Front Air Bag Nondeployments in Frontal Crashes Fatal to Drivers or Right-Front Passengers

Abstract

Objective : Public concern has arisen about the reliability of front air bags because Fatality Analysis Reporting System (FARS) data indicate many nondeployed air bags in fatal frontal crashes. However, the accuracy of air bag deployment, the variable in question, is uncertain. This study aimed to provide more certain estimates of nondeployment incidence in fatal frontal crashes.

Methods : Fatally injured passenger vehicle drivers and right-front passengers in frontal crashes were identified in two U.S. databases for calendar years 1998–2006 and model years 1994–2006: FARS, a census of police-reported fatal crashes on public roads, and National Automotive Sampling System/Crashworthiness Data System (NASS/CDS), a probability sample of tow-away crashes. NASS/CDS contains subsets of fatal crashes in FARS and collects detailed data using crash investigators. Front air bag deployment coding for front-seat occupant fatalities was compared in FARS and NASS/CDS, and case reviews were conducted.

Results : Among FARS frontal deaths with available deployment status (N = 43,169), front air bags were coded as not deployed for 18 percent of front occupants. In comparison, NASS/CDS (N = 628) reported 9 percent (weighted estimate) nondeployment among front occupants killed. Among crashes common to both databases, NASS/CDS reported deployments for 45 percent of front occupant deaths for which FARS had coded nondeployments. Detailed case reviews of NASS/CDS crashes indicated highly accurate coding for deployment status. Based on this case review, 8 percent (weighted estimate) of front occupant deaths in frontal crashes appeared to involve air bag nondeployments; 1–2 percent of front occupant deaths represented potential system failures where deployments would have been expected. Air bag deployments appeared unwarranted in most nondeployments based on crash characteristics.

Discussion : FARS data overstate the magnitude of the problem of air bag deployment failures; steps should be taken to improve coding. There are inherent uncertainties in judgments about whether or not air bags would be expected to deploy in some crashes. Continued monitoring of air bag performance is warranted.

Keywords:

• Air bag
• FARS
• NASS/CDS
• Traffic accidents
• Motor vehicle deaths
• Frontal crashes

ACKNOWLEDGMENTS

This work was supported by the Insurance Institute for Highway Safety. Partial support was provided by the Crash Injury Research and Engineering Network center at the University of Maryland School of Medicine in Baltimore. From the Insurance Institute for Highway Safety, we are grateful to Joe Nolan for providing expert advice on the case reviews and to Laura Strouse for providing research assistance. From the University of Maryland, Baltimore, we are grateful to Shiu M. Ho for providing advice on matching and computer programming and to Joseph P. Lloyd for providing technical advice on vehicle features and crash investigation.

Notes

^a FARS: Frontal defined as 11, 12, 1 o'clock principal impact point (or initial impact point among 335 deaths where principal was missing); NASS/CDS: Frontal defined as principal area of damage from collision deformation classification.

^b First-generation air bags: rigid barrier test (model years 1994–1997); sled-certified air bags: sled test (model years 1998–2005); certified advanced air bags: certified as advanced and compliant with federal standards for occupant crash protection (model years 2003–2006).

^c Total also includes air bags that did not fall into air bag generation categories, such as those tested using rigid barriers after model year 1997.

^d Percentages exclude missing air bag deployment data and inactivated air bags.

^a χ²= 651.54, 1 df; p< 0.001.

^a χ²= 221.36, 12 df; p< 0.001.

^a NASS/CDS case weights were multiplied by 1.415 to address underrepresentation of deaths in NASS/CDS (based on ratio of FARS to NASS/CDS front occupant deaths).

^b Percentages exclude missing data.

^c Included 5 deaths in which vehicles had caught fire post-crash and NASS/CDS investigators judged that deployment had occurred, but extensive damage made photographs difficult to interpret by IIHS reviewers.

^a Longitudinal delta-V calculated by NASS/CDS program.