You are here

Article Text

Article menu

Download PDFPDF

Original Article
Motor vehicle mismatch: a national perspective
  1. Samuel P Mandell1,
  2. Christopher D Mack2,
  3. Eileen M Bulger1
  1. 1Department of Trauma Surgery, Harborview Medical Center, Seattle, Washington, USA
  2. 2Harborview Injury Prevention and Research Center, University of Washington, Seattle, Washington, USA
  1. Correspondence to Samuel P Mandell, HIPRC Box 359960 Seattle, WA 98104-1520, USA; mandells{at}u.washington.edu

Abstract

Objective To determine the RR of the severe injury associated with light truck vehicle (LTV) versus passenger vehicle (PV) mismatch following motor vehicle collisions across the USA.

Methods This was a retrospective cohort study with the primary outcomes of Injury Severity Score (ISS) >8 and body region Abbreviated Injury Score (AIS) >2. The National Automotive Sampling System Crashworthiness Data System (NASS CDS) was searched for occupants in frontal and side impact crashes from 1993 to 2007. Occupants in PVs struck by LTVs were compared to PV occupants struck by another PV. Poisson regression was used to estimate the RR of severe injury after adjusting for driver age, driver gender, and change in velocity during the crash (∆v). Because 21% of cases were missing ∆v, multiple imputation was used to estimate the missing values. NASS CDS weights were used to estimate the risk of severe injury nationally.

Results PV occupants in front impact crashes with an LTV as the striking vehicle were at increased risk of severe injury compared to those struck by another PV (RR 1.37, 95% CI 1.09 to 1.73). A similar increase risk was observed in side impact crashes (RR 1.34, 95% CI 1.12 to 1.62). Increased risk of injury was also identified in several body regions.

Conclusions Motor vehicle mismatch crashes are associated with a significant increase in risk of severe injury for PV occupants in the USA. Addressing vehicle compatibility remains an important issue for occupant safety.

  • MVTC
  • driver
  • passenger

https://doi.org/10.1136/ip.2009.026047

Statistics from Altmetric.com

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    Introduction

    Over the last three decades the United States vehicle fleet has undergone significant changes. Pickups, minivans, vans, and sport utility vehicles (SUVs), together classified as light truck vehicles (LTVs), make up a larger portion of vehicles on the road today.1 According to data from the US Department of Energy, new LTV sales rose considerably from 1980 to 2007 (figure 1), with the most dramatic increase in the largest vehicle categories.2

    Figure 1
    Figure 1

    New light truck vehicle (LTV) sales 1980–2007, in thousands of vehicles.

    These changes have led to safety concerns. Fatalities of occupants in passenger vehicles (PVs) struck by LTVs have been increasing while those of PV occupants struck by other PVs have been declining.1 3 One suggested reason for this trend is the incompatibility of LTVs and PVs. Compatibility is composed of the combined effects of the struck vehicle's ability to protect occupants and the striking vehicle's aggressivity. Aggressivity of a vehicle class has been defined as driver fatalities in the collision partner divided by the number of crashes involving the striking vehicle class.4 The National Highway Traffic Safety Administration (NHTSA) has found that the most aggressive vehicles on the road in the 1990s were LTVs.4 While PVs have become safer, many safety features have been engineered for crashes involving other PVs. When an LTV strikes a PV, the increased weight and vehicle stiffness serve to protect occupants of LTVs and increase their aggressivity, while the lower frame height of PVs places occupants at greater risk.1 4–11

    A case series of PV–LTV crashes within the Crash Injury Research and Engineering Network (CIREN) found that occupants in PVs struck from the side by LTVs had more frequent head and thorax injuries. These injuries were attributed to the bumper of the LTV overriding the side door reinforcements of the PV. In frontal impacts, severe extremity injuries were more frequent in occupants of PVs compared with LTVs.6 Other studies have shown that PV drivers are at increased risk of severe injury when struck by LTVs, but have been limited to drivers and have not identified specific injury patterns.12 13 One case–control study identified mismatch as a risk factor for spine injury.14 To date, no study has examined all these injury patterns on a national level. Further, the number of injuries that may be due to motor vehicle mismatch on US roadways has not been determined.

    The purpose of this study was twofold. First, to determine the RR of moderate to severe injury associated with LTV versus PV mismatch across the USA. Second, to estimate the proportion of those injuries associated with characteristics of the striking vehicle for occupants of PVs that are struck by LTVs compared to those struck by PVs. Assuming the associations are causal, these estimates would help to determine the number of severe injuries that could be prevented by addressing mismatch.

    Methods

    Data source

    The National Automotive Sampling System Crashworthiness Data System (NASS CDS) is a nationwide crash data collection program sponsored by NHTSA. A CDS crash must be police reported, involve a harmful event (property damage and/or personal injury) resulting from a crash, and involve at least one towed PV or LTV in transport on a traffic way. All injuries are documented and scored using the Abbreviated Injury Severity (AIS)-90 coding.15 16 Injury scores are available for each body region. Ten sampling strata are used, and crashes resulting in fatalities and serious injuries are over-sampled. Currently, data are collected in 27 primary sampling units throughout the USA on vehicles involved in approximately 5000 automobile crashes annually. Weights calculated from the probability of being sampled allow the data to represent all police-reported, tow-away crashes in the USA.

    Analysis

    This was a retrospective cohort study in which the exposure was being struck by a LTV and the main outcome measure was moderate to severe injury. We searched the CDS for occupants of PVs who were involved in frontal or side impact crashes from 1993 to 2007. A frontal impact was defined as a being struck with a principal direction of force between 11 o'clock and 1 o'clock (12 o'clock denoting directly ahead of the vehicle), while a side impact was either between 2 o'clock and 4 o'clock or 8 o'clock and 10 o'clock. Occupants were excluded if they were involved in a crash with more than two vehicles, or if a rollover occurred. Only occupants seated near the impact of the struck vehicle were studied. For front impacts, occupants included were the driver and front seat passenger, and for side impacts, they were nearside front or rear seat occupants. Cohorts were then identified as either occupants in PV–PV crashes if the striking vehicle was a PV or occupants in a PV–LTV if the striking vehicle was an LTV. We also examined classes of LTV such as SUVs, vans, and pickup trucks separately. Moderate to severe injury was defined as overall injury severity score (ISS) >8, or an AIS >2 for individual body regions.

    All data were analysed using Stata V.10.1. RRs were estimated directly using Poisson regression. Applying Poisson regression to binary outcome data can overestimate the SE of the relative risks; however, this is overcome by using robust standard errors.17 18

    NASS CDS probability sampling was accounted for using the svy family of commands in conjunction with Poisson regression. The weight provided with the CDS data was used to produce national estimates. However, the highly variable survey weights yielded inefficient estimates when applied to small subsamples. Using the method described by Korn and Graubard, the efficiency of the survey weights was calculated using the equation inefficiency ≈(CV)2/(1+(CV)2), where CV is the coefficient of variation. The inefficiency in the NASS CDS was approximately 90%. To place this in context, 75% inefficiency would lengthen the CI around an RR from 0.1 to 0.2. Given that a meaningful RR on a national scale may be a small increase, such as from 1.0 to 1.5, 90% inefficiency could significantly increase the variance and the CI. Therefore, the survey weights were truncated at the 95th percentile.19–21 This increase in efficiency comes at some cost. It is likely that the most heavily weighted crashes will be under-represented in the weighted sample. Inferences drawn from the data may, therefore, be most applicable to a subset of more severe crashes, as these tend to have lower sampling weights.

    Crash severity is also an important predictor of outcome in motor vehicle crashes and is related to the velocity of each vehicle at the time of the collision. Often, this information is not known. Crash investigators therefore calculate the relative change in the velocity of the vehicles, known as Δv. This is done in three ways. First, if both vehicles in the crash are available, crash data is entered into a reconstruction program that then estimates the Δv.22 However, it is not always possible to examine both vehicles in a crash; in such cases, the program estimates the Δv based on a how the observed amount of vehicle damage corresponds to what would occur in a crash into a solid barrier, known as the barrier equivalent speed. Finally, when crash simulation software cannot be used, the investigator can estimate the Δv based on crash scene observations. This occurs with significant over- or under-ride of the two vehicles, such as in mismatch crashes. The stiffest portions of the vehicles do not line up and the vehicles do not contact bumper to bumper. Severe deformation can then occur in softer areas of the struck vehicle. Using the deformation at the bumper level would under estimate Δv, but using the intrusion in the softer areas may overestimate Δv. To correct for this, crash investigators can average the two areas of intrusion when estimating Δv.22 Δv therefore captures aspects of frame height and stiffness related to mismatch. Unfortunately, neither frame height nor stiffness is coded for in CDS and their effects cannot be individually examined. Because of the importance of Δv as a validated predictor of crash severity, it was included as a covariate in an adjusted model.23

    Information from total Δv, barrier equivalent speed, and investigator-estimated Δv, were combined into a single categorical variable. This ensured that all known information for each crash was used. However, even after this, Δv was missing for 21% of the occupants in the CDS. Rather than discard cases without Δv, which could introduce significant bias, the missing values were imputed by chained equations as described by Royston.24 25 This method treats the dataset as a simple random sample. The predictors used included the outcomes of ISS, severe injury based on AIS >2 for each body region, speed limit, curb weight, number of injured, maximum injury severity in the vehicle, driver zip code, relationship of the crash to an intersection, number of lanes, light conditions, traffic flow, direction of force, extent of damage, magnitude of intrusion, age, sex, ejection, entrapment, treatment at a medical facility, and maximum AIS. Five imputations were performed to impute the missing values of Δv. Subsequent analyses were then performed on each imputed dataset and regression results were combined using the mim command.

    Aside from Δv, previous studies have identified other important covariates in motor vehicle crashes. Men and younger drivers, are more likely to engage in high-risk behaviours.26 At the same time, women in a crash may be at increased risk of injury or death.27 28 For both genders, seatbelt use remains an important measure for mitigating risk of death and injury and must be taken into account.27 A multivariate regression model using driver age, driver gender, occupant age, airbag availability, model year, and occupant restraint use was used to estimate an adjusted RR for striking-vehicle type. A restrained occupant was defined as an individual properly wearing a lap, lap and shoulder, or automatic belt (depending on belt availability), or a child in a proper child restraint system.

    Finally, in order to estimate the proportion of injuries that could be attributed to motor vehicle mismatch, the attributable risk per cent (AR%=1−RR/RR) was calculated.

    Results

    In the study period of 1993–2007, there were a total of 76 625 occupant records in CDS. Of these, 7698 were crashes where an LTV struck a PV, with 5017 frontal impacts and 2681 side impacts. Nationally, this represents an estimated 1 710 000 PV occupants in PV–LTV (‘exposed’) crashes compared with 3 590 000 occupants represented in PV–PV (‘unexposed’) crashes (table 1).

    View this table:
    Table 1

    Number of impacts by exposure type

    Mean age was similar in both groups. There were more female occupants than male occupants in both the exposed and unexposed cohorts. The distribution of sex within the cohorts, however, was the same, with 56% being women and 44% being men (table 2).

    View this table:
    Table 2

    Demographics of exposed groups

    In frontal crashes, PV occupants struck by an LTV had an increased risk of moderate to severe injury (adjusted RR 1.37) compared to those struck by a PV. These occupants also had an increased risk of head, thorax, abdominal, upper extremity, and lower extremity injuries (table 3). After adjustment, RRs for all of the injuries were statistically significant, except for head injuries. While not statistically significant, vans as the striking vehicle appear to carry the highest risk.

    View this table:
    Table 3

    Crude and adjusted RR of injury for occupants in frontal impact PV–PV crashes compared to PV–LTV crashes; adjustment includes driver age, driver gender, and Δv

    Using the adjusted RR of 1.37 for severe injury in frontal crashes yielded an attributable risk per cent of 27% of severe injuries potentially due to mismatch (table 4). For thoracic injuries, 29% of severe injuries could be attributed to motor vehicle mismatch. Abdominal injuries had the highest AR% due to mismatch at 55%. Both upper and lower extremity injuries could be attributed to mismatch in crashes with AR% of 25% and 36%, respectively.

    View this table:
    Table 4

    Attributable risk (AR) due to mismatch in frontal crashes

    Examining side impacts also demonstrated an increased risk of severe injury to PV occupants struck by LTV (adjusted RR 1.34). Severe head injury, thoracic, and lower extremity injuries also showed a statistically significant increase in risk (table 5). Following adjustment for covariates, only overall severe injury and head injury remained significant. Pickups carried the most risk for overall injury; however, for head injury, vans had the highest risk.

    View this table:
    Table 5

    Crude and adjusted RR of injury for occupants in side impact PV–PV crashes compared to PV–LTV crashes; adjustment includes driver age, driver gender, and Δv

    With an adjusted RR of 1.34, severe injury for side impacts with LTVs had an AR% of 25%. This indicates that one quarter of severe injuries in side impact mismatch crashes were potentially due to the mismatch between vehicles (table 6). For head injuries, 34% of injures to occupants of PV in mismatch crashes were potentially attributable to vehicle mismatch.

    View this table:
    Table 6

    Attributable risk (AR) due to mismatch in side impact crashes

    Discussion

    In 2004, Acierno et al performed a case review looking at the patterns of injury found in mismatch crashes. Based on their results of PVs struck by an LTV from the front, lower extremity, head, and chest injuries were relatively common. For side impacts, head and chest injuries predominated.6 The findings here for front impacts are consistent with what would be expected based on that case series, showing increased risk for overall injury, head, chest, and lower extremities. However, for side impacts, there was no statistically significant increase in risk of thoracic injury, while there were increased risks for overall severe injury and head injuries.

    One explanation for these different findings may be increased safety measures. In 2003, a collaboration of automakers agreed to work together to make LTV more compatible with smaller cars in crashes. In particular, the goal was to make sure that the stiff, energy absorbing structures within the vehicles were aligned in crashes. As early as 2004, reductions in death rates were being noted in vehicles which had more compatible structures.29 However, given our adjustment for model year, this cannot be the entire explanation. Unfortunately, the automakers have recently decided to disband their voluntary cooperative effort. This is of concern as this study demonstrates that there are still many injuries associated with mismatch between vehicles.

    While efforts to make vehicle geometry more compatible may lead to the reduction of some injuries, the weights of the vehicles have not substantially changed.7 Weight has been linked to both better occupant protection and higher vehicle aggresivity.10 30 It may be that the mass of the striking vehicle is the major contributor of injury in some mismatch crashes.

    Additionally, it may be that different types of LTVs are responsible for varying amounts of injury. A Canadian study has demonstrated that occupants of vehicles struck by pickups, for example, have a significantly increased risk of death when compared to those struck by SUVs.13 Our results suggest similar differences.

    A cohort study of crashes in New Zealand in 2005–06 found that after adjusting for driver behaviour, age, and distance driven, that SUVs did not pose an increased safety risk compared to PVs on New Zealand roads.31 Similarly, in this study, adjusting for driver age and gender as proxies for risk taking behaviour, and Δv, the relative risks of injury were considerably lower and in several cases became not statistically significant.

    Despite the encouraging data that the risks posed by LTV may be decreasing, this national sample still estimated a significant increase in the risk of injury due to vehicle incompatibility. Further, this study was limited to incompatibility between larger LTVs and smaller PVs. New laboratory crash tests, however, have demonstrated that similar incompatibility may occur between standard PVs and the mini and micro cars that are increasing in popularity.32 This would indicate that by focusing on LTVs alone, this and previous studies have only been examining part of the problem and the true risks of injury due to mismatch would be greater if these PV–PV crashes were evaluated for mismatch. Such an evaluation may be difficult within the NASS CDS, as the actual frame geometry and stiffness are not recorded.

    There are several limitations to this study. As has already been discussed, frame geometry and stiffness were not available in the database. However, due to the way that Δv was estimated by crash investigators, it may have contained elements of both geometry and stiffness. This ‘contamination’ of Δv may be adjusting the regression model for some aspects of the exposure of interest. In addition, while Δv is a validated proxy for crash severity, it is not completely accurate. Bias may also be introduced because Δv was estimated differently given different crash situations.

    The weighting system of the national sample may also limit the broader application of this study. Due to the oversampling of severe crashes in the NASS CDS and the relatively large population from which the sample is drawn, the range of weights was extreme (0–72 000). While the truncation of the weights increased the efficiency and the precision of point estimates, this may come at the cost of accurately representing the overall population of crashes on US roadways. It is possible that the results over represent more severe crashes.

    Despite these limitations, mismatch is still associated with a significant amount of severe injury in the USA. As the auto manufacturers have reduced their efforts to improve vehicle compatibility, independent and government research in this area will take on more importance. In the future, efforts to classify the risk of individual types of vehicles should be undertaken. In addition, the risks of crashes between classes of vehicles other than PV and LTV should be examined.

    Lowering the stiffness of structural aspects of LTV to make them more compatible with cars as well as increasing door reinforcements may help improve compatibility. These features have already been introduced by the auto industry, but strict regulations mandating them have not followed.7 29 Promotion of these efforts as well as examination of ways to decrease the influence of vehicle mass will help ensure even safer passenger vehicles in the future.

    What is already known on the subject

    • There are increasing numbers of light truck vehicles on US roadways.

    • Fatalities of occupants in passenger vehicles struck by light truck vehicles are rising.

    What this study adds

    • The risk of injury to occupants of passenger vehicles struck by light truck vehicles using national data has been quantified.

    • The proportion of injuries that could be prevented by addressing vehicle mismatch has been estimated.

    Acknowledgments

    The authors would like to thank Frederick Rivara, Thomas Koepsell, and Robert Kaufman for their advice and assistance.

    References

    1. ↵
      1. IIHS
      . Vehicle incompatibility in crashes [Status Report]. Arlington, VA, USA: Insurance Institute for highway safety, 2005;40.
    2. ↵
      1. Davis S,
      2. Diegel S,
      3. Boundy R
      . Transportation energy databook. 27 edn. Oak Ridge, TN: U.S. Department of Energy, Office of Energy Efficiency and Renewable Energy, 2008.
    3. ↵
      1. Summers S,
      2. Hollowell W,
      3. Prasad A
      . NHTSA's Research Program for Vehicle Compatability Paper #307. Eighteenth International Conference on the Enhanced Safety of Vehicles. Nagoya, Japan, 2003.
    4. ↵
      1. Gabler HC,
      2. Hollowell W
      . NHTSA's Vehicle Aggressivity and Compatability Research Program: Paper No. 98-S3-O-01. Sixteenth International Conference on the Enhanced Safety of Vehicles. Windsor, Canada, 1998.
    5. ↵
      1. Loftis K,
      2. Martin RS,
      3. Meredith JW,
      4. et al
      . Investigating methods for determining mismatch in near side vehicle impacts—biomed 2009. Biomed Sci Instrum 2009;45:256–61.
      OpenUrlPubMed
    6. ↵
      1. Acierno S,
      2. Kaufman R,
      3. Rivara FP,
      4. et al
      . Vehicle mismatch: injury patterns and severity. Accid Anal Prev 2004;36:761–72.
      OpenUrlCrossRefPubMedWeb of Science
    7. ↵
      1. Baker BC,
      2. Nolan JM,
      3. O'Neill B,
      4. et al
      . Crash compatibility between cars and light trucks: benefits of lowering front-end energy-absorbing structure in SUVs and pickups. Accid Anal Prev 2008;40:116–25.
      OpenUrlCrossRefPubMedWeb of Science
    8. ↵
      1. Kahane C
      . Vehicle weight, fatality risk and crash compatability of model year 1991-99 passenger cars and light trucks. Washington, DC: Department of Tansportation, 2003 Contract No: DOT HS 809 662.
    9. ↵
      1. Martin JL,
      2. Lenguerrand E
      . A population based estimation of the driver protection provided by passenger cars: France 1996–2005. Accid Anal Prev 2008;40:1811–21.
      OpenUrlCrossRefPubMedWeb of Science
    10. ↵
      1. Mayrose J,
      2. Jehle DV
      . Vehicle weight and fatality risk for sport utility vehicle-versus-passenger car crashes. J Trauma 2002;53:751–3.
      OpenUrlPubMedWeb of Science
    11. ↵
      1. Wood DP
      . Safety and the car size effect: a fundamental explanation. Accid Anal Prev 1997;29:139–51.
      OpenUrlCrossRefPubMedWeb of Science
    12. ↵
      1. Toy EL,
      2. Hammitt JK
      . Safety impacts of SUVs, vans, and pickup trucks in two-vehicle crashes. Risk Anal 2003;23:641–50.
      OpenUrlCrossRefPubMedWeb of Science
    13. ↵
      1. Fredette M,
      2. Mambu LS,
      3. Chouinard A,
      4. et al
      . Safety impacts due to the incompatibility of SUVs, minivans, and pickup trucks in two-vehicle collisions. Accid Anal Prev 2008;40:1987–95.
      OpenUrlCrossRefPubMedWeb of Science
    14. ↵
      1. Cobb JD,
      2. MacLennan PA,
      3. McGwin G Jr.,
      4. et al
      . Motor vehicle mismatch-related spinal injury. J Spinal Cord Med 2005;28:314–19.
      OpenUrlPubMedWeb of Science
    15. ↵
      AAAM abbreviated injury scale, 2005 revision. Barrington, IL: Association for the Advancement of Automotive Medicine, 2005.
    16. ↵
      1. Revell M,
      2. Pynsent PB,
      3. Abudu A,
      4. et al
      . Trauma scores and trauma outcome measures. Trauma 2003;5:61–70.
      OpenUrlAbstract/FREE Full Text
    17. ↵
      1. Cummings P
      . Methods for estimating adjusted risk ratios. Stata Journal 2009;9:175–96.
      OpenUrlWeb of Science
    18. ↵
      1. Zou G
      . A Modified poisson regression approach to prospective studies with binary data. Am J of Epidemiol 2004;159:702–6.
      OpenUrlAbstract/FREE Full Text
    19. ↵
      1. Korn E,
      2. Graubard B
      . Analysis of large health surveys: accounting for sampling design. J R Stat Soc 1995;158:263–95.
      OpenUrl
    20. ↵
      1. Rubin DM,
      2. Downes KJ,
      3. O'Reilly AL,
      4. et al
      . Impact of kinship care on behavioral well-being for children in out-of-home care. Arch Pediatr Adolesc Med 2008;162:550–6.
      OpenUrlCrossRefPubMedWeb of Science
    21. ↵
      1. Rubin DM,
      2. O'Reilly AL,
      3. Luan X,
      4. et al
      . The impact of placement stability on behavioral well-being for children in foster care. Pediatrics 2007;119:336–44.
      OpenUrlAbstract/FREE Full Text
    22. ↵
      NHTSA. National automotive sampling system 1997 crashworthiness data system data collection, coding, and editing manual. Washington, DC: Department of Transportation, 1997.
    23. ↵
      1. Gabauer DJ,
      2. Gabler HC
      . Comparison of roadside crash injury metrics using event data recorders. Accid Anal Prev 2008;40:548–58.
      OpenUrlPubMedWeb of Science
    24. ↵
      1. Raghunathan TE
      . What do we do with missing data? Some options for analysis of incomplete data. Annu Rev Public Health 2004;25:99–117.
      OpenUrlCrossRefPubMedWeb of Science
    25. ↵
      1. Royston P
      . Multiple imputation of missing values. Stata J 2004;4:227–41.
      OpenUrl
    26. ↵
      1. Turner C,
      2. McClure R
      . Age and gender differences in risk-taking behaviour as an explanation for high incidence of motor vehicle crashes as a driver in young males. Inj Control Saf Promot 2003;10:123–30.
      OpenUrlCrossRefPubMed
    27. ↵
      1. Bedard M,
      2. Guyatt GH,
      3. Stones MJ,
      4. et al
      . The independent contribution of driver, crash, and vehicle characteristics to driver fatalities. Accid Anal Prev 2002;34:717–27.
      OpenUrlCrossRefPubMedWeb of Science
    28. ↵
      1. Ulfarsson GF,
      2. Mannering FL
      . Differences in male and female injury severities in sport-utility vehicle, minivan, pickup and passenger car accidents. Accid Anal Prev 2004;36:135–47.
      OpenUrlCrossRefPubMedWeb of Science
    29. ↵
      IIHS. Having improved compatibility, automakers disband cooperative effort to address this issue [status report]. Arlington, VA, USA: Insurance Institute for highway safety, 2009;44.
    30. ↵
      1. Broyles RW,
      2. Clarke SR,
      3. Narine L,
      4. et al
      . Factors contributing to the amount of vehicular damage resulting from collisions between four-wheel drive vehicles and passenger cars. Accid Anal Prev 2001;33:673–8.
      OpenUrlCrossRefPubMedWeb of Science
    31. ↵
      1. Keall MD,
      2. Newstead S
      . Are SUVs dangerous vehicles? Accid Anal Prev 2008;40:954–63.
      OpenUrlCrossRefPubMedWeb of Science
    32. ↵
      IIHS. Special issue: car size, weight, and safety [status report]. Arlington, VA, USA: Insurance Institute for highway safety, 2009;44.

    Footnotes

    • Funding SPM is funded by the Eastern Association for the Surgery of Trauma Wyeth/EAST Foundation Scholarship.

    • Competing interests None.

    • Provenance and peer review Not commissioned; externally peer reviewed.