Documentation:FIB book/Spine Injuries in Passenger Rollover Head-First Impacts
Overview
Introduction
Motor vehicle accidents can have a significant impact on a person's quality of life, particularly when it comes to spinal cord injuries. Rollover accidents are one type of vehicle accident that can be especially dangerous. Using the Abbreviated Injury Scale (AIS), cervical spine injuries with an AIS 3 or greater severity score were 4 times more likely in rollovers than frontal and side crashes and the cervical spine is the third-most injured body region in rollovers (Hu et al. 2008). Even though the rate of spine injuries or fractures is higher than that of a spinal cord injury in a rollover event, the rate of spinal cord injuries is around 6.6% (Fakharian et al., 2022) in all rollover events. With the number of rollover accidents increasing every year, it's therefore important to understand how the cervical spine responds in these situations and the likelihood of severe injuries. While there are Anthropomorphic Test Devices (ATDs) designed for frontal and side-impact crashes, there are currently no verified ATDs for rollover testing. In safety standards such as the Federal Motor Vehicle Safety Standards (FMVSS), there is only consideration of the likelihood of a rollover event for a specific vehicle during testing. Some researchers have explored the biofidelity of ATDs in rollovers, focusing on the interaction between the head/neck and the roof, but more work is needed in this area which we will elaborate on in the following sections.
Importance and Significance
Data from the National Highway Traffic Safety Administration (NHTSA) shows that fatal motor accidents due to automotive rollovers have been trending upward since 2012. The following graph illustrates the recent increase in fatalities involving both rollovers and head-on collisions, in all NHTSA regions:
These statistics have not discouraged the growing consumerism of larger vehicle types or automakers’ drive to meet these demands. In fact, in the same timeframe, the market share of SUVs has increased from 21% to 31%. (Data for the Automotive Trends Report 2023) A causation can be drawn between the increasing number of fatalities involving rollovers and the increased market share of SUVs due to the significantly greater propensity for SUVs to roll over due to their higher center of mass. Furthermore, it is stated that on average 35% (El-Menyar et al., 2014) of incapacitating and fatal injuries occur during rollovers of heavy trucks annually in the US.
Epidemiological data shows that a significant proportion of severe injuries caused by automotive rollovers are spine injuries. 1.30% of rollover accidents lead to AIS 2+ cervical spine injuries. This is among the most common severe injury modes, only trailing head injuries which have an injury rate of 1.71% (Funk et al., 2012). This establishes the fact that there is an alarming rise in the number of automotive rollovers due to consumer trends, which may cause an increase in the number of AIS 2+ spinal cord injuries.
Previous Work
The scientific literature examining the effects of rollover crashes on the spine is aware of the fact that almost ⅔ of all traffic fatalities come from occupants in light passenger vehicles and approximately ⅓ of those fatalities result from rollovers (Wu et al., 2018) so rollovers are extremely high-risk mode of crash injury. Rollovers can result in occupant injury through ejection and contact with either road or car components, through contact with the vehicle's roof, or a combination of both. Given the roof contact places a supraphysiological load on the head and neck, cervical spine injuries may occur. The current scientific literature aims to analyze the risk of C-spine injury in rollover scenarios since motor vehicle collisions have been reported to be the leading cause of automotive-related cervical spine fractures in adults in the US and Canada (Morgan et al., 2023).
.png)
The Roberts et. al study uses a buck test setup in a simulated environment called the Dynamic Rollover Test System (DRoTS) as can be seen in Figure 1A. The study’s primary aim was to establish the biofidelity of ATDs such as the Hybrid III (and HIII with a pedestrian pelvis) amongst others in a rollover with a focus on head/neck interaction with the roof since these ATDs are built for frontal impact crashes and (Zhang et al. 2013) showed that different ATDS are unable to replicate fidelity in rollovers. However, this study provides us with important insight into the head-neck responses in 8 ATD tests and 4 Post Mortem Human Subjects (PMHS) tests. The buck, as shown in Figure 1a, was rollover over about its Centre of Mass onto a translating roadbed surface. The buck itself was instrumented with accelerometers and angular rate sensors as well as potentiometers to record impact kinematics and vertical roof deformation. The structure of the roof was made to be repeatable and comparable to real-world deformation patterns and it even matches the FMVSS 216 standard, and this was validated using Finite Element Analysis (FEA). The 4 cadavers used were all male and had the following anthropometry in Table 1 so the median was close to the 50th% Male.
The seating position for all ATDs and PMHS was standardized by the University of Michigan Transportation Research Institute (UMTRI) standard driving position. For the PMHS, after each test, a C-spine collar was placed to ensure further C-spine injuries were not caused during removal from the buck or transportation to the post-test scan center. The following ATDs were used for the 8 tests:
- THOR
- WorldSID
- Polar II
- Hybrid 3 (HIII)
- Hybrid 3 Pedestrian (H3P)
- Modified ATD for Rollovers
- Hybrid III ATD with a pedestrian pelvis instead of the molded seated pelvis traditionally used with the Hybrid III (H3PP)
,_b._The_initial_condition_with_ATD_in_an_upright_position._(Roberts_et_al).png)
The testing conditions were the same for all tests with an average vehicle pitch angle of -1.26±0.31 degrees, average roll angle of 142.63±1.13 degrees, an average roll rate of 243.56±3.28 degrees/second, average roll bed velocity (or translational velocity as such) of 9.86±0.11m/s and an average vertical velocity of 1±14 0.05 m/s. From the 4 cadaver tests, 3 of them sustained C-spine injuries as can be seen in Table 2.
.png)
However, only PMHS 679 sustained a bony fracture at any level of the C-spine. Presumably, the highest AIS code would be a 4 for this specimen. Interestingly, for the 8 ATD tests (2 of the 6 ATD tests had tests repeated) all calculated values of Nij were below 1 which is the upper limit (or Injury Assessment Relative Value) for neck (C-spine) injury limit (refer to Table 3).
.png)
Also note, that the only PMHS which sustained the fracture was the tallest specimen and had the shortest distance to the roof surface and the only PMHS without any injuries had a maximum level of roof deformation. However, the main conclusion to draw from this in the context of 50th percentile PMHS and ATDs is that even though all ATD tests resulted in Nij<1.0 critical value, 3 out of 4 of them sustained C-spine injuries even though Nij is built to predict both bony and soft tissue injuries such as the ones seen in PMHS results. So, the biofidelity of ATDs is poor in rollover scenarios and head-first impacts and Nij seems to be a poor indicator of C-spine injuries and the overall roof deformation level seems to be an insufficient predictor of injury, rather the presence of any level of roof deformation seems to be a good predictor. There are confounding variables such as bone mineral density and neck musculature impact and finally, the exact head and C-spine column curvature at impact likely affects the injury pattern. The Dobbertin et al study also aims to further delve into this correlation between roof crush and the risk of head, neck and spine injury using population-level or epidemiological data without active testing. The data source was the NHTSA’s National Automotive Sampling System-Crashworthiness Data System (NASS-CDS) but it was further filtered for non-ejection-related injuries in rollovers, i.e. injuries resulting from contact with an overhead vehicle component as determined by investigators. Even though ejection was an important confounding variable, occupants wearing a three-point manual, three-point automatic, manual lap seatbelt, or no seatbelt were all included in the analysis, and the NHTSA investigators' on-site judgment on whether the injuries were resultant of ejection or contact with objects outside was used as exclusion criteria. The CDS system employs the use of the Abbreviated Injury Scale (AIS) and since there is a clear association between AIS and a 30-day post-crash survival rate, this is a good metric to use. In this study, codes 0 (unknown or no injury) and 7 (unknown severity) were excluded to reduce miscategorization. This study used the AIS codes to derive the New Injury Severity Score (NISS), which is a modified version of the ISS and uses the three most severe AIS scores regardless of body region. Since the scope of the study was limited to head, neck, and spine injuries, the NISS only used the sum of squares of the AIS scores of an occupant’s head, neck (all tissue between head and thorax excluding spinal column), and spine (cervical, thoracic and lumbar) injuries. The independent variables (or changed parameters) included the magnitude of the roof crush vertically in 10cm increments as well as the age, seatbelt usage, and the deployment of the front airbag. Of note, since rollovers have much lower velocity impacts than front/side crashes, the DeltaV was not considered in this study. By cross-sectional and matched case-control statistics, the HNS-NISS (Head, Neck, Spine-NISS) score was evaluated with a cut-off of 9 (or 3*AIS3 or greater) injuries. The following Figure 3 summarizes the results.
We can see the highest injury odds result from no seatbelt usage and front airbags deployed scenario and this undergoes linear increase with an increase in roof crush level. The same linear increase is seen with the lowest injury risk scenario whereby seatbelts were used and front airbags did not deploy. This pattern was noted in both weighted and unweighted multivariate models. Hence, seatbelt usage is crucial in lowering injury risk in rollovers too. It is difficult to isolate from this data only the damage to the Cervical spine specifically but it would be a fair assumption to extrapolate the pattern to C-spine injuries and deduce that the risk of C-spine injuries follows the general pattern considered the axial force pattern on the C-spine in head-first impacts.
The dynamic, complex nature of rollovers also presents a delicate nuance to any tests designed to simulate testing. The Heller et. al study was monumental in studying the occupant dynamics throughout the entire rollover duration in a steer-induced maneuver. The study primarily aimed to study the general impact of Rollover-Activated Side Curtain Airbags (RSCAs) which are mandated to be installed in all new cars, as well as seatbelt pretensioners. Two HIII 50th percentile male ATDs with pedestrian pelvises were positioned in upright postures in the front two passenger seats in a pickup truck, equipped with the RSCA and pretensioner. A similar setup was used for a sedan, with two HIII ATDs in passenger seats but the sedan was not equipped with either RSCAs or a seatbelt pre-tensioner. ATDs were secured using a three-point seatbelt restraint system and instrumented with head accelerometers, 6-channel upper and lower neck load cells as well as chest and lumbar spine sensors.
The sedan was accelerated to a speed of 105 kph before inducing a rollover that lasted 3-½ revolutions in the passenger side leading direction over a 29 m distance. The driver-side ATD made contact with the front window glass (which made contact with the ground and shattered) followed by partial ejection and the peak Nij (or Neck Injury IARV) was 1.13 in compression-extension during torso-augmentation whereby the torso keeps moving towards the neck even though the neck is obstructed in moving due to physical obstructions.
.png)
The pickup truck was accelerated to a speed of 116kph before it was released to rollover and the RSCA and pretensioner both deployed at a 32.3-degree roll angle, so before a full rollover and without causing significant neck load to the ATDs. During the entire roll sequence, the Nij (or Neck Injury IARV) never exceeded the critical value of 1, and compressive neck loading never exceeded the critical 4000N level, but the ATD head (and face directly) was in contact with the roof of the vehicle.
This Heller et. al study further underscored the conclusion that compressive neck (or C-spine) injuries in rollovers are a result of neck loading produced by the motion of a torso mass towards a stopped head (i.e. diving injuries which when in a specific head-neck alignment are referred to as torso augmentation).
With the advent of computational FE models, a lot of the financial challenges to designing repeatable, robust physical test setups to study rollovers can be circumvented. Hu et. al (2008) study analyzed the changes in neck injury risk depending on different impact orientations, velocities, friction, etc as well as passive and active muscle forces. The FE human head-neck model was based on a 29-year-old male with a height of 1.74m and weight of 75kg. The final model was composed of 32,135 elements and 23,933 nodes in the Ansys LS-DYNA environment. The different material properties used in the model can be seen in Table 4.
Problems and Challenges
The primary challenge is the lack of a specialized Anthromorphic Test Device (ATD) for rollover testing. The Hybrid III is built for frontal impact testing, the BIOSID and ES-2 dummies are built for side impact testing, the BIORID is built for rear impact testing and so none of these ATDs cannot mimic PMHS kinematics (or biofidelic response) of Post-Mortem Human Surrogates (PMHS) or cadaveric samples in rollover crashes (Funk et al., 2012). The secondary challenge is the variability of the injury pattern and severity due to small changes in the boundary (or initial) conditions of the test setup. In other words, there is no consensus in the biomechanics community as to what test setup accurately and comprehensively replicates a realistic rollover-like condition for the specimens inside the vehicle.
The use of cadaveric samples is common in the literature revolving analysis of head-first impacts (HFI) and spine injury. Experimental studies have shown using full-body PMHS specimens results in significantly varying injury patterns with variations in initial conditions (Morgan et al., 2023). Alternatively, an approximated mass can be used as a substitute for the torso in a setup known as the Torso-Surrogate Mass (TSM) boundary condition which should allow consistent geometry and biomechanics when the spine is in compression. This has been used in multiple studies and an example included a quasi-static (very slow) loading pattern (Myers et al.,1991) with varying degrees of head boundary conditions: fully constrained, fully unconstrained in the sagittal plane (rotation+translation), unconstrained laterally only, and found significant repeatable differences in injuries.
.png)
The difference in biofidelic response between ATDs and PMHS in head-first impact has already been shown in the previous section (Refer to Previous Work), but the Morgan et al. study aims to validate the latest computational studies built off of the TSM boundary condition experimental models. The latest studies have advanced modelling for bone, muscle, and ligament and also consider multivariate parameters important in an HFI, such as Coefficient of Friction, impact velocity, etc. Morgan et al. compare these computation models, both Full Body (FB) and TSM, with the Global Human Body Models Consortium 50th percentile male ATD (GHBMC M50-O). As can be seen in Figure 4, the GHBMC M50-O ATD was used both as an FB model and the head and neck were separated as a TSM model, and in the image, the red represents muscle while gray represents gray tissue. Of note, the FB model was free to move and was unconstrained. The exact boundary conditions in the simulation for each TSM and FB were based on the actual GHBMC TSM and FB models. The curvature index of the spine was computed as the simulations ran to determine impact kinematics. The total force between the head and impact plate was also measured and any hard tissue failure can be predicted easily in the model. At the same time, the elements in the bones of GHBMC M50 are deleted beyond principal strain indicating bone injury. The different measurements for each computational simulation can be seen in table 5.
We can see that for HFI, the FB condition reduced axial load significantly (79 corroded elements mostly in the C2 transverse) vs TSM boundary condition (1121 corroded elements mostly at anterior C2) even though the head-to-plate force is not drastically different. This suggests that the rigid assumption of the thoracic spine in the TSM model leads to decreased kyphosis and this means that more loads are transferred to the neck. Hence, the resultant compression-extension failure patterns between the two models show that the initial boundary conditions are highly determinative of the ultimate injury risk and no two rollovers are quite the same even if broadly speaking, the impact velocity or the number of rolls etc are similar.
As outlined in the Heller et al. study, newer vehicles equip Rollover-activated Side curtain airbags (RSCAs) and seatbelt pretensioners which affect the impact dynamics drastically for occupants in rollovers. To reiterate, this study used a non-production prototype RSCA in a pickup truck on both the driver's side and the passenger's side. It also used a sedan with its side curtain airbags removed. Two HIII 50th percentile male ATDs were positioned in upright seated postures in each of the two vehicles. An automatic steer controller was used to cause a steer-induced maneuver resulting in a rollover. Only the ATD in the sedan without the RSCAs experienced Nij values and peak compressive forces exceeding the IARV as well as partial ejection.
.png)
As can be seen in Figure 5 with the red dots signifying the passenger and driver ATD in sedan only. However, we can see new features like this further complicate analysis of the spinal injury in rollovers literature since most new cars have these features yet most of the literature is not fully cognizant of these additional safety features. In this case, the RSCA and pre-tensioner were unable to prevent head-roof contact for the ATDs and if anything, may not decrease peak neck forces during this contact period since lateral movement is restricted by the airbag.
This presents a new dimension to the already somewhat contentious findings being drawn from rollover studies. There are newer and more advanced safety features in modern vehicles, especially physical features such as side airbags or perhaps roof structures made of a certain alloy that is less stiff and may have a lower coefficient of friction. These mean that the initial boundary conditions are constantly changing for rollovers happening at roughly the same velocity or in the same dynamic manner and the findings from the existing literature cannot be successfully translated or extrapolated to these vehicles and the occupants in them. While the FMVSS standards are constantly being updated as well(Heller et all) to incorporate more of these safety features, we are not fully aware of the far-reaching and full effect of these measures and how they play into the vehicle safety narrative.
Future Research
Ultimately, the data and literature presented in the preceding sections present a general overview of head-first rollover epidemiology and associated test data from PMHS and ATDs. However, with the rapidly evolving automotive landscape - with an increasing market share of SUVs, it is crucial to continue growing the epidemiological knowledge base, so automakers and policy markers can be alerted early to a potential increase in fatalities or injuries caused by automotive rollovers. There is also a knowledge gap in similar studies dedicated to electric vehicles.
The primary research needs to focus on the development of modifications or the creation of specialized ATDs for rollover scenarios only. These ATD’s biofield responses in vertex head-first impact (HFI) scenarios must be compared with existing PMHS data within controlled initial boundary conditions. These ATDs must also be biofidelic when anthropomorphically scaled to match different physiological parameters such as body mass, height bone mineral density etc. If possible, the soft tissue structures replicated on the ATDs should be up to contemporary standards as well by the existing head, neck and spine literature. This can help with controlling for the effect of neck musculature in rollover impact and injury studies.
Similarly, the epidemiology of automobiles with new technologies aimed at reducing the severity and likelihood of rollover fatalities is not easily accessible or collated. Examples of these new technologies include seatbelt pre-tensioners and rollover-activated side curtain airbags (RSCAs). This highlights the need to not only conduct studies on recent data but also to pay particular attention to newer vehicle models and types - to provide automakers with real-world data on the efficacy of their safety improvements.
Studies on the epidemiology of rollovers and PMHS tests generally compare and present data on high AIS injuries to the head, spine, and thorax. While these injury types constitute a bulk of the rollover fatalities, non-fatal injuries involving musculature have not been studied in detail although most rollovers lead to non-fatal injuries. A potential cause of this lack of study is the challenge of doing accurate studies on musculature without PMHS. Specifically, an area that should be developed for cervical spine injuries is neck musculature injuries and impacts due to rollovers.
Lastly, there is a clear correlation between taller occupants in rollovers and the severity of injury due to closer proximity to the roof. This has not been studied with all other factors controlled for and is a potential for further studies. This could be used to validate the ATDs specialized for rollovers currently in development.
Conclusion
The impact of rollovers on spinal cord injuries is a critical aspect affecting individuals' quality of life. The prevalence of cervical spine injuries in rollovers, as discussed at the beginning of this report, emphasizes their heightened risk compared to frontal and side crashes. Despite the higher incidence of spine injuries over spinal cord injuries, the latter remains a significant concern.
The escalating number of rollover accidents, evident in the Dobbertin et al. study analysis, underlines the pressing need for comprehensive research. Fatalities associated with rollovers, coupled with the rising market share of SUVs, underscore the urgency for improved safety measures. The correlation between roof crush magnitude and the likelihood of head, neck, and spine injuries further accentuates the need for advanced safety standards.
Existing studies utilizing Anthropomorphic Test Devices reveal limitations in biofidelity, especially in predicting cervical spine injuries during rollovers. The inadequacy of Injury Assessment Relative Value (Nij) as an indicator and the complexity of factors affecting injury patterns necessitate a nuanced approach to evaluating safety mechanisms. Furthermore, the introduction of new safety features such as Rollover-Activated Side Curtain Airbags and seatbelt pre-tensioners adds complexity to injury dynamics, demanding continuous adaptation of safety standards.
Future research directions should prioritize the development of specialized Anthromorphic Test Devices tailored for rollover scenarios. Bridging the gap between epidemiological knowledge and advancements in vehicle technologies, particularly in electric vehicles, is crucial. Additionally, exploring non-fatal injuries involving musculature and considering the impact of occupant height on injury severity are avenues for further investigation. In the ever-evolving automotive landscape, a holistic understanding of rollover injuries remains imperative for enhancing safety protocols and minimizing the impact on individuals' lives.
References
Data for the Automotive Trends Report. U.S. Environmental Protection Agency. (2023, November 22). https://www.epa.gov/automotive-trends/data-automotive-trends-report
Dobbertin, K. M., Freeman, M. D., Lambert, W. E., Lasarev, M. R., & Kohles, S. S. (2013). The relationship between vehicle roof crush and head, neck and spine injury in rollover crashes. Accident Analysis & Prevention, 58, 46–52. https://doi.org/10.1016/j.aap.2013.04.020
El-Menyar, A., Latifi, R., Parchani, A., Peralta, R., Tuma, M., Zarour, A., Abdulrahman, H., Al-Thani, H., Asim, M., & El-Hennawy, H. (2014). Epidemiology, causes and prevention of car rollover crashes with ejection. Annals of Medical and Health Sciences Research, 4(4), 495–502. https://doi.org/10.4103/2141-9248.139279
Fakharian, E., Mohammadzadeh, M., Saberi, H., Fazel, M., Rejali, M., Akbari, H., Mirzadeh, A., & Mohammadzadeh, J. (2022). Spinal injury resulting from car accident: Focus to prevention. Asian Journal of Neurosurgery, 12(02), 180–184. https://doi.org/10.4103/1793-5482.152110
Funk, J. R., Cormier, J. M., & Manoogian, S. J. (2012b). Comparison of risk factors for cervical spine, head, serious, and fatal injury in rollover crashes. Accident Analysis & Prevention, 45, 67–74. https://doi.org/10.1016/j.aap.2011.11.009
Heller, M., Sharpe, S., Newberry, W., Dibb, A., Zolock, J., Croteau, J., Carhart, M., Kerrigan, J., & Clauser, M. (2015). Occupant kinematics and injury response in steer maneuver-induced furrow tripped rollover testing. SAE International Journal of Transportation Safety, 3(2), 164–215. https://doi.org/10.4271/2015-01-1478
Hu, J., Yang, K. H., Chou, C. C., & King, A. I. (2008). A numerical investigation of factors affecting cervical spine injuries during rollover crashes. Spine, 33(23), 2529–2535. https://doi.org/10.1097/brs.0b013e318184aca0
McElhaney, J. H., & Myers, B. S. (1993). Biomechanical aspects of cervical trauma. Accidental Injury, 311–361. https://doi.org/10.1007/978-1-4757-2264-2_14
Morgan, M. I., Corrales, M., Cripton, P., & Cronin, D. S. (2023). Effect of Torso Boundary Conditions on Spine Kinematic and Injury Responses in Head-First Impact Assessed with a 50th Percentile Male Human Body Model. SAE International Journal of Transportation Safety, 11(2), 151–156. https://doi.org/10.4271/09-11-02-0014
Roberts, C. W., Toczyski, J., & Kerrigan, J. R. (2019). Cervical spine injury in rollover crashes: Anthropometry, excursion, Roof Deformation, and ATD prediction. Clinical Biomechanics, 64, 42–48. https://doi.org/10.1016/j.clinbiomech.2018.04.004
Thompson-Bagshaw, D. W., Quarrington, R. D., Dwyer, A. M., Jones, N. R., & Jones, C. F. (2023). The structural response of the human head to a vertex impact. Annals of Biomedical Engineering, 51(12), 2897–2907. https://doi.org/10.1007/s10439-023-03358-z
Wu, J., Summers, S., Ridella, S., Lee, E., Kang, T., & Myers, J. (2018, November 30). Occupant injuries related to rollover crashes and ejections from recent Crash data. TRID. https://trid.trb.org/view/1757664
Zhang, Q., Kerrigan, J., Lessly, D., Seppi, J., Riley, P., Lockerby, J., Overby, B., Sowers, C., & Crandall, J. (2013). Whole-body kinematics: Response comparison of the hybrid III and hybrid III pedestrian ATD in DRoTS rollover tests. International Research Council on the Biomechanics of Injury Conference, 330–367. https://www.scopus.com/record/display.uri?eid=2-s2.0-84896669189&origin=inward