The unformatted htm text below is from		Copyright © 2004 by Leonard Evans

12 Airbag benefits, airbag costs 

This  html version contains only the text (no figures, tables, equations, or summary and conclusions).  To check printed book appearance see pdf version of Chapter 1 or pdf version of Chapter 16.

Introduction
     No safety device has consumed more attention and resources than the airbag. The airbag mandate, the requirement that vehicles be equipped with airbags, has been at the center of US safety policy since the 1970s. The cost and complexity of airbags, and the controversy surrounding them, calls for more detailed analyses than was devoted to any of the other occupant protection devices covered in Chapter 11. Chapter 15 discusses the profound impact airbags and the airbag mandate have had on overall US safety policy since the 1970s. This chapter discusses airbags as devices, with particular emphasis on their costs and benefits.
     While airbags were originally intended to be primary occupant protection devices, all vehicle manufacturers now explicitly state that they are supplemental devices aimed at enhancing the effectiveness of the primary occupant protection device, the lap-shoulder belt. Here the term airbag refers only to frontal airbags.
     The main component of the airbag system is a strong fabric bag folded and stored in a module on the steering column for the driver, and in the dashboard for the passenger. When onboard sensors detect a frontal crash of severity exceeding a set threshold, equivalent to a delta-v of about 10 mph, detonators deploy the airbag. High pressure chemically-produced gasses force the bag out of the module and inflate it sufficiently rapidly that it is in place in front of the occupant before the occupant has had time to move forward appreciably in response to the crash forces.
     A plethora of technical and policy subjects relating to airbags is covered in a massive and rapidly expanding literature (see, for example, the summary in Ref. ). Despite so much literature, many of the most basic questions still lack confident answers. The question of whether the benefits of airbags are commensurate with their considerable costs has received scant attention. This question constitutes a major portion of the present chapter. Even after over
ten million airbag deployments, it is still not known with confidence whether airbags provide a net decrease or net increase in risk of different severity injuries. However, it is well established that when a crash occurs, airbags reduce fatality risks to belted or unbelted drivers, as summarized in Table 11-4 (p. 286).
Overview of frontal airbags
     Before estimating airbag benefits and costs, some overview information on airbags in the US is presented. Because many key quantities are changing, the benefit-cost comparison must be locked to some specific time. July 2003 is selected. In situations in which data are not available for July 2003, projections will be made from available data.
Number of airbags on US roads in July 2003
Data on the growth of airbags in the US vehicle fleet provide the following estimates for July 2003: ,

Number of airbag deployments
     NHTSA estimates that 520,300 airbags deployed in 1996. In July 1996 there were an estimated 74.6 million airbags on US roads,2 leading to a deployment rate of 6.97 deployments per 1,000 airbags per year.
Elsewhere NHTSA estimates 3.8 million deployments from the 1980s to 1 October 1999. From the data in Ref. 2, we estimate exposure as 606.9 million airbag-years. This implies 6.26 deployments per 1,000 airbags per year.  Let us take the average of these as the best estimate, leading to an airbag deployment rate of 6.6 deployments per 1,000 airbags per year. Applying this rate to the number of airbags leads to the following estimates for the number of airbag deployments in 2003:
Airbag deployment and non-deployment in fatal crashes. Figure 12-1 shows the growth of deployments in fatal crashes as recorded in FARS data. Deployments in fatal crashes are closely proportional to the growth of airbags in the fleet - indeed for the nine years plotted, the probability that a crash was fatal given that the airbag deployed varied only between 1.06% and 1.19%, with an average of 1.12%. For 2002, the latest year plotted, 16,682 of the 1.54 million deployments occurred in fatal crashes. (The estimate for 2003, which we require later, is that about 19,000 of the 1.70 million deployments are expected to occur in fatal crashes).
     Similarly stable was the probability of death, given that the airbag deployed in a fatal crash. Over the nine years plotted this probability varied only between 42.9% and 44.8%, with an average of 43.6%. The highest value of 44.8% was for the most recent year, 2002, in which 7,467 occupants were killed in 16,682 deployments. An increasing trend in this probability would follow if second generation airbags either reduced the number of life-saving deployments because of higher thresholds, or reduced the effectiveness when they did deploy because of the reduced power. Preliminary data indeed suggest lower effectiveness for second generation airbags.
For 2003, about 8,300 occupants are expected to be killed in crashes in which their airbags deploy. The data in Fig. 12-1 include 37,223 occupants killed in seats at which airbags deployed, so by mid 2003 well over 40,000 occupants had died in seats protected by airbags.
     The total number of driver and right-front passenger fatalities in cars and light trucks remained relatively unchanged from 1994 through 2002 even as the percent of drivers with airbags increased from 13% to 60% and the percent of passengers with airbags increased from 3% to 50%.2 This finding alone is sufficient to reject the claim that airbags would prevent 12,100 fatalities, as promised in the documentation used to justify the airbag mandate. (p 34298)
Deaths caused by deploying airbags. In order to provide protection the airbag must inflate in the short time between the detection of the crash and the occupant beginning to move forward appreciably in response to crash forces. The only way things can happen in a short time is for them to happen quickly. The limited time available makes it necessary for the surface of the airbag to move rapidly towards the occupant, reaching speeds of around 150 mph. The goal is that when the occupant first contacts the airbag it should already be inflated. However, if the occupant is in the space into which the airbag inflates, he or she will be struck at up to 150 mph rather than striking the vehicle interior at a speed that could be as low as 10 mph. The impact from an inflating airbag poses a major risk of death or serious injury. People of any size are at risk if any part of their body is in the space into which an airbag deploys, as might happen if they were reaching to retrieve a dropped object. This risk was understood and named out of position since the 1970s. Drivers of short stature sitting at their most comfortable distance from the steering wheel are out of position. Concerns about the injuries that inflating airbags might cause children appeared in the technical literature from the 1970s.8,
NHTSA reports that, as of July 2003, there were 231 confirmed deaths caused by airbag deployments in crashes that would otherwise not have been life threatening. Most of those killed were children. However, 77 drivers were killed, 58 of them female, and 28 of these of height 62 inches or less.10
     Beyond these specifically identified cases, about 40,000 occupants have been killed in crashes in which their airbags deployed. While it is not known how many of these were killed by deploying airbags, it is absurd to think that the number is zero. Some unknown number of occupants (say, N1) die in crashes that they would have survived if no airbag had been present. The number (say, N2) who survive because of the airbag is likewise unknown. Effectiveness estimates address only the difference N2 - N1, but provide no information on the values of either N1 or N2. If credence is given to the large numbers of saved by the airbag anecdotal claims, then there must be a correspondingly large number of killed by the airbag cases to balance most of these, otherwise net effectiveness would be far higher than the values found in large-scale epidemiologic studies (Table 11-4, p. 286. See also p. 327).
Direction of impact. While airbags are designed to deploy only in frontal crashes, deployments in fact occur for impacts in many directions. Fig. 12-2 presents driver fatalities in FARS 2002 for drivers of cars and light trucks according to the principal impact point, the point associated with most harm. The logarithmic scale is used because of the wide variation, from 3,154 driver deaths at 12 o'clock to 36 at 5 o'clock. For all 16,682 airbag deployments in FARS 2002 (driver or passenger, any injury outcome):
     Thus, 25% of deployments occur in crashes in which the principal impact is not by any criterion even approximately frontal. These include the side and rear crashes shown in Fig. 12-2 plus a number of categories not shown, including principal impact at top or undercarriage, and non-collisions.
The principal impact and initial impact variables in FARS relate to the region of damage on the vehicle (Fig. 3-16, p. 55). The direction of force is not normally known, as it would require a detailed post-crash investigation to determine it. The data above are not materially different if the initial impact point is used instead of the principal impact point. These variables have identical values for over 90% of the vehicles in FARS 2002.
Fatalities when airbags do not deploy
FARS 2002 codes 4,770 drivers of cars or light trucks killed in seats for which airbags were available and which had principal impact point at 12 o'clock. The FARS coding for these 4,770 fatalities is:

    Thus the airbag did not deploy for 719/4,770 = 15% of drivers killed in frontal crashes in seats with airbags. This is a lower bound, because unknown deployment cases will include many non-deployments. Over the wider definition of frontal, clock points 10, 11, 12, 1 and 2, there were 6,357 drivers killed, 4,235 with deployments and 1,139 with non-deployments. A central problem for airbag system design is setting deployment thresholds. Lower thresholds lead to more deployments with the potential that the airbag might produce serious or fatal injuries in minor crashes. As the threshold is increased the airbag becomes unavailable for more crashes in which it has the potential to reduce injury severity. Regardless of what threshold is chosen, it is inevitable that there will be crashes that would have had better outcomes if the threshold had been different.
Airbag benefits, airbag costs
      Estimating the benefits and costs of airbags involves many factors and much complexity. A 1984 NHTSA benefit-cost analysis, which was reevaluated in 2002, included over 50 inputs. More information could invite yet more complexity making it even more difficult to identify the factors of primary importance. For example, a major cost of airbags is the cost of replacing
them after deployment. There is a specific estimated parts replacement cost for vehicles of every make and model year. The labor-cost component for replacement varies throughout the US and between one type of repair business and another. Whether a vehicle is repaired or scrapped depends on these
same factors. It is, for example, estimated that nearly all vehicles more than seven years old are scrapped if they are involved in a crash in which their airbag deploys.
     The benefit-cost comparison presented here is based on my paper that avoided excessive detail by focusing on average values and ignoring minor factors. For example, based on material in the literature,14, it was assumed that half of the airbags that deploy are replaced, and that the same replacement cost applied to all.
Another reason for avoiding detail is that the values of so many key quantities, particularly the effect of airbags on injury risk, are highly uncertain. It serves little purpose to embark on a detailed calculation requiring a complex chain of assumptions to estimate less central quantities that are determined with adequate precision by simple approximate estimates. In the same spirit, we assume that cost estimates published for 2000 apply to July 2003 without fine-tuning to account for inflation. All monetary values are expressed in dollars without regard to whether they are for 2000 or 2003. This makes it easier to retain a clear connection with the original sources, and the cost values given in Chapter 2.
Benefits and costs expressed in dollars
      Airbags are installed for one purpose - injury reduction. (The term injury includes fatal injury unless the context implies otherwise.) Only monetary costs are included in the quantitative comparison of benefits and costs, although there are additional non-monetary costs that will be discussed later. In comparing benefits and costs it is necessary to use a common metric. This is facilitated by the NHTSA study (discussed in Chapter 2) that finds that traffic crashes cost the US $231 billion in 2000. Table 12-1 shows the injury portion of the total costs listed in Table 2-4 (p. 25). Fatal crashes always involve property damage, and crashes without injuries can incur injury costs in connection with, for example, ambulances driving to the crash site and diagnostic tests confirming no injury. All injury harm is converted to a dollar cost. For example, the lifetime economic cost to society for each fatality is estimated at just under a million dollars, over 80 percent of which is attributable to lost workplace and household productivity. As the original authors present reasoned discussion for the difficult decisions that are necessary for such conversions,17 their results are accepted as the basis for estimating benefits of airbags.

Table 12-1. Estimates of the injury component of the total economic costs of motor vehicle crashes in 2000.

Injury-risk changes due to airbags
      The effectiveness of airbags in reducing fatal injury risk was addressed in Chapter 11. Here we use the average value of 10% for drivers, whether belted or unbelted, given in Table 11-4, p. 286. We assume that driver effectiveness estimates apply also to right-front passenger. The literature indicates lower effectiveness for passengers, but this is one of the many details that are not included in the present calculations.
     While fatal injuries conceptually involve only a yes or no determination, non-fatal injuries lie along a severity continuum. Accordingly, effectiveness must relate to some range of injuries, such as is categorized by the Abbreviated Injury Scale (AIS). Effectiveness estimates are sought for specific AIS levels, whereas the cost estimates are given according to MAIS (p. 22). The uncertainty in the effectiveness estimates is sufficiently great that ignoring the distinction between AIS and MAIS will not introduce material additional uncertainty.
     Another difference between injuries and fatalities is that no data file comparable to FARS exists for injuries. This makes determining effectiveness for injuries even more difficult than for fatalities. The best estimates rely on the National Automotive Sampling System Crashworthiness Data System, NASS (Chapter 2).
A 2000 study using 1993-1996 NASS data found that airbag deployment reduced driver fatality risk and risk of the most severe injuries (AIS>4). However, airbag deployment was found to increase the probability that a driver (particularly a woman) sustained AIS 1 to AIS 3 injuries. The results were presented in terms of the delta-v at which airbag deployment produced a net increase or decrease in risk for female and male drivers rather than an effectiveness for different AIS levels. A 2002 study using 1995-2000 NASS data found that front-seat occupants whose air bags deployed had increased risk of AIS>2 injury. These studies were based on cases in which
airbags deployed, which may tend to bias effectiveness estimates downward. We therefore rely on the results of a later study using the same 1995-2000 NASS data to compare outcomes for occupants in vehicles with and without airbags. The results of this study for belted and unbelted occupants are presented in the top plot in Fig. 12-3 in the format of Ref. 15. Even after 10 million deployments, estimates of airbag effectiveness for injuries remain highly uncertain.
Benefits of airbags
      In order to estimate the benefits of airbags we use the effectiveness values in Table 12-2. For AIS = 5 (or MAIS = 5) or we use the same 10% value found for fatalities (typically, about half of AIS = 5 injuries prove fatal). For AIS = 4 we select a value half that for fatalities - somewhat more than midway between the values for AIS = 3 and AIS = 5. The values for AIS 1 toAIS 3 injuries are from Fig. 12-3, and discussed in more detail in Ref. 15.
To complete the estimates of effectiveness we need to estimate the percent
of all road users protected by airbags. Of all 2001 traffic fatalities, 61% were drivers and 15% right-front passengers, percents that remain relatively unchanged from year to year. The previously used data2,3 estimate 64% of all driver seats and 55% of all passenger seats were protected by airbags in July 2003. Thus the driver airbag is protecting 0.61 0.64 = 39% of all road users and the right-front passenger airbag 0.15 0.55 = 8% of all road users.
The conclusion is that the benefits of airbags in calendar year 2003 are:
for drivers $1.60 billion
for right-front passengers $0.34 billion
total 2003 benefits $1.94 billion
Airbags reduce the total number of US fatalities by 0.10 (39% + 8%) = 4.7%. The total annual $1.94 billion benefits of airbags reduce the annual $231 billion cost of traffic crashes by about 1%.
Costs of airbags
     There are a number of costs of airbags in addition to the original purchase cost of the devices. First we address a major annual cost associated with a fleet of vehicles equipped with airbags.
Cost of replacing airbags after deployment. Airbag systems differ from most safety equipment in that after they do what they are supposed to do, the complete system must be replaced. In 1998 NHTSA published the following estimates of the costs of replacing airbags:
Driver $400 to $550.
Passenger $480 to $1,300 without windshield replacement
$1,130 to $3,350 with windshield replacement.
There are many changes since these estimates, especially increases in labor costs, and the increase in dual compared to driver-only systems. In July 2003, I telephoned a number of automobile repair businesses in four states to obtain estimates for replacing airbags in the four highest volume cars sold in the US. The responses varied in precision - in some cases to the nearest cent, while in others nothing more precise than "$2,400-$3,000, depending on the vehicle". Based on the responses, I concluded that a typical cost of replacing a dual system was $2,000 without windshield replacement. Estimates for driver-only replacement did not materially differ from half of the cost for the dual system. Therefore, I will assume a replacement cost of $1000 for any driver airbag, and the same amount for replacing a passenger airbag without windshield replacement. I assume that the windshield is replaced for half of passenger airbag replacements at a cost of $400, so that the following values will be used:
replacing driver airbag $1,000
replacing passenger airbag $1,200
These estimates are based on information for the four highest selling cars, with selling prices less than the average car (or light truck). Replacement parts generally cost more for more expensive vehicles. Airbag replacement costs for luxury cars can be up to $6,000 for dual systems - so actual average replacement costs are likely to be higher than the assumed values.
We assume that after an airbag deploys either it is replaced, or the vehicle is scrapped. In many cases the additional cost of replacing the airbag will lead to the decision to scrap rather than repair. This additional cost due to the presence of the deployed airbag will be ignored. These assumptions lead to the replacement cost estimates in Table 12-3.

Table 12-3. Estimates of airbag replacement costs for 2003.

Table 12-4. Comparison of benefits and replacement costs of airbags for 2003.

Cost of installing airbags. We assume the same cost estimates as used in an earlier study,16 although these were criticized as being too low. These were that the driver-only system cost $278 and the dual system $410. I could find no more specific or current estimates of cost. This amount represents about 2% of the cost of the typical $20,000 vehicle. A breakdown of all costs summing to the purchase cost of the vehicle could be informative. Using the assumed costs (and splitting the dual system cost equally between driver and passenger) leads to the estimates in Table 12-5. Driver airbags cost consumers $30 billion. The total cost to consumers of the airbags on the roads of the US in July 2003 is $54 billion. This exceeds the current Gross Domestic Product of more than half of the member countries of the United Nations (estimated by converting data in Ref. to 2003 dollars).

Table 12-5. Estimates of the purchase cost to consumers of the 257 million airbags on US roads in July 2003.

Total costs of airbags. In order to express the total costs of airbags on an annualized basis, it is necessary to amortize the one-time purchase cost over the expected life of the vehicle. There are complex procedures to do this that involve assumptions about discount rates, etc. In keeping with the simpler structure of the present calculation, we assume that the initial purchase cost converts to an annual cost of 10% of the purchase cost over an assumed ten-year life span of the vehicle. This simple procedure provides a lower annual cost than a more detailed computation. We thus assume that the $30 billion spent to purchase the driver airbags on the roads in July 2003 is equivalent to an annual expenditure of $3 billion. Table 12-6 shows the total annual costs of keeping the airbags on the roads in 2003.

Table 12-6.

Benefit-cost comparison
     Table 12-7 compares the annual costs and annual benefits of airbags on the roads in July 2003. The cost of the driver airbag exceeds the benefit by a factor of two. For the passenger airbag, the cost exceeds the benefit by more than a factor of 8.

Table 12-7. Comparison of estimates of annual costs (Table 12-6) to estimates of annual benefits (Table 12-2).

Comparison with prior estimates. The 1997 study found that airbags were clearly not cost effective for passengers, but might be for drivers.16 The major difference between that study and the one presented here is that the effectiveness estimates used here were not available then. The earlier study used an effectiveness of 11% for fatalities, the best estimate available at the time, and not materially different from the 10% used here. Given that there were no estimates of effectiveness in injury reduction, the earlier authors made the then plausible assumption that effectiveness for injuries was the same as the 11% value for fatalities. Later research (discussed above) shows that airbag effectiveness in reducing injuries is not nearly so high.
      The reason for such instability in benefit versus costs analyses is because effectiveness is so close to zero. Consider the contrast with safety belts. An error of 10 percentage points in the 42% estimated effectiveness would change estimated benefits, but not enough to affect policy materially. However, if an effectiveness close to zero can be determined only to an uncertainty of 10 percentage points, the difference between a +10% effectiveness and a -10% effectiveness has dramatic consequences. An AIS = 1 effectiveness of +8% rather than the -2% used here would have had a massive influence on the calculations, as would an equally likely -12% effectiveness.
Second generation airbags. In response to the many deaths and injuries caused by airbags, new design concepts keep being introduced. After 1998 so called second generation airbags appeared, so that some portion of the airbag fleet in 2003 consisted of such airbags. The effectiveness and cost estimates were all based on earlier first generation airbags.
Design changes include setting higher crash thresholds before deployment. This certainly reduces inflation-caused injuries in low severity crashes, and also reduces replacement costs. However, it also reduces the number of cases in which the airbag provides benefits, especially as airbags already do not deploy in over 15% of cases in which occupants are killed in frontal crashes.
    Another change was reducing deployment forces - so called depowering. Lower power airbags reduce inflation injuries, but also provide less protection. In the limit one can depower an airbag so much that it hurts nobody, but also helps nobody. Depowering very likely reduces the net benefits.
    The changes seem all in the direction of making the airbag less effective, thus decreasing its already low benefit to cost ratio. Preliminary data suggest lower fatality-risk reduction from second generation airbags.6
Replacement costs of passenger airbags can be reduced by suppressing deployment when no passenger is present. However, even perfect technology that suppressed all passenger airbag deployments when no passenger was present, while at the same time never suppressing deployment when one was present, would still leave the benefit to cost ratio for passenger airbags well below that for driver airbags. This is because passenger airbags inherently prevent fewer injuries because of lower passenger-seat occupancy, yet the passenger and driver airbags have similar purchase costs.
It will be many years before we have even the meager knowledge about effectiveness of second generation airbags that we now have for first generation airbags.
Other airbag costs
     William Haddon, a giant in the history of US injury control, discusses the nature of injuries in the broadest terms as the transfer of energy in such ways and amounts and such rapid rates as to harm people. He lists 10 strategies to reduce risks. The first is to prevent the marshalling of the form of energy in the first place. The airbag constitutes a topsy-turvy violation of this principle, by injecting yet more energy into an event in which energy is the source of harm. It is implausible to expect that 1.7 million annual airbag deployments, each an explosive event, will not cause human harm. The additional explosive energy released to inflate the airbag, in common with most sources of energy, produces its own set of injuries. For example, crashes generate much noise, but nothing approaching that produced by an airbag at the ears of an occupant.
Additional injuries caused by airbags
     Beyond inflation-produced blunt trauma injuries, deploying airbags have been associated with many injuries that are unlikely to occur without airbags, including hearing loss, , eye injuries, and asthmatic attacks. , In one case a woman passenger in a vehicle with no passenger airbag suffered ear injuries that had a devastating effect on her quality of life. A driver-side airbag deploying in a low-severity crash caused her injury. She had no crash-related trauma - her only harm was from the airbag. These injuries are, in principle, included in the injury effectiveness estimates of airbags, but some are of a nature that might be missed in the usual processes of AIS coding.
Drivers sitting in ways they would not choose, or looking rearwards at children in rear seats, are actions likely to increase crash risk. Such actions will not change the effectiveness of airbags, as defined in Chapter 11, which is the risk reduction, given the crash, but they will increase harm by increasing the number of crashes. Behavior changes induced by airbags are discussed in more detail below.  Rescue crews must exercise additional care to protect trapped occupants and themselves against the risk that an undeployed airbag might deploy. NHTSA advises:
Although it is rare, an air bag can suddenly deploy during rescue operations, creating a hazardous operating condition, causing further injury, and delaying medical assistance to victims. While every crash poses unique conditions, there are some procedures that will help minimize risks.   Rescue workers are provided with over 1,500 words of instructions - an additional training cost, and further illustration of the non-passive nature of airbags.
    Injuries caused by airbags at center of airbag design dilemmas. Efforts to reduce airbag injuries confront fundamental dilemmas. In order for the airbag to accomplish its primary mission it must deploy from its module at high speed, yet it is this speed that causes harm. Lowering the speed reduces the harm caused, but also the harm it is designed to prevent. Likewise, increasing the threshold crash speed reduces injuries caused, but also injuries prevented. Weight-measuring sensors in seats are under consideration to suppress deployment for at-risk short drivers. These may still allow the airbag to kill short overweight drivers, but suppress deployment for tall slim drivers. Each additional device adds cost, complexity, and requires selecting thresholds. Regardless of what threshold values are chosen, there will be many cases in which a higher value, and many in which a lower value, would have produced a better outcome.
Additional monetary costs
     As airbags increase the purchase cost of vehicles, their owners face higher replacement costs if their vehicle is stolen or destroyed (without crashing). If the owner purchases comprehensive insurance, such potential losses translate into higher premiums. The insurance industry was an enthusiastic supporter of airbags, as might be expected because insuring more expensive items commands higher premiums. In pursuit of their support for airbags, the insurance industry promised that airbags would reduce premiums because of substantially reduced injury costs. The high annual cost of replacing deployed airbags overwhelms any such considerations, and must inevitably generate higher net premiums.
Additional disposal costs are associated with scrapping airbag-equipped vehicles because of the explosive nature of airbags. As with any complex system, there is likely to be some maintenance or inspection cost over the life of the vehicle, as acknowledged by NHTSA4 and prior benefit-cost studies.16 If permission to disconnect is obtained, then a cost to disconnect (in addition to the original purchase cost) is paid. The consumer is obliged to pay twice to get nothing.
Comfort and convenience
      The need to avoid the dangers of deploying airbags has led motorists to do things contrary to their preferences.
Children in rear seats. Placing children in rear seats in situations in which they would otherwise be in front seats inhibits interactions between driver and child that have been traditionally pleasurable and beneficial to both. The child is denied the better view available from the front seat - and may consequently grow up knowing less about driving, with possibly adverse effects on safety. A parent driver with a busy schedule that permits limited time with his or her child is reducing the quality of in-vehicle time if the child is consigned to a rear seat. Absent the airbag, there is no differential safety advantage for a child compared to an adult sitting in a rear seat. Even the most caring individuals would hardly confine their adult companions to rear seats to enhance safety. Passengers travel in front because it is more pleasant, even though it is more dangerous in a crash. One study finds that an adult sitting in a front compared to in a rear seat has a 35% higher risk of being killed, while another study finds the risk higher by 64%. Note how enormous the difference is compared to any risk reductions associated with airbags. No campaigns have been mounted to get adult passengers out of front seats.
    Short drivers and sitting comfort. Short individuals who adjust their seats to positions they would not otherwise choose in order to avoid airbag risks suffer a comfort cost. Pedal extenders or other ancillary devices to compensate for being unable to reach controls are sources of additional discomfort and inconvenience, and for multiple drivers, may need to be removed or installed for each change in driver.
Equity and ethics
      While airbags are estimated to reduce fatality risk to the total population of front-seat car occupants, they do not provide equal protection for all. They provide negligible benefits for drivers age 70.24 What raises larger issues is evidence that they increase risk to large identifiable sectors of the population, even beyond the risk increases they pose to children in front seats.
Gender differences. Taking into account injuries at all levels from fatal to minor, the effects of airbags on net harm to belted drivers were estimated as:

     A major contributor to the difference was that females were substantially more likely to receive AIS ³ 3 upper-extremity injuries. If one assumes that car driver crash rates for males are twice those for females, the effectiveness for the total population would be (2 11.6 - 1 9.2)/3 = 4.7%. Thus, while the device provides an overall benefit, this benefit arises by reducing risks to males while increasing risks to females. The higher injury risk to females is found consistently in other studies,19 while fatality studies report inconsistent effectiveness dependence on gender.24,
Of the 77 drivers NHTSA identified as killed by airbags in low severity crashes, 75% were female. That is, for every male killed, three females were killed. For all drivers of cars and light trucks, FARS shows that for every male driver killed, 0.42 female drivers were killed. Thus females are over represented as fatalities caused by airbag inflation by a factor of 3.0/0.42 = 7.1. Of the female drivers killed, 48% were 62 inches or less (about 20% of females are 62 inches or less). Short females are more than 15 times as likely to be killed by airbags as average drivers. It was unmistakably determined that the airbag was the source of the death because the crashes were of such low severity as to not pose serious injury risk. If these deaths had been caused in an identical manner, but the crashes had been of higher severity, the deaths would have entered FARS in the usual way, and would have been incorrectly attributed to crash trauma. The conclusion is inescapable that many of the fatalities that in fact occur at the lower end of normal fatal crash severities are caused by airbags and not by crash trauma, and that the victims are preferentially short females. The net effectiveness reflects the difference between lives saved, preferentially large males, and lives taken, preferentially small females. Small females are being knowingly killed in order to save large males, a situation that society would hardly tolerate in any context other than airbags.
      Airbags fail medical ethical standards. At the core of medical ethics is the admonition, First, do no harm. The airbag clearly fails this standard. Airbags on the roads are known to place short females at increased risk, yet there is no high priority effort to deactivate them at public expense. Instead, the US government places hurdles in the way of owners who want to pay legitimate businesses to deactivate these devices that they were forced to buy. I believe it is unprecedented in any democracy for a citizen to have to petition government, and be required to make a convincing case, for permission to remove a device known to increase her risk of harm. Even if the petition is successful, which is not guaranteed, she is required to continue to be exposed to the risk of harm while administrative procedures are completed, and while she finds a business willing to disconnect the airbag.
     A medication that kills some patients is likely to be quickly banned. The following three arguments would not be presented to defend it, and if they were, they would be rejected. 1. All the patients taking it are already sick. 2. It saves more patients than it kills. 3. If patients do not want it, they do not have to take it. The airbag exists in a different ethical universe without any convincing reasons why this should be so. It has already killed 231 people, nearly all of them young and healthy. Yet the US government compels the unwilling to purchase it, and keep it in their vehicles.
The airbag has no parallel with vaccinations that are known to cause a few deaths when administered to large populations. The crucial difference is that a vaccination would never be given to any individual if it was known that this would increase the risk of death to that particular individual. Patients are not given drugs if it is known that they will have an allergic reaction to them.
     As for second generation airbags, we simply do not know. Finding out by compelling citizens to act as unwilling guinea pigs in a large scale experiment is outside the realm of anything that could be contemplated for a new untested drug, or modification to an unsuccessful (or even successful) drug.
Fundamental flaw in estimating benefits of airbags
     The term effectiveness is often incorrectly interpreted to represent the change in casualties with and without the device. This is not so, because effectiveness measures the change in risk, given that a crash occurs. If the device affects crash risk, then the change in casualties will differ from the effectiveness. The change in casualties is identical to the effectiveness only under the assumption that the device does not affect driver behavior.
This assumption is demonstrably false for airbags. Airbags generate the clearest overt behavior change of placing children and infants in rear seats. A driver may crash because of distraction from a child in a rear seat. If the driver is injured but not killed, this non-fatality will likely be counted as a fatality prevented in the effectiveness estimates (see also Fig 15-8, p. 399).
     Short drivers are advised to sit further from the steering wheel than they would otherwise choose. Sitting in a less comfortable, more tiring, and less natural driving position flies in the face of conventional advice for safe driving. Sitting further from the steering wheel makes it more difficult to steer and to brake, and likely increases total braking reaction time. When short drivers adjust their seats rearwards, their view of the road becomes more restricted. The combined effects of not sitting in the preferred location may increase crash risk.
When it became clear in 1997 that airbags were killing short ladies, a number of short ladies told me "When I discovered that my airbag could kill me, I started to drive more cautiously." If one accepts this statement, it is hard to dispute the corresponding conclusion that a belief that airbags dramatically reduce risk must lead to less cautious driving. For decades the public was exposed to images suggesting that airbags provided near total protection in crashes. Thousands of slow motion deployments were shown on television, conveying an impression that the occupant moved forward towards a gentle caress by a soft cushion. If knowledge of bullets came only through slow motion pictures, one might conclude that all one had to do to avoid being hit was to step leisurely away from the bullet's path when you observed it approaching you. In the bullet and airbag cases, the slow motion pictures conceal the near instantaneous nature, noise, and violence of the event.
     If beliefs about airbags led to an undetectable 3% increase in average speed, an initial 100 potentially fatal crashes would increase to 113 (based on the 4th power relationship discussed in Chapter 9). The 10% effectiveness of the airbag would prevent 11 of these, leaving 102 fatalities, an increase of 2 over the original 100. Thus, instead of reducing fatalities by 10%, the airbag would increase fatalities by 2%. In Chapter 11 we addressed the possibility that belt-wearing could lead to behavior changes, and mentioned a test-track experiment suggesting that the same drivers increased speed by about 1% when belted compared to when unbelted. Behavior changes associated with airbags are expected to be larger than those associated with belt-wearing, so a 3% effect is plausible. A 3% increase in speed would reduce the fatality reductions from universal belt wearing from 42% to 35%, an important reduction, but not one that would have crucial policy implications. Because of the lower effectiveness of the airbag, a 3% increase in speed turns a fatality decrease into a fatality increase.
      All estimates of lives saved by airbags assume that there are no behavior changes associated with airbags. Logically, such an assumption cannot be exactly true. Empirically, there have been no measured speed increases associated with airbags, and given the difficulties of such measurements, there are unlikely to be. However, the information available makes it very likely that moderately higher risk-taking is associated with the mistaken belief that airbags provide far more protection than they do. Behavior changes smaller than can be observed can cause airbags to increase the number of casualties even though they reduce the risk of injury in a crash. Estimates of lives saved by airbags all assume no behavior changes, and accordingly are more in the nature of logical upper limits rather than best estimates. It seems to me just as likely that airbags have increased fatalities as that they have decreased fatalities.
"Saved by the airbag"
      There are innumerable saved by the airbag reports. The evidence is in many cases simply that there was a severe crash, the airbag deployed, and the occupant survived. There are more than 8,000 cases per year in which there was a severe crash, the airbag deployed, and the occupant died. It is as unreasonable to claim that the dead were killed by the airbag as to claim that the survivors were saved by the airbag. Given the 10% effectiveness of the airbag in reducing fatality risk, it is only after a detailed post-crash examination that one can conclude whether an airbag prevented or caused a death.
What happens to airbag benefits if belt use increases?
     The costs in Table 12-1 were based on what occurred in the US in 2000, a time when about half of fatally injured occupants were unbelted. If the half who were unbelted had been wearing belts, 42% of them would not have died. This is equivalent to a 21% reduction in total driver and front passenger fatalities. If all occupants were belted, then the 9% effectiveness for belted occupants rather than the 10% for all occupants (Table 11-4, p. 286) would apply. The benefit of airbags in reducing driver fatalities would therefore be multiplied by a factor
(1-0.21) (9/10) = 0.71, so the estimated $1.57 billion benefit would decline to $1.12 billion. Additional reductions in benefits at other injury levels are likely to be approximately proportional, suggesting that universal belt use would reduce the benefits of airbags by about 30% of the values estimated.
     Safety belts provide far greater benefits than airbags at minimal cost. A very approximate estimate of the benefits of moving from current to universal belt use can be obtained immediately from Table 12-2. Assume that belt effectiveness is 42% for all injury levels, and that half of all those injured were belted. The transition to universal use can be considered numerically equivalent to adding a new device with an effectiveness of 21% to all cars and light trucks. The result is that achieving 100% belt use would provide benefits of $17.9 billion for drivers of cars and light trucks and $4.4 billion for right-front passengers. The total benefit of moving to universal belt use, $22.3 billion, is more than 11 times the $1.9 billion benefit from airbags in 2003.
     Studies from Transport Canada estimate that during the eleven-year period 1990-2000, belts prevented 11,690 deaths and airbags 313. , Over this period benefits were estimated (in Canadian dollars) at $17.5 billion for belts and less than $0.5 billion for airbags.
Other issues
     While over $60 billion has been paid for airbags (those on the roads plus those already retired), only minuscule resources have been assigned to better determine the benefits and costs associated with them. Even after 10 million deployments, no reliable estimates of how the device affects different levels of injuries have been published in peer-reviewed literature. No ongoing benefit-cost studies are being performed. The simple analysis presented here was supported entirely out of my own pocket. Spending one hundredth of one percent of the cost of airbags on research evaluating their in-use performance could provide more confident answers to many key questions.
The airbag is not worth anything near what it costs. As belt use increases it becomes worth still less. If wiser safety policy leads to fewer crashes, the airbag becomes worth even less.
Even if airbags did not have innumerable problems, including killing occupants in minor crashes, it is still indefensible public policy to compel consumers to purchase items that provide less benefit than they cost. The present US airbag mandate requiring that vehicles be fitted with airbags should be rescinded. Vehicle manufacturers should be permitted to offer them as options, giving consumers freedom of choice. Government's role should be to generate and disseminate reliable information to help consumers make informed choices.

Summary and conclusions (see printed text)

References for Chapter 12 - Numbers in [ ] refer to superscript references in book that do not correctly show in this html version.  To see how they appear in book see pdf version of Chapter 1 or pdf version of Chapter 16.

[1]   Kent RW, editor.  Air Bag Development and Performance – New Perspectives from Industry, Government, and Academia.  SAE Special Publication PT-88.  Warrendale, PA: Society of Automotive Engineers; March 2003.

[2]   Ferguson SA.  An update on the real-world experience of passenger airbags in the United States. Airbag 2000+.  Fourth International Symposium and Exhibition on Sophisticated Car Occupant Safety Systems, 30 November - 2 December, 1998.  Karlsruhe, Germany: Fraunhofer-Institut Fur Chemische Technologie (ICT); 1998.

[3]   Highway Loss Data Institute.  Unpublished data. Estimated number and percent of vehicles in fleet with airbags.  Arlington, VA; August 2003.

[4]   National Highway Traffic Safety Administration.  FMVSS No. 208, Advanced air bags, preliminary economic assessment, Chapter VI, Technology, Costs, and Leadtime.  Washington, DC: Office of Regulatory Analysis & Evaluation, Plans and Policy; August 1998.            

http://www.nhtsa.dot.gov/airbag/PrelimEconAssess/chpt06.html

[5]   National Highway Traffic Safety Administration.  Safety Fact Sheet, 11/02/99.                 

http://www.nhtsa.gov/airbags/factsheets/numbers.html

[6]   Augenstein JS, Digges KH.  Using CIREN data to assess the performance of the second generation of air bags.  SAE paper 2004-01-0842.  Warrendale, PA: Society of Automotive Engineers; 2004.  (Also included in: Air Bags and Belt Restraints.  SAE special publication SP-1876; 2004).

[7]   Federal Register, Vol. 42, No 128, Part 571 – Federal Motor Vehicle Standards: Occupant protection systems, Docket No. 75-14, Notice 10, p. 34289-34305; 5 July 1977

[8]   Patrick LM, Nyquist GW.  Airbag effects on the out-of-position child.  SAE paper 720442.  Warrendale, PA: Society of Automotive Engineers; 1972.

[9]   Aldman B, Anderson A, Saxmark O.  Possible effects of air bag inflation on a standing child.  Proceedings of the 18th Annual Conference of the American Association for Automotive Medicine, Toronto, Canada; 12-14 September 1974.

[10] Counts for air bag related fatalities and seriously injured persons.                                     

http://www-nrd.nhtsa.dot.gov/pdf/nrd-30/NCSA/SCI/2Q_2003/HTML/QtrRpt/ABFSISR.htm

[11] Cases from National Center for Statistics and Analysis Special Crash Investigation Program.  Drivers who sustained fatal or serious injuries in minor or moderate severity air bag deployment crashes. 

http://www-nrd.nhtsa.dot.gov/pdf/nrd-30/ NCSA/SCI/2Q_2003/HTML/SummTab/AdultD.htm

[12] National Highway Traffic Safety Administration.  Final regulatory impact analysis, Amendment of FMVSS 208, passenger car front seat occupant protection.  Washington, DC; 11 July 1984.

[13] Thompson KM, Segui-Gomez M, Weinstein MC, Graham JD.  Validating benefit and cost estimates: The case of airbag regulation. Risk Analysis. 2002; 22: 803-811.

[14] Werner J, Sorenson W. Survey of air bag involved accidents: An analysis of collision characteristics, system effectiveness and injuries.  SAE Paper 940802.  Warrendale, PA: Society of Automotive Engineers; 1994.

[15] Evans L.  Airbag benefits, airbag costs.  SAE paper 2004-01-0840.  Warrendale, PA: Society of Automotive Engineers; 2004.  (Also included in Air Bags and Belt Restraints.  SAE special publication SP-1876, 2004).

[16] Graham JD, Thompson KM, Goldie SJ, Segui-Gomez M, Weinstein MC.  The cost-effectiveness of airbags by seating position.  J Am Medical Assoc. 1997; 278: 1418–1425

[17] Blincoe LJ, Seay AG, Zaloshnja E, Miller TR, Romano EO, Luchter S, Spicer RS.  The economic impact of motor vehicle crashes, 2000.  Report DOT HS 809 446.  Washington, DC: US Department of Transportation, National Highway Traffic Safety Administration,; May 2002.

http://www.nhtsa.dot.gov/people/economic/EconImpact2000/

[18] Association for the Advancement of Automotive Medicine.  The abbreviated injury scale. AAAM; 1990.

[19] Segui-Gomez M.  Driver air bag effectiveness by severity of the crash.  Am J Pub Health. 2000; 90: 1575–1581

[20] McGwin G Jr, Metzger J, Jorge E, Alonso JE, Rue LW III.  The association between occupant restraint systems and risk of injury in frontal motor vehicle collisions.  J Trauma. 2003; 54: 1182–1187

[21] McGwin G Jr.  Airbags and the risk of injury in frontal motor vehicle crashes.  2003.  Submitted for publication.

[22] Larkin GL, Weber JE.  Cost effectiveness of air bags in motor vehicles (letter).  J Am Medical Assoc. 1998; 279: 506.

[23] United Nations, Department of Economic and Social Affairs, Statistics Division.  GDP of 167 countries in current international dollars. http://unstats.un.org/unsd/cdbdemo/cdb_years_on_top.asp?srID=29923&crID=&yrID=1990

[24] Kahane CJ.  Fatality reduction by air bags: Analysis of accident data through early 1996.  Report DOT HS 808 470.  Washington, DC: National Highway Traffic Safety Administration; 1996.

[25] Haddon W Jr.  On the escape of tigers: An ecologic note.  Am J Public Health. 1970; 60: 2229-2234.    http://www.siu.edu/~ritzel/courses/313s/tiger/tigerarticle.htm

[26] Yaremchuk K, Dobie R.  The otologic effects of airbag deployment.  J Occupational Hearing Loss. 1999; 2: 67-73.

[27] Buckley G, Setchfield N, Framption R.  Two case reports of possible noise trauma after inflation of air bags in low speed car crashes.  Brit Med J. 1999;318, 499-500.

[28] Duma SM, Kress TA, Porta DJ, Woods CD, Snider JN, Fuller PM, Simmons RJ. Air bag induced eye injuries: A report of 25 cases. J Trauma. 1996; 41: 114-119.

[29] Gross KB, Kelly NA, Reddy S, Shah NJ, Grain TAK. Assessment of human responses to non-azide air bag effluents. Proceedings of the 43rd Stapp Car Crash Conference, SAE paper 99SC26. Warrendale, PA: Society of Automotive Engineers; 1999.

[30] Gross KB, Haidar AH, Basha MA, Chan TL, Gwizdala CJ, Wooley RG, Popovich J.  Acute pulmonary response of asthmatics to aerosols and gases generated by airbag deployment.  Am J Respir Crit Care Med. 1994; 150: 408-414.

[31] Saving lives, wrecking ears.  U.S. News and World Report; 26April 1999, p.72.

[32] National Highway Traffic Safety Administration.  Rescue procedures for airbag-equipped vehicles.  Campaign Safe & Sober.                                              

http://www.nhtsa.dot.gov/people/outreach/safesobr/16qp/procedures.html

[33] Evans L, Frick MC.  Seating position in cars and fatality risk.  Am J Pub Health. 1988: 78; 1456-1458.

[34] Smith KM, Cummings P.  Passenger seating position and the risk of passenger death or injury in traffic crashes.  Accid Anal Prev. 2004; 36: 257–260

[35] Dalmotas DJ, Hurley J, German A, Digges K.  Air bag deployment crashes in Canada.  Paper 96-S1O-05.  15th Enhanced Safety of Vehicles Conference, Melbourne, Australia; 13-17 May 1996.

[36] Cummings P, McKnight B, Rivara FP, Grossman DC.  Association of driver air bags with driver fatality: A matched cohort study.  Brit Med J. 2002; 324: 1119–1122

[37] Janssen W.  Seat-belt wearing and driver behavior: An instrumented-vehicle study.  Accid Anal Prev. 1994; 26: 249-261.

[38] Transport Canada.  Estimates and lives saved among front seat occupants of light-duty vehicles involved in collisions attributable to the use of seat belts and air bags in Canada.  Road Safety and Motor Vehicle Regulation, Fact Sheet RS 2001-03 E TP13187E; October 2001.                  

http://www.tc.gc.ca/roadsafety/menu.htm

[39] Stewart DE, Arora HR, Dalmotas D.  An evaluation of the effectiveness of supplementary restraint systems (“air bags”) and conventional seat belts: Estimates of the numbers of lives saved among front seat outboard occupants of light-duty vehicles involved in collisions attributable to the use of seat belts and the fitment of supplementary restraint systems (“air bags”) in Canada, 1990-1997.  Transport Canada Publication No. TP13187 E.  Ottawa, Ontario; 1998.