Eric D. Zemper, Ph.D.
Exercise Research Associates,
Department of Physical Medicine
& Rehabilitation,
Abstract
Objective:
To prospectively measure the relative risk of cerebral concussion among
those with a history of concussion compared with those having no previous
concussions, utilizing a population of high school and college football
players.
Design: A representative
national sample of high school and college football players was followed for
two football seasons over a two-year period (1997-1998) as part of a national
football injury surveillance project.
There were a total of 15,304 player-seasons and over 1 million
athlete-exposures to the possibility of injury in practices and games; 975
(6.4%) of the player-seasons had a history of concussion in the previous five
years.
Results:
There were 572 concussions recorded, 161 among those with a previous
history (16.5%) and 411 among those with no history (2.9%). Relative risk for individuals with a previous
history of concussion is 5.8 times greater than for individuals with no history
(95% CI: 4.8, 6.8).
Conclusion:
This large prospective cohort study indicates the risk of sustaining a
cerebral concussion is nearly six times greater for individuals with a history
of previous concussion than for individuals with no such history.
Key Words:
Brain concussion, TBI (traumatic brain injury), relative risk, athletic
injuries, football
INTRODUCTION
There has been a common presumption for many years that an individual who has incurred a cerebral concussion is at greater risk for sustaining another such injury. However, the relative risk for sustaining a concussion (the ratio of the risk among those with a history of concussion to the risk among those with no history of concussion) has not been adequately defined.1,2 There have been occasional reports that mention relative risks ranging from three3,4 to six,5 but a commonly quoted figure is from a study by Gerberich et al.,6 which found a relative risk of four times higher. However, their study was a retrospective mail survey of high school football players and coaches, which presented methodological problems, primarily with regard to accuracy of recall of the number and severity of concussive incidents. In addition, while other parts of this paper dealt with non-loss of consciousness concussive injuries, the calculation of relative risk considered only head injuries severe enough to cause loss of consciousness. As noted by others,7-10 concussion covers a spectrum of signs and symptoms that do not necessarily include loss of consciousness.
The American Academy of
Neurology defines a concussion as a traumatically induced alteration in mental
status (e.g., confusion, amnesia) that may or may not involve loss of consciousness.11 A more extensive definition was proposed in
2001 by the Concussion in Sport Group, a committee formed at the first
International Symposium on Concussion in Sport, which met in Vienna, Austria,
and was organized by the International Hockey Federation, the Federation Internationale de Football Association, and the
International Olympic Committee. This group defines concussion as “a complex
pathophysiological process affecting the brain, induced by traumatic
biomechanical forces,” and then lists five common features that may be used in
defining the nature of a concussive head injury, including that it may or may
not involve loss of consciousness.12
These definitions were developed in an attempt to be more inclusive of
what have been recognized as signs and symptoms of milder concussion that were
not included in the long-standing definition of cerebral concussion published
over 35 years ago by the committee on head injury nomenclature of the Congress
of Neurological Surgeons.13
With regard to concussion in a sport setting, McCrory et al. have questioned whether there truly is an increased risk of concussion for those with a previous history of concussion.2 In any collision sport they presume risk of concussion to be directly proportional to the amount of playing time, and the more time exposed, the more chance for a concussion. Therefore, they propose that the risk of repeat injury may only reflect the cumulative amount of exposure.
There is a need for studies addressing this issue of the existence and magnitude of the relative risk for those with a history of concussion. A prospective study of the general population to define the relative risk of a second concussion would be challenging because the incidence of this injury is small enough that a large population must be followed for a considerable period of time. In addition, many milder concussions go unreported to and untreated by medical personnel. These difficulties can be avoided by studying a special population with a relatively high risk of concussion compared to the general population, and who are under constant observation by medical personnel during their participation in this high risk behavior: high school and college football players. While within the brain the pathophysiology of concussion and thus the intrinsic risk of concussion in these individuals is the same as for the general population, this sub-population regularly participates in an activity with a higher extrinsic risk for sustaining this injury, providing a convenient “laboratory” for studying the specific question of the intrinsic risk of a second concussion.14
As part of a general study of football injuries, data were collected over two fall football seasons (1997-1998) from a large national sample of high school and college football players. For the purpose of further defining the relative risk of a second concussion, specific items of information were included in the data collection process that made it possible to do detailed studies of head injuries in this population. From these data it is possible to calculate the relative risk of concussion among individuals with a previous history compared with those with no previous history of concussion.
METHODS
Football injury data used
for this study were collected by the Athletic Injury Monitoring System (AIMS)
from a national sample of 56 high schools and 42 colleges for the 1997 season,
and 33 high schools and 43 colleges during the 1998 season. The sampling frame for this study was all
high schools and colleges in the
The high school and college samples for this study were distributed by four geographic regions, and the college sample also was distributed by size of program (National Collegiate Athletic Association and National Association of Intercollegiate Athletics divisions). Based on geographic region and size of program of NCAA and NAIA members sponsoring football during the period of this study, a c2 test of goodness-of-fit showed no significant difference between the sample distribution and the actual distribution (c2=4.701, a=0.05, df=11, CV=19.675). This indicates the college sample was representative of the entire country by region and size of program, and therefore the results presented here for the collegiate level can reasonably be projected to the country as a whole.
The nature of the high school sample precludes as rigorously representative a sampling as possible with the colleges. The study protocol required a medically trained person on site (specifically, a certified athletic trainer), and only a relatively small proportion (~15%) of the more than 13,000 high schools that sponsor football teams in the country have athletic trainers on staff (compared with essentially all colleges). Very few of the athletic trainers who are working at the high school level are in smaller schools. Under these conditions the most reasonable approach was to sample a large number of schools (albeit a small proportion of the total), keeping the sample representative by geographic region. The high schools in this sample were statistically representative by geographic region (c2=1.570, a=0.05, df=3, CV=7.815) and, with over 7000 total player-seasons (one player participating for one football season) over the two seasons, the total high school sample was considered adequate for the purposes of this study.
At
the beginning of each season certified athletic trainers at each participating
school completed a form indicating how many players had a history of concussion
anytime during the previous five years.
This information was based on the known medical history of each player
while at the high school or college, and on detailed medical histories obtained
for each player when they enter a collegiate program. Individuals in the high
school and college programs in this study are under continuous medical
surveillance while participating in football or any other sport, and their
medical records are kept current by the certified athletic trainers from the
time they enter the school sports program. Combined with data collected
throughout each season from individual injury reports on concussions, which
included information on whether or not the concussed player had a history of
concussion, these data allow calculation of relative risk and 95% confidence
intervals.
Utilizing simple, easy-to-use forms, data on exposure to the possibility of injury in practices and games, and on any injury that kept a player from participating for one day or more (including any head injury whether on not time loss was involved), were returned on a weekly basis throughout each season by the athletic trainers at each participating school. As data arrived each week, it was logged in and screened for completeness and consistency by the author and subsequently entered into the computer database. A common problem with surveillance systems of this sort is incomplete data submission from the field. To prevent this problem in this study, the author maintained frequent mail and telephone contact with the athletic trainers, obtaining corrections of any inconsistencies in the reported data and reminding them to submit missing data. Through this process of “preventive maintenance” the response rate for data submission was extremely high. For the college sample 99.3% of the weekly reports were submitted, and for the high school sample the response rate was 98.5% over the two seasons.
This
study used the definition of concussion (noted previously) and the grading
system for severity of concussion developed by the
The AIMS injury surveillance project, from which the data used in this report were derived, received approval from the ExRA human subjects review committee as exempt research involving collection of existing publicly available data, documents or records, with the information recorded in such a manner that subjects cannot be identified directly or through identifiers linked to the subjects.
RESULTS
General. During
the two seasons there were 572 reported concussions in this sample (240 HS and
332 college). This injury comprised about 10% of the
total reported injuries (Table 1).
Approximately 4.1% of the sample population incurred concussions during
this study. With an estimated 1.08
million participants in organized football in this country, this projects a
total of approximately 44,000 concussions occurring each year for football
players in this country. This is
considerably less than the estimate of 250,000 commonly seen in the literature based
on the Gerberich study,6
but it does agree with recent, more refined estimates.20, 21
Table 1 Summary
of Exposure and Injury Rates for a National Sample of High School and College
Football Players, 1997-1998 Seasons |
|||||
|
# Player- Seasonsa |
# Athlete- Exposuresb |
Total Injuries /1,000 A-E |
Total # Concussions |
Concussions/
1,000 A-E |
High School |
7197 |
475,589 |
4.93 |
240 |
0.50 |
College |
8107 |
582,659 |
5.91 |
332 |
0.57 |
Combined HS + Coll. |
15,304 |
1,058,248 |
5.49 |
572 |
0.54 |
a A Player-Season is one
player participating in one football season. The total number of individual players in this study will be less than
this number, because many players were participating in both seasons. b An Athlete-Exposure (A-E) is one player participating in one game or
practice, where he is exposed to the possibility of being injured. |
Injury rates are calculated by AIMS based on total athlete exposures, utilizing the formula:
((#
injuries reported) ÷ (# exposures)) x 1000 = injury
rate per 1000 Athlete-Exposures
(An
Athlete-Exposure is one player participating in one game or practice where he
is exposed to the possibility of being injured.) The concussion rates at the high school and
college levels were quite similar, with a rate of 0.50 concussions per 1000
Athlete-Exposures (A-E) for high school players and 0.57 per 1000 A-E for
college players. The combined rate is
0.54 concussions per 1000 A-E. This is
equivalent to one concussion in every 1850 times a player participates in a
game or practice. For a team of 100
players (generating 100 A-E in every daily practice and every game, assuming
they all played in the game), this would be one concussion in every 18-19 days
of activity. A less accurate and less
useful “rate” still commonly seen in sports medicine literature is the number
of injuries per 100 players. For
concussions in this sample, there were 3.33 concussions per 100 high school
players per season and 4.10 concussions per 100 college players per
season. While this appears to indicate
the rate for college players is over 20 percent higher, because college players
typically have more exposures (practices and games) during a season, the data
from this study show the more accurate rates based on athlete-exposures for
high school and college players are in reality quite similar. This is one illustration of how attempting to
make comparisons using injuries per 100 players as a rate can be misleading.22
Relative Risk. From the information collected at the beginning of each season on how many players had a history of concussion, plus data from the individual injury forms indicating whether or not a concussion being reported occurred in a player with a previous concussion, it is possible to calculate relative risk. A summary of the data used in these calculations is presented in Table 2.
Table 2 Summary of Concussion History and
Concussion Experience of a National Sample of High School and College
Football Players, 1997-1998 Seasons |
|||||||
|
# Team- Seasons |
# Player- Seasons |
# w/ Hx |
# w/ no Hx |
Total # Conc. |
# Conc. w/ Hx |
# Conc. w/ no Hx |
High School |
89 |
7197 |
296 |
6901 |
240 |
53 |
187 |
College |
85 |
8107 |
679 |
7428 |
332 |
108 |
224 |
Combined HS + Coll. |
174 |
15,304 |
975 |
14,329 |
572 |
161 |
411 |
At the high school level the sample for the calculation of relative risk consisted of 7197 player-seasons. Of those, 296 player-seasons (4.1%) had a history of concussion sometime during the previous five years. There were a total of 240 concussions recorded in this sample, 53 among players with a previous history, and 187 in players with no previous history. Using this information, the relative risk is calculated as follows:
Relative Risk = (53/296) = 0.1791 = 6.6 (95% CI = 5.0, 8.8)
(High School) (187/6901) 0.0271
Among those high school players with a previous history of concussion, 18% sustained a new concussion, while among those with no previous history less than 3% sustainedthis injury. At the high school level the relative risk of sustaining a concussion during a given season among individuals with a previous history of concussion is over six times greater than for those with no previous history of concussion.
At the college level the sample for this calculation consisted of 8107 player-seasons. A total of 679 player-seasons from this sample (8.4%) had sustained a concussion within the previous five years. Out of the total of 332 concussions recorded during the two seasons for this sample, 108 occurred in players with a previous history and 224 occurred in players with no previous history. The relative risk therefore is calculated as follows:
Relative Risk = (108/679) = 0.1591 = 5.3 (95% CI = 4.3, 6.6)
(College) (224/7428) 0.0302
Sixteen percent of college players with a previous history of concussion sustained a new concussion, while 3% of those with no previous history sustained concussions. At the college level the relative risk of sustaining a concussion during a given season among those individuals with a history of concussion during the previous five years is more than five times greater than for those with no previous history of concussion.
Combining the high school and college data results in the following calculation of risk:
Relative Risk = (161/975) = 0.1651 = 5.8 (95% CI = 4.8, 6.8)
(411/14329) 0.0287
Therefore, in this large national sample of high school and college football players over two seasons, the risk of sustaining a concussion is 5.8 times greater for individuals with a history of concussion during the previous five years than for those with no previous history of concussion. From a different perspective, approximately 1 in 35 individuals with no previous history will sustain a concussion during a given season, while approximately 1 in 6 with a previous history will sustain a new concussion.
Table 3 Grade of Cerebral Concussion by History of
Concussion |
|||
Grade * |
No History |
History |
Total |
1 |
268 (65.2%) |
101 (62.7%) |
369 (64.5%) |
2 |
132 (32.1%) |
52 (32.3%) |
184 (32.2%) |
3 |
11 (2.7%) |
8 (5.0%) |
19 (3.3%) |
(History vs No History comparison: c2=3.338, a=0.05,
df=2, CV=5.991) * |
As
shown in Table 3, the distribution of concussions by grade is essentially the same
between those with no history and those with a history of concussion (c2=3.338, a=0.05, df=2,
CV=5.991). Therefore, a history of
previous concussion does not appear to impact the severity of a new
concussion. Since Grades 1 and 2 do not
involve loss of consciousness, these data also show that over 95% of the
concussions in this sample did not involve loss of consciousness. To investigate the possibility of player
position being associated with greater risk, an analysis of concussions by
position played was performed. It showed
there were no differences by position between those with a history and those
with no previous history of concussion, with the exception of offensive tackles
with a history having significantly fewer than expected concussions and
quarterbacks with a history having more than expected (based on the
distribution by player position among those with no history of concussion).
Discussion
Cerebral concussion is a common problem that in the past has received little attention from those in the general population or from athletes,23,24 despite growing concerns in the medical community. Among the concerns is the presumed increased risk for those with a previous history of concussion. However, the existence and magnitude of this risk has not been adequately defined. The data presented here do indicate there is an increased risk, and the risk is about 5.8 times higher for those with a previous concussion in this sample population of football players. Since this sample included only young males, in the strictest sense a statistical generalization is not possible to young females or older adults. However, regardless of the extrinsic risk of the activity that causes a concussion (playing football or falling off a chair), within the brain the pathophysiological mechanism of concussion presumably will be the same for everyone, including young females and older individuals. Therefore, it should be reasonable to make the logical projection that this increased intrinsic risk of a second concussion will be similar for the general population as well. These results replicate and extend previously reported results from data collected by this project from a smaller national sample of college football teams during the 1988-90 seasons,5 which showed a relative risk of 5.95.
McCrory et al.2 have suggested that there may not be an actual increased risk, but it is rather a reflection of cumulative time in risky activity (playing in a contact sport). This can be addressed by looking at the relative risk for high school and collegiate football players. If their suggestion is correct, it would be expected that collegiate players, who have been involved in the sport for a longer period, will have a higher relative risk than high school players. Since the calculated relative risk for the high school players is greater than for collegiate players, these data do not support their premise.
Currently there is a great deal of discussion among sports medicine practitioners regarding when it is appropriate to return an individual to participation in a sport following a concussion, with no complete agreement on specific criteria.7-11 The results of this study, indicating there is indeed a greater risk of repeat concussions among those who have incurred a concussion, may have an impact on such on-the-field clinical decisions. At the very least, these results should highlight the need for team physicians and athletic trainers to reconsider the common tendency to send a player back in as soon as they can “see straight.”
This study did not collect information on the number of previous concussions incurred by those with a history of concussion, or when they occurred (other than within the previous five years). Therefore, it is not possible to determine whether there is a dose-response relationship between number of previous concussions, or the time since the previous concussion, and the risk of a new concussion. These are factors that should be investigated in the future.
A study of this nature is dependent
upon the accuracy of the reported histories of previous concussions. While the individuals in college programs are
under continuous medical surveillance while participating in football, and
their medical records are presumably accurate, the reported histories of
concussion from the high school level could be open to more question. Realistically,
at the high school level the trainers are reporting only what is in the
individual's medical record while attending that school and participating in
sports; the trainers were not asked to get self-reports from the athletes. If some of these athletes had actual
histories of non-sport concussion or concussion during the year or two prior to
entering high school that did not get recorded, they would be counted among
those with no history. If such an
individual incurred a concussion and it was tallied among those with no history
instead of among those with a history, the result would be to artificially
raise the numbers of concussions among those with no history and reduce the
numbers among those with a previous history.
This results in a calculated relative risk that
is lower than it should be. Therefore,
the results presented here of a relative risk of 5.8 should be considered a
conservative number that is a minimum relative risk.
Conclusion
The data presented here from a large prospective cohort study, utilizing a sub-population of individuals involved in an activity with high extrinsic risk for concussion (football players), indicate there is an increased intrinsic risk of a second concussion among those with history of concussion during the previous five years. These individuals are 5.8 times more likely to suffer a new concussion than those with no history of concussion.
Funding source: The injury surveillance project, from which the data used in this study were drawn, was funded by Riddell Inc.
REFERENCES
1) Salcido R, Costich JF. Recurrent traumatic brain injury. Brain Inj 1992;6:293-298.
2) McCrory P, Johnston KM, Mohtadi NG, Meeuwisse W. Evidence-based review of sport-related concussion: basic science. Clin J Sport Med 2001;11:160-165
3)
Annegers JF, Grabow JD,
4)
Guskiewicz KM,
Weaver NL,
5) Zemper ED. Analysis of cerebral concussion frequency with the most commonly used models of football helmets. J Athl Training 1994; 29:44-50.
6) Gerberich SG, Priest JD, Boen JR, Straub CP, Maxwell RE. Concussion incidence and severity in secondary school varsity football players. Am J Public Health 1983; 73:1370-1375.
7) Cantu RC. Cerebral concussion in sport: management and prevention. Sports Med 1992: 14:64-74.
8) Collins MW, Lovell MR, McKeag DB. Current issues in managing sports-related concussion. JAMA 1999; 282:2283-2285.
9) Kelly JP, Rosenburg JH. Diagnosis and management of concussion in sports. Neurology 1997; 48:575-580.
10) Wojtys EM, Hovda D, Landry G, et al. Current concepts; concussion in sports. Am J Sports Med 1999; 27:676-687.
11)
12) Aubry M, Cantu R, Dvorak J, et al. Summary and agreement statement of the first International Conference on Concussion in Sport, Vienna 2001. Br J Sports Med 2002; 36:6-10.
13) Committee on Head Injury Nomenclature of the Congress of Neurological Surgeons. Glossary of head injury including some definitions of injury to the cervical spine. Clin Neurosurg 1966; 12:386-394.
14) Ruchinskas RA, Francis, JP, Barth, JT. Mild head injury in sports. Applied Neuropsychology 1997; 4(1):43-49.
15) Thompson N, Halpern B, Curl WW, Andrews JR, Hunter SC, McLeod WD. High school football injuries: evaluation. Am J Sp Med 1987; 15:117-124.
16) Zemper ED. Injury rates in a national sample of college football teams: a two-year prospective study. Physician Sportsmed 1989; 17:100-113.
17) Zemper ED, Pieter W. Cerebral concussion in taekwondo athletes. In: Hoerner EF, ed. Head
and Neck Injuries in Sports, ASTM STP 1229
18) Zemper ED. A four-year prospective study of preventive knee braces and
MCL injuries in a national sample of college football players. In: Hoerner EF, ed. Safety
in American Football
19) Zemper ED. A two-year prospective study of prophylactic knee braces in a national sample of college football players. Sports Training, Medicine and Rehabilitation 1990;1:287-296.
20) Powell JW, Barber-Foss KD. Traumatic brain injury in high school athletes. JAMA 1999; 282:958-963.
21) Maroon JC, Lovell MR, Norwig J, Podell K, Powell JW, Hartl R. Cerebral concussion in athletes: evaluation and neuropsychological testing. Neurosurgery 2000; 47(3):659-669.
22) Zemper ED. Epidemiology of athletic injuries. In: McKeag
DB, Hough D. Primary Care Sports Medicine
23) Goldstein M. Traumatic brain injury: a silent epidemic [editorial]. Ann Neurol 1990; 27:327.
24) Kelly JP. Traumatic brain injury and concussion in sports. JAMA 1999; 282:989-991.