American Journal of Epidemiology Advance Access originally published online on December 12, 2007
American Journal of Epidemiology 2008 167(5):546-552; doi:10.1093/aje/kwm336
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
ORIGINAL CONTRIBUTIONS |
Using Head-on Collisions to Compare Risk of Driver Death by Frontal Air Bag Generation: A Matched-Pair Cohort Study
1 National Study Center for Trauma and Emergency Medical Systems, Epidemiology and Preventive Medicine, University of Maryland School of Medicine, Baltimore, MD
2 War-Related Illness and Injury Study Center, Washington DC VA Medical Center, Washington, DC
Correspondence to Dr. Elisa R. Braver, National Study Center for Trauma and EMS, 701 West Pratt Street, Room 526, Baltimore, MD 21201 (e-mail: elisabraver{at}gmail.com).
Received for publication June 26, 2007. Accepted for publication October 23, 2007.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
US air bag regulations were changed in 1997 to allow tests of unbelted male dummies in vehicles mounted and accelerated on sleds, resulting in longer crash pulses than rigid-barrier crashes. This change facilitated depowering of frontal air bags and was intended to reduce air bag–induced deaths. Controversy ensued as to whether sled-certified air bags could increase adult fatality risk. A matched-pair cohort study of two-vehicle, head-on, fatal collisions between drivers involving first-generation versus sled-certified air bags during 1998–2005 was conducted by using Fatality Analysis Reporting System data. Sled certification was ascertained from public information and a survey of automakers. Conditional Poisson regression for matched-pair cohorts was used to estimate risk ratios adjusted for age, seat belt status, vehicle type, passenger car size, and model year for driver deaths in vehicles with sled-certified air bags versus first-generation air bags. For all passenger-vehicle pairs, the adjusted risk ratio was 0.87 (95% confidence interval: 0.77, 0.98). In head-on collisions involving only passenger cars, the adjusted risk ratio was 1.04 (95% confidence interval: 0.85, 1.29). Increased fatality risk for drivers with sled-certified air bags was not observed. A borderline significant interaction between vehicle type and air bag generation suggested that sled-certified air bags may have reduced the risk of dying in head-on collisions among drivers of pickup trucks.
accidents, traffic; air bags; automobiles; government regulation; seat belts
Abbreviations: FARS, Fatality Analysis Reporting System; NHTSA, National Highway Traffic Safety Administration; SUV, sport utility vehicle
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
Frontal air bags reduce the risk of deaths in frontal crashes among drivers and adult right-front passengers (1–7). As of January 1, 2007, the National Highway Traffic Safety Administration (NHTSA) estimated that 22,855 lives had been saved by frontal air bags between 1987 and 2006 (8). Yet, frontal air bags also caused children and other small occupants to die or receive severe injuries in low-speed crashes that likely would have resulted in no injuries or minor injuries in the absence of an air bag (1, 4, 9, 10). Nonuse of seat belts, improper belt use, placing rear-facing child seats in front, or leaning toward air bags contributes to air bag–induced injury (4, 6, 11). NHTSA reported 175 air bag–related deaths among children, 90 among drivers, and 13 among adult front-seat passengers as of January 1, 2007 (8).
In response to the deaths and serious injuries caused by frontal air bags, NHTSA promulgated a temporary rule, effective for model year 1998, that gave automobile manufacturers the option of testing frontal crash performance for unbelted male dummies by using 30-mph sled tests rather than the previously mandated 30-mph head-on crash tests into rigid barriers (1 mile = 1.609 km) (12). Sled tests simulate crashes, without intrusion into the vehicle, by attaching a test vehicle to a moving platform. Allowing sled tests of frontal crash performance using unbelted male dummies facilitated depowering of frontal air bags because sled tests had a longer crash pulse, resulting in more time to inflate and less inflation energy, than rigid barrier tests, which required frontal air bags to deploy rapidly with high force to restrain dummies (13). Most automobile manufacturers elected to use sled certification starting in model year 1998, accounting for about 70 percent of all new passenger vehicles sold that model year (4).
Some safety organizations opposed the rule because they feared that depowered air bags could reduce protection of large occupants, particularly unbelted men in high-speed crashes (12). NHTSA also expressed concern about a potential trade-off between having fewer air bag–related deaths among small occupants but more deaths among unbelted adults and provided estimates of additional adult deaths that would occur from depowering air bags (4, 12).
Studies using different methods have examined the effects of depowering air bags. NHTSA's (8) ongoing surveillance system shows a dramatic drop in air bag–induced deaths beginning in model year 1998 and then a further drop in later model years. Air bag–induced deaths per registered vehicle are reduced among sled-certified vehicles (4, 14), partially attributable to more children being transported in rear seats. Risk of driver deaths per registered vehicle was similar for drivers with sled-certified air bags relative to those with first-generation air bags (4, 13), with the exception of drivers of pickup trucks (13).
Olson et al. (15) conducted matched-pair cohort analyses of the effects of air bag generation, computing driver versus passenger mortality risk within the same cars among vehicles with first-generation air bags, sled-certified air bags, and no frontal air bags. Fatality risk was equivalent for occupants of vehicles having first-generation and sled-certified air bags, except that children aged 0–5 years in the front seat of cars with sled-certified air bags had a lower risk than those in cars with first-generation air bags. The study estimated the impact of newer styles of frontal air bags on overall risk of front-occupant deaths in motor vehicle crashes rather than the specific effects on deaths in frontal crashes (15).
Head-on collisions are high-risk events that constitute strenuous tests of the real-world performance of occupant restraint systems. Other researchers have used head-on collisions to compare the risks for drivers in vehicles with superior versus poor performance in crash tests (16) and the independent and combined effectiveness of air bags and seat belts in reducing driver deaths (2). Head-on collisions between first-generation and sled-certified passenger vehicles may be the most direct measure of the relative protection afforded by redesigned air bags.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
A matched-pair cohort study of two-vehicle, head-on collisions resulting in at least one death was undertaken to assess drivers' risk of death in vehicles with sled-certified air bags relative to first-generation air bags. Fatal head-on collisions involving first-generation versus sled-certified air bags were identified by using the Fatality Analysis Reporting System (FARS), a census of fatal crashes on public roads in which a road user died within 30 days of the event, for 1998–2005 (17). The Vindicator software of the Highway Loss Data Institute (18) decodes vehicle identification numbers and identifies makes, models, and model years of vehicles. The institutional review board of the University of Maryland, Baltimore determined that the study was exempt from oversight because FARS contains no unique identifiers.
Matched-pair cohort studies are those in which both members of a pair experienced a common event, such as a crash, and either one or both members had the outcome of interest (in this instance, death) (15, 19–21). Risk ratios from matched-pair cohorts are equivalent to relative risks and can be computed by using outcome-centered databases such as FARS or disease registries (20, 21).
Sled certification was ascertained from NHTSA's brochures, websites, and the 1998–2000 National Automotive Sampling System/Crashworthiness Data System manual (22–24), in addition to a survey of automobile manufacturers. All members of the Alliance of Automobile Manufacturers and the Association of International Automobile Manufacturers were sent questionnaires about air bag technologies for all makes and models of passenger vehicles sold in the United States for model years 1998–2005. At the request of some manufacturers, we provided a sample of vehicle identification numbers (first 12 digits of 17) from FARS so they could respond to our survey. All but one manufacturer participated. Manufacturers' responses and other information sources were compared; the few inconsistencies that could not be resolved were referred back to manufacturers.
First-generation air bags were studied for vehicles with standard frontal air bags for drivers during model years 1994–1997, as identified from an unpublished Highway Loss Data Institute database of vehicle features (Matthew Moore, Highway Loss Data Institute, personal communication, 2006). Vehicles were studied if they were passenger cars, pickup trucks, sport utility vehicles (SUVs), or minivans. Sled-certified vehicles for model years 1998–2005 had to be identified as such by manufacturers or NHTSA. Vehicles with frontal air bags that were certified as advanced and compliant with Federal Motor Vehicle Safety Standard 208 were excluded, as were Mercedes automobile models, which largely were not sled certified and had advanced features.
Conditional Poisson regression models for matched-pair cohorts were used to estimate risk ratios adjusted for driver age, seat belt status, vehicle type, and passenger car size for driver deaths in vehicles with sled-certified air bags relative to vehicles with first-generation air bags (20, 21). Driver age was a dichotomous variable (age 60 years or older versus 15–59 years) so that effects of newer air bags on older drivers, who are fragile, could be examined. Car size classifications (mini/small, midsize, and large/very large) were the same as those used by the Insurance Institute for Highway Safety and incorporate vehicle length, width, and weight (13). Separate regression models were built for all passenger vehicles combined and for head-on collisions involving only passenger cars. Interaction terms between air bag generation and driver age, gender, seat belt status, vehicle type, and car size were included in models to determine whether there were significant differences in effects of air bag generation for categories of these variables.
When estimating effects of vehicle type on risk of death among pairs of drivers involved in the same head-on collisions, these effects were equivalent to effects of mismatches in vehicle types that occurred among paired vehicles. This also was true when estimating effects of different car sizes among paired vehicles, which reflected pairs in which cars of different sizes collided.
Air bags and other vehicle safety features evolved and improved during the study period. To account for the protective effects of younger vehicle age and improved vehicle designs among later models of sled-certified vehicles, regression models examined sled-certified vehicles for model years 1998–1999, 2000–2001, and 2002–2005.
Chi-square tests were performed to elucidate the relation of gender to vehicle type and to explore the relation between air bag generation, vehicle type, and risk of dying. To evaluate protection afforded by newer air bag designs, frontal air bag presence rather than air bag deployment was the exposure of interest. SAS version 9.1 software was used for all data analyses (25).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
A total of 1,572 two-vehicle, head-on collisions between passenger vehicles with first-generation and sled-certified frontal air bags in which at least one driver died were identified in FARS during 1998–2005. Most fatal head-on crashes occurred on roads with speed limits of 55 mph (89 km/hour) or higher; many were on roads with 40–50 mph (64–80 km/hour) speed limits (table 1). Cars were the most common vehicle types, followed by pickup trucks, SUVs, and minivans. Most sled-certified vehicles in head-on collisions were model years 1998–2000, which reflects the longer time period they were at risk. Drivers of first-generation air bag–equipped vehicles tended to be younger and were slightly more likely to be unbelted.
|
Pickup trucks and SUVs were considered together in multivariate models because they showed almost the same risk ratios relative to cars. Minivans were associated with a nonsignificant 30 percent lower risk ratio compared with cars and were analyzed separately.
Main effects of potential confounding variables are presented in table 2. As expected, risks of dying were significantly higher for drivers aged 60 years or older, those who failed to use seat belts, and drivers of cars colliding with pickups/SUVs, after adjusting for all variables in the model, including air bag generation (table 2).
|
For all passenger vehicle pairs combined, the crude risk ratio for driver deaths in sled-certified vehicles was 0.71. After adjustment for driver age, seat belt status, and vehicle type, the risk ratio increased to 0.87 (95 percent confidence interval: 0.77, 0.98). Interaction terms indicated no significant interactions between air bag generation and driver age, gender, or seat belt status and a borderline significant interaction between sled certification and pickups/SUVs. When death rates by vehicle type were examined, this interaction appeared to be due to significantly lower death rates among drivers of pickup trucks with sled-certified air bags (27 percent) than of pickups with first-generation air bags (48 percent) (two-sided p < 0.001). Only a 9 percent difference in death rates was observed between sled-certified and first-generation passenger car drivers. Including interaction terms for vehicle type increased the adjusted risk ratio to nearly 1.0. No consistent differences were observed between the later and earlier model years of sled-certified air bags when all vehicle types were included in models.
Among the separate matched pairs of fatal, head-on car-car collisions, drivers of mini/small and midsize cars had significantly higher risks of death compared with drivers of large cars (table 3). Among car drivers with sled-certified air bags compared with those having first-generation air bags, a crude risk ratio of 0.90 and an adjusted risk ratio of 1.04 (95 percent confidence interval: 0.85, 1.29) were observed, controlling for driver age, seat belt status, and car size. When regression models examined sled-certified cars of earlier and later model years, no significant differences were observed in the adjusted risk ratios; however, model years 1998–1999 for sled-certified cars were associated with a nonsignificant elevated adjusted risk ratio of 1.25 (95 percent confidence interval: 0.91, 1.71). Interaction terms were insignificant for models with cars-only matched pairs, suggesting no differential effects of sled-certified air bags by age, gender, seat belt use, or car size among car drivers.
|
Gender was not included as a covariable in the final versions of car-only or all-vehicle-type models because interaction terms for gender and air bag generation were not significant. Gender was strongly associated with vehicle type, with significantly fewer females driving pickup trucks (two-sided p < 0.001).
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
Drivers involved in the same head-on collision generally experience similar crash forces; thus, head-on collisions between passenger vehicles with first-generation air bags and those with sled-certified air bags are a good measure of the relative performance of the air bag generations. This matched-pair cohort study's findings are supportive of a conclusion that sled certification has not reduced the protectiveness of frontal air bags for drivers involved in high-severity events. Some of the findings for drivers of sled-certified vehicles may have been attributable to concurrent improvements in vehicle structure and seat belt performance.
No significant differences regarding protection from death by sled-certified air bags were noted by seat belt status, driver age, or gender. A borderline significant interaction between pickups/SUVs and air bag generation was observed. In addition, death rates among drivers of pickup trucks with sled-certified air bags were substantially lower than among those with first-generation air bags. Hence, sled certification may have resulted in improved frontal crash protection for drivers of pickup trucks.
In assessing the overall effect of sled certification, one plausible conclusion could be that there was no significant change for drivers of passenger cars and that observed benefits primarily accrued among drivers of pickup trucks. This finding contradicts a study reporting increased risk of driver death per registered pickup truck with sled-certified air bags (13). The earlier finding (13) may have been an artifact of registration-based analyses, which, unlike head-on collisions, are affected by changes in crash or travel patterns within makes and models and restriction to vehicles whose platform did not change after sled-certifying air bags. Possibly, more aggressive first-generation air bags in light trucks, because of their stiffer structure, might explain the observed differential benefits of depowering by vehicle type.
We expected to observe stronger protective effects of sled-certified air bags in later than in earlier model years because of improvements in air bag designs, seat belt technology, and vehicle crashworthiness (16). Small numbers hampered our ability to detect differences among model years. There was a nonsignificant elevated risk for drivers of sled-certified cars for model years 1998–1999 in head-on collisions with cars with first-generation air bags not reported in other studies (4, 13).
Because frontal air bags are designed to protect occupants in frontal crashes, evaluating their performance in frontal crashes is essential. Examining all impacts combined answers the question of how well frontal air bags reduce occupant deaths overall but makes it more difficult to detect changes in frontal crash performance and does not consider recent improvements in side structure and side impact air bags, which are present in some sled-certified vehicles but not in first-generation vehicles.
The major limitation of this study was small numbers in some categories. Conditional Poisson regression can result in overly wide confidence intervals when numbers are small (19). Seat belt load limiters and pretensioners affect crash forces, but numbers of vehicles with these features were too small for analysis. Furthermore, we were unable to determine whether large drivers received reduced protection from sled-certified air bags.
The strengths of the study include the use of multiple data sources for air bag characteristics. Some researchers have used model year 1998 as a surrogate for sled certification, whereas we excluded vehicles manufactured after model year 1997 if they were not sled certified. Using matched-pair cohorts in head-on collisions controlled for crash forces exerted upon drivers, and the multivariate models controlled for major confounding factors.
Consistent with previous research, this study generally did not observe an increased risk of death for adult drivers with sled-certified air bags. The exception was a nonsignificant increase in fatality risk for drivers of model year 1998–1999 cars colliding with other cars. Crashes in which deaths occur are high-severity events (13), typically at high speeds, so that concerns about whether sled-certified air bags provide equivalent protection in high-speed crashes appear to be unfounded. The findings suggesting lower risks of driver death in pickup trucks/SUVs with sled-certified air bags are intriguing but need further study. The majority of pickup truck drivers are male, which suggests that sled-certified air bags likely are not compromising protection among large occupants involved in crashes.
|
ACKNOWLEDGMENTS |
|---|
This study was sponsored by the Blue Ribbon Panel for Evaluation of Advanced Airbag Technology, with funding received from the Alliance of Automobile Manufacturers. The views in this study are those of the authors and do not necessarily reflect those of the sponsor or funding source.
The authors thank the Alliance of Automobile Manufacturers and Association of International Automobile Manufacturers for providing contact information to enable their members to be surveyed and for encouraging their members to respond to the survey. They are also grateful to the Insurance Institute for Highway Safety for sharing information from their vehicle features database and data on additional vehicle characteristics.
Conflict of interest: Financial support was received from the Alliance of Automotive Manufacturers, a trade association representing automakers, via the Blue Ribbon Panel for Evaluation of Advanced Airbag Technology. The Alliance of Automotive Manufacturers did not provide advice or comments regarding the study design, data analysis, or interpretation. Oversight of the study was performed by the Blue Ribbon Panel, which consists of representatives from the traffic safety research community, the National Transportation Safety Board, academia, medical institutions, and the insurance industry.
|
References |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
- Braver ER, Ferguson SA, Greene MA, et al. Reductions in deaths in frontal crashes among right front passengers in vehicles equipped with passenger air bags. JAMA (1997) 278:1437–9.
[Abstract/Free Full Text] - Crandall CS, Olson LM, Sklar DP. Mortality reduction with air bag and seat belt use in head-on passenger car collisions. Am J Epidemiol (2001) 153:219–24.
[Abstract/Free Full Text] - Cummings P, McKnight B, Rivara FP, et al. Association of driver air bags with driver fatality: a matched cohort study. BMJ (2002) 324:1119–22.
[Abstract/Free Full Text] - Kahane CJ. An evaluation of the 1998 –1999 redesign of frontal air bags. (2006) Washington, DC: National Highway Traffic Safety Administration, US Department of Transportation. (NHTSA technical report DOT HS 810 685). (http://dmses.dot.gov/docimages/pdf99/431569_web.pdf).
- Lund AK, Ferguson SA. Driver fatalities in 1985 –1993 cars with airbags. J Trauma (1995) 38:469–75.[Web of Science][Medline]
- National Highway Traffic Safety Administration. Fifth/sixth report to Congress: effectiveness of occupant protection systems and their use. (2001) Washington, DC: US Department of Transportation. (Report DOT HS-809-442). (http://www-nrd.nhtsa.dot.gov/Pubs/809-442.pdf).
- Zador PL, Ciccone MA. Automobile driver fatalities in frontal impacts: air bags compared with manual belts. Am J Public Health (1993) 83:661–6.
[Abstract/Free Full Text] - National Highway Traffic Safety Administration, National Center for Statistics and Analysis Special Crash Investigations Program. In: Counts of frontal air bag related fatalities and seriously injured persons (2007) Washington, DC: US Department of Transportation. (http://www-nrd.nhtsa.dot.gov/departments/nrd-30/ncsa/sci.html#).
- Update: fatal air bag-related injuries to children—United States, 1993 –1996. MMWR Morb Mortal Wkly Rep (1996) 45:1073–6. (Erratum in MMWR Morb Mortal Wkly Rep 1997;46:40).[Medline]
- Durbin DR, Kallan M, Elliott M, et al. Risk of injury to restrained children from passenger air bags. Traffic Inj Prev (2003) 4:58–63.[CrossRef][Medline]
- Newgard CD, Lewis RJ. Effects of child age and body size on serious injury from passenger air-bag presence in motor vehicle crashes. Pediatrics (2005) 115:1579–85.
[Abstract/Free Full Text] - National Highway Traffic Safety Administration. Federal motor vehicle safety standards; occupant crash protection. 49 Code of Federal Regulations, Part 571. (1997) Washington, DC: US Department of Transportation. (http://www.nhtsa.dot.gov/cars/rules/rulings/n114.html).
- Braver ER, Kyrychenko SY, Ferguson SA. Driver mortality in frontal crashes: comparison of newer and older airbag designs. Traffic Inj Prev (2005) 6:24–30.[CrossRef][Medline]
- Ferguson SA, Schneider L, Segui-Gomez M, et al. The blue ribbon panel on depowered and advanced airbags—status report on airbag performance. Annu Proc Assoc Adv Automot Med (2003) 47:79–81.[Medline]
- Olson CM, Cummings P, Rivara FP. Association of first- and second-generation air bags with front occupant death in car crashes: a matched cohort study. Am J Epidemiol (2006) 164:161–9.
[Abstract/Free Full Text] - Farmer CM. Relationships of frontal offset crash test results to real-world driver fatality rates. Traffic Inj Prev (2005) 6:31–7.[CrossRef][Medline]
- National Highway Traffic Safety Administration. In: Fatality Analysis Reporting System (1998–2005) Washington, DC: US Department of Transportation. (http://www-nrd.nhtsa.dot.gov/departments/nrd-30/ncsa/FARS.html).
- Highway Loss Data Institute. Vindicator software (2006) Arlington, VA: Highway Loss Data Institute. (http://www.hldi.org/research/advisories/iihs_advisory_9.pdf).
- Cummings P, McKnight B, Greenland S. Matched cohort methods for injury research. Epidemiol Rev (2003) 25:43–50. (Erratum in Am J Epidemiol 2004;160:300).
[Free Full Text] - Cummings P, McKnight B, Weiss NS. Matched-pair cohort methods in traffic crash research. Accid Anal Prev (2003) 35:131–41.[CrossRef][Web of Science][Medline]
- Greenland S. Applications of stratified analysis methods. In: Modern epidemiology—Rothman KJ, Greenland S, eds. (1998) 2nd ed. Philadelphia, PA: Lippincott-Raven.
- National Highway Traffic Safety Administration. Buying a safer car. (1997–2006) Washington, DC: US Department of Transportation. (http://www.safercar.gov/pages/ResourcesLinksBSC.htm).
- National Highway Traffic Safety Administration. Equipment and safety. (http://www.safercar.gov/).
- National Highway Traffic Safety Administration. National Automotive Sampling System/Crashworthiness Data System. (2000) Washington, DC: US Department of Transportation. (http://www-nass.nhtsa.dot.gov/NASS/CDS/DataColl/man2000.pdf).
- SAS Institute, Inc. SAS version 9.1 software (2003) Cary, NC: SAS Institute, Inc. (http://support.sas.com/91doc/docMainpage.jsp).
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  C. M. Olson, P. Cummings, and F. P. Rivara RE: "USING HEAD-ON COLLISIONS TO COMPARE RISK OF DRIVER DEATH BY FRONTAL AIR BAG GENERATION: A MATCHED-PAIR COHORT STUDY" Am. J. Epidemiol., April 1, 2008; 167(7): 885 - 886. [Full Text] [PDF]
|
|
|
|
|
|
E. R. Braver, J. A. Kufera, M. T. Alexander, M. Scerbo, K. Volpini, and J. P. Lloyd THE AUTHORS REPLY Am. J. Epidemiol., April 1, 2008; 167(7): 886 - 887. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||