Accident Analysis and Prevention 39 (2007) 2227

Lateral impact injuries with side airbag deploymentsA descriptive study

Narayan Yoganandan , Frank A. Pintar, Jiangyue Zhang, Thomas A. Gennarelli
Department of Neurosurgery, Medical College of Wisconsin, VA Medical Center, Milwaukee, WI, United States
Received 26 September 2005; received in revised form 13 April 2006; accepted 24 May 2006

Abstract
The present study was designed to provide descriptive data on side impact injuries in vehicles equipped with side airbags using the United States
National Automotive Sampling System (NASS). The database was queried with the constraint that all vehicles must adhere to the Federal Motor
Vehicle Safety Standards FMVSS 214, injured occupants be in the front outboard seats with no rollovers or ejections, and side impacts airbags be
deployed in lateral crashes. Out of the 7812 crashes in the 19972004 weighted NASS files, AIS 2 level injuries occurred to 5071 occupants.
There were 3828 cases of torso-only airbags, 955 cases of torsohead bag combination, and 288 inflatable tubular structure/curtain systems. Side
airbags were not attributed to be the cause of head or chest injury to any occupant at this level of severity. The predominance of torso-only airbags
followed by torsohead airbag combination reflected vehicle model years and changing technology. Head and chest injuries were coupled for the
vast majority of occupants with injuries to more than one body region. Comparing literature data for side impacts without side airbag deployments,
the presence of a side airbag decreased AIS = 2 head, chest, and extremity injuries when examining raw data incidence rates. Although this is the
first study to adopt strict inclusionexclusion criteria for side crashes with side airbag deployments, future studies are needed to assess side airbag
efficacy using datasets such as matched-pair occupants in side impacts.
2006 Elsevier Ltd. All rights reserved.

Keywords: Lateral impact; Side airbags; Head and chest injuries; Descriptive study

1. Introduction in frontal impacts with frontal airbags, no such systematic eval-

uations of side impacts with side airbags have been published,
Airbags are used in modern motor vehicles for enhancing to the best knowledge of the authors (Pintar et al., 2000). Only
occupant safety during impact. Frontal airbags have been in a few studies have begun to appear in the literature with no
the vehicle fleet for over a decade, and the United States fed- unified conclusions on side airbag responses. McGwin et al.
eral requirements are promulgated through Safety Standards (2003) used the United States National Automotive Sampling
FMVSS 208 (NHTSA, 2005). In contrast, side airbags are more System (NASS) database for the years 19972000 to determine
recent than frontal airbags and are intended to primarily pro- the association between side airbags and risk of injury in motor
tect the occupant during lateral crashes. They are installed as vehicle collisions with near-side impact. The authors assumed
thorax or torso-only airbag, torsohead airbag (combination that all vehicles with side airbags as optional equipment were
airbags), or separate torso and head airbags (inflatable tubu- equipped with the technology and concluded that vehicles with
lar structure or curtain). Although federal regulations do not side airbags had a risk of injury similar to occupants of vehicles
exist in the United States for side airbags, such systems are without side airbags. This assumption is a major limitation of the
gaining popularity because of public awareness for safety and study, and its validity was not discussed. Using NASS database
their potential injury mitigating characteristics (NHTSA, 1999, for the year 2000, another study conducted an analysis of 187
2003). While many studies are conducted to investigate injuries occupants with airbag deployments out of which 62 were in side
impacts (Bazarian et al., 2004). Although this study concluded
that side airbags may be effective in preventing cranial trauma,
Corresponding author at: Department of Neurosurgery, Medical College of
less than 1% of occupants were in vehicles equipped with side
Wisconsin, 9200 West Wisconsin Avenue, Milwaukee, WI 53226, United States.
airbags. These analyses, albeit brief, indicate the need to con-
Tel.: +1 414 384 3453; fax: +1 414 384 3493. duct a more detailed study specific to side airbag deployments.
E-mail address: yoga@mcw.edu (N. Yoganandan). Therefore, the purpose of the present investigation is to focus

0001-4575/$ see front matter 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.aap.2006.05.014
 N. Yoganandan et al. / Accident Analysis and Prevention 39 (2007) 2227 23

on a descriptive case series of lateral impact-induced injuries in

Raw data
vehicles with side airbag deployments.

Low extremities

10
10
11
2. Methods

Weighted

3321
652
336
The NASS database was interrogated with the constraint that
all vehicles adhere to the Federal Motor Vehicle Safety Stan-

Raw data
dards, FMVSS 214. Case selection criteria was such that the
occupant should be involved in a side impact collision with the

13
8
1
principal direction of force between 50 and 130 for passengers,

Up extremities
and 230 and 310 for drivers, and the primary impact should be

Weighted
in the lateral direction resulting in deployment of a side airbag.

467
388
163
All types of airbag systems, i.e. torso alone, side or head curtain
or inflatable tubular structure, or a combination of torso and side

Raw data
curtain or head, were included. Other selection criteria included
passenger cars, light trucks, and vans. Only outboard front seat

5
5

2
1
occupants, driver and passenger, were included in the study.

Weighted
Rollovers and full ejection events were excluded. The 1990 ver-

Spine
sion of the Abbreviated Injury Scale was used (AIS, 1990) for

354
522

132
9
injury coding. Briefly, the grading system for injury levels is as
follows: (0) no injury; (1) minor; (2) moderate; (3) serious; (4)

Raw data
severe; (5) critical; (6) maximum; (7) unknown. Body regions
represented the head, face, neck, chest, abdomen, spine, and

6
6
4
upper and lower extremities. In addition, the source of injury

Abdomen

Weighted
and confidence in injury assignments was extracted from case-

246
3042
141
by-case description of injuries in the database. The sourcing of
injuries is a separate variable in NASS. The assignment of the
injury source and the confidence levels are done by specialists

Raw data
at the Zone center responsible for quality controlling the work
of the field investigator that documented the crash. The confi-
3
6
6
6
3

3
dence levels are initially indicated by the field investigator, but
Weighted

these are subject to correction by the specialists at the Zone

Chest

179
3237
378
288
61

318
center. AIS = 1 level data are presented to a limited extent, and
since they are less clinically significant, descriptions are lim-
ited to occupants sustaining AIS 2 injuries. Weighted data are
Raw data

included in the results and discussion.

5

3. Results
Weighted
Number of occupants sustaining injuries as a function of body region
Neck

223

NASS data for the years 19972004 were used in the study.
With the inclusionexclusion criteria specified earlier, a total of
Raw data

7812 side crashes had impact-related injuries (all AIS levels).

The actual number of sampled cases was 68. Of the 7812 crashes,
18

7214 were left and 598 were right side impacts. Lap and shoulder
Weighted

belts were used by 7346 occupants, and no belt restraints were

used by the remaining 466 occupants. Limiting the analysis to
47
1040
Face

AIS 2 injuries, 4572 crashes occurred to the left side, and 499
crashes occurred to the right side, for a total of 5071 impacts.
Raw data

When data were categorized based on airbag type, there were

3828 cases of torso only airbags, 955 cases of torsohead bag
3
10
6
4
1
1
1

combination, and 288 inflatable tubular structure/curtain sys-

Weighted

tems. Regarding the location, 4724 airbags deployed from the

Head

seat, 237 from the door and roof side rail, 59 from the door, 49
174
441
401
315
42
84
209

from the roof side rail, and nine from the seat back and roof side
rail.
AIS level
Table 1

Table 1 provides a summary of raw and weighted data based

AIS 1
AIS 2
AIS 3
AIS 4
AIS 5
AIS 6
AIS 7

on AIS level and body region. As discussed later, one raw data
24 N. Yoganandan et al. / Accident Analysis and Prevention 39 (2007) 2227

Fig. 1. Bar chart showing the number of occupants sustaining AIS 2 injuries Fig. 3. Bar chart showing the number of occupants sustaining only injuries to
as a function of body region. one body region at the AIS 2 level.

point had a weighting factor of 2798, and all the remaining data
had weighting factors less than 250. AIS 1 level is included for
comparison purposes, and as indicated in Section 2, the follow-
ing weighted results are applicable to AIS 2 trauma. The most
commonly involved body region was the chest (4282 occupants),
followed by the abdomen (3183 subjects), and head (1492 occu-
pants). No occupants sustained neck injuries. Fig. 1 shows the
distribution as a function of body region. Limiting the analysis to
include AIS 3 trauma, 1408 side impact cases were identified.
The head and chest were identified to be the most frequently
affected body regions with 1051 and 1045 occupants. Fig. 2
shows the distribution as a function to body region.
Out of the 5071 occupants with AIS 2 trauma, 727 occu- Fig. 4. Bar chart showing the number of occupants sustaining only one injury
pants had injuries to one body region. While head, chest, spine, to any body region at the AIS 2 level.
and lower and upper extremities were involved, no cases were
found with only neck, abdomen, or facial injuries. The dis- and in the remaining 42 subjects, the right side B-pillar structure
tribution of these injuries is shown in Fig. 3. Of these 727 was the source.
subjects, 511 occupants sustained one injury to any body region. Out of the 5071 occupants with AIS 2 trauma, 4344 occu-
In this subgroup of single injury to any body region, head was pants had injuries to more than one body region. In this subset,
the most frequently injured body region followed by the upper 1214 occupants had head injuries at the AIS 2 level, and 422
extremities, spine and lower extremities (Fig. 4). Out of the 278 torso airbags, 735 torsohead airbags, and 56 torso and inflatable
occupants sustaining a head injury in the group of 511 occu- tubular structure/curtain airbag systems were deployed in side
pants, all occupants wore lap and shoulder belts. Thorax only crashes. Out of these 1214 occupants with head injuries, chest
bags deployed in 220 crashes and thoraxhead airbags deployed injuries occurred to 1195 subjects. A majority of head injuries
in the remaining 58 crashes. The confidence in injury was cer-
tain in 25 and probable in 253 cases. A non-contact source was
attributed to be the cause of head (brain) injury in 236 occupants,

Fig. 5. Bar chart showing the number of occupants with head injury as a function
Fig. 2. Bar chart showing the number of occupants sustaining AIS 3 injuries of AIS level. Data obtained from the group of occupants with AIS two plus
as a function of body region. injuries to more than one body region and focusing only on AIS 2 head trauma.
 N. Yoganandan et al. / Accident Analysis and Prevention 39 (2007) 2227 25

were at the AIS three and four levels, and the distribution as a to offer torso protection. Torsohead systems, i.e. combination
function of AIS is shown in Fig. 5. In the same subset of 4344 airbags, followed torso-only airbags with 955 cases. Because the
occupants, 4225 subjects sustained chest injuries at the AIS 2 function of the combination system is to protect the torso and
level, and 3308 torso airbags, 791 torsohead airbags, and 126 head, they have additional volume. Occupant kinematics may
torso and inflatable tubular structure/curtain airbag systems were be different between the two systems. The objectives of the two
deployed in side crashes. AIS = 2 level injuries were most pre- different airbags in separate torsocurtain systems are to protect
dominant (Table 1). Injury sources for chest trauma included the torso and head with independent kinematic controls. There-
interior surfaces of the vehicle and pillar, and for head, they fore, these different side airbag systems offer unique features
included the side roof rail or header. In none of these cases, the in their form and function for occupant protection. The general
side airbag, regardless of bag type, was attributed to injury. trend in the United States market is a shift toward separate torso
and head airbag systems instead of combination airbags. For the
4. Discussion model year 2004, torso side airbags outnumbered the combina-
tion system by three to one.
Studies with side airbag deployments have largely remained Examining injuries, occupants with trauma to more than
anecdotal. Langwieder et al. (1998) reported minor trauma from one body region at the AIS 2 level, chest injuries occurred
a single crash in a vehicle with thorax airbag and inflatable tubu- along with head injury in >90% of the subjects. This finding
lar structure. Kirk and Morris (2003) while reporting data from suggests a coupling phenomenon between the two body com-
10 cases with different impact directions concluded that further ponents. Chest injuries did not occur in isolation at AIS 3
studies of airbag deployments are essential. In an analysis of levels. A similar conclusion was advanced in another study
10 frontal impacts, side impacts, and rollover crashes, Dalmotas wherein an association was found between cranial and cervi-
et al. (2001) stated additional field collision data on side airbag cal column trauma (Yoganandan et al., 1990). In this previous
systems are needed. From a study of four side impacts with side study, cervical injuries were attributed to external load transfer
airbags, Bauer et al. (2000) stated a proper statistical sampling from head impact within the interior component of the vehicle.
could not be achieved at this time reflecting the limited nature In the present context, lateral impact loading was delivered to
of field data. In contrast to these individual and limited case the torso secondary to interaction with the interior of the vehi-
series, the present study provides descriptive information using cle. This finding was based on injury source identification in the
weighted samples and forms the largest dataset in the literature database. Although side airbags might have served to distribute
focusing only on lateral impacts with deployed side airbags. the load to the chest, one possibility is that occupant positioning
Side airbags began to enter the vehicle fleet as an optional may have been less than optimal to fully realize injury mitigating
feature in 1990s, reflecting the emerging nature of the technol- characteristics of side airbags. It is well known in impact biome-
ogy. No cases with side airbags were found in the present study chanics literature that positioning affects load transfer, injury
prior to the year 1997. Different types of airbag systems have mechanism, and tolerance (Yoganandan et al., 2002; Maiman
been introduced including torso-only, torsohead (combination et al., 2002). The effects of subject positioning on head and
bag), torsocurtain or, inflatable tubular structure/curtain. These torso injuries with side airbags have not been investigated and
bags deploy from the door, seat, and side rail components of the current motor vehicle safety standards (FMVSS 214) do
the vehicle. Only 59 side airbags deployed from the door. The not directly address the issue. While NASS is not ideal for this
door-mounted torso only airbag is being installed less frequently determination, other databases such as Crash Injury Research
because of its response to out-of-position occupants. Laboratory Engineering Network may be used. The latter database contains
investigations have shown their deleterious effect specifically in detailed kinematics and medical information to better evaluate
the out-of-position scenario demonstrating increased craniocer- airbag responses (Yoganandan et al., 2005). As discussed, this
vical loads and chest deflections (Pintar et al., 1999). The side topic needs further consideration for an improved assessment of
airbag out-of-position injury technical working group remarked: side airbag efficacy.
there have not been enough deployments to assess the out- It is possible to compare the findings with data obtained from
of-position injury risk of side airbags from accident data, and cases without side airbag deployments. Hassan et al. (1999)
viewed that new systems should be designed according to these reported that 90% of AIS 2 severity side crashes in 19881996
recommendations for further limiting out-of-position occupant NASS files occurred at a delta-V of less than 39 km/h. Analysis
injury risk largely because new technology is emerging that is using the present data indicated a similar delta-V (37 km/h) to be
expected to meet the guidelines while still providing side impact associated at the same severity. Although this finding does not
protection (Lund, 2000). Dinas and Fildes (2002) reported that appear encouraging, several limitations should be recognized.
a significant number of occupants were seated out-of-position For example, different types of airbags intended to protect dif-
while traveling on the road and that a number of these were ferent regions of the occupant were included in the analysis;
seated in a manner that may possibly result in injury from the torso-only bag for chest and curtain-only bag for head. Vehi-
deployment of a side airbag. A need thus exists to evaluate cles were not FMVSS 214 complaint in the previous study, and
positioning issues in side impacts. injuries included body regions not intended for protection from
In the present study, out of 5071 crashes with AIS 2 injuries, the deploying side airbag. NASS data during the earlier years
a majority were torso-only side airbags without supplementary represented lower proportions of belted occupants while in the
systems such as curtain. The primary function of these airbags is present 19972004 study, majority of crashes involved belted
26 N. Yoganandan et al. / Accident Analysis and Prevention 39 (2007) 2227

Table 2
Comparison of data (% injuries) from the present study with literature
Description Head Chest Abdomen Up extremities Low extremities

AIS = 2 raw data 24.4 14.6 14.6 19.5 24.4

AIS = 2 weighted data 8.7 63.8 60.0 7.7 12.9
Hassan et al. US data 37 18 9 29 23
Hassan et al. UK data 30.6 16.1 11.8 28.3 31.1
AIS = 36 raw data 29.3 36.6 9.8 2.4 26.8
AIS = 36 weighted data 16.6 14.3 2.8 3.2 6.6
Hassan et al. US data 9 13 3 3 6
Hassan et al. UK data 15.3 22.8 9.3 3.3 24.2

occupants. Although the role of such restraint systems in side 5. Conclusions

crashes has been reported in earlier literature, their effects on
modern vehicles (some have pretensioning belts) with or without Using NASS database for the years 19972004, this study
side airbag deployment have not been studied (Mills and Hobbs, identified injuries in 7812 side impact crashes with side airbag
1984). Most importantly, this is not a matched-pair comparison deployments. The majority of side airbags were torso-only fol-
of the two groups. These issues underscore the need to further lowed by torsohead combination systems. Head and chest
investigate side impacts with side airbag deployments from epi- injuries were coupled for the vast majority of occupants with
demiological and laboratory biomechanical perspectives for a injuries to more than one body region. In no case was the side
better assessment of their efficacy. airbag attributed to be the source of AIS 2 injuries. Chest
Regarding injuries to specific body regions, a comparison of injuries did not occur in isolation at AIS 3 levels. Although
data from the present study with findings from the Hassan et this is the first study to adopt strict inclusionexclusion crite-
al. study indicated the following (Table 2). Using weighted data ria for side crashes with side airbag deployments, future studies
as a basis, at the AIS 2 level, head injuries in the presence of are needed to assess side airbag efficacy using datasets such as
a deploying side airbag appeared to be reduced compared to matched-pair occupants in side impacts.
side impacts without side airbags from the United States and
United Kingdom databases. This trend was also true for upper
Acknowledgements
extremity trauma. Chest and abdominal injuries showed a dra-
matic increase with the presence of a side airbag. However, a
This study was supported in part by the VA Medical Research
single raw data point with a weighting factor of 2800 contributed
and NIH grant AG024443. The authors gratefully acknowledge
to the increase; all other cases had weighting factors less than
the assistance of Mr. Dale Halloway.
250. When raw data were used as a basis for comparison, as
shown in Table 2, both head and chest injuries demonstrated a
decreasing tendency in side airbag deployed lateral impacts, a References
positive trend towards occupant safety and vehicle crashworthi-
ness. When AIS = 36 level trauma were considered, raw data AIS, 1990. The Abbreviated Injury Scale. Arlington Heights. American Asso-
showed increases in head and chest injuries with the present side ciation for Automotive Medicine, IL.
Bauer, P., Lange, W., Messner, G., Rauschner, S., Pieske, O., 2000. Compar-
airbag ensemble, suggestive of increased survival with serious ison of real world side impact/rollover collisions with and without thorax
injuries. However, as discussed earlier, this interpretation must bagsfirst field experience study. AAAM, Chicago, IL.
be supported with additional analysis. Bazarian, J., Fisher, S., Flesher, W., Lillis, R., Knox, K., Pearson, T., 2004.
As indicated earlier, the dataset available for this study was Lateral automobile impacts and the risk for traumatic brain injury. Ann.
small and weighted occupants are based on this low sam- Emerg. Med. 44, 142152.
Dalmotas, D., German, A., Tylko, S., 2001. The Crash and Performance of
ple size. Confidence decreases when weighted estimates from Side-Mounted Airbag Systems. ESV, NHTSA, Amsterdam, Netherlands.
such selected samples are extracted from a probability-sampled Dinas, A., Fildes, B., 2002. Observations of Seating Positions of Front
database; NASS is no exception. Consequently, conclusions Seat Occupants Relative to the Side of the Vehicle. AAAM, Tempe,
from this study should be viewed with this important sample AZ.
size recognition. Confidence in the results can be increased Hassan, A., Mackay, M., Padmanaban, J., Davis, M., 1999. Comparsion of real
world side impact collisions which ouccrred in the United Kingdom and
by adding samples from other datasets such as the Coopera- United States. AAAM, Stiges, Spain.
tive Crash Injury Study from the United Kingdom. Because Kirk, A., Morris, A., 2003. Side Airbag Deployments in the UKInitial Case
of sample size limitations, it is difficult to conclude that side Reviews. ESV, NHTSA, Nagoya, Japan.
airbags reduce injuries, and therefore, to quantify and address Langwieder, K., Hummel, T., Anslem, D., 1998. The Effect of Airbags on
their efficacy, additional data are needed which may be obtained Injuries and Accident Costs. ESV, NHTSA, Windsor, Canada.
Lund, A.K., 2000. Recommended Procedures for Evaluating Occupant Injury
by coalescing with sources such as CCIS and/or including spe- Risk from Deploying Side Airbags. International Standard Organization.
cific cases from Crash Injury Research Engineering and Network Maiman, D., Yoganandan, N., Pintar, F., 2002. Preinjury cervical alignment
studies. These are avenues for future research. affects spine trauma. J. Neurosurg. Spine 97 (1), 5762.
 N. Yoganandan et al. / Accident Analysis and Prevention 39 (2007) 2227 27

McGwin, G., Metzger, J., Porterfield, J., Moran, S., Rue, L., 2003. Pintar, F., Yoganandan, N., Gennarelli, T., 2000. Airbag Effectiveness on Brain
Association between side airbags and risk of injury in motor Trauma in Frontal Crashes. AAAM, Chicago, IL.
vehicle collisions with near-side impact. J. Trauma 55 (3), 430 Pintar, F., Yoganandan, N., Maltese, M., Samaha, R., Eppinger, R., 1999. Three-
436. year-old child out-of-position side airbag studies. In: Proceedings of the
Mills, P.J., Hobbs, C.A., 1984. The probability of injury to car occupants Stapp Car Crash Conference, San Diego, CA.
in frontal and side impacts. In: Stapp Car Crash Conf, Chicago, IL, Yoganandan, N., Baisden, J., Maiman, D., Pintar, F., 2005. Type II odontoid
pp. 223232. fracture in frontal impact. J. Neurosurg. 2, 481485.
NHTSA, 1999. Evaluation of FMVSS 214 side impact protection dynamic per- Yoganandan, N., Haffner, M., Maiman, D.J., Nichols, H., Pintar, F.A., Jentzen,
formance requirement, Department of Transportation, p. 238. J., Weinshel, S., Larson, S.J., Sances Jr, A., 1990. Epidemiology and injury
NHTSA, 2003. Department of Transportation, http://www.nhtsa.dot.gov. biomechanics of motor vehicle related trauma to the human spine. SAE
NHTSA, 2005. Department of Transportation, http://www.nhtsa.dot.gov. Trans. 98 (6), 17901807.