines, which wereboth greater than those for the offensive and defensive skillplayers {P < .01). The duration of impact was longer inoffensive line players than in all other athletes, followed bythe defensive line, who had longer-duration impacts thanthe offensive and defensive skill players (P < .01). Theoffensive and defensive skill players did not differ forduration of impact {P > .01).

On evaluating differences between the calculated vari-ables and impact location, we noted main effects for allvariables (P < .01). The post hoc assessment for maximumhead jerk indicated that impacts to the top of the helmetgenerated more jerk than all other impact locations (P >

2000

"„ 1600

S.S 1200

800

400

.01). Impacts to the side of the helmet produced less jerkthan al! other locations (P < .01). No difference was seenbetween impacts to the front or back of the helmet (P >.01). Impact force was greater at the top of the helmet thanat all other locations (P < .01). with differences occurringat all locations. In descending order, the greatest force wasgenerated when blows occurred to the top, front, back, andside. Impact impulse was also greatest at the top of thehelmet (P < .01), with differences occurring at all helmetlocations. The greatest impulse values occurred in the sameorder: top, front, back, and side. Compared with all otherlocations, the duration was shortest {P < .01) for impactsto the top of the helmet and longest {P < .01) for impactsto the side of the helmet. Impacts to the front and back ofthe helmet did not differ in duration (P > .01).

Last, the correlation analyses indicated strong relation-ships between the resultant linear acceleration and themaximum jerk (r,,iH649) = 0.90. P < .01), maximum force(̂ (18649) = 0.98, P < .01), and impulse (rs(i8649) = 0.85, P< .01). Significant but weaker correlations were notedbetween rotational acceleration and jerk (/V;KO4<)) = 0.55. / '< .01), force (is(i8649) = 0.68, P < .01), and impulsei''s(i8649) = 0.67. P < .01). Significant weak relationshipswere seen between impact duration and linear-0.12. P < .01) and rotational acceleration (0.10. P< .01).

DISCUSSION

Our goal was to describe head impacts in interscholasticfootball athletes across a sea.son of play. In addition to thepreviously reported assessments of resultant linear and

Fron

tn ff

J£Utom

an5

iI

=

1til

•a<n

ve Lin

a.o

e

Fro

ntri

ji:

otnm

ffens VE

M

1

„ S

ide

k

Top "co

it

Def

J£O(0

m

ens

i

=m«

ve Lir e

Oit

Det 2 B

ack

s

i

i

i

ai•nif)

veSk

Top

Fro

nt -seum

m

AH=1

S

i

=

i

(U•D

CO

ayers

-

Top

Figure 6. Resultant rotational acceleration by player position and impact location.


Table 2. Calculated Variables by Session Type, Player Position, and Impact Location (Mean ± SD)

Variable

Session type«PracticesGames

Player positionOffensive lineOffensive skillDefensive lineDefensive skill

Helmet impact locationFrontBackSideTop

OverallAll sessions, players, and

locations

Maximum Head Jerk, m/s3

10.04 ± 9.8710.81 ± 9.97

9.73 ± 10.4011.01 ± 11.5210.31 ± 7.2410.25 ± 8.29

9.85 ±8.1610.01 ± 11.388.14 ± 6.74

15.42 ± 13.26

10.38 + 9-92

Impact Force, N

1281.67 ± 966.741357.72 ± 1090.63

1314.14 ± 931.681318.22 ± 1257.571332.08 ± 783.491242.98 ± 939.21

1285.87 ± 932.531262.57 ± 100.821093.96 ± 756.571767.35 i 1418.18

1314.88 ± 1023.36

Impulse, N s

4.73 ± 3.284.91 ± 3.90

4.94 ± 3.114.76 ± 4.434.81 ± 2.714.50 ± 3.56

4.79 ± 3.434.64 ± 3.254.42 ± 3.155.65 ± 4.74

4.81 ± 3.57

Impact Duration, ms

9.06 ± 3.028.90 ± 3.00

9.19 ± 2.908.86 ±3.168.98 ± 2.888.83 ±3.17

9.04 ± 2.779.01 ± 3.039.84 ±3.527.78 ± 2.71

8.99 ± 3.01

For all variables, magnitudes during games were greater than during practices (P < .01).

rotational acceleration., we also sought to calculate andprovide a statistical analysis of variables thought to betterdescribe these impacts. The tiiost notable finding from thisinvestigation is that the mean linear acceleration resultingfrom impacts recorded during games (24.76.Í,') and practices{23.26,ti) exceeded that reported at the collegiate level. Usingidentical methods. Mihalik et al'-* reported that the averagelinear acceleration of a collegiate football team was 22.25,1,'across all session types, and BroHnson et al'^ found a meanlinear acceleration of 20.9,ÍÍ in a similar collegiate sample.I'lirthcnnore. Mihalik et aM-̂ removed impacts of less thanI OÍ; from the data set, deeming them inconsequential.Employing the same technique here would further increasethe mean linear acceleration to 24.98,? across all sessions.Whether the small dilTerences in linear acceleration areclinically meaningful is not entirely clear. That is, we areunable to discern if these differences may influence the riskfor concussivc injury, but we speculate that the increasedmagnitude would likely escalate the potential for injury in thehigh school athlete. This potential is of particular importanceconsidering that medical coverage is available at fewer than50% of high schools.''' where the risk for catastrophic injuryappears to be the highest.-"

The distribution and magnitude of impacts across helmetlocations also differed by the level of play. In collegiatefootball athletes.'^ blows to the front ofthe helmet were10% less frequent and resulted in }g less force than in ourinterscholastic players. Conversely, impacts to the top andback ofthe helmet iti the collegiate athletes occurred morefrequently but resulted in slightly less linear acceleration (1to 2,1,'). Most concerning in the high school athletes were theimpacts to the top of the head, which yielded the greatestlinear acceleration (Figure 5) and impact force magnitude(Table 2). The increased impaet intensity to this area ofthehelmet likely elevates the risk of concussion but alsoincreases the propensity for more severe cervical injury.^iThis finding highlights the need for coaching propertackling techniques, such that the athlete keeps his headup and avoids contact with the top of the helmet. Anexplanation of the diflerences between collegiate andinterscholastic football athletes is beyond the scope of thisinvestigation. However, we hypothesize that physical

maturation and the associated neck strength and endur-ance discrepancy between the levels of play may be afactor. On average, collegiate athletes are reported to weigh15 kg more than our athletes but stand only 3 em taller.--Thus, collegiate athletes may have a more developedmusculature system that is better able to control headmotion after impact.

Using a variety of methods, prior investigators haveattempted to identify the resultant linear accelerationnecessary to cause a concussion. Initial attempts to quantifyhead impacts were completed by placing a single triaxialaecelerometer within the padding of football helmets. Reid etaP ' reported impacts ranged from 40 to 230,i,' in a collegiatefootball player, with a single concussion resulting from a188.Í,'linear acceleration. Naunheimet al-'' reported the meanpeak linear acceleration ot interscholastic football impacts as29.2g but no concussions occurred. These studies providedinitial assessments of football impacts but were limited bytheir small sample sizes; also, the reported linear accelera-tions likely represent motion of the helmet and not theathlete's head. More reeently. in a sample of concusslveimpacts in 13 collegiate football athletes, the H ITSquantified mean linear acceleration at IO2.77,t».-'̂ In perhapsthe most e.xtensive attempt to establish a threshold forconcussion, researchers^f» acquired video footage of 31concussive blows that occurred during professional footballgames. After the impacts were reconstructed in a laboratorysetting with Hybrid III models, the mean linear accelerationwas noted to be 98.Í,'. The authors proposed that a linearaeceleration of 70 to 75^ was necessary to induce concus-sion.-f" Using these criteria, our data included 271 impacts( 1.4%) that exceeded the 7O.ti level and 78 impacts (0,4%) thatexeeeded the 98,? level, slightly fewer impacts than previouslywere reported at the high sehool level, i** However, only 5concussive injuries were diagnosed (although not included inour analyses), suggesting that the magnitude of impactreported by the NFL Mild Traumatic Brain InjuryCommittee may not apply to the high school athlete or thatother variables need to be considered when establishing athreshold for eoncussion (or both).-'

Despite the difTerences in the mean linear accelerationsoccurring at dilTerent levels of the sport, the diflerences in


impact magnitude among the high school athletes may belinked to the probability of concussive injury. The incidenceof concussion is not equal across all player positions: Thehighest injury rates in high school football occur inquarterbacks (1.3 per 100 team-game positions) and runningbacks (0.74 per 100 team-game positions),*^ and widereceivers sustain more severe injuries.^ Our classificationmethod would designate these athletes as offensive skillplayers, who, we found, sustained the greatest magnitude oflinear acceleration after an impact (Figure 5). The highermagnitudes of impacts may result from the full-speed, open-field impacts that commonly take place at these playerpositions. The highest injury rates at the collegiate leveloccurred to the oíTcnsivc line (0.95/1000 athlete-exposures)and defensive skill players (0.93/1000 athlete-exposures), butthese findings do not correspond with our data.- Our datasuggest that high school offensive and defensive line athletesin these positions sustained the lowest-magnitude impactsbut also experienced the greatest number of impacts duringgames and practices (Figure 4). Impact frequencies of ourdata by position group show that the \5 offensive anddefensive line players (44% of all players) sustained 11 035impacts ( 57% of all hits), whereas the i 9 skill players ( 56% ofall players) sustained 8189 impacts {43% of all hits). Thehigher impact frequency endured by the linemen likely resultsfrom their involvement in every play, whereas the magnitudeof these impacts is likely lowest due to the short distancebetween the offensive and defensive line players and thesubsequent low impact speed. Furthermore, given the natureof high school football, in which the number of players islimited, many athietes in our sample played both offensiveand defensive positions, possibly increasing impact exposureand injury risk.

In assessing the calculated variables, the offensive anddefensive line players were estimated to have the largest headmasses. The role that head mass has in the concussion injurymechanism is not thoroughly understood, but finite elementanalysis models have been used to provide indirect evidencesupporting the relationship between increased head mass andgreater intracranial pressure after impact.2« Increasedpressure is thought to be related lo the degree of structuraldamage in the cerebral tissue after impact.-'' Conversely,smaller head mass is associated with higher linear androtational acceleration values,-^" which may make the athletemore susceptible lo injury. Notably, the offensive skillplayers enrolled in this investigation had the smallest headmasses and the largest linear accelerations and head ierksafter impact. The change in head velocity of these athleteswas 11.01 m/s, exceeding the 7.2 m/s in concussed profes-sional athletes reported by the NFL Mild Traumatic BrainInjury Committee. Furthermore, the duration of impactexperienced by the concussed NFL athletes was 15 millisec-onds.-*̂ - almost twice as long as the average we recorded.Finally, the use of impulse as a variable to quantify impaetshas been reported previously in 100 concussed rugbyathletes.^ I Using 2-dimensional analyses of the rugbyimpacts, the mean impulse of the impacts was estimated at29 N-s. A direct comparison with our data cannot be made.because we did not include any concussive impacts, but ournonconcussive impacts fell well below this value. A disparitybetween the NFL and rugby impact reconstructions and thedata collected here is evident. The reasons for thesedifferences, however, are not completely apparent. Estimat-

ing head mass from only anthropométrie variables may notnecessarily represent the player's efTective mass at the time ofimpact. The effective mass at impact is equivalent to thecombined mass of the head and body linked through thetensed neck musculature and acting as a single unit. Theadditional mass results in a lower resultant acceleration,although any neck motion at impact would lessen theinfluenceof the body mass. Additional dilVerencesin findingsinclude our method, which permitted the tracking of allimpacts (concussive and nonconcussive) during all sessions(practices and games). The method employed by the NFL^''and rugby-" investigations was limited by including onlyconcussive impacts captured on video that were laterreconstructed or evaluated.

Further investigations of head impacts resulting inconcussion are needed to establish the relevance of thecalculated variables to injury biomeehanics. The strongrelationships between linear acceleration and jerk, force,and impulse indicate the potential for redundancy whenexamining impact biomeehanics. The high correlationslikely result from the calculated variables being founded inthe reeorded linear acceleration. The relationship betweenthese variables and rotational acceleration is not as strong,indicating that rotational acceleration provides novelinformation when describing and quantifying an impact.Similarly, the weak relationships between impact durationand linear and rotational acceleration also suggest thecontribution of unique information.

Despite our unique findings, certain limitations arepresent. Although we collected a myriad of impacts, thelarge number of data points also may have influenced someof our significant findings. Further, when comparing theseresults with those at the collegiate level, one shouldconsider the different distributions of player positionsand the time frame at which those data were collected.Finally, future investigators should attempt to quantify theplaying time and skill level ofthe athletes. These variableswill likely play a role in the number of impacts and,potentially, the severity of impacts incurred.

This project is the fu'st report in an ongoing analysis ofhead impacts and concussions incurred by interscholasticfootball players. Our intent is to disseminate descriptivebiomechanieal information related to impacts sustainedduring a season of interscholastic football. The findingsfrom this investigation demonstrate clear dilTcrences betweenimpacts occurring on the interscholastic football field andthose occurring at the collegiate and professional levels.Through ongoing data collection and analyses, we believethat a better understanding of football-related impacts canbe provided for the benefit of sports medicine professionalsand so thai equipment manufacturers can take the appro-priate steps to improve safety equipment. Evaluating impactsto the head in sports other than football will prove morechallenging, but the HITS has been adapted for soccer,boxing, and ice hockey athietes. Our understanding ofimpact forces and biomeehanics in sports in which headgearis not commonly worn during practices and compelilions (eg.soccer, basketball) will, unfortunately, lag behind.

The ultimate goal of this research is to develop criteria thatwill allow the HITS or a similar system to be used asa sidchnediagnostic tool for concussive injuries. However, until moreinvestigations can be conducted to belter understand all thebiomechanieal components associated with sport concussion.


the HITS remains an instrument of research. In a limitedcapacity, the HITS may be used as "an extra set ofeyes" toidentity those athletes who have sustained significant blows tothe head and warrant follow-up evaluation by sports medicineprofessionals. The clinical e.xamination. coupled with lests ofconcussion-related symptoms, neurocognitive function, andpostural control, remains the most sensitive assessment of thiscomplex injury.^-

ACKNOWLEDGMENTS

This piiijcui could iioi have been completed without the strongsupport and a.ssistance from the following individuals: ScottHamilton and the llniiy Rockets foolball (cam; .1. R. Burr: JamiePing; John Storsvcd. HSD. ATC; Susan Mantel. MD; and RickGreenwald. PhD.

REFERENCES

1. Langlois JA. Rutland-Brown W, Wald MM. The epidemiology andimpuc't of IniiHTialit; brain injury: a brief overview. J l/fiiil TrtiunuiRchahil. :iH)(i:2l(5);.'í75-378.

2. Gu.skicwicz KM. McCrcii M. Miirshall SW. et ai. Cumulative eflcclsassiK'ialed wiih reeurreiU concussion in collcgiiitc looiball players: ihcNCAA Concussion Study. JAMA. 20Ü3;290(I9):2549-2555.

?. Guskicwiez KM. Weaver NL. Padua DA. Giirrett WE Jr. Epidemi-ology of concussion in collcgiiite and high school foolball players.Am J Spons Mi-(l. 2()O():2S(5):M3-6.S(}.

4. Pellman EJ. Powell JW. Viano DC. ct al. Concussion in professionallooiball: epidcmiological features of game injuries and review of (heliterature, part .1. Ni'uro.sw^cry. 2004:54(1 ):81 94.

5. Powell JW, Barber-Foss KD. Traumatic brain injury in high schoolüthicles. JAMA. 1999;2K2( 10):958-963.

(>. McCrea M. Hamineke T. Olsen G, Leo P, Guskiewicz K. Unreporledconcussion in high school looiball players: implications for preven-tion. Clin J Sport Mvil. 2OO4:I4(1|:I3 17.

7. Gerberich SG. Priest JD. Bocn JR. Sträub CP. Maxwell RE.Concussion incidences and severity in secondary school varsitylootball players. Am J Public Heiilih. 1983;73(12):I37O-I375.

8. Guskiewicz KM. Bruce SL. Cantu RC, et al. National AthleticTrainers' Association position statement: management of sporl-rclaied concussion. J Aihl Train. 20O4:39(.'():280-297.

'). McCrea M. Kelly JP, Klugc J. Ackley B. Randolph C. Standardizedassessment oi concussion in football players. Neurology. 1997;48(3):586 588.

10. Riemann BL, Guskiewicz KM. EfFects of mild head injury onpostural stability as measured through clinical balance testing. J AihlTrain. 20()0;.15(l):19-25.

11. CYisco JJ. Clui JJ. Greenwald RM. An algorithm lor esiinniiingac-œlcralion magnitude and impact location using multiple iionorthog-onal single-axis accelerometers. ./ ßiornnh Big. 2()04:l26(6):849 S54.

12. t)uma SM. Mimoogian SJ, Bussone WR, et al. Analysis of real-timehead accelerations in collégiale football players. Clin J Sport Med.2005:I5(l):3-8.

13. Brolinsori PG. Manoogian S. McNeely D. Goforth M. Greenwald R.Duma S. Analysis of linear head accelerations from collegiate lootballimpacts. Curr Sporls Mvil Rep. 2lH)6:5( 1 ):23-28.

14. Mihalik JP. Bell DR. Marshall SW. Guskiewic/ KM. Measurement ofhead impacts in collegiate football players: an investigation of positionaland cvcnt-iypc dilTerences. NíWimirí-cry. 2007:61 (6): 122̂ ) 1235.

15. Schncbel B. Ciwin JT. Anderson S, Gatlin R. In vivo study of headimpacts in football: a comparison of Niitional Collegiate AthleticAssociation Division I versus high school impacts. Nairosiirgcrr.2007:6()(3):49()^')5.

16. Kleiven S. Infltience of impact direction on the human head in predictionof subdural hematoma. J Niiiroirciumu. 2{K)3;2()(4):365 379.

17. Zatsiorsky VM. Kinetics of Human Motion. Champaign, IL: HumanKinetics: 2002:577.

18. Barlh JT. Freeman JR. Broshek DK. Varney RN. Acceleration-deceleration sport-related concussitMi: ihc graviiy of it all. ./ .•l//i/Train. 2(Hll:.16(3):253-256.

19. Glier R. High schools try teamwork to put trainers on sile. hUp://www.nylimes.com/2007/12/27/rashion/27ritness.htnil. Published De-cember 27, 2007. Accessed December 21. 2008.

20. Cantu RC, Voy R. Second impact syndrome: a risk in any contadsport. Physician Spori.snwtl. I995:23(fi):27-.34.

21. Swarlz EE. Floyd RT. Ccndoma M. Cervical spine functionalanatomy and the biomechanics of injury due lo compressive loading.J Allil Train. 2OO5:4O(3):I55-161.

22. McCrea M, Guskiewic/ KM. Marshall SW. et al. Acute ciTects andrecovery lime following concussion in collegiate football players: iheNCAA Concussion Study. JA.MA. 20O3;29O( I9):2556 2563.

23. Reid SE, Tarkington JA, Fpsiein HM. O'Dca TJ. Brain tolerance toimpact in football. Siirg Gynccol Oh.-iU-i. 197I;I33(6):929 9.16.

24. Naunhcim RS. Standeven J. Richter C. Lewis LM, Comparison ofimpact data in hockey, football, and soccer. J Traumn. 2()l)(l:48(5):938-941.

25. Guskicwiez KM. Miiialik JP, Shankar V. et al. Measurement of headimpacts in collegiate football players: relationship between headimpact biomechanics and aeute clinical ouicome after concussion.Neurosurgi'ry. 2007;6l(6):1244 1252.

26. Pellman EJ. Viano DC. Tucker AM. Casson IR. Waeckerle JF.Concussion in professional football: reconstruction of game imjiactsand injuries. Ncurosurgery. 2003:35{4):7yy SI4.

27. Greenwald RM. (Jwin JT. Chu JJ. Crisco JJ. Head impact severitymeasures for evaluating mild traumatic brain injury risk exposure.NeuriKsurgery. 200S;62(4):789 798.

28. Kleiven S. von Hoist H. Consequences of head size following trauma10 Ihe human head. J Blonmh. 2002:35(2):153 160.

29. Lissner HR. Lebow M, Evans FG. Experimental studies on therelation between acceleration and intracranial pressure changes Inman. Siirfi Gym-col Obslcl- l%0:111:329 338.

30. Viano DC. Cas.son IR. Pellman LJ. ei al. Concussion in professionalfootball: comparison with boxing head impacts, part 10. Ni'iiro.siir-gcry. 2OO5;57(6):II54-1I72.

31. Mclntosh AS, McCrory P. Comerford J. The dynamics of concussivehead impacts in rugby and Australian rules football. Med Sti SportsExerc. 2(IOO:32( 12): 1980-1984.

32. BroglioSP. Macciocchi SN, Ferrara MS. Sensitivity of the concussionassessment battery. /Vi-i/mviH-tjcrr. 2{H)7;6O(6):tO5O 1057.

Steven P. Bw^liu. PhD. A TC. fontrihuted to conception ami design; acc/ui.'iirion ami anaiysi.s ami infcrpretation ofthe liata: ami (¡rafting,crifkal révision, anüßiuil approval of llic arlicic. Jacob J. So.mojf PliD, conirihiileil lo anal) sis and inierprelalitin oj the ikiUi ami i/rafiin^,critical revision, and fuial approval of the article. SuntiHoon Shin. MS. contrihuled to anaiysi.s and inierprelalion of the dala and (Iraflini;ami final approval of lite article. Xiimin^ He, PhD. contrihuleil to analvsi.'i ami inicrprcialion ofthe ilaia ami draftini;. ailical revision, amifinal approval of the arliele. Christopher Alcaraz. MD, and JerracI Zimmerman, MD, lontrilmteil lo conception ami desii^n: aequisition of¡he ilata; and drafting, crilieal revision, and final approval of the article.

.4iiäre.ss eorrespondeme io Steven P. Broglio. PhD. ATC. Department of Kinesiology and Community Health, 906 S Goodwin Avenue,I'rhana. IL 6IH0I. Address e-mail to hrogli<>«''uiiu\edu.


head impacts during high school football: a biomechanical … › 2010 › 09 › ... ·...

Documents