History of airbag safety benefits and risks

Abstract

Objective

The history of airbags was reviewed for benefits and risks as they became a supplement to lap-shoulder belts. Sled and crash tests were evaluated and field data was analyzed for airbag effectiveness. Field data on airbag deaths and studies on mechanisms of deployment injury were analyzed. The history was reviewed as airbags evolved from the early 1970s to today.

Methods

Airbag benefits were determined from NHTSA crash tests with unbelted and belted dummies in 40, 48, and 56 km/h (25, 30, and 35 mph) frontal impacts with and without airbags. The literature was reviewed for testing of passive restraints with and without airbags. Recent NCAP tests were compared with earlier tests to determine the change in occupant responses with seatbelts and supplemental airbags in modern vehicles. 1994–2015 NASS-CDS field data was analyzed for MAIS 4 + F injury. Risks were compared for belted and unbelted occupants in frontal impacts by delta V. Airbag risks were identified from field deployments and research. The 1973–76 GM fleet had two deaths due to the occupant being out-of-position (OOP). The mechanisms of injury were determined. From 1989–2003, NHTSA investigated 93 driver, 184 child passenger, and 13 adult passenger airbag deaths. The data was reviewed for injury mechanisms. Second generation airbags essentially eliminated OOP airbag deaths. More recently, three suppliers were linked to airbag rupture deaths. The circumstances for ruptures were reviewed.

Results

The risk for serious head injury was 5.495% in drivers and 4.435% passengers in 40–48 km/h (25–30 mph) frontal crash tests without belts or airbags. It was 80.5% lower at 1.073% in drivers and 82.0% at 0.797% in passengers with belts and airbags in 35 mph NCAP crash tests of 2012–20 MY vehicles. NASS-CDS field data showed a 15.45% risk for severe injury (MAIS 4 + F) to unbelted occupants and 4.68% with belted occupants in 30-35 mph frontal crash delta V with airbags, as deployed. The reduction in risk was 69.7% with belt use and airbags deploying in 96.1% of crashes. There were benefits over the range of delta V. Two airbag deaths were studied from the 1970s GM fleet of airbags. The unbelted driver death was caused by punchout force with the airbag cover blocked by the occupant and membrane forces as the airbag wrapped around the head, neck or chest with the occupant close to the inflating airbag. The unbelted child death was from airbag inflation forces from pre-impact braking causing the child to slide forward into the deploying airbag. Research showed that unrestrained children may have 13 different positions near the passenger airbag at deployment. NHTSA investigation of first generation airbag deaths found most driver deaths were females (75.3%) sitting forward on the seat track, close to the driver airbag. Seatbelt use was only 30%. Most child deaths (138, 75.4%) involved no or improper use of the lap-shoulder belts. Of these, 115 deaths involved pre-impact braking. Only 37 (20.2%) children were in child seats with 29 in rear-facing and 8 in forward-facing child seats. Eight child deaths (4.4%) occurred with lap-shoulder belt use. Airbag designs changed. More recently, Takata airbags were related to at least 24 deaths by airbag rupture prompting a recall; the successor company Joyson had an airbag recall. ARC airbags have experienced a chunk of the inflator propelled into the driver during deployment with several deaths leading to a recall.

Conclusions

Airbags are effective in preventing death and injury in crashes. They provide the greatest protection in combination with seatbelt use. NHTSA estimated airbags saved 28,244 lives through 1-1-09 while causing at least 320 deployment deaths, which has prompted improved designs, testing, and recalls.

Keywords:

• Airbags
• airbag deployment death
• airbag injury
• OOP
• out of position occupants
• passive restraints
• child injury
• airbag rupture

Introduction

Origins of airbags

In the 1950s, there was widespread use of motor vehicles with crashes causing death and injury. Figure A1 in the Supplementary Appendix shows the death rate was 7.24 per 100 million miles traveled in 1950 (NHTSA 2022). Most occupants were unrestrained. Concepts were developed to control forces on the occupant by lengthening the stopping distance with impact on the interior. Energy absorbing (EA) components were designed to limit force from the interior in a frontal crash. They included an EA steering system (Marquis 1967), HPR (high penetration resistant) windshield (Rieser and Chabal 1967), and padded dashboard (instrument panel). The industry developed these features and many were introduced in 1967 MY (model year) vehicles. Seatbelts were known to prevent ejection and injury from motor-sport racing experiences. Testing showed that seatbelts restrained an occupant by forces from anchors on frame structures that coupled the occupant to the vehicle. The use of seatbelts lowered forces on the occupant and increased ride-down (Katoh and Nakahama 1982).

Figure 1. Dummy responses in 53.1 km/h (33 mph) delta V frontal sled tests (data from Seiffert et al. 1974).

Seatbelts were installed in vehicles, although use rates were very low in the late 1960s and 1970s. They consisted of separately adjustable lap and shoulder belts without a retractor. They were inconvenient. An NHTSA study of restraint use in 19 metropolitan areas from July to December 1979 found only 2% of 1–4 year olds (yo), 3.3% of 5–12 yo, 3.7% of 13–19 yo and 6.9% of adults were restrained by seatbelts (Morris et al. 1980). The low use rates spawned automatic or passive restraints that protected without action by the occupant. Automatic restraint concepts initially included airbags with a manual lap belt and then passive belts. The airbag inflated a fabric bag in the space between the occupant and the interior early in a frontal crash. There were concepts for inflatable dash and windshield panels and for the steering wheel, instrument panel, and roof rail airbags. The target was to protect an unrestrained occupant in a 40 km/h (25 mph) barrier crash.

Airbag concepts

Linderer filed an application 10-6-51 that was granted on 11-12-53 as DE 896312 C for a device to protect occupants in vehicles from injuries in a crash. The concept included an inflatable container in a folded state in front of an occupant that unfolded and inflated to prevent injury by manual or automatic triggering. The concept included a relief valve adjusted by a spring to control the inflation pressure to soften the impact. Hetrick filed an application 8-5-52 that was granted on 8-19-53 as US 2649311 A for a safety cushion assembly for automobiles. The concept was an inflatable cushion in the steering wheel and on the dashboard triggered by a fast-opening valve that inflated the cushion in a crash by accumulated air in a reservoir.

Bertrand filed two applications in the US. One was file 10-5-55 and granted on 5-13-58 as US 2,834,606 for a safety device for passengers consisting of a normally deflated bag that was inflated downward from the roof rail in a crash in front of the passenger to avoid bodily harm. The second was file 11-5-56 and granted on 5-13-58 as US 2,834,609 for a passengers safety device for a vehicle consisting of a high pressure tank of air and folded bags in the roof that were quickly inflated downward in a crash with a means of deflating the airbag by the occupant after the crash.

By the late 1950s, there were isolated research and development projects on inflatable restraints, although it would take more than a decade for sufficiently broad-based programs to implement airbags in vehicles. The coordinated effort involved sensing the crash, logic to trigger the airbag, refinements with aggressive driving and crash testing and finalizing the size, shape, packaging, and reliability of the airbag.

Passive restraint mandate

In 1968, NHTSA required seatbelts in passenger vehicles (Federal Register 1966). FMVSS 208 (Federal Motor Vehicle Safety Standard) required passenger vehicles provide a seatbelt at every forward-facing (designated) seating position. There were no crash test requirements to evaluate the protection of vehicle occupants. Because of the low use rates of seatbelts, NHTSA started considering automatic crash protection in motor vehicles. There were two types of automatic crash protection, passive belts or airbags (Martin and Lundstrom 1975). Johannessen (1987) summarized passive belt concepts and Carr (1978) and Romeo (1980) summarized airbag designs. Patrick (1975) considered active and passive restraints and concluded mandatory seatbelt use was the most beneficial. FMVSS 208 changed many times over the next 16 years until the Secretary Dole decision in 1984. Kratzke (1995) summarized the regulatory history of automatic crash protection in FMVSS 208.

NHTSA (1970) published a final rule requiring automatic crash protection for all passenger cars starting in 1973 and light trucks and vans in 1974. Compliance was determined in a crash test with dummies in the front outboard seats. NHTSA received many petitions for reconsideration and court challenges. They postponed the effective date until 1975. In a 1972 court decision (Chrysler v DOT), the court overturned the automatic protection requirements because the specifications for a crash test dummy were inadequate. The option for passive seatbelt or airbags fostered the attitude of an either or restraint by passive seatbelts or airbag. Glenn (1974) summarized NHTSA research on passive belt sled tests with volunteers and dummies as part of an evaluation of automatic restraints.

NHTSA (1972) mandated an ignition interlock requirement that the seatbelt be buckled before the driver could start the vehicle in 1973 MY vehicles. The goal was to increase seatbelt use. The public resisted the belt-starter interlock requirement and disabled the system. On October 27, 1974, President Ford signed a law prohibiting the seatbelt interlock and NHTSA deleted the interlock option from FMVSS 208. In 1974, NHTSA proposed reinstating the automatic protection requirements for front outboard seating positions starting in 1976. On December 6, 1976, Secretary Coleman called on auto manufacturers to join with the Federal government in a demonstration program so that approximately 500,000 cars with passive restraint systems would be offered for sale at a reasonable cost to consumers in the 1979 and 1980 MY (model year).

In 1977, Secretary of Transportation, Brock Adams, reexamined the decision on automatic protection, and later in the year, NHTSA published a final rule reinstating automatic protection requirements for passenger cars starting in the 1982 MY for vehicles with >290 cm (>114″) wheelbase and the 1983 MY for vehicles with >254 cm (>100″). Six petitions for reconsideration of the rule were filed and there were a number of legal challenges. In 1981, Secretary Lewis and Administrator Peck rescinded automatic protection requirements because of economic conditions and the likely insignificant safety benefits that would result if vehicles provided automatic protection by means of detachable automatic belts. In 1983, the Supreme Court decided the 1981 rescission of the automatic protection requirement was unlawful.

VW passive shoulder belt and knee bolster

Seiffert and Borenius (1972) raised performance issues with inflatable restraints in small passenger vehicles because of higher deceleration forces and more difficult time constraints. The complexities focused VW attention on developing an automatic shoulder belt with knee bolster to satisfy passive restraint requirements. Seiffert et al. (1974) reported on frontal sled tests at 53.1 km/h (33 mph) delta V with the Part 572 dummy. Three restraint configurations were compared, including manual lap-shoulder belts 2-pt shoulder belt with knee bolster, and airbag with knee bolster in the driver and passenger seats.

Figure 1 shows Part 572 dummy responses in 53.1 km/h (33 mph) delta V frontal sled tests (Seiffert et al. 1974). Manual lap-shoulder belts resulted in higher HIC than with either passive restraint. The combination of the shoulder belt and knee bolster resulted in low HIC and chest acceleration. VW produced vehicles with the shoulder belt attached to the door and knee bolster (Miller et al. 1978; States et al. 1977; Seiffert 1979). Other manufactures developed different passive belt systems, including motorized shoulder belts, 3-pt systems with the shoulder belt mounted on the door and other concepts.

1973–76 General Motors fleet of driver and passenger airbags

Campbell (1972) described the 1973–76 General Motors driver and front passenger airbags. The driver airbag was 56 cm (22″) in diameter, extended about 25.4 cm (10″) rearward of the steering wheel rim, and had a volume of 79.3 L (2.8 ft³). It ignited sodium azide to produce nitrogen gas inflating the airbag. The passenger airbag was more complex. It included a sodium azide gas generator augmented by Argon compressed gas in a hybrid generator. The cushion was 410 L (14.5 ft³) to restraint two passengers. The passenger airbag had an inner tubular knee airbag to restrain the lower extremities.

The crash sensing and triggering logic was described with variable inflation levels for the passenger airbag (Campbell 1972). The passenger airbag was dual staged with a low output at about 19.3 km/h (12 mph) and a second stage at about 29.0 km/h (18 mph). In a 48.3 km/h (30 mph) frontal barrier impact, the satellite sensor on the bumper triggered at 7 ms, which released the stored gas and activated one of the solid fuel generators producing “low level” inflation, which provided restraint of two occupants in a 29.0 km/h (18 mph) barrier impact. If the central accelerometer detected a higher severity crash, the second stage was fired at 31 ms in a 48.3 km/h (30 mph) barrier impact. The second-stage, “high level” inflation increased the restraint capability. Other crash conditions and severities were evaluated, including a 40 km/h (25 mph) impact into a pole. The author also reviewed volunteer and animal testing.

Louckes et al. (1973) further described the airbag module, sensing system, and triggering logic for the driver airbag. The main performance was to sense a 30 mph barrier impact and inflate the airbag in <40 ms. The airbag for the driver was stored in the hub of a specially designed steering wheel (Patrick et al. 1972). Figures 41–47 in Louckes et al. (1973) showed high-speed photos of the opening of the vinyl cover as gas was generated and pressure increased inside the driver airbag module. The cover bulged prior to the doors opening at the front. As the doors opened, the airbag was visible. Later images showed the airbag gaining a fully inflated shape in the space in front of the occupant. A specially designed clock spring maintained electric contact to the sensing system and allowed the steering wheel to rotate with continuous electrical contact to the airbag. A knee restraint was added to the lower instrument panel to restrain the driver’s lower extremities.

Klove and Oglesby (1972) discussed special problems during the development of the frontal airbags for the driver and front passengers. They included controlling the energy from the airbag to the occupant, which caused high rebound forces and testing for out-of-position (OOP) children in various positions on the passenger seat.

Smith et al. (1972) conducted 40 static and 35 dynamic airbag deployment tests of the General Motors airbag with volunteers at Holloman Air Force Base, New Mexico, sponsored by NHTSA. The sled tests were at 24.3 km/h (15.1 mph) with 8.6 g up to 50.7 km/h (31.5 mph) with 21.7 g. The testing found no serious injury. Smith et al. (1974) conducted additional sled tests of the General Motors driver airbag with volunteers at Southwest Research Institute, San Antonio, Texas. Forty (40) sled tests were conducted with volunteers and 32 with a dummy at 20.9-48.3 km/h (13-30 mph) in eight increasing levels. There was no serious injury to the volunteers.

Wilson (1974) described more details on the crash testing during the development of the General Motors driver and passenger airbags. The crash testing included (1) the 95th driver and right front passenger with seat full rear, (2) the 50th driver and right front passenger in the mid-track position, (3) the 95th driver and 5th passenger with the seat full rear, (4) 5th driver and 95th passenger with seat full forward, (5) three 50th in panic brake positions, and (6) 95th driver, 3 year old center passenger and 95th right passenger with seat full rear, (7) 3 year-old male, right front passenger standing on the floor with hand on the instrument panel, (8) 3 year-old male right front passenger with seat full rear, (9) single 95^th in the right-front seat and (10) three 50th occupants on the front seat in 24.1 km/h (15 mph) and 48.3 km/h (30 mph) barrier impacts. Martin and Lundstrom (1975) summarized the early experiences with the General Motors test fleet of airbags and discussed the alternatives of passive seatbelts.

Mertz (1988) summarized the design and crash performance of 11,321 driver and passenger airbags in 1973–76 General Motors vehicles. The airbags were designed to restrain a 95^th Hybrid III seated full rear on the tracks. The goal was to fully inflate the airbag before the occupant moved forward enough to interact with the inflating airbag in a 48 km/h (30 mph) frontal barrier impact. Breed and Castelli (1988) described the 5″, −30 ms rule where the sensor must trigger the airbag 30 ms before the occupant moves 5″ (12.7 cm) forward in the crash. The airbag pressure was selected to restrain the torso of the occupant without the use of a lap belt, although the restraint system included a manual lap belt. For an unbelted driver, a knee bolster was installed on the lower instrument panel to provide lower extremity restraint. The passenger airbag was mounted in the lower instrument panel in front of the right front passenger. It was designed to restrain the right front and center front passengers. Lower torso and leg restraint was provided by a tubular knee airbag inside of the larger airbag. The passenger airbag had dual staged deployment depending on crash severity to restrain the right-front and center passenger. Manual lap belts were provided for each seating position. This was an upgrade from an earlier police-vehicle fleet of Chevrolets with airbags but no lap belts.

Mercedes proposes airbags supplement seatbelts

Reidelbach and Scholz (1979) described the Mercedes-Benz passive restraint system which consisted of a seatbelt with a pretensioner to eliminate belt slack with an emergency locking retractor and supplemental airbag deployed by solid propellant gas generators, knee bolsters and electronic crash sensor with dual level triggering. They determined the combination of manual lap-shoulder belts, airbags and pretensioners provided optimum restraint. They showed how the conventional lap-shoulder belt provided the primary restraint and how the airbag provided supplemental protection for the head. By this approach, the driver airbag was designed to attenuate the impact of the head, instead of the occupant’s whole upper torso. The approach recommended by Mercedes Benz rallied support within the government and industry, and airbags took on the name SRS (supplemental restraint system). The combination of manual belts and supplemental airbag rapidly became the preferred restraint system (Struble 1998). It followed the successful implementation of 3-pt belts with a locking retractor (Bohlin 1967).

NHTSA mandates airbags

On July 17, 1984, Secretary of Transportation, Elizabeth Dole, signed a rule providing a phase-in of automatic protection in cars beginning the 1987 MY (NHTSA 1984; Dole 1985). Vehicles manufactured for the 1990 MY were required to provide automatic protection. During the phase-in of automatic protection, the rule encouraged manufacturers to install airbags instead of automatic belts by providing a 1.5 car credit for vehicles with driver airbags and any type of automatic protection for the passenger. The rule had incentives to encourage States to pass mandatory seat belt use laws and included a provision that the automatic protection requirements would be eliminated if the Secretary of Transportation determined by April 15, 1989 that enough States had enacted mandatory seatbelt use laws meeting the criteria specified in FMVSS 208.

In 1985, NHTSA published a notice proposing occupant protection by manual seatbelts and airbags be evaluated according to the same crash tests used to evaluate automatic protection. The proposal was adopted requiring 10% of vehicles in 1987 MY have passive restraints with 25% in 1988 MY, 40% in 1989 MY and 100% in the 1990 MY (NHTSA 1984). Williams et al. (1989) found that most manufacturers opted for passive belts to meet the 10% requirement in 1987 MY vehicles. Reinfurt et al. (1991) found misuse of passive seatbelts involving passive shoulder belts without the lap belt buckled and detaching the passive belt system. NTSB (1994) found severe neck injuries and decapitation without lap belt use and a passive shoulder belt. Augenstein et al. (2000) found a pattern of liver laceration in drivers wearing the passive shoulder belt without the lap belt fastened.

After a period of passive belts in vehicles, driver airbags became standard equipment with passenger airbags following. Manual lap-shoulder belts were the primary restraint and the airbags supplement the seatbelt by distributing load on the face and upper torso. In 1991, President Bush signed a law requiring passenger vehicles for the 1998 MY provide airbags at the driver and right-front passenger positions supplementing manual lap-shoulder belts as the means for automatic protection offered in vehicles by the 1998 MY and light trucks and vans by the 1999 MY.

Airbags as standard equipment

“First generation” airbags

By the late 1980s and early 1990s, airbags were available in vehicles. There was enough field data to determine the effectiveness in frontal crashes. Evans (1991) provided a comprehensive analysis of the effectiveness of occupant restraints using the matched-pair comparison method. The airbag only was 17 ± 4% effective in preventing fatalities in front outboard seats. The combination of lap-shoulder belts and airbag was 46 ± 4% effective in preventing fatalities. The effectiveness of the combination was higher than with only lap-shoulder belts at 41 ± 4%. For completeness, the shoulder belt only in passive belts was 29 ± 8 effective. He also determined the effectiveness of seatbelts by crash direction in two components. One was the effectiveness of lap-shoulder belts preventing ejection and the other was the effectiveness preventing interior impact. For example, lap-shoulder belts were 43% effective in preventing fatality with 9% of the effectiveness from the seatbelts preventing ejection in frontal crashes. In contract, lap-shoulder belts were 77% effective in preventing fatalities in rollovers with 63% of the effectiveness from seatbelts preventing ejection. Seatbelts were effective in all crash types.

Lund and Ferguson (1995) evaluated driver fatalities in 1985–1993 vehicles with airbags. They found driver deaths were reduced 24% in frontal crashes compared to vehicles with manual belts only and 16% in all crashes. Based on vehicle registration data, fatality rates were reduced by 23% for vehicles with airbags and manual belts compared with the same vehicles with only manual belts and 16% for all crashes. Braver et al. (1997) found that right-front passenger fatalities were 18% lower in frontal crashes of cars with airbags and 11% lower in all crashes. However, children less than 10 years old had a 34% increased risk of dying in frontal crashes. The airbags were called “first-generation” airbags referring to vehicles certified for the 48.3 km/h (30 mph) barrier crash test. It became apparent that driver and passenger airbags were causing serious and fatal injuries in some low-severity crashes, crashes where minor or no injury was expected.

NHTSA’s in-depth Special Crash Investigations (SCI) determined that many of the deaths involved young children in the right-front passenger seat close to the airbag module at deployment (Kleinberger and Summers 1997). The close proximity was related to a lack of seatbelt use or improper seatbelt use with the child moving close to the airbag by pre-impact braking. There were cases of infants in rear-facing child seat in the right-front passenger seat. The back of the child restraint was close to or against the airbag at deployment (Kleinberger and Summers 1997). Small female drivers sitting close to the steering wheel were injured in the chest and neck by the deploying airbag (Kahane 2006).

In October 1995, national education campaigns by NHTSA, automakers and insurers started publicizing the dangers of children riding in the front seats with passenger airbags. They advised parents to place children in rear seats. The campaigns were highly effective (Kindelberger and Starnes 2003; Viano and Parenteau 2021). Drivers were advised to sit at least 25.4 cm (10″) away from the steering wheel. Letters were sent to owners of vehicles equipped with airbags to fix labels to the dashboards or sun visor warning of the dangers of sitting close to a deploying airbag. Airbag warning labels were required in all new vehicles.

“Sled certified” airbags

In 1997, NHTSA amended the requirements in FMVSS 208 for frontal crash performance to temporarily allow a 48.3 km/h (30 mph) sled test with unbelted dummies as an alternative to the 48.3 km/h (30 mph) frontal barrier crash test (NHTSA 1997). This was an interim step to reduce the inflation velocity of airbags while NHTSA considered broader changes to the standard. The airbags were called “sled-certified.”

Automakers reduced airbag inflation rates, because the 48.3 km/h (30 mph) sled deceleration had a more gradual rise to peak deceleration than the pulse in a barrier crash test. This reduced the risk for injuries to occupants close to the module at deployment. The airbags were also called “depowered” airbags. The depowered airbags used less gas-generating propellant or stored gas volume for airbag inflation. Sled certified airbags had more time to inflate and had less inflation energy, because the barrier crash tests of unbelted dummies required more rapid deployment of the airbag with higher forces to restrain the dummy.

By 1998, 84% of new vehicles had “depowered” airbags because of the interim rule (Kahane 2006). Peak tank-inflation pressure was 16% lower and the rate of pressure rise was reduced 30%. There were a number of other airbag design changes during the following years including (a) reduction in airbag volume, (b) reduction in rearward deployment distance through internal tethers, (c) recessing driver airbags below the steering-wheel rim, (d) improved folding techniques that reduced airbag fabric deployment speeds and (e) shift from pyrotechnic inflators to hybrids including stored gas.

Braver et al. (2008) studied deaths among drivers and right-front passengers in frontal crashes with redesigned airbags compared to first generation airbags. They found a 65% reduction in children 0–4 years old and a 46% reduction in children 5–9 years old in vehicles with sled-certified or depowered airbags. There was no significant differences in risk for adults, although the relative risks were lower in the sled-certified vehicles.

“Advanced” or “second generation” airbags

NHTSA (2001) issued an advanced airbag rule for September 2003. It included a wide array of test requirements for OOP (out-of-position) protection of children and small females. The 5th Hybrid III (small female dummy) was added to the frontal barrier crash test. The 48.3 km/h (30 mph) sled test or 48.3 km/h (30 mph) barrier test was added with unbelted dummies replacing the 40 km/h (25 mph) barrier test. The lower speed test allowed airbags to deploy with lower energy than first generation airbags. The phase-in of “advanced” airbags or “second generation” airbags improved the protection of belted and out-of-position occupants (NHTSA 2001). The testing included full-overlap, head-on, offset frontal and OOP (out-of-position) conditions with dummies ranging from the 50^th Hybrid III, 5th Hybrid III belted and unbelted, and 1 yo (year old), 3 yo and 6 yo Hybrid III child dummies in OOP conditions. Smith et al. (2003) described some of the testing methods.

Advanced or second generation airbag tests required the system to distinguish between adult and child passengers and modify or suppress deployments characteristics to match the occupant size, belt use status, seat track position and crash severity. When an infant or small child was in the right-front passenger seat or a small woman with the seat track forward, the airbag system had to suppress deployment or deploy the airbag with low injury risks. Advanced airbags had sensors measuring crash severity, rear-facing child seats and occupant seatbelt use, weight and proximity to the airbag. Variable-force airbag deployment was achieved by staging the deployment energy in phases with increasing crash severity. Advanced airbags had deployment thresholds at lower crash severity for unbuckled occupants.

The 2001 rule specified a 40 km/h (25 mph) barrier impact and offset-frontal impact with unbelted 50th Hybrid III dummies. The 40 km/h (25 mph) barrier impact involved shorter deceleration pulses and higher forces than the 48 km/h (30 mph) sled test. The rule was controversial. Some safety organizations sued NHTSA, but the court upheld the agency. By the 2007 MY, all passenger vehicles were required to have advanced airbags. The rule also specified a 56 km/h (35 mph) barrier impact with belted 50th Hybrid III dummies by the 2010 MY. This was an increase from the 48 km/h (30 mph) crash test required for the 2007 MY.

The Alliance of Automobile Manufacturers funded a three-year program of data collection of frontal crashes by an independent third party. A panel of experts was established as the Blue Ribbon Panel (BRP) for the evaluation of advanced airbags. Ferguson and Schneider (2008) summarized the findings of the Blue Ribbon panel. The found data from the NHTSA SCI indicate that airbag deployment driver deaths were down from 80 per 100 million registered vehicle years in 1990–91 to 1 per 100 million registered vehicle years in 2002–03 in low-speed frontal crashes. There was a decline in passenger deaths due to the airbag. The decline in airbag deployment deaths was due to several factors including (a) depowered airbags, (b) increased seatbelt use, (c) changes in behavior with occupants sitting farther away from the steering wheel and (d) parents placing children in rear seats because of public information campaigns about airbag inflation risks (Kahane 2006).

Ferguson and Schneider (2008) noted there was convincing evidence that advanced airbags and public education reduced airbag-deployment child deaths. The risk of fatality in frontal crashes for children in the right-front seat was reduced by about half with the greatest reductions for the youngest children. Parents were still being advised to place their children in rear seats to avoid injuries from deploying airbags, even with the lower fatality rates. There was no evidence of a loss of protection in frontal crashes for lap-shoulder belted adult drivers and passengers with sled certified or advanced airbags. There were continuing risks with unbelted occupants.

Braver et al. (2010) determine if changes in airbag design affected occupant protection. Frontal crash mortality rates were compared for occupants in vehicles with sled certified with and without advanced features and advanced airbags. Mortality rates were 12% lower (5–19%, 95th CI) than for drivers without advanced features, including unbelted men and drivers younger than 60 years old. The mortality rate was 21% higher (4–39%, 95th CI) for belted drivers in advanced airbags. They concluded that advanced airbags were protective for some occupants. Teoh (2014) continued the evaluation of sled certified and advanced airbag field performance. He concluded that advanced features in sled-certified airbags were beneficial for drivers and right-front passengers.

The field studies of sled certified and advance airbags found significant reductions in small driver and child passenger deaths compared to first generation airbags. OOP risks were essentially eliminated with advance airbags. The changes did not adversely affect the protection of normally seated occupants. The combination of seatbelts and advance airbags protected occupants in crashes. The restraints in vehicles have continued to improve. Today, most seatbelts have dual pretensioners with the retractor pretensioner firing first and an anchor pretensioner firing with a slight delay tightening the belts on the occupant. The retractor includes a load-limiter that allows webbing to pay out under controlled load. Advance airbags are smart systems with sensor input on the crash severity, seat track position, and occupant weight. The passenger airbag includes occupant detection to sense a child, small occupant or child seat.

The objective of this study was to review the history of airbag benefits and risks as they became a supplement to lap-shoulder belts. It analyzed sled and crash testing and field effectiveness showing safety improvements with airbags. Field data on airbag deployment deaths and mechanisms of injury were analyzed. The history was reviewed as airbags evolved from the early 1970s to today and complements other work (Kratzke 1995; Stocke 1998; Kent et al. 2005).

Methods

This study reviewed the benefits of airbags by analyzing crash test responses with unbelted and belted dummies and field accident data on injury in frontal crashes with and without seatbelt use. The testing included various restraint configurations. The field data involved belted and unbelted occupants with airbag deployments, as occurred, in frontal crashes analyzed by delta V.

This study also reviewed the risks of airbags by analyzing field accident data and individual cases with airbag deployment injury and death to understand the circumstances and mechanisms of injury. It summarized the research on the biomechanics of airbag injury and the redesign of airbags to lower the risks of deployment injury. It concludes with an evaluation of airbag rupture deaths leading to recalls.

Benefits of airbags

NHTSA’s crash test database (https://www.nhtsa.gov/research-data/research-testing-databases#/vehicle) was searched for frontal crash tests with unbelted dummies and no airbags and lap-shoulder belted dummies with airbags. The literature was reviewed for frontal crash and sled tests that compared different restraint configurations. Matched tests were compared to determine the safety benefits of airbags and seatbelt use in frontal crashes.

NHTSA unbelted crash tests

NHTSA’s crash test database was searched for early barrier impacts in the 1970s–80s with unbelted drivers and right-front passengers. Four tests were found at 41.4 ± 2.0 km/h (25.7 ± 1.2 mph) and four tests at 48.0 ± 0.2 km/h (29.8 ± 0.1 mph). The responses of Part 572 and Hybrid III dummies were summarized for the head, chest, and femurs.

GM airbag and manual lap belts

Wilson (1974) summarized crash tests with the GM air cushion restraint system (ACRS) in 1973–76 General Motors vehicles. Mertz (1988) reported on additional crash test results. The testing involved 30 mph crash tests with airbags and either lap belted or unbelted dummies in the driver and passenger seats. The biomechanical responses were compared with 40 mph crash tests with airbags and unbelted dummies.

Ford airbag and manual belts

Maugh (1985) reported on crash tests for the 1984 Tempo/Topaz. The tests were at 30 mph with airbags and belted or unbelted Part 572 driver. They also ran a 35 mph airbag test with a belted Part 572 dummy. The biomechanical responses were summarized.

NHTSA NCAP frontal crashes tests

NHTSA’s crash test database was searched and eight NCAP tests were selected at 35 mph with airbags and lap-shoulder belted 50th Hybrid III in the driver seat and 5th female Hybrid III in the passenger seat. The vehicles were 2012–2020 passenger cars with advance airbags and dual pretensioning lap-shoulder belts. The biomechanical responses were summarized and compared with the unbelted crash tests at 25–30 mph (40–48 km/h) without airbags. The biomechanical responses were converted to injury risks for each body region with Logist functions. The risk r(x) was: r(x) = [1 + exp(α-βx)]⁻¹, where α and β are parameters for a best-fit relating injury risk to biomechanical response x (Parenteau et al. 2021).

Field data on severe injury (MAIS 4 + F) in frontal crashes

Viano and Parenteau (2022) determined the risk for severe injury (MAIS 4 + F) by crash type, seatbelt use, and crash severity (delta V) using 22 years of NASS-CDS from 1994 to 2015 with all light vehicles and occupants 15+ years old. They used 9 increments of delta V from <16–72+ km/h (<10–45+ mph). Crashes with frontal damage were selected. Injury risks for belted and unbelted occupants were calculated by dividing the number of severely injured (MAIS 4 + F) by the number of exposure (MAIS 0 + F) occupants using weighted data. The frontal crash risks for belted and unbelted occupants were plotted with the effectiveness of seatbelt use by delta V and airbag deployment, as occurred. This provides the most relevant field data on the performance of airbags and seatbelts.

Frequency of airbag deployment in frontal crashes

Viano and Parenteau (2010) determined the frequency of airbag deployment by delta V for lap-shoulder belted occupants in frontal crashes using the 1997–2007 NASS-CDS. The crashes involved a vehicle being towed from the scene.

Risks of airbags

Driver airbag death

The first driver airbag death in the U.S. led to research on the mechanisms of fatal injury. Static testing of the airbag was reviewed that showed high forces on the chest with the occupant against or near the inflating airbag. The inflation caused high forces from the airbag punching out of the module. With the cover blocked by the occupant, the cover bulged from internal pressure and the constraint on normal inflation of the airbag. Sled testing was reviewed that showed the dynamics of the driver out-of-position (OOP) and in the path of the inflating airbag.

Passenger airbag death

The first passenger death from the airbag was an infant sleeping on the passenger seat. The vehicle braked before the impact causing the child to slide forward into the path of the inflating airbag. Static deployments and sled testing were reviewed that demonstrated the mechanism of death with children near the deploying airbag. The research determined a range OOP positions leading to a matrix of tests to assess the performance of passenger airbags with out-of-position (OOP) children and small occupants.

1973–76 General Motors airbags field performance

By the end of 1980, 216 deployment accidents were investigated with 216 drivers, 86 right-front occupants, and 11 center-front occupants with 13 children <10 yo. The cases were reviewed. The injuries were classified as expected or unexpected based on the severity of the crash and damage to the vehicle and interior. The cases with unexpected injury were studied for the mechanism of the airbag deployment injury.

Child OOP airbag deaths

Research was conducted on the possible position of children with pre-impact braking. Most of the studies involved unbelted children. Thirteen positions were selected to cover the field accident exposure. The OOP positions were evaluated with child dummies and anesthetized animals to assess the risk for death and injury and to establish standard practices for testing airbags with child and small occupants.

First generation airbag performance

Special Crash Investigation (SCI) conducted by NHTSA involved the collection of field accident data on airbag deployment deaths and injuries. By 2009, there were 296 confirmed deaths by airbag deployment. The data was reviewed for drivers, adult passengers, and child passengers either in or not in rear-facing child seats. The databases were analyzed by occupant age, sex, height, weight, and injury mechanism. The crashes were analyzed by pre-impact braking and severity.

Advance airbag ruptures

In 2003, the first death of a driver occurred from fragments from a Takata airbag canister in a 2004 Subaru Impreza. The fragments were propelled toward the driver opening a new risk with airbags from a defect in the manufacture or durability of the airbag with the canister rupturing and propelling metal fragments toward the occupant. The size and velocity of the fragment were enough to penetrate the skin and injure vital organs. Death and injury from airbag ruptures is tracked NHTSA’s Office of Defects. As of 2023, 24 deaths have been reported and the cases were reviewed.

Other airbag suppliers have experienced rupture fragments from the canister killing and injuring occupants. A manufacturing problem occurred with Joysen airbags causing a recall. A chunk of metal has broken off ARC inflators during deployment propelling the fragment toward the occupant. Several cases have been reviewed and the dislodged fragment may be from an inadequate weld during the manufacture of the airbag

Results

Benefits of airbags

NHTSA unbelted crash tests

Four tests at 41.4 ± 2.0 km/h (25.7 ± 1.2 mph) and four tests at 48.0 ± 0.2 km/h (29.8 ± 0.1 mph) were found with unbelted drivers and right-front passengers in 1983–87 vehicles. Figure 2 summarizes the biomechanical responses for the driver (top) and right-front passenger (bottom) at the two severities. The dummy used was a mix of Hybrid III and Part 572. On average, the peak chest acceleration was at or above the 60 g tolerance, while the HIC, head acceleration and femur load were below. There were few other studies with unbelted responses in crash or sled tests.

Figure 2. Unbelted dummy responses in NHTSA 41 km/h (25 mph) and 48 km/h (30 mph) frontal barrier impacts.

GM airbag and manual lap belts

Wilson (1974) summarized the crash test performance of the GM air cushion restraint system (ACRS) with lap belt that went into the 1973–76 General Motors vehicles after a 1,000 vehicle field trial program in late 1972. Additional crash test results were summarized by Mertz (1988). Figure 3 shows the driver (top) and passenger (bottom) responses of the Part 572 dummy in frontal barrier crash tests at 30 and 40 mph. The chest and head responses were generally below tolerance even in the 40 mph barrier impacts. The femur loads were above tolerance in the highest severity test with the lap belt.

Figure 3. Dummy responses in frontal barrier crash tests (data from Wilson 1974; Mertz 1988).

Ford airbag and manual belts

Ford worked with Eaton on automatic inflatable restraints for occupant protection (Kemmerer et al. 1968; Frey 1970). This led to an experimental design for the 1981 Lincoln Town Car. Ford refined the design for the 1984 Tempo/Topaz by integrating the airbag in a steering wheel module with a unique wheel and column, five crash sensors, and a diagnostic module (Maugh 1985). The airbag had internal tethers to control the shape. Figure 4 shows driver dummy responses in frontal barrier crash tests with a driver airbag. The HIC, chest acceleration, and femur loads were below tolerances, even unbelted in 30 mph tests. Maugh (1985) modified the vent size to improve performance in the 35 mph NCAP test with good results.

Figure 4. Dummy responses in frontal barrier crash tests with driver airbag (data from Maugh 1985).

NHTSA NCAP frontal crashes tests

Eight vehicle crash tests at 35 mph were summarized with a lap-shoulder belted 50th Hybrid III in the driver seat and a 5th female Hybrid III in the passenger seat. The vehicle were a 2012–2020 passenger vehicle with advance airbags and dual pretensioning lap-shoulder belts. Figure 5 compares dummy responses to drivers and front passengers in 25–30 mph frontal crashes with unbelted occupants and no airbags (left bars) to those in 35 mph NCAP frontal crashes with belted occupants and airbags (right bars). There is a dramatic reduction in dummy responses with belts and airbags.

Figure 5. Dummy responses to drivers and front passengers in: (left bars) 25–30 mph frontal crashes with unbelted occupants and no airbags and (right bars) 35 mph NCAP frontal crashes with belted occupants and airbags.

Figure 6 compares the risk for severe injury and death (MAIS 4 + F) to drivers and front passengers. The left bars show unbelted drivers and passengers without airbags in 25–30 mph frontal crashes to those in 35 mph NCAP frontal crashes with belted occupants and airbags (right bars). The risk for severe head injury in unbelted drivers in 40–48 km/h (25–30 mph) frontal impacts is 5.495% compared to 1.073% in modern vehicles in 56 km/h (35 mph) NCAP tests with belts and airbags. This is an 80.5% reduction in risk. For the passenger, the head injury risk in 40–48 km/h (25–30 mph) frontal impacts was 4.435% compared to 0.797% in modern vehicles with belts and airbags. This is an 82.0% reduction in risk. There were similar reductions for other body regions.

Figure 6. Risk for MAIS 4 + F injury to drivers and front passengers in crash tests: (left bars) 25–30 mph frontal crashes with unbelted occupants and no airbags and (right bars) 35 mph NCAP frontal crashes with belted occupants and airbags.

Field data on severe injury (MAIS 4 + F)

Figure 7 (top) shows the risk for severe injury (MAIS 4 + F) to belted and unbelted occupants with airbags as deployed in frontal crashes by delta V (Viano and Parenteau 2022). For 30–35 mph delta V frontal impacts, the risk for severe injury is 15.45% with unbelted occupants and 4.68% with belted occupants with airbags as deployed. The reduction in risk is significant with belt use. Figure 7 (bottom) shows the effectiveness of seatbelt use in lowering injury risks over the range of crash severities. For 30–35 mph delta V frontal impacts, the effectiveness is 69.7%.

Figure 7. (top) risk of MAIS 4 + F injury from NASS-CDS for belted and unbelted occupants with airbags, as deployed, in frontal crashes by delta V and (bottom) effectiveness of seatbelt use in preventing severe injury and death (data from Viano and Parenteau 2022).

NHTSA Estimate of lives saved by airbags

NHTSA (2009) estimated that airbags saved 28,244 lines through 1-1-09. This includes 23,127 drivers with 9,128 belted and 13,999 not belted and 5,117 front-right passengers with 2,234 belted and 2,883 not belted.

Frequency of airbag deployment in frontal crashes

Figure 8 shows the frequency of frontal crashes by delta V (change in velocity) for lap-shoulder belted occupants using 1994–2015 NASS-CDS with n = 440,001/yr (Viano and Parenteau 2022). The crashes involve a vehicle being towed from the scene and do not include lower severity “fender bender” accidents. It includes the frequency of airbag deployment for lap-shoulder belted occupants in frontal crashes using 1997–2007 NASS-CDS (Viano and Parenteau 2010). The data shows why airbag performance is so important in minor-to-moderate severity crashes. 34.4% of all frontal crashes had a delta V < 16 km/h (10 mph) with airbag deployment in 35.0% of the crashes. 37.2% of the frontal crashes had a delta V of 16–24 km/h (10–15 mph) with airbag deployment in 62.8% of the crashes. The data shows that 71.6% of frontal crashes had a delta V < 16 km/h (15 mph). Pre-impact braking and lack of seatbelt use are factors in the occupant being close to the airbag in minor-to-moderate severity crashes. Overall, the airbag deployed in 64.2% of all towaway frontal crashes. 31.3% of all deployments occurred in <24 km/h (15 mph) and 49.6% in <32 km/h (20 mph) frontal crashes.

Figure 8. Frequency of frontal impacts from by delta V for lap-shoulder belted occupants from 1994 to 2015 NASS-CDS (Viano and Parenteau 2021) and frequency of airbag deployment from 1996 to 2007 NASS-CDS (data from Viano and Parenteau 2010).

Risks of airbags

First driver airbag death

The first death caused by an inflating driver airbag occurred to Dr. French. He was a physician on his way home from a late shift at the hospital. He apparently fell asleep at the wheel, went up a curb and struck a tree. He was found dead slumped over the steering wheel after the crash. He was unbelted. The case was unusual because one of the two doors for the airbag on the steering wheel did not properly opened. General Motors Research Laboratories was asked to investigate the crash. Horsch and Culver (1979) determined that the driver was against the airbag at deployment and blocked the normal opening of the airbag doors.

Punchout force

Supplementary Figure A2 shows the force of the driver airbag inflation into a body block with three different initial separations from the airbag. With a normal separation of 17.8 cm (7.0″), the peak force was 4.23 kN (951 lb). With 7.7 cm (2.6″) separation, the peak force was 11.57 kN (2,601 lb). With the body block against the airbag and no separation, the peak force was 20.0 kN (4,496 lb). The static tests demonstrated the high forces of inflation with the body block against or in close proximity to the inflating airbag. Horsch and Culver (1979) determined the fatal injury was from the punchout force with the driver partially against the airbag at inflation.

Horsch and Culver (1979) also conducted deceleration sled tests at 37.4 km/h (23.2 mph) to study the occupant in proximity to the inflating airbag. Figure 9 summarizes the results. With the occupant normally seated with 267 mm (10.5″) separation from the EA steering wheel. The dummy moved forward impacting the steering wheel causing column compression. The chest compression was 46 mm (1.81″) with a maximum velocity of 5 m/s (11.2 mph) and 3 ms spinal acceleration of 39 g. With the occupant normally seated and the airbag triggered at 9 ms (a normal crash timing), the chest compression was 18 mm (0.71″) with a maximum velocity of 1 m/s (2.2 mph) and the 3 ms spinal acceleration was 32 g. The IARV (injury assessment reference values) for chest compression is 50 mm (2.00″) in the 50^th Hybrid III and the spinal acceleration IARV is 60 g (Mertz et al. 2016). The tests showed low injury risks compared to IARV with a normal seating position without and with airbag deployment, although the airbag lowered risks.

Figure 9. Hybrid III responses with punchout forces in 37.4 km/h (23.2 mph) deceleration sled tests with the driver at varying separations from the airbag deployment (data from Horsch and Culver 1979).

By delaying the firing of the airbag, the occupant was progressively closer to the airbag when it fired. With a 55 ms trigger time, the occupant was leaning forward with a 43 mm (1.70″) separation. The chest compression was 58 mm (2.29″) with a maximum velocity of 15 m/s (33.6 mph) and 3 ms spinal acceleration was 64 g. The rate of chest compression significantly increased with the delayed deployment. With a 65 ms trigger time, the occupant was against the airbag at inflation. The Hybrid III chest bottomed against the spine with >84 mm (>3.31″) chest compression, 23 m/s (51.5 mph) chest deflection velocity, and 113 g spinal acceleration. With a 72 ms trigger time, the airbag deployed 7 ms after the chest contacted the steering wheel. The chest bottomed out with >84 mm (>3.31″) with a deflection velocity of 15 m/s (33.6 mph) and 93 g spinal acceleration. All the tests with separation of 0–43 mm involved very high chest deflection and spine acceleration with responses above IARV. The rate of chest deflection was 15–23 m/s (33.6–51.5 mph), extremely high velocities.

The risk for chest injury was very low in a normal seating position without and with airbag deployment. It transitioned to very high risks for fatality when the occupant was close to or against the airbag at inflation. The proximity deployments showed extremely high risks for fatal injury from the deploying airbag.

The testing showed that with normal deployment, the inflation gases increase the pressure in the airbag module until they reach the pressure to open the doors allowing the airbag to expand. With the increasing volume in the inflated airbag, the pressure was about 180 kPa (26.1 psi) when the airbag broke open the doors on the module and the airbag inflated. As the airbag reaches full inflation, the pressure in the airbag was 17–69 kPa (3–10 psi) and then it deflated through large vent holes at the back and porous airbag fabric, which allowed gas to escape deflating the airbag. With occupant loading, the venting of gas increased as energy was absorbed by occupant restraint.

Figure 10 shows that the high pressure inside the module caused the cover to bulge outward with high-rate deformation of the chest in contact or near the airbag punchout force. With the occupant blocking the airbag cover, the pressure inside the module increased to 600 kPa (87.1 psi) as gas was generated in the restricted volume inside the module. The bulge is at high pressure and caused punchout forces with the occupant against the module. The sled tests confirmed the high forces of punchout with the occupant blocking or in close proximity to the doors of the inflating airbag. The airbag deployment death was caused by punchout force from the constrained airbag.

Figure 10. Punchout force causing the cover to bulge with high velocity compression of the chest (modified from Horsch and Culver 1979 and Melvin et al. 1993).

The airbag was packaged in a sealed container with only two doors at the front to allow the airbag to expand toward the occupant. The door opening allowed gas pressure to escape the container as it filled the airbag. Punchout referred to the initial pressurization and swelling of the module before the airbag inflated out of the steering wheel hub. The airbag is initially folded behind the module cover and the available volume is constrained by the internal volume of the module. With inflation, gas flows into the folded airbag, the internal pressure increases and the module swells increasing stresses on the module cover until it opens and the airbag inflates toward the occupant. In this phase, the airbag has the area of the module cover. The punchout phase ends as the airbag breaks-out of the module, rapidly adding volume and “unloading” pressure in the airbag. Fatal chest injury by the airbag was related to internal pressure in the module. It was not from an impacting mass; it was from high-rate deformation of the chest. If there was available volume for the airbag to expand by the generated gas, there was low internal pressure in the airbag and with reasonable restraining load on the occupant.

Lau et al. (1993) found that the viscous response was the most sensitive biomechanical response associated with OOP risks for chest injury by the driver airbag. The viscous response (VC) is the product of the instantaneous velocity (V) and the instantaneous compression © of the chest deformation. VC_max is the peak viscous response (Viano and Lau 1985, 1988; Lau and Viano 1986, 1988). Internal injury to the heart and vital organs often occurred with minimal rib fractures because of the high-rate, impulsive loading with low compression (Lau and Viano 1988). VC_max occurred within 15 ms of airbag ignition during the punchout force of the airbag. This was about half the time it took to reach maximum chest compression (Lau et al. 1993). VC_max coincided with peak airbag pressure and force on the chest. Lau et al. (1993) also noted that punchout loading was reduced by allowing the airbag to inflate from other parts of the container not in contact with the driver during OOP deployments. Loading by the inflating airbag was reduced by a compliant steering column supporting the module. The amount and rate of generated gas had only a marginal effect on the injury from the punchout force. Even an inflator with inadequate gas output to protect a properly seated occupant had sufficient energy to cause severe injury with the occupant in contact with the inflating module by the punchout force from the bulging cover of the module and initial inflation.

Wrap-Around force (membrane tension)

Horsch et al. (1990) continued to study airbag injury with an OOP driver. The testing found significant deployment loads with the driver near the steering wheel airbag. Figure 11 shows a second type of OOP inflation injury. It is from wrap-around forces with membrane tension as the inflating airbag wrapped around the chest or neck and face of the occupant. The high tension forces in the fabric of the airbag loaded the occupant.

Figure 11. Membrane force with the airbag wrapped around the neck-face or chest with high velocity loading (modified from Horsch and Culver 1979 and Melvin et al. 1993).

Membrane forces refer to loading after the airbag escapes from the module and the inflating airbag wraps around the occupant. The force is a result of airbag pressure acting on the occupant and usually includes tension forces from airbag wrap-around. Membrane force can be as great as punchout forces but differ from punchout by a larger loading area and lower airbag pressure.

Lau et al. (1993) studied driver OOP injury. They found the punchout and membrane loading caused injury by high-rate deformation of the body from the velocity of the expanding airbag on the body region. Membrane forces caused injury by high compression over a larger area. The wrap-around loading was most severe in small drivers, sitting forward on the seat tracks. Melvin et al. (1993) found that the 5th Hybrid III female dummy had a greater risk of neck and base of skull injury from membrane forces from airbag inflation than the 50th Hybrid III male dummy for a range of seating positions. The membrane loading was associated with the inflating airbag as it wrapped around the neck, chin and face. The tension in the airbag fabric increased as the airbag inflated due to internal airbag pressure, not from inertial forces from the airbag. The force was sufficient to dislocate the atlanto-occipital (AO) joint and fracture the base of the skull with brainstem and spinal cord injury. Agaram et al. (2001) and Bass et al. (1999) described other OOP inflation tests with neck and chest loading of the Hybrid III.

The wrap-around effect was reduced by a biased flap of fabric covering the folded airbag. As the airbag unfolded, the fabric contacted the occupant if they were close to the module. The flap redirected the airbag to one side of the neck, chin or face diverting membrane forces from wrapping around the occupant.

Field performance of the 1973–76 General Motors fleet of airbags

Mertz (1988) summarized the initial field experience with 11,321 driver and passenger airbags. In 1973, General Motors sold 1,000 Chevrolet Impalas with airbags to the Government. At the time, they were called air cushion restraint systems (ACRS). The vehicles were a field trial program (FTP) to gain experience with airbags. In 1974–76, General Motors sold 10,321 Buicks, Oldsmobiles, and Cadillacs with driver and passenger airbags.

Pursel et al. (1978) used a matched case methodology to estimate the effectiveness of the ACRS and other occupant restraints. The method found crashes of similar severity with different restraints used by matched occupants. The risk was compared for different occupant restraints. The effectiveness of the ACRS was 17.8% (−23% to +56%, 90% CI) in preventing AIS 3 injury in deployment crashes and 5.6% (−23% to +33%, 90% CI) for AIS 2 injury.

Mertz (1988) conducted a case review of airbag deployment injuries in 1973–76 General Motors vehicles. By December 1980, there were 216 deployment accidents involving 216 drivers, 86 right-front occupants and 11 center-front occupants. There were 13 children less than 10 years old out of the 97 passengers (13/97 = 13.4%). Seatbelt use was not reported. The crashes and injuries were reviewed to determine if the injury was expected or unexpected. Most of the AIS 3+ injury occurred in minor to moderate severity frontal impacts without intrusion or distortion of the occupant compartment. There were unexpected injuries based on the type and severity of the crash that may have occurred with the occupant close to the airbag at deployment. There was a 41 year old male drive unexpectedly killed by chest or neck injuries in a moderate severity frontal impact (case LA128). This is likely the case that was studied by Horsch and Culver (1979).

Child deaths from passenger airbags

A 1 month old unbelted male was unexpectedly killed by the passenger airbag. The infant was laying unbelted on the passenger seat. Braking prior to the crash caused the child to slide forward before the passenger airbag deployed. The child was exposed to high forces of deployment in close proximity to the deploying airbag and died of severe head injuries.

Child OOP airbag risks

Airbag deployment injury to children close to the module at deployment was anticipated prior to the production of passenger airbags and studied during the development of the 1973–76 General Motors airbags. Patrick and Nyquist (1972) conducted out-of-position (OOP) tests with passenger airbag deployment. Five positions were evaluated (1) standing on the floor near the airbag, (2) crouched on the seat with face on the airbag, (3) kneeling on the seat with arms and elbows on the instrument panel, (4) sitting on the floor and (5) normally seated with the chest 15.2 cm (6″), 30.5 cm (12″) or 45.7 cm (18″) forward of the aft end of the inflated airbag. They concluded that with the head in the path of the deploying airbag, brain injury occurred and with the chest in the path of the inflating airbag, fatal organ injuries occurred. The passenger airbag had sufficient force to propel a 13.2–15.5 kg (29–34 lb) occupant over the front seatback. Aldman, et al. (1974) reached a similar conclusion of risk for death with a children within 10–15 cm (4″–6″) of an inflating passenger airbag. They also found the rear-facing child seat resting against the passenger instrument panel was forcefully propelled by the inflating airbag with a volume of 190 L (6.7 ft³).

Takeda and Kobayashi (1980) described child OOP positions and injury with the passenger airbag mounted at different heights on the instrument panel. Montalvo et al. (1982) determined factors influencing the position of children with pre-impact braking. They found a child moved off the front edge of the seat during hard braking and was near or against the instrument panel in different positions. The hard-braking tests identified 13 positions for the children. Stalnaker et al. (1982) determined the influence of pre-impact braking on the position of unrestrained, anesthetized animals sitting, kneeling or standing during the hard braking. Mertz et al. (1982) conducted OOP deployments with anesthetized animals in a range of passenger positions. The airbag caused severe injuries to major organs of the head, neck and torso confirming deployment risks for children near or against an inflating airbag.

The child OOP research provided a basis for SAE to specifying guidelines for child OOP testing in J1980 to assess injury risks with passenger airbags (SAE 2011, superseding SAE 1990). NHTSA adopted many of the OOP conditions in the advance airbag rule to ensure low risks with OOP.

Other unexpected injuries with 1973-76 airbags

There was a high frequency (9/97 = 9.3%) of passenger leg fractures in minor to moderate crashes without intrusion (Mertz 1988). The fractures were unexpected because of low femur loads of 2.09–5.38 kN (470–1,210 lb) in sled testing at 48.3 km/h (30 mph) of the airbag. Seven of the nine fractures were to the tibial plateau consistent with a downward load on the knee joint by the tubular knee airbag. Testing with an instrumented lower extremity showed that with the legs angled to the instrument panel, large compressive force occurred in the tibia of one leg explaining the field injury. The high rate of tibial plateau fractures led to the removal of the tubular knee airbag inside the passenger airbag.

There were hand and arm fractures with airbag deployment. The fractures occurred as the airbag impacted the hand or the hand was flung against the interior by the inflating airbag. In some cases, the hand was flung backward into the face of the driver.

Leading edge velocity of airbags

Powell and Lund (1995) measured the leading-edge velocity of driver airbags in 1993 MY vehicles. The results showed that airbags have a maximum velocity of 245 ± 41 km/h (67.9 ± 11.2 m/s or 152 ± 25 mph) and excursion of 387 ± 59 mm (15.2″ ± 2.3″). The airbags had various fold patterns and many used tethers inside the airbag to control the shape (Supplementary Table A1).

NHTSA SCI (Special Crash Investigation) of first generation airbags

By the early 1990s, front airbags became standard equipment in passenger vehicles. Manual lap-shoulder belts were provided and the State encouraged the use of seatbelt by passing laws requiring the driver and right-front passenger buckle up when using a motor vehicle. NHTSA started receiving notification of unexpected deaths and injuries from airbag deployment.

From 1972–90, NHTSA conducted Special Crash Investigations (SCI) of crashes involving airbag deployments. In 1990, SCI shifted from investigating all airbag crashes to investigating only airbag-related serious injury and death, airbag success stories, the interaction between airbags and child safety seats, airbag nondeployment crashes and inadvertent airbag deployments. The focuses changed as airbag technology developed. In 1991, SCI confirmed the first driver airbag deployment fatality (CA9109). In 1993, the first child airbag deployment death was confirmed (CA9307). In 1996, SCI expanded to investigate all airbag related life threatening injury or death, and they started publishing summary tables for each confirmed airbag deployment death and serious injury to adults, children up to 12 years old not in rear facing child safety seats (RFCSS) and infants in RFCSS.

Chidester and Roston (2001) and Kindelberger et al. (2003) provided updates on airbag deployment deaths and serious injury from SCI. There were a number of SCI reports with cases tabulated by NHTSA with airbag deployment death and serious injury. The last report was published in 2009, when NHTSA concluded the special investigation of airbag deaths (NHTSA 2009). There were 296 confirmed deaths related to first generation airbag deployments in 1989–2003 MY vehicles.

Figure 12 shows there were 93 airbag-related driver deaths. The cases are shown by vehicle MY and million vehicle years. The majority of the deaths occurred in 1990–1997 MY vehicles. The greatest rate was in 1991 MY vehicles with 0.342 driver deaths per million vehicle years. By 2000, airbag-related deaths diminished with the introduction of sled-certified and advanced driver airbags.

Figure 12. Airbag-related driver deaths by vehicle model year (MY) in SCI (data from NHTSA 2009).

There were 93 drivers with 70 (75.3%) females and 23 (24.7%) males. The crashes had a 21.7 ± 7.2 km/h (13.5 ± 4.5 mph) delta-V with a median severity of 20.9 km/h (13 mph). Pre-impact braking occurred in 27 crashes with 43 no braking and 23 unknown. The cause of death was primarily to the head, neck or chest with 30 (32.3%) cases of brain injury or skull fracture, 8 (8.6%) with cervical spine injury, 47 (50.5%) with chest injury, including heart rupture and aortic laceration, and 6 (6.5%) cases with head and chest injury, 1 (1.1%) with head and neck injury and 1 (1.1%) with head, neck and chest injury.

Twenty-five (25) of the head or neck injuries occurred at the base of the skull or the upper neck connection to the head. This is consistent with the airbag wrapping around the face and neck with high forces at the base of the skull. Seatbelts were used by only 28 drivers (30%) while 58 (62%) were unbelted, 4 (4%) were unknown belt use and 3 (3%) had belt misuse reported. The height of the drivers was classified. There were 41 the size of th^e 5th Hybrid III (<64″), 45 the size of the 50^th Hybrid III (64″-70″) and 5 the size of the 95th Hybrid III (>70″). For the females, there were 33 drivers and 6 passengers under <157.5 cm (62″) tall that qualified as short-statured females.

Figure 13 shows there were 184 children killed by the passenger airbag. The cases are shown by vehicle MY and million vehicle years. The majority of the deaths occurred in 1993–2001 MY vehicles. The greatest rate was in 1995 MY vehicle with 0.539 child deaths per million vehicle years. By the 2004 MY, child airbag deployment deaths were essentially eliminated by sled certified and advanced passenger airbags. The majority of children (138, 75.4%) were unrestrained or improperly restrained. Of these, 115 of the deaths involved pre-impact braking with 19 no braking and 4 unknown. Thirty-seven children (37, 20.2%) were in child seats with 29 in rear-facing and 8 in forward facing child seats and one unknown. Only 8 children (4.4%) were lap-shoulder belted. For the lap-shoulder belted children, 3 deaths were with pre-impact braking and 5 were without braking. There were 8 children killed by the driver airbag. 54.4% of the children were female and 45.6% were male. The average reported height was 108.9 ± 18.8 cm (42.9 ± 7.4) and weight was 20.5 ± 7.6 kg (45.2 ± 16.8 lb). The delta V of the crashes averaged 19.4 ± 5.7 km/h (12.1 ± 3.6 mph) with a median severity of 18.5 km/h (11.5 mph).

Figure 13. Airbag-related child deaths not in RFCSS by vehicle model year (MY) in SCI (data from NHTSA 2009).

There were 13 adult passenger deaths with the first in a 1992 MY vehicle, 1 in 1993 MY, 1 in 1994 MY, 3 in a 1995 MY, 3 in 1997 MY, 2 in 1998 MY, 1 in a 2000 MY and the last one in 2004 MY vehicle. Seven (53.9%) of the adult passengers killed by the airbag were not restrained, 5 (38.5%) were lap-shoulder belted and 1 (7.7%) was misusing the belts. Chidester and Mynatt (2015) provided an update on SCI cases with improvement in airbags from the first generation to sled certified (depowered) to advanced airbags. Deaths and injuries from airbags were significantly reduced. By the early 2000s, airbag deaths from punchout and membrane forces were essentially eliminated with depowered and advanced airbags. Over the same period of time, NHTSA estimated that airbags saved the lives of 23,127 drivers with 9,128 lap-shoulder belted and 13,999 unbelted and 5,117 right-front passengers with 2,234 lap-shoulder belted and 2,883 unbelted. By 2010, Supplementary Figure A1 shows the death rate was 1.11 per 100 million miles traveled (NHTSA 2022). This was an 84.7% reduction from 1950.

There were 29 infant deaths in rear-facing child safety seats (RFCSS) by the passenger airbag. The average age was 3.4 ± 2.2 months old. There were 14 male and 15 female deaths. The average weight was 7.4 ± 2.2 kg (16.2 ± 4.8 lb) and height was 62.4 ± 8.0 cm (24.6 ± 3.1 inch). Only 10 (41.7%) of the 24 known cases involved proper use of the RFCSS with 6 (25%) case of misuse, typically the child seat on the lap of a passenger. The average delta V was 20.1 ± 6.9 km/h (12.5 ± 4.3 mph). Seventeen (65.4%) of the known cases involved pre-impact braking. Nine (34.6%) had no braking. All the infant deaths were related to brain injury with 16 cases of skull fracture (59.3% of 27 cases).

Airbag ruptures propelling fragments

Field accidents have not identified new patterns of driver or right-front passenger death from OOP deployment with the occupant close to or against advanced airbags, OOP conditions. There has emerged new risks with airbag ruptures propelling fragments of the airbag canister toward the occupant. The size of the fragments vary from about 1–10 mm or more and the speeds are sufficient to penetrate the body.

Takata airbags

Takata airbags are widely used in motor vehicles. There have been cases with ruptures of components in the canister that are propelled with the airbag deployment toward the occupant. The fragments have enough mass and velocity to penetrate the skin of the occupant resulting in death and serious injury.

Figure 14 shows the Takata airbag deaths from the rupture of the canister by vehicle MY (model year) and CY (calendar year) of the crash. NHTSA is aware of 24 deaths associated with fragments from a Takata airbag. The first airbag rupture death was on 5-27-09. A female driver died in a 2001 Honda Accord crash from cannister fragments from the airbag. Since then, there have been at least 23 other cases with a Takata airbag rupture, 17 in the US, 6 from Malaysia, and 1 from Australia. The average age of the vehicle is 10.9 ± 4.1 yo (range from new to 15 yo) indicating the age of the vehicle and durability of the airbag are factors.

Figure 14. Takata driver airbag rupture deaths by vehicle MY and CY of crash.

There is likely an under-reporting of cases worldwide. A review of the literature identified a few Journal articles related injuries and death from airbag fragments. Fouda and Almaged (2022) reported an airbag fragment from a 2006 Honda Civic killed a 31 yo male in the Kingdom of Bahrain. This would make the 25th fatality. McCrary et al. (2016) reported on a crash with a 2006 Ford Mustang where a computed tomography angiography (CTA) scan weeks later revealed a metallic foreign body in the prevertebral soft tissues of the left posterior pharyngeal. Surgery removed the fragment 2 months after the accident. They noted the case was the first medical report of a metal object from an airbag deployment penetrating the neck. Vadysinghe et al. (2023) reported a fatality with a 2 × 2 cm fragment from the airbag lodged in C4–C5 and transection of the right carotid artery. The Takata airbag defect has resulted in the recall and replacement of driver and passengers airbags, although the replacement of airbags has been slow with so many older vehicles with a 2nd, 3rd or more owners.

Joyson and ARC airbags

Joyson is the successor company of Takada. They have experienced airbag manufacturing issues with deployment fragments in crashes causing recalls. There are cannister ruptures with ARC airbags. Several cases have occurred with penetrating injury from a chunk of the cannister breaking off during deployment and propelling into the occupant.

Estimate of airbag deployment deaths

The field data provides a rough estimate of at least 320 airbag deployment deaths, which has prompted improved designs, testing and recalls.

Discussion

In 1966, NHTSA mandated the installation of seatbelts in motor vehicles by the 1968 MY, but very few occupants used the belts while riding in the vehicle. This led the Government to pursue automatic restraints (passive belts or airbags) that fostered a debated about the preference of belts or airbags for occupant protection. It took Mercedes-Benz to clarify that occupant protection needed seatbelts and airbags with manual use of lap-shoulder belts and supplemental airbags. NHTSA adopted the position in Secretary Dole’s decision in 1985. This focused efforts on State laws requiring the use of seatbelts and on manufacturers installing airbags for the driver and right-front passenger. Airbags became standard equipment with President Bush’s decision in 1991. Today, seatbelt use is frequent, except in fatal crashes, and advanced airbags are standard equipment in vehicles.

In 1973–76, a limited fleet of driver and passenger airbags were sold by General Motors. The initial 216 airbag deployment crashes resulted in one driver death from the airbag in a minor-to-moderate severity crash where no serious injury was expected. Research showed that the driver death was from punchout force from the inflating airbag with the driver against the steering wheel, blocking the normal space for the airbag deployment. The mechanism of death was high-rate chest compression with a viscous response (VC) sufficiently high to cause heart rupture, aortic laceration, and fatal injury.

Figure 15 shows human tolerance to chest compression (C) and viscous response (VC). Crushing injury is related to chest compression with a tolerance of 38% C for a 50% risk of serious injury. With higher velocities of chest compression, the tolerable compression decreased. The viscous response (VC) assesses the combination of the amount (C) and rate (V) of chest deflection, which is related to absorbed energy by high-speed deflection (Viano and Lau 1985). The tolerance for high-rate compression is VC = 1.00 m/s for a 50% risk of serious injury. Bir and Viano (1999) analyzed 41 anesthetized animal tests with different rates of chest compression. The viscous response was correlated with heart rupture at VC = 2.69 ± 0.86 m/s. The testing showed VC correlated with the risk of severe injury at VC = 2.29 ± 1.01 m/s and ventricular fibrillation at VC = 2.20 ± 0.84 m/s. The testing by Horsch and Culver (1979) had a viscous response of VC = 2.63 ± 0.69 m/s with the chest of the Hybrid III in contact with the inflating airbag (Horsch et al. 1990; Lau et al. 1993). The inflation was so severe in some of the tests that the sternum bottomed out against the spine of the Hybrid III. The average deflection was 75.4 ± 6.5 mm with bottoming out at 84 mm. Powell and Lund (1995) found inflation velocities of 245 ± 41 km/h (67.9 ± 11.2 m/s or 152 ± 25 mph) in 1993 MY driver airbags, consistent with high-rate loading of the occupant in contact with the inflating airbag. Additional research identified a second mechanism of injury from membrane tension with the inflating airbag wrapping around the occupant while deploying. Tension in the airbag fabric was sufficient to cause fatal neck, head, and chest injuries while the airbag was unfolding (Lau et al. 1993; Melvin et al. 1993).

Figure 15. Tolerance to chest compression and viscous response with OOP airbag testing of the Hybrid III dummy and anesthetized animals with punchout force on the chest.

NHTSA’s SCI investigated airbag deployment deaths of drivers and right-front passengers in vehicles equipped with first generation airbags. The cases involved a high frequency of female driver deaths (75.3%) in minor-to-moderate severity crashes. Smaller females have the seat forward on the tracks and sit close to the steering wheel. The understanding of punchout force and membrane tension with smaller females led to standardized test conditions the 5th female Hybrid III dummy in OOP positions near the inflating airbag (Miller and Phillips 2001). The 5th Hybrid III 151.4 cm (59.6″) tall and weighs 46.7 kg (102.8 lb). The OOP tests assessed the risk for injury in various seating positions with the small female blocking the normal airbag deployment. The conditions were considered in the advance airbag rule (NHTSA 2001).

The field experience with the 1973–76 General Motors airbags also resulted in a child-passenger airbag death in a crash where no serious injury was expected. The infant death confirmed earlier concerns that unrestrained children (and short adults) moved forward with pre-impact braking and were close to the instrument panel when the airbag deployed. Research showed the seat was often forward on the tracks with children and small passengers. The risk was in minor-to-moderated speed crashes because of the higher number of crashes with low delta V.

The driver deaths in SCI had an average delta V of 21.7 ± 7.2 km/h (13.5 ± 4.5 mph) with a median severity of 20.9 km/h (13 mph). The child deaths not in RFCSS had an average delta V of 19.4 ± 5.7 km/h (12.1 ± 3.6 mph) with a median severity of 18.5 km/h (11.5 mph). The majority of SCI deaths were in crashes with delta V < 24 km/h (15 mph) indicating the importance of airbag performance in low-speed crashes where the driver and passenger may be OOP and near the airbag during inflation.

In contrast, only 3.37% of frontal crashes had a delta V > 48.3 km/h (30 mph) and only 1.84% with delta V > 56.3 km/h (35 mph). Most federal and consumer crash tests are conducted at more than 48.3 km/h (30 mph) into a rigid or deformable barrier. The NCAP crash test of NHTSA is a 56.3 km/h (35 mph) rigid barrier impact with a typical delta V of 62.8 km/h (39 mph) due to restitution (rebound). Injury risks increase with delta V, but the OOP airbag inflation injuries were amplified by the high frequency of low-speed crashes where there are opportunities for OOP when the airbag deploys.

NHTSA’s SCI of passenger airbag deaths found 94.4% of children were not restrained (60.0%), were improperly restrained (17.5%) or were on the lap of an adult (16.9%). 77.5% of the deaths involved pre-impact braking with 18.8% no braking and 3.8% unknown braking. The cases confirmed that deployment of first generation airbags in minor-to-moderate severity crashes put children and smaller occupants at risk by pre-impact braking and lack of seatbelt restraint allowing the occupant to be near the deploying airbag.

The research identified the likely positions of children with pre-impact braking (Patrick and Nyquist 1972; Montalvo et al. 1982; Stalnaker et al. 1982). Static and sled testing showed the injury mechanisms for children and small occupants by OOP deployments of the airbag (Mertz et al. 1982). The understandings led to a multi-faceted approach lowering the risk for passenger airbag deaths. Supplementary Figure A3 shows the test conditions for child OOP and Supplementary Figure A4 shows the conditions for adult passenger OOP (SAE 2011). The research led to (1) seat track sensors to monitor the forward position on the track, (2) seat sensors to detect children, child seats, and small occupant, (3) low-risk airbag deployment designs with a biased flap or other countermeasures, (4) strategies to suppress deployment if the occupant was near the airbag and (5) an array of tests to confirm low risks of airbag injury. These approaches were considered in the advance airbag rule (NHTSA 2001).

The airbag deployment deaths to drivers and passengers were essentially eliminated with advanced airbags in vehicles. Figures 12 and 13 show that airbag-related deaths essentially stopped in 2003 with the phase-in of advanced airbags. Since then, there have been issues with advanced airbag manufacturing and durability with the deployment of cannister fragments through the airbag fabric penetrating the occupant with injury and death. Airbags are a successful part of the occupant protection system in vehicles. NHTSA (2009) estimated airbags saved 23,127 driver lives and 5,117 right-front passenger lives, although at least 320 airbag deployment deaths have occurred over the same period of time.

Limitations

There are many limitations to the type of review and analysis conducted in this study: (1) The history of airbags and automatic restraints is complex and occurred over decades. While the most important milestones have been covered, a deeper evaluation and other perspectives may provide additional insights on the evolution of airbags. Strother and Morgan (1974) discuss the early work at NHTSA. (2) The 1973–76 fleet of driver and passenger airbags involved General Motors vehicles. I was aware of the development and the issues of OOP considered before the introduction of the systems in vehicles. There were a number of people from different divisions of General Motors working on the development of airbags, including Fisher Body Division, Saginaw Division, Oldsmobile Division, Delco Electronics, Inland Division, SRDL (Safety Research and Development Laboratory), Engineering Staff and General Motors Research Laboratories, among others. The collaboration of many people brought airbags into production. Other companies had development work, but very little of it was published. (3) There were comments about an early 1970s fleet of Mercury Marquis with front airbags. A search of the literature did not find a technical paper describing the system. It would be helpful to understand the airbag design and the reasons the program was discontinued by Ford. (4) Some crash and sled tests were conducted with the newly developed 50^th Hybrid III dummy that had improved biofidelity of the head, neck, and chest. Other tests were conducted with older style dummies representing the 5th and 95th adult, 6 yo and 3 yo children and infants. The lack of biofidelity in the older dummies limited the evaluations of OOP risks. The family of Hybrid III dummies developed in the late 1970s through the 1990s. It now includes the 5th female, 50th male, 95th male, 10 yo, 6 yo, 3 yo and infant dummies used in OOP testing. (5) During the literature review, a number of law journal articles were evaluated on automatic restraints, airbags, airbag-related injuries, and airbag defects. While most of the articles were well written, many of the technical facts were incorrect or misunderstood. Caution is suggested in considering the technical aspects of law review articles, even if the narrative linking bits of incorrect history is excellent writing. (6) While the government and safety advocate perspectives were included in the study, my background was with General Motors. I have an industry perspective as well as a public health view of the history of airbags. My perspective may differ from others. (7) The paper has not gone into details on the passenger occupant detection (POD) or weight sensor technologies in the right-front passenger seat or airbag suppression technologies and logic in advanced airbags, for example, see Anishetty et al. (2000). It has also not covered modern crash sensing, for example see Sala and Wang (2003), Gabler and Hinch (2007). (8) The field data on the 1973–76 fleet of General Motors airbags was limited. Seatbelt use was not reported, but the two airbag deaths involved an unbelted occupant. The field data from the SCI investigations of first generation airbag deaths included information about the crash, occupant, and injury. There were no photographs available of the vehicle or interior. This type of information would be helpful. (9) The injury mechanisms with first generation airbags inflating against an occupant were studied with anesthetized animals to understand the injury types and timing with high-rate airbag loading. The studies with dummies explored design changes to reduce the risks of OOP injury. There were different causes of death and injury related to airbag inflation. The most relevant causes and biomechanics were summarized in this study. (10) Airbag deployment causes minor-to-moderate severity injuries by contact with the unfolding airbag, including skin abrasions on hands, arms and neck from the inflating airbag. Eye abrasions have occurred from the unfolding airbag. Burns have occurred from hot gases or with the occupant against the hot inflator after the crash. These types of injury and others were not addressed in the study. (11) The airbag is a part of FMVSS 208, the crash protection standard. There are many other standards that affect the protection of occupants in frontal impacts. They were not evaluated in this study. Kahane (1984, 2015) provided comprehensive evaluations of the effectiveness of the safety standards over more than 30 years.

Acknowledgments

The author conducted occupant protection and biomechanics research at the General Motors Research Laboratories during the development of first generation airbags. He managed the research on airbag-related injury and designs to reduce OOP inflation injuries with advanced airbags. This study is his understanding of the history based on the published research and his recollection of the history managing the injury biomechanics and occupant protection research at General Motors. Others may have relevant literature and memory from different perspectives.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

All data generated and analyzed during this study are included in the published article and the references. Additional information on the data and studies can be requested of the author at [email protected].

Additional information

Funding

The author(s) reported there is no funding associated with the work featured in this article.