The Epidemiology of Traumatic Brain Injury: A Statewide Study of Hospitalized Maryland Drivers

Abstract

Most previous studies of the incidence of traumatic brain injury (TBI) are 15–20 years old, and none have specifically addressed brain injuries attributable to motor vehicle crashes. This study is based on a statewide surveillance of all hospitalized drivers in Maryland for the three-year period 1994–1996. The incidence of brain injury has been characterized for this population in terms of driver, vehicle, and crash-related factors. In addition, those characteristics most predictive of serious brain injury have been identified. Findings from these analyses reveal that predictors of serious injuries are somewhat different than those for the total incidence of TBI.

Traumatic brain injury accounts for one third of all injury deaths in the United States (Sosin, 1989). It is estimated that each year in the U.S., for every person who dies of a head injury, five are admitted to hospitals, and an additional 26 seek outpatient medical treatment (Kraus, 1993). Thus, in 1990 there were approximately 75,000 fatalities, 366,000 patients hospitalized, and 1,975,000 who required medical attention for a head injury. Estimates of TBI incidence, severity, and cost reflect the fact that these injuries cause enormous losses to individuals, their families, and society. In a previous Maryland report based on hospital discharge records, MacKenzie et al. reported an incidence rate of 132/100,000 cases (Mackenzie, 1989).

The death rate from brain injuries declined 22% from 24.6 per 100,000 U.S. residents in 1979 to 19.3 per 100,000 in 1992 (Sosin, Sniezek, Waxweiler, 1995). During this period, a 25% decline in motor vehicle-related fatality rates was counterbalanced by a 13% increase in firearm-related brain injury fatalities, which occurred between 1984 and 1992. Although there is some information on fatalities due to brain injuries, data on nonfatal injuries are relatively sparse, and many of these studies were conducted more than 15 years ago. However, based on data from the National Hospital Discharge survey, annual rates of hospitalization related to TBI decreased approximately 48% from 1980 through 1995 (NCHS, 1997).

There have been 10 publications, mostly between 1980 and 1991, related to the overall incidence of head injury in the United States (Andersen, Kalsbeck, Hartwell, 1980; Annegers et al., 1980; Fife et al., 1986; Jagger et al., 1984; Kalsbeck et al., 1981; Klauber, Marshall, et al., 1978; Klauber, Barrett-Connor et al., 1981; Whitman et al., 1984; Kraus et al., 1984; MacKenzie et al., 1989). Seven of these reports show transport-related events as the single largest cause of new cases of brain injury, the two exceptions being in urban areas, where assaults and firearms constituted the single largest cause. Most of these studies have reported on the percentage distribution of injuries by major external cause (e.g. transport, falls, etc); overall, approximately half of head injuries are associated with “transport.” Four of the incidence studies describe motor vehicle-related brain injuries by status (i.e. occupant, motorcyclist, pedestrian, or bicyclist). Of those, three of the four show that about 75% of patients with head injuries were occupants of cars or trucks (Annegers, 1980; Jagger, 1984). No reports, however have focused on the characteristics of the crash and/or the driver, as they relate to the nature and severity of the brain injury. In only one report was the external cause verified using police reports (Kraus et al., 1984); however, characteristics of the crashes/vehicles were not included in that paper.

Other studies, not population-based, have related crash characteristics to motor vehicle-related brain injuries. In an analysis of injury patterns associated with direction of impact among drivers admitted to trauma centers, no statistically significant difference was noted between brain injuries in frontal vs. left lateral crashes (Dischinger, 1993). Such differences were noted, however, in an in-depth crash reconstruction study conducted on 145 trauma patients (Siegel et al., 1994; 1994). Despite no difference in the overall incidence of brain injuries in lateral vs. frontal collisions, those with severe brain injuries in that study were more likely to have been involved in lateral crashes; this was especially so for those wearing seatbelts. However, this difference was primarily detected using admission Glasgow Coma Scale (GCS) scores (Teasdale and Gennett, 1974), which are not available from the hospital discharge data. Severe brain injuries were associated with the side window frame and A pillar intrusion contacts with the head. Many of the injuries, both in frontal and lateral collisions, were incurred at changes in velocity (delta v) of less than 30 mph.

In a more recent analysis of crash reconstruction data on 200 trauma patients, it was noted that in frontal crashes, airbags with or without seatbelts reduced the incidence of severe brain injury (Loo, Siegel, Dischinger et al.,1996). When airbag protection was present, it prevented brain injuries caused by impact contacts and reduced the incidence of these injuries resulting from vehicular compartment intrusions. However, in lateral collisions, belt use did not protect against brain injuries.

The aforementioned crash reconstruction studies provide valuable insight into the causative factors involved in motor vehicle-related brain injuries. These studies were conducted in conjunction with trauma centers; thus, they are representative of the subgroup of drivers with serious multisystem injuries admitted to these specialized hospitals. Such studies are important, as they provide details on the forces, contact points, and intrusions which are not available from routine data sources such as police reports. In addition, these in-depth studies, with prospective data collection while the patient is hospitalized, provide more clinical accuracy and detail than that available in hospital discharge records. However, these studies are not population-based, and it is therefore not possible to determine the relative importance of various factors in the determination of risk of TBI. The purpose of the current study is to examine the epidemiology of brain injuries in a population of all drivers admitted to Maryland hospitals during the period 1994–1996, comparing characteristics of the drivers, their crashes and their injuries.

METHODS

SOURCE OF THE DATA

In order to examine injuries for all drivers hospitalized in the state of Maryland during the period 1994–1996, data were obtained from two sources, hospital discharge records from the Health Services Cost Review Commission (HSCRC), and police reports from the Maryland Automated Accident Reporting System (MAARS). These data were linked, using probabilistic linkage techniques, to obtain data on all drivers of cars, trucks, or vans admitted to acute care hospitals. The uniform hospital discharge abstract data were obtained from all 52 non-federal acute care hospitals in the state. This estimate, by definition, excludes outpatient cases and deaths which occurred either at the scene, in transport, or in an emergency department.

DATA LINKAGE

All patients with an ICD-9 code between 800.0–959.0 were selected from the hospital discharge records; there were 131,191 eligible cases. From the police reports, all drivers (N=545,105) were selected. Passengers were not included, as one of the key linkage variables, date of birth, is not available from the police report. The two databases were then linked, using probabilistic linkage techniques (Jaro, 1989; Jaro, 1995). This technique is based on a computation of odds ratios for the variables in question; thus, not all variables hold the same weight with regard to the probability of a match. Following linkage, the resulting database included data on 7,750 hospitalized drivers of automobiles, station wagons, limousines, light trucks, vans and recreational vehicles.

DEFINITION OF BRAIN INJURY

Crash and driver characteristics were compared for drivers hospitalized with and without traumatic brain injury (TBI). In addition, among drivers with TBI, those with serious vs. minor/moderate injuries were compared. Traumatic brain injuries were, using CDC criteria (Thurman DJ, Sniezek JE, et al., 1995) defined as those with fracture of the vault or base of the skull, other and unqualified and multiple fractures of the skull, and intracranial injury, including concussion, contusion, laceration, and hemorrhage (ICD-9 codes of 800.0–801.9, 803.0–804,9, and 850.0–854.1). ICD-9 codes are available from the hospital discharge data; using ICDMAP-90, these codes were “translated” to obtain drivers’ Abbreviated Injury Scale (AIS), Maximum AIS (MAIS) and Injury Severity Score (ISS) (Mackenzie, 1997). “Serious” TBI injuries were defined as those with a MAIS of 4 or 5. If a patient had more than one brain injury, the injury with the higher AIS score was selected.

DEFINITION CRASH-RELATED FACTORS

Data on characteristics of the crashes, seatbelt use, etc., were obtained from the police report. The “first harmful event” from the police report was used to describe the actual crash circumstance. Point of impact was derived from the coded diagram, as shown below in Figure 1. Frontal crashes were represented by codes 16, 01, 02, and 03, while lateral collisions included codes 04, 05, 06, 07, 15, 14, 13, and 12.

Figure 1.

Police Report Vehicle Diagram

ANALYSES

Two sets of analyses were conducted: (1) to determine the incidence of TBI in the total population of hospitalized drivers, and (2) to determine, among drivers with TBI, the proportion of those with serious brain injuries. For each set of analyses, comparisons were based on Pearson’s chi-square statistic for categorical variables, and Wilcoxon’s rank-sum test for non-normal, unbounded continuous data. A probability level below 0.05 for a two-sided test was considered statistically significant. In addition, multiple logistic regression analyses were conducted to determine the relative importance of the various driver and crash characteristics related to the incidence and severity of brain injury. The Hosmer and Lemeshow goodness-of-fit test was applied to assess the fit of each regression model.

RESULTS

Overall, there were 7,750 drivers of cars, trucks, or vans hospitalized during the three-year period; 2,925 drivers (37.8%) had a TBI. The estimated annual incidence rate of TBI serious enough to require hospitalization of drivers in Maryland was 74/100,000 population.

The majority of drivers (59.8%) were men, and more than half (58.0%) were less than 40 years of age. Two thirds were culpable for their crash, and 16% had been drinking or using drugs, according to the police. Two-thirds were wearing their seatbelts at the time of the collision, and only 4.9% had a deployed airbag. The median ISS score was 5, with a median hospital stay of 2 days, and median hospital charges of $3,482. Drivers incurring hospital costs of $100,000 or more, while representing only 1% of the study population, accounted for 16% of the total charges. Most patients (87.4%) returned home following their discharge from the hospital; 2.2% died of their injuries.

As shown in Table 1, drivers with brain injuries, as compared to those without brain injuries, were significantly more likely to be male, drinking/using drugs, at fault, and unbelted. In addition, drivers in the TBI group were significantly more likely to be younger than 40 years of age (65.1% vs. 53.4%, p<0.001), to require rehabilitation services (12.1% vs. 9.5%, p<0.001), and to have died (3.4% vs. 1.4%, p<0.001).

Table 1.

Characteristics of Drivers^*

	TBI		Non-TBI
	(N=2,925)		(N=4,825)
	n	(%)	n	(%)	p
Male	1,892	64.7	2,744	56.9	<0.001
Age
<=19 yrs	443	15.2	459	9.2
20–39	1,460	49.9	2,129	44.2
40–59	633	21.6	1,265	26.2
60+	389	13.3	967	20.1	<0.001
Drinking/Drugged	568	21.2	575	12.8	<0.001
Culpable for Crash	1,974	73.6	2,773	64.0	<0.001
Belted	1,495	59.1	3,200	74.4	<0.001
Air Bag Deployed	130	4.4	247	5.1	NS
Hospital Disposition
Home	2,471	84.5	4,300	89.1
Died	100	3.4	68	1.4
Rehab./Other	354	12.1	457	9.5	<0.001

^*Totals and percentages may vary due to missing values.

Of the 7750 collisions (data not shown), the majority (61%) occurred during the daytime (6a.m.–6p.m.), and approximately one third were on weekends (6p.m. Friday to 6a.m. Monday). Most (59.8%) occurred in clear/cloudy weather. Two or more vehicles were involved in 65.6% of crashes, while 32.5% involved collisions with fixed objects or parked cars. The primary point of impact was frontal/offset frontal (66.1%), followed by lateral impacts (26.1%). The predominant types of vehicles involved were cars, station wagons, and limousines (78%); the remainder were light trucks, vans, and multipurpose vehicles.

Drivers with TBI were significantly more likely, compared to the rest of the drivers, to be involved in a crash at night, on a weekend, and in inclement weather (Table 2). Also, drivers with brain injuries were significantly more likely to have been involved in fixed object or parked vehicle collisions (37.5% vs. 30.8%, p<0.001). They also had more lateral collisions than other drivers (29.0% vs. 24.3%, p<0.001). There was no difference in the type of vehicle, for drivers with and without TBI. Finally, although the majority of vehicles were categorized by the police as disabled or destroyed, vehicles of drivers with TBI were significantly more likely to have been so labeled.

Table 2.

Characteristics of Vehicles/Crashes^*

	TBI		Non-TBI
	(N=2,925)		(N=4,825)
	n	(%)	n	(%)	p
Day Time	1,615	55.2	3,121	64.7	<0.001
Weekend	925	34.5	1,388	31.3	0.004
Weather
Clear/Cloudy	1,575	54.0	3,049	63.5
Inclement	1,343	46.0	1,755	36.5	<0.001
“First Harmful Event”
Other Vehicle	1,743	61.4	3,198	68.1
Parked Vehicle	121	4.3	140	3.0
Fixed Object	943	33.2	1,304	27.8
Overturn	30	1.1	51	1.1	<0.001
Point of Impact
Frontal	1,798	64.3	3,101	67.2
Lateral	812	29.0	1,122	24.3
Rear	151	5.4	331	7.2
Overturn	37	1.3	62	1.3	<0.001
Vehicle Type
Automob/Stn
Wag./Limo.	2,288	78.2	3,748	77.7
Light Trk/Van/MVP	637	21.8	1,077	22.3	NS
Vehicle Damage Extent
None/Functional	340	12.1	804	17.5
Disabling/Destroyed	2,479	87.9	3,784	82.5	<0.001

^*Totals and percentages may vary due to missing values.

Overall, 37.7% of hospitalized drivers (N= 2925) had a traumatic brain injury. Of this group, the majority of injuries (79.2%) were minor in nature (MAIS 1 or 2) with 12.7% of moderate severity (MAIS =3), and 8.1% severe (MAIS 4 or 5), based on the maximum AIS score for the brain injury. For 151 drivers, no AIS score was available; of these, 148 (98%) had unspecified intracranial injuries (ICD9 854) and 3 (2%) had multiple fractures involving the skull or face, with other bones (1CD9 804). Since the level of severity was not known, no AIS score could be assigned to these patients and they were therefore excluded; thus, analyses involving injury severity were based on 2774 drivers.

Figure 2 shows the distribution of the total incidence of TBI for male and female drivers. Incidence rates declined significantly (p<0.001) with age for both genders. However, rates of severe injury (i.e. the proportion of brain injuries with MAIS of 4 or 5), shown in Figure 3, were fairly constant (approximately 8% for males and 6% for females) until approximately the age of 60, when they increased for both male and female drivers. For both all brain injuries and serious brain injuries, males had higher rates than females, for all age groups.

Figure 2.

Incidence of TBI By Age (N=7,750)

Figure 3.

Rate of Severe TBI By Age (N=2,774)

The association between selected driver characteristics and TBI incidence/severity is shown in Table 3. As also shown in Figures 2 and 3, the incidence and severity were greater for men than women drivers. In additions, as shown in Figures 2 and 3, the incidence declined with age, but the highest proportion of serious injuries was noted among drivers aged 60 and over. Drinking/drugged drivers had a significantly higher incidence of TBI, but there was no difference in the level of severity, as compared to non-intoxicated drivers. In addition, drivers who were at fault for their crash had a significantly higher incidence of TBI; there was no difference, however, in severity by the culpability of the driver. Drivers wearing seatbelts had both a significantly lower incidence of TBI, and a significantly lower level of severity among those with a TBI. There were no significant differences in either incidence or severity for drivers with airbags (with or without seatbelts), although the number of such cases remains relatively small.

Table 3.

Incidence and Severity of TBI by Driver Characteristics^*

	Proportion of All Drivers (N=7,750)		Proportion of Drivers with TBI (N=2,774)
Characteristics	With TBI		With MAIS 4/5
	(%)	p	(%)	p
Gender
Male	40.8		8.9
Female	33.2	<0.001	6.8	<0.001
Age
<= 19 years	49.1		7.9
20–39	40.7		7.5
40–59	33.4		6.6
60+	28.7	<0.001	13.1	<0.001
Drinking/Drugged
Yes	49.7		7.9
No	35.1	<0.001	7.9	NS
Culpable for Crash
Yes	41.6		8.3
No	31.2	<0.001	7.2	NS
Belted
Yes	31.8		5.9
No	48.4	<0.001	10.8	<0.001
Air Bag Deployed
Yes	34.5		11.3
No	37.9	NS	8.0	NS
Hospital Disposition
Home	36.5		3.8
Died	59.5		59.8
Rehab./Other	43.7	<0.001	23.4	<0.001

^*Totals and percentages may vary due to missing values.

Ranked ISS scores for drivers with TBI were significantly higher than that for other drivers (medians 6 vs. 4, p<0.001). Length of hospital stay and hospital costs did not differ between groups. Drivers with serious brain injuries, however, did not fare as well as those with mild brain injuries in terms of ISS (medians 25 vs. 6, p<0.001), length of stay (medians 7 vs. 2, p<0.001) or hospital cost (medians 17,277 vs. 3,186, p<0.001).

Table 4 shows associations between vehicle/crash characteristics and brain injury. Although, as shown in Table 2, more than half of brain injuries were attributable to two-car collisions, the risk of brain injury was highest for fixed object and parked car collisions, with an incidence of 42.0% and 46.4%, respectively (see Table 4). The most severe brain injuries, however, resulted from vehicles which overturned. As noted earlier, the majority (66%) of drivers were injured in frontal or offset frontal collisions; however, the highest incidence of TBI (42%) was found in lateral collisions. The most serious brain injuries were attributable to lateral and rear-end collisions, although the number of drivers with known AIS scores involved in rear-end crashes are relatively small (N=147), compared with frontal (N=1699) and lateral (N=772) collisions.

Table 4.

Incidence and Severity of TBI by Crash/Vehicle Characteristics

	Proportion of All Drivers (N=7,750)		Proportion of Drivers with TBI (N=2,774)
Characteristics	With TBI		With MAIS 4/5
	(%)	p	(%)	p
Day Time
Yes	34.1		7.9
No	43.5	<0.001	8.4	NS
Weekend
Yes	40.0		5.8
No	36.5	0.004	9.5	<0.001
Weather
Clear/Cloudy	34.1		8.1
Inclement	43.4	<0.001	8.1	NS
“First Harmful Event”
Other Vehicle	35.3		8.0
Parked Vehicle	46.4		6.1
Fixed Object	42.0		8.3
Overturn	37.0	<0.001	10.0	NS
Point of Impact
Frontal	36.7		7.2
Lateral	42.0		10.4
Rear	31.3		10.2
Overturn	37.4	<0.001	8.3	0.03
Vehicle Type
Automob/Stn Wag./Limo.	37.9		7.8
Light Trk/Van/MVP	37.2	NS	9.4	NS
Vehicle Damage Extent
None/Functional	29.7		7.4
Disabling/Destroyed	39.6	<0.001	8.1	NS

^*Totals and percentages may vary due to missing values.

Based on these findings, two sets of multiple logistic regression analyses were conducted; results of these analyses are shown in Tables 5 and 6. The first analysis was based on the total population of 7750 drivers, and the dependent variable of interest was the incidence of TBI. Of these, 5641 drivers had complete data and were included in subsequent regression analyses. Statistically significant (p<0.05) risk factors for the incidence of TBI included: male gender, drivers at fault, driving at night and lateral crashes (see Table 5). In addition, there was a significant (p=0.003) interaction term for age by driver condition and a marginally significant (p=0.08) interaction term for belt use by driver condition. Variables that did not remain in the regression model included vehicle type, airbag, first harmful event, weekend crash, and weather conditions. The graph in Figure 4 provides a visual description of the age by driver condition interaction. Except for the very young and very old, the curve for intoxicated drivers is consistently higher than that of the non-intoxicated group. In fact, for the 20 to 29 year old group, the rate of TBI for intoxicated drivers remains around 50%, whereas the rate for non-intoxicated drivers decreases by approximately 25%. Analysis of the belt use by driver condition interaction term indicates that the rate of TBI was slightly higher among drivers who were drinking and belted as opposed to those who were drinking and unbelted (data not shown).

Table 5.

Results of multiple logistic regression on incidence of TBI (N=5641)^*

Variable	Parameter estimate	p	Odds Ratio (OR)	95% CI of OR
				Lower	Upper
Intercept	−0.02	0.85	-	-	-
Male	0.20	<0.001	1.22	1.09	1.38
Age (years)	−0.01	<0.001	0.99	0.98	0.99
Intoxicated	−0.49	0.03	0.61	0.40	0.94
At Fault	0.30	<0.001	1.34	1.18	1.53
Belted	−0.58	<0.001	0.56	0.49	0.64
Daytime	−0.15	0.02	0.86	0.76	0.97
Lateral (vs. Frontal)	0.32	<0.001	1.38	1.21	1.57
Belt X Intoxicated	0.28	0.08	1.32	0.97	1.80
Age X Intoxicated	0.02	0.003	1.02	1.01	1.03

^*Goodness-of-fit p=0.88 indicating that the model fits the data well.

Table 6.

Results of multiple logistic regression on incidence of serious TBI (N=1975)^*

Variable	Parameter estimate	p	Odds Ratio (OR)	95% CI of OR
				Lower	Upper
Intercept	−2.26	<0.001	-	-	-
Age ≥ 60 years	0.54	0.03	1.72	1.04	2.78
Intoxicated	−0.29	0.46	0.75	0.32	1.53
Belted	−0.80	<0.001	0.45	0.31	0.64
Weekend	−0.50	0.02	0.61	0.39	0.93
Crash into fixed object+	−0.09	0.71	0.92	0.58	1.42
Lateral (vs. Frontal)	0.43	0.02	1.54	1.06	2.23
Rear (vs. Frontal)	0.63	0.06	1.87	0.92	3.52
Crash into fixed object+
X Intoxicated	0.80	0.09	2.23	0.91	5.96
Age ≥ 60 years
X Weekend	0.82	0.07	2.28	0.91	5.50

^*Goodness-of-fit p=0.57, indicating that the model fits the data well.

⁺Includes parked vehicles

Figure 4.

Driver Conditions By Age Groups (N=5,641)

Only drivers with TBI were included in the second logistic regression model. For this analysis, the outcome variable of interest was “serious” TBI, i.e., a brain injury with a MAIS of 4 or 5. Of the 2774 drivers with a TBI, 1975 had complete data and were included in regression analyses. As shown in Table 6, a slightly different set of characteristics was found to be predictive for serious TBI. Individual risk factors that were significantly associated with the outcome measure included unbelted drivers and lateral crash. Marginally significant interaction terms included first harmful event by driver condition (p=0.09) and age by weekend (p=0.07). Characteristics that were removed from the model after adjustment for all variables included vehicle type, airbag, weather conditions, gender, driver at fault, and time of day. Analysis of the interaction terms indicated that (1) among intoxicated drivers, the rate of serious TBI was twice as high in crashes involving a stationary object (i.e. parked vehicle or fixed object) than in crashes involving moving vehicles, and (2) for crashes that occurred on a weekend, the rate of serious TBI was four times as likely to occur among drivers aged 60 or greater than among younger drivers.

DISCUSSION

Although there is no easy way to fully describe the human costs of traumatic brain injury, there are some estimates of monetary costs. In 1985 dollars, it is estimated that the annual expenditures for TBI total $37.8 billion dollars, $4.5 billion for direct annual expenditures, and $33.3 billion for indirect annual costs (Max et al., 1991). Prevention efforts must rely on a better understanding of the epidemiology of these injuries, a large proportion of which result from motor vehicle collisions. Many of the studies in the literature are 15–20 years old, and do not address specifics of the mechanism of injury for the total group, nor the details of the crash characteristics for those which were due to motor vehicle collisions.

Findings from the current study reveal that TBI rates were significantly higher among young drivers, male drivers, and drivers involved in lateral collisions. In addition, drivers who were perceived by the police as having been drinking, those without seatbelts, and those deemed to be culpable for the collision, also had high rates of TBI. Risk factors for serious brain injury included drivers over the age of 60, drinking/drugged drivers, unbelted drivers, weekend collisions and fixed object crashes.

As mentioned previously a limitation of this study is the fact that details of the crash are obtained from police reports and not from detailed crash reconstructions. Thus, data on actual impact forces, as well as contact points and intrusions, are not available. However, these findings do represent population-based data, as they include all injured drivers admitted to Maryland hospitals for a three-year period. Interestingly, these epidemiologic findings are in fairly good agreement with those from trauma center-based crash reconstruction studies. For example, from both sources, serious brain injuries were found to be associated more with lateral, as compared to frontal, crashes.

In a recent report on “Reducing the Burden of Injury” (Institute of Medicine, 1999), the authors note that, “despite significant strides in the past decade, the biomechanical properties of the brain and the biologic response of the brain to injury are not well characterized.” Since such a large proportion of brain injuries result from motor vehicle crashes, a first step would seem to be a better understanding of the current epidemiology of motor vehicle-related brain injuries--i.e. who is at risk, and what types of collisions are involved? With the changing mix of vehicles on the highway, and the introduction of new safety equipment, such as side airbags, this is obviously a dynamic process, requiring ongoing analysis of the most recent data available, in order to monitor changes in injury patterns. An example in point is a recent analysis by Siegel et al., showing that in lateral crashes between sedans and full-sized sport utility vehicles, both the incidence and severity of brain injuries appear to rise (1999 (in press). The next step, i.e. understanding the actual forces, contact points, and intrusions associated with motor vehicle injuries, should be based on studies of real world crashes. Based on findings from these combined efforts, biomechanics researchers would have a stronger scientific basis upon which to conduct their experimental work on the response of the brain to different types of collision insults.

There are several primary prevention strategies which might be effective for preventing TBI in this population, given our findings. These strategies would need to combine educational, legislative, and environmental approaches to the problem. Obviously, there is a need to decrease drunk driving, especially among young males. There is also a need to educate drivers, especially young male drivers, about the importance of wearing seatbelts. However, educational strategies have, in the past, been shown to be of limited effectiveness, especially for adolescents (Robertson, 1992).

Airbags should be a more effective intervention for brain injury, in that the protection they afford is passive in nature, and can therefore be implemented across large groups of persons at risk, with maximum benefit. Thus, this technology will benefit both the risky young driver and the older driver at higher risk of serious brain injury. As pointed out by Loo et al. (1996), findings from real world crashes reveal that “the most important effect of the airbag is to reduce the severity of brain injury”, although the overall incidence of brain injury may not be altered, at least among hospitalized patients (Dischinger, 1996). This finding of reduced severity was based on GCS scores, obtained in the clinical setting; as mentioned previously, such information is not routinely available in hospital discharge records. The introduction of side airbags may prove to be an even more effective intervention, given the fact that the incidence of TBI is high in lateral collisions. Unfortunately, it will probably be years before the most high-risk drivers, who frequently drive older vehicles, have access to modern vehicle restraint systems.