Benefits of Australian Design Rule 69 (full frontal crash protection) and airbags in frontal crashes in Australia

Abstract

In-depth data at MUARC was used to evaluate the Australian Design Rule 69 (ADR69) - Full frontal dynamic crash requirement, as well as the effectiveness of frontal airbag deployment on injury risk and associated cost of injury. ADR69 was introduced in Australia in mid-1995 and was based largely on the US equivalent FMVSS-208. The results indicate reductions in excess of 90% in the likelihood of sustaining AIS 2+ injuries in body regions where frontal airbags would be expected to benefit. The average injury cost savings for drivers of post-ADR69 manufactured vehicles was found to be up to AUD$19,000 depending on body region considered. Limitations and implications of these findings are discussed.

Research conducted by the Monash University Accident Research Centre (MUARC) in the early 1990’s identified a high frequency of head and chest injuries associated with contact to the steering wheel, instrument panel, and the seat belt sustained by drivers and front passengers in the event of frontal crashes [Fildes, Lane, Lenard & Vulcan, 1991]. Following a comprehensive research program [see MUARC 1992; Seyer, 1992, 1993 as examples] conducted by the Department of Transport and Regional Services (DoTARS), Australian Design Rule 69 (ADR69) was introduced in a 4-year phased-in program from 1 July 1995.

ADR69 was based largely upon US Federal Motor Vehicle Standard FMVSS208 with the specification that Hybrid III test dummies be restrained with a seatbelt. This modification was based on the extremely high (90%+) seat belt wearing rates among front seat occupants in Australia [ARUP, 1995]. The purpose of the ‘ADR69 Full Frontal Impact Occupant Protection’ requirement was to “…specify vehicle crashworthiness requirements in terms of forces and accelerations measured on anthropomorphic dummies in outboard front seating positions in full frontal test crashes so as to minimise the likelihood of injury to occupants of those seating positions” [ATSB, 2003]. Manufacturers were required to meet the performance standards specified by ADR69 as shown in Table 1. A 5-second seat-belt warning device was also mandated.

Table 1.

ADR 69 full frontal impact occupant protection criteria

Test parameter	Performance criteria
Head	HIC not to exceed 1000 over 36ms
Sternum	Compression not to exceed 76.2mm
Thorax	Chest deceleration not to exceed 60g
Femur	Axial force not to exceed 10kN
Barrier	To conform to SAE document J850
Speed	48.3km/h (30mph)

Although there were a number of ways in which manufacturers could meet the dynamic full frontal crash protection standard, many introduced frontal airbags since earlier studies [e.g., MUARC 1992] anticipated that airbags would provide much of the injury reduction benefit associated with ADR69. Since ADR69 was mandated, frontal airbags in Australia have been progressively introduced as standard equipment for the driver position for many models, and more recently introduced to the front outboard passenger position for some vehicle models [Morris et al., 2001].

The effect of frontal airbag deployment on injury risk has been extensively studied worldwide since the introduction of airbags in vehicles in the 1980’s. Studies using real-world crash data conducted in Europe and Australia have shown that airbags are largely beneficial in reducing injury risk and severity, particularly of the head and face, while reporting an increase in the incidence of typically low severity upper extremity injuries [refer Morris et al., 2001].

In examining airbag effectiveness rates globally, it is important to note the differences in airbag design in the US compared to Europe and Australia. In the US, airbags are mandated as “primary restraint systems” where they must provide protection for unrestrained occupants. In contrast, airbags in Australia and Europe are designed to perform as “supplementary restraint systems” and to be used in conjunction with seat belts. Despite differences in design purpose between European and Australian airbags, and US airbags, it is accepted that airbags provide optimal occupant protection when used in conjunction with a seat belt [see Morris et al., 2001]. For a full discussion of the theory behind restraint systems, general airbag development and modern airbag design, as well as results of field studies of airbag effectiveness, the reader is referred to Morris et al. (2001).

MUARC was commissioned by the ATSB to conduct a case-control study during 1995 to evaluate the impact of ADR69 on passenger car safety. Research examining the effectiveness of ADR69 was reported recently for the ATSB [Morris et al., 2001]. Briefly, the report found reductions in head, face, neck and chest injuries among occupants of airbag-deployed vehicles. In addition, using the HARM method, the mean HARM per driver was 60% higher in non-airbag vehicles compared to airbag equipped and deployed vehicles. While the findings of the report were extremely positive, the level of statistical control exercised to account for differences in crash and occupant characteristics between the airbag and non-airbag group was limited.

This study aims to re-evaluate the effectiveness of ADR69 and frontal airbag deployment on injury risk, severity and cost associated with belted drivers involved in frontal crashes in Australia using more sophisticated analysis techniques than used previously. This paper uses drivers involved in frontal crashes and studied in-depth by researchers of MUARC for the period 1989 – 2002, with the year of manufacture ranging from 1986 – 2000. Using advanced regression techniques, this paper will report differences in the likelihood of injury and injury cost benefits for each body region, adjusted for influential differences in driver and crash characteristics between the groups of interest.

METHOD

IN-DEPTH DATABASES

This paper draws on in-depth data of three crashed vehicle studies conducted by MUARC: the Crashed Vehicle File (CVF) conducted from 1989 – 1993; the study funded by FORS (now Australian Transport Safety Bureau, ATSB) to evaluate ADR69 and conducted from 1995 – 2000, and the current Australian National Crash In-Depth Study (ANCIS)1 from 2000 onwards.

The CVF and ANCIS projects enrolled patients admitted to hospital as a consequence of a traffic crash. The criterion for inclusion to the FORS study was involvement in a tow-away crash with the Researchers receiving voluntary notifications from registered tow-truck operators with a nominal spotters fee being paid. The value of combining the three datasets is that tow-away and hospitalised crashes are combined and analysed to fully evaluate ADR69 and airbag effectiveness in reducing injuries across a broad spectrum of crash severities. It was possible to combine the separate databases due to the use of common core data points, as well as the inclusion of EBS to control for crash severity (and other variables) in regression models. Sample weights designed to provide a fully representative view of injuries across the fleet were unavailable. A brief description of the collection procedures is presented below.

ETHICS APPROVALS

Ethical considerations demanded that the ‘case’ only proceed if the crash-involved occupant and the owner of occupants of the vehicle (if different) consented to participate in the study. The Monash University Standing Committee on Ethics involving Research on Humans (SCERH), as well as Institutional Ethics Committees at the study hospitals, approved the conduct of each in-depth database.

VEHICLE INSPECTION PROCEDURES

Vehicle inspections were conducted in accordance with standard international practice [National Automotive Sampling System-NHTSA, 1989]. Vehicle damage was coded as per ‘SAE Recommended Practice J224b’. Delta-V and EBS were determined using Calspan Reproduction of Accident Speeds on the Highway Version 3 (CRASH3).

INJURY DATA

Injury data were gathered on each occupant involved in the collision. For hospitalized occupants, injury details were recorded from medical records of the treating hospital. Participants were also administered a structured interview by a Research Nurse. For persons killed in the crash, injury details were obtained from coronial records. For occupants not requiring hospital treatment, a Research Nurse conducted a structured telephone interview to gather crash and injury details. All injuries, whether self-reported or medically verified, were coded according to the Abbreviated Injury Scale (AIS), 1990 revision [AAAM, 1998]. The Injury Severity Score (ISS) was calculated for each case, and acts as a global index of injury severity [AAAM, 1998]. The ISS is derived from the AIS and ranges from 0 (uninjured) – 75 (unsurvivable).

THE CONCEPT OF HARM

HARM is a metric for quantifying societal injury costs from road trauma and involves a frequency and a unit cost component. The HARM metric has been used in a number of studies at MUARC as a means of estimating societal benefits from the introduction of new countermeasures [MUARC, 1992; Fildes, Fitzharris, Koppel et al., 2002] as well as a means for quantifying the financial benefits to society in evaluation studies [Fildes, Deery, Lenard, et al., 1996]. HARM benefits associated with drivers of pre- and post-ADR69 manufactured vehicles are presented in the RESULTS.

In its most general form, HARM is used as a measure of the total cost of road trauma. Injury costs by body region and injury severity were reported earlier and were determined using the human capital method [MUARC, 1992]. Included within the HARM estimates are treatment, rehabilitation, loss of productivity and wages, pain and suffering allowances and administration costs. The HARM values were originally based on total societal crash costs originally published by Steadman and Bryan (1988). For the purpose of this paper, HARM values have been re-factored (by 2.5) to reflect more recent estimates of road crash costs published by the BTE (2000) that were 2.5 times higher than those estimated by Steadman and Bryan (1988). The proportional differences do however remain the same. Table 2 shows the cost of injury by body region and AIS severity.

Table 2.

Average cost of injury (HARM) by body region and AIS severity (AUD$‘000)

	AIS INJURY SEVERITY
Body Region	Minor	Moderate	Serious	Severe	Critical
	AIS 1	AIS 2	AIS3	AIS 4	AIS 50
Head	5.25	24.50	100.75	232.25	820.50
Face	5.25	24.50	100.75	133.00	272.25
Neck	5.25	24.50	100.75	133.00	272.25
Chest	3.75	20.75	58.00	94.25	136.75
Abdomen-Pelvis	3.75	20.75	58.00	94.25	136.75
Spine	3.75	20.75	135.50	1167.50	1396.00
Upper Ex	5.25	36.00	85.25	N/A	N/A
Lower Ex	3.75	36.00	108.25	160.00	272.25

Note: AIS6 – maximum for all regions, $830.75 (‘000)

INCLUSION CRITERIA

The inclusion criteria for this study include: driver only; passenger vehicles (small, medium, large); frontal impact defined by CDC; EBS (km/h) being known; no rollover in the collision; collision partner known; vehicles manufactured pre-1995 (referred to as ‘pre-ADR69’, 1986–1994) and post-1995 (referred to as ‘post-ADR69’, 1996–2000) with vehicles manufactured in 1995 excluded from the analysis due to ADR69 being introduced mid-1995; airbag status known; at least one AIS1 injury sustained with all injury data known; seat belt worn, plus age, gender, height and weight of the driver were known. Fatalities were excluded from the analysis due to very few being present in the databases.

STATISTICAL ANALYSIS

Descriptive analysis of vehicle and driver characteristics is presented, split on drivers of pre-ADR69 and post-ADR69 vehicles. Analysis of vehicle and driver characteristics was undertaken to explore potential differences between the two groups and to describe the sample. To test for differences in sample group means, t-tests and 2-way ANOVA’s were used [Keppel, 1991], while chi-square tests were used to test for differences in distributions between the two groups [Siegel & Castellan, 1988].

Logistic regression was used to determine the relative odds, or likelihood, of drivers sustaining an AIS 2+ injury for each body region between drivers of pre- and post-ADR69 manufactured vehicles [Hosmer & Lemeshow, 2000]. Regression modelling allows for the statistical control of differences in crash and occupant characteristics associated with drivers across the ADR69 groups, while also examining the effect of airbag fitment and deployment and the ADR69 regulation itself. The combined impact of the ADR69 regulation and airbag deployment for AIS 2+ injury risk was also determined. Potential confounders in the analysis were assessed and adjusted for if required, and included differences in EBS; collision partner; age; gender; driver weight, and driver height. The interaction between ADR69 group and airbag fitment was examined to determine whether the injury reduction benefits of airbag deployment differed across the ADR69 groups.

The difference in average harm between drivers of pre- and post-ADR69 manufactured vehicles was estimated using a Poisson regression model compensating for variance over-dispersion [Tabachnick & Fidell, 2000]. A Poisson regression model was considered appropriate as the HARM measure is essentially cost weighted injury counts. The Poisson error structure of the regression model is appropriate to reflect the count nature of the data, while the cost weighting leads to the variance over-dispersion for which the model also accounts. To control for differences in average crash severity between the two groups, EBS was included as a covariate in the regression model, as was vehicle market class. It was not possible to separate the effect of ADR69 group and airbag deployment in the Poisson cost model, and therefore the cost effectiveness of post-ADR69 compared to pre-ADR69 vehicles is presented. Consequently, the cost of injury for the analysis contains an unspecified airbag effect.

Analysis was conducted using Stata Intercooled V8.2, and SAS V.8. A p-value of <=0.05 was used to assess statistical significance. Institutional ethics committees approved the data collection activities. This paper used de-identified data and was therefore exempt from ethics committee approval.

RESULTS

VEHICLE AND CRASH CHARACTERISTICS

A total of 285 cases in the dataset satisfied the study entry criteria. Table 3 shows that of the 285 drivers, 129 (45.3%) were drivers of post-ADR69 vehicles (1996–2000), while the remaining 156 (54.7%) were drivers of pre-ADR69 vehicles (1986–1994). Of drivers in post-ADR69 manufactured vehicles, 82.2% (106) had a frontal airbag system deploy, compared to 30.7% (48) in the pre-ADR69 group, χ²(1)=75.1, p<0.001. The correlation between ADR69 group and airbag system status was 0.51 (p<0.001)

Table 3.

Number & percent of drivers by ADR69 and airbag status

Airbag status	Pre-ADR69		Post-ADR69
	Freq.	%	Freq.	%
Fitted & deployed	48	30.7	106	82.2
Not fitted, or Fitted& not deployed	108	69.3	23	17.8
TOTAL	156	100.0	129	100.0

The majority of drivers in both ADR69 groups were occupants of large passenger vehicles (Table 4). There were significantly more drivers in large vehicles and fewer in small vehicles among the pre-ADR69 group compared to the post-ADR69 group, χ²(2)=7.5, p=0.02.

Table 4.

Number of drivers by ADR69 status & vehicle group

Vehicle market class	Pre-ADR69		Post-ADR69
	Freq.	%	Freq.	%
Small	21	13.50	32	24.8
Medium	9	5.80	11	8.5
Large	126	80.70	86	66.7
TOTAL	156	100.0	129	100.0

Cars and passenger car derivatives represented approximately 57% of collision partners, while poles or trees represented approximately 28% of collision partners (Table 5). There were a smaller number of other collision partners, and there was no difference in the distribution of collision partners between the ADR69 groups, χ²(4)=4.1, p=0.4.

Table 5.

Number of drivers by collision partner & ADR69 group

Collision partner	Pre-ADR69		Post-ADR69
	Freq.	%	Freq.	%
Car / Ute	88	56.4	74	57.4
SUV, van	10	6.4	7	5.4
Pole / tree	40	25.6	40	31.0
Truck / bus	13	8.3	4	3.1
Roadside object	5	3.2	4	3.1
Total	156	100.0	129	100.0

CRASH SEVERITY

The crash severity, indexed as EBS (km/h), differed between the pre- and post ADR69 groups (refer Figure 1). The mean EBS for the pre-ADR69 drivers (41.2 km/h, SD=14.9 km/h) was significantly higher than for drivers of post-ADR69 vehicles (34.9km/h, SD=17.3km/h), t(283)=4.8, p< 0.001. The EBS distribution (see Figure 1) between the two groups was also statistically different, χ²(5)=39.3, p<0.001. The median EBS for the pre-ADR69 cases was 44.1 km/h while for drivers of the post-ADR69 vehicles the median EBS was 31.3 km/h. As the distribution of crash severity differs between the ADR69 groups, EBS must be included in all regression models in order to account for this difference.

Figure 1.

Number of drivers by EBS category and ADR69 group

DRIVER CHARACTERISTICS

Of the 285 drivers, 66% were male (188) and 34% were female (97). Table 6 indicates that there was no difference in the distribution of male and female drivers across the two samples, χ²(1)=0.2, p=0.6.

Table 6.

Number of drivers by gender by ADR69 status

Gender	Pre-ADR69		Post-ADR69
	Freq.	%	Freq.	%
Male	101	64.7	87	67.4
Female	55	35.3	42	32.6
Total	156	100.0	129	100.0

The overall mean age of pre-ADR69 drivers (40.2 yrs, SD=14.6; 39 yrs) and post-ADR69 drivers (38.4 yrs, SD=14.4 yrs, Median: 36 yrs) did not differ (p>0.05). The median values were also similar to the mean indicating a relatively normal distribution. Furthermore, there was no difference in the mean age of males and females either within or between the two ADR69 groups (p>0.05). Table 7 gives the age distribution indicating no difference between the two groups, χ²(5)=2.3, p=0.8.

Table 7.

Number of drivers by age category and ADR69 group

Age category	Pre-ADR69		Post-ADR69
	Freq.	%	Freq.	%
17 – 24	25	16.0	22	17.1
25 – 34	38	24.4	37	28.7
35 – 44	36	23.1	31	24.0
45 – 54	31	19.9	25	19.4
55 – 64	14	8.9	7	5.4
65 – 90	12	7.7	7	5.4
Total	156	100.0	129	100.0

The overall mean height for males was 178cm (SD=6.1, Median: 178; Range 155–196) and for females was 164.5cm (SD=7.3, Median: 165; Range 149–180). A two-way ANOVA indicated that while the mean height of males and females was statistically different [F(1,281)= 236.4, p<0.001], there was no differential height difference for males and females between the two ADR69 groups. The median height values are also similar to the mean indicating relatively normal distribution. Analysis of driver height using 5cm intervals indicated that the height distributions of the pre-ADR69 drivers and post-ADR69 drivers was evenly matched, χ²(5)=4.7, p=0.4.

The overall mean weight for males (82.4kg (SD=13.9, Median: 81kg; Range 50–175kg) was greater than for females (66.2kg, SD=13.3, Median: 64; Range 35–102), [F(1,281)= 89.2, p<0.001]. A 2-way ANOVA again indicated no differential weight difference for males and females between the two ADR69 groups. The median weight values are also similar to the mean indicating a relatively normal weight distribution. Analysis of driver weight using 10kg intervals indicated that the weight distribution of the pre-ADR69 drivers and post-ADR69 drivers was evenly matched (χ²(5)=7.5, p=0.1).

INJURY OUTCOMES

The overall injury severity as indexed by the injury severity score (ISS) of the pre-ADR69 drivers and post-ADR69 drivers did not differ, t(283)=0.8, p=0.4. The mean ISS for drivers of pre-ADR69 vehicles was 6.0 (SD=6.7, Range: 1–34, Median: 4) compared to 5.3 (SD=7.5, Range: 1–41, Median: 2) for drivers of post-ADR69 vehicles. Analysis indicated no difference in the proportion of drivers’ injuries being classified as minor or major trauma (ISS<15) between the ADR69 groups (ISS>15), χ²(1)=0.1, p=0.9. Approximately 9% of drivers in each group sustained an ISS of greater than 15.

The distribution of Maximum Abbreviated Injury Scale score (MAIS) for drivers of pre- and post ADR69 vehicles differed significantly in Table 8, χ²(4)=11.9, p<0.01. A higher proportion of drivers of post-ADR69 vehicles sustained AIS 1 (minor) injuries and a correspondingly lower proportion of AIS2+ injuries compared to drivers of pre-ADR69 vehicles. For injuries of AIS 3 and higher, there was no difference between the groups.

Table 8.

MAIS distribution for pre-& post-ADR69 drivers

MAIS	Pre-ADR69		Post-ADR69
	Freq.	%	Freq.	%
Minor (1)	71	45.5	81	62.8
Moderate (2)	55	35.3	23	17.8
Serious (3)	25	16.0	20	15.5
Severe (4)	4	2.5	4	3.1
Critical (5)	1	0.7	1	0.8
TOTAL	156	100.0	129	100.0

INJURY RISK ANALYSIS

Table 9 shows the percent of drivers who sustained an AIS 2+ injury by body region. Also presented are adjusted relative odds ratios of drivers of pre- & post-ADR69 vehicles, the impact of airbag deployment and their combined effect, adjusted for covariates. Adjusted relative odds ratio estimates are presented for three comparisons:

Table 9.

Percent of drivers sustaining AIS 2+ injuries per body region and adjusted relative odds ratios of pre- & post-ADR69 vehicles, airbag deployments, and the combined effect.

Body region	Pre-ADR69 (n=156)	Post-ADR69 (n=129)	95% CL (L-U)	P
Head	23.1%	3.9%
Post-ADR69	OR: 0.20		0.05–0.76	0.02
Airbag deployment	OR: 0.15		0.04–0.52	0.003
ADR69+Airbag	OR: 0.03		0.01–0.15	<0.001
Face	11.5%	1.5%
Post-ADR69	OR: 0.12		0.02–0.83	0.03
Airbag deployment	OR: 0.09		0.01–0.91	0.04
ADR69+Airbag	OR: 0.01		0.001–0.18	0.001
Neck	4.5%	1.5%
Post-ADR69	OR: 0.44		0.07–2.77	0.4
Airbag deployment	OR: 0.46		0.07–3.03	0.4
ADR69+Airbag	OR: 0.20		0.03–1.47	0.1
Chest	26.9%	16.3%
Post-ADR69	OR: 0.98		0.41–2.33	0.9
Airbag deployment	OR: 0.23		0.09–0.56	0.001
ADR69+Airbag	OR: 0.22		0.09–0.53	0.001
Abdomen / Pelvis	4.5%	6.2%
Post-ADR69	OR: 2.58		0.67–9.80	0.1
Airbag deployment	OR: 0.27		0.06–1.31	0.1
ADR69+Airbag	OR: 0.71		0.17–2.97	0.6
Spine	2.5%	6.9%
Post-ADR69	OR: N/A
Airbag deployment	OR: N/A
ADR69+Airbag	OR: 2.34
Upper Extremity	14.7%	13.9%
Post-ADR69	OR: 0.92		0.39–2.21	0.8
Airbag deployment	OR: 0.96		0.40–2.33	0.9
ADR69+Airbag	OR: 0.89		0.37–2.18	0.8
Lower Extremity	19.9%	17.0%
Post-ADR69	OR: 1.50		0.64–3.51	0.3
Airbag deployment	OR: 0.60		0.25–1.39	0.2
ADR69+Airbag	OR: 0.90		0.39–2.10	0.8

1. For drivers of post-ADR69 vehicles relative to drivers of pre-ADR69 vehicles, irrespective of airbag deployment status (referred to as ‘ADR69’ in Table 9);

2. For drivers of airbag deployed vehicles relative to drivers of vehicles without an airbag deployment, irrespective of ADR69 status (referred to as ‘Airbag’ in Table 9), and

3. For drivers of ADR69 compliant vehicles with an airbag deployment relative to drivers of pre-ADR69 manufactured vehicles without an airbag deployment (referred to as ‘ADR69+Airbag’ in Table 9), and is presented to highlight their combined impact on injury risk.

Critically, in interpreting the ‘ADR69’ result (Comparison 1) the estimate is the relative odds, or likelihood, of injury irrespective of frontal airbag deployment status, and as such includes post-ADR69 drivers with and without frontal airbag deployment. For the ADR69 comparison, inclusion of the airbag variable in the logistic regression model simply adjusts for differences, if any, in airbag effectiveness between the pre- and post-ADR69 groups, but does not remove the impact of the airbag on injury risk. Consequently, the ADR69 result contains an unspecified impact of airbag deployment on injury risk. Similarly, in interpreting the ‘airbag’ result (Comparison 2), the odds ratio is the relative odds or likelihood of injury irrespective of ADR69 status, and consequently contains an unspecified impact of the ADR69 regulation within the airbag estimate itself.

The inability to separate the individual influence of the ADR69 regulation and frontal airbags separately has arisen because drivers within both pre- and post-ADR69 groups experienced frontal airbag deployments, and also to a lesser and unknown extent that some vehicles in the pre-ADR69 sample would likely meet the ADR69 design regulation. In more general terms, the introduction of ADR69, while not mandating the fitment of frontal airbag systems, provided the impetus for many manufacturers to do so, hence the difficulty in isolating the impact of the impact of the ADR69 regulation from the impact of the airbag itself. Comparison 3, ‘ADR69 + Airbag’ is the combined benefit compared to pre-ADR69 vehicles without an airbag, and represents the improvement in vehicle safety through advances in design technology due to ADR69 changes and airbag fitment.

The relative odds ratios describe the difference in odds, or likelihood, of sustaining the specified injury between drivers of post-ADR69 vehicles relative to drivers of pre-ADR69 manufactured vehicles for the specified AIS 2+ body region. A p-value of <0.05 indicates a statistically reliable reduction (<1) or increase (>1) in the odds of injury for each of the three specified comparisons. Each model has been adjusted for influential covariates either related to the injury of interest, or to adjust for differences between the comparison groups, e.g., EBS. All models were adjusted for EBS, vehicle market class and collision partner. Other covariates are noted within each body region. For ease of presentation, each body region will be considered individually.

AIS 2+ HEAD INJURY RISK

Approximately 23% of drivers of pre-ADR69 vehicles sustained an AIS 2+ head injury, compared to 4% of drivers of post-ADR69 (refer Table 9). The relative odds ratio associated with drivers of post-ADR69 vehicles relative to drivers of pre-ADR69 vehicles was 0.20 (80% reduction), with 95% confidence intervals stating the relative odds, or likelihood of AIS 2+ head injury might be as low as 0.05 to as high as 0.76 (0.05–0.76, p=0.02). Similarly, the relative odds of sustaining AIS 2+ head injuries associated with a frontal airbag deployment compared to drivers of vehicles without an airbag deployment was 0.15 (0.04–0.52, p=0.003, 85% reduction). The relative odds ratio for drivers of post-ADR69 vehicles with an airbag deployment was 0.03 that of drivers of pre-ADR69 vehicles without an airbag deployment (0.01–0.15, p<0.001, 97% reduction). These results suggest that the frontal airbag is the single most effective change impacting upon head injury risk.

AIS 2+ FACE INJURY RISK

A higher proportion of drivers of pre-ADR69 vehicles (11.5%) sustained an AIS 2+ face injury compared to post-ADR69 drivers (1.5%). The relative odds ratio of drivers of post-ADR69 vehicles sustaining an AIS 2+ face injury relative to that of drivers of pre-ADR69 vehicles was 0.12 (0.02–0.83, p=0.03). The relative odds of sustaining AIS 2+ facial injury associated with a frontal airbag deployment was 0.09 (0.01–0.91, p=0.04) compared to no deployment, while the relative odds of sustaining AIS 2+ face injuries for drivers of post-ADR69 vehicles with an airbag deployment relative to drivers of pre-ADR69 vehicles without a frontal airbag system deployment was 0.01 (0.001–0.18),

AIS 2+ NECK INJURY RISK

4.5% of drivers of pre-ADR69 vehicles compared to 1.5% of post-ADR69 drivers sustained an AIS 2+ neck injury. Although not statistically significant, the relative odds of injury associated with drivers of post-ADR69 vehicles was 56% lower than for drivers of pre-ADR69 manufactured vehicles (OR: 0.44, 0.07–2.77, p=0.4), while a 54% reduction in the likelihood of injury associated with airbag deployment was observed (OR: 0.46, 0.07–3.03, p=0.4). The relative odds ratio for drivers of post-ADR69 vehicles with an airbag deployment compared to drivers of pre-ADR69 vehicles without an airbag deployment was 0.20, representing an 80% reduction in the likelihood of sustaining AIS 2+ neck injuries (0.03–1.47, p=0.1). It is probable that the failure to detect statistically reliable reduction is due to low sample numbers.

AIS 2+ CHEST INJURY RISK

27% of drivers of pre-ADR69 vehicles compared to 16% of post-ADR69 drivers sustained an AIS 2+ chest injury. While the ADR69 estimate is not significant (0.98, 0.41–2.33, p=0.98), the effect of a frontal airbag deployment is a 77% reduction in the likelihood of sustaining an AIS 2+ chest injury versus non-deployment (OR: 0.23, 0.09–0.56, p=0.001), and clearly drives the benefits associated with post-ADR69 vehicles with an airbag deployment compared to drivers of pre-ADR vehicles without an airbag deployment (OR: 0.22, 0.09–0.53, p = 0.001). Notably, the risk of chest injury was influenced, and therefore adjusted for, driver age, gender and height.

AIS 2+ ABDOMEN / PELVIS INJURY RISK

4.5% of drivers of pre-ADR69 vehicles sustained an AIS 2 (moderate) or higher injury of the abdomen and pelvis, compared to 6.2% of drivers of post-ADR69 drivers. Drivers of post-ADR69 manufactured vehicles were 2.6 times more likely to sustain this injury type than drivers of pre-ADR69 vehicles, although this difference was not statistically significant (0.67–9.80, p=0.1). Conversely, the relative odds of injury associated with a frontal airbag deployment were 0.27 (0.06–1.31, p=0.1). The combined benefit of ADR69 vehicles plus airbag system deployment is indicated by the relative odds of injury being 29% lower (OR: 0.71,0.17–2.97, p=0.6) than for drivers of pre-ADR69 vehicles without a frontal airbag system being activated. The latter two findings, though not statistically significant, may be indicative of benefits attributable to airbag system deployment, including seat belt technology (for instance, pretensioners and load limiters), in reducing the injury risk to the abdomen and pelvis.

AIS 2+ SPINE INJURY RISK

2.5% of drivers of pre-ADR69 vehicles sustained an AIS 2 (moderate) or higher injury of the spine, compared to 6.9% of drivers of ADR69 compliant vehicles, indicating an increase in injury risk. In the regression analysis the presence of an interaction between ADR69 group and airbag deployment status means that the main effects of ADR69 and airbag system deployment cannot be unambiguously interpreted. The interaction model indicates that drivers of post-ADR69 vehicles with an airbag deployment showed a trend toward having an increased likelihood of sustaining spine AIS 2+ spine injuries injury than drivers of pre-ADR69 vehicles without an airbag deployment (OR: 2.34, 0.32–16.90, p=0.4).

AIS 2+ UPPER EXTREMITY INJURY RISK

14.7% of drivers of pre-ADR69 vehicles sustained an AIS 2 (moderate) or higher injury of the upper extremity, compared to 13.9% of drivers of post-ADR69 vehicles. The relative odds ratio associated with post-ADR69 vehicles was 0.92 (0.39–2.21, p=0.8; 8% reduction), 0.96 (0.40–2.33, p=0.9) associated with frontal airbag deployment, and 0.89 (0.37–2.18, p=0.8) for post-ADR vehicles with an airbag deployment compared to drivers of pre-ADR69 vehicles without an airbag deployment. As previous research identified an increased injury risk associated with airbag deployments, AIS 1 injuries were examined, and it was evident that the relative odds of sustaining minor upper extremity injury is twice as likely for drivers of post-ADR69 vehicles with an airbag deployment than for drivers of pre-ADR69 vehicles without an airbag (OR: 2.13, 1.14–4.01, p=0.02). These results demonstrate that while there has been an increase in the odds, or likelihood, of sustaining injuries of the upper extremity among drivers of post-ADR69 - airbag deployed vehicles, this is driven by increases in AIS1 (minor) injuries rather than more serious AIS 2 injuries and higher.

AIS 2+ LOWER EXTREMITY INJURY RISK

20% of drivers of pre-ADR69 vehicles sustained an AIS 2 (moderate) or higher injury of the lower extremity, compared to 17% of drivers of post-ADR69 vehicles, indicating no difference in the likelihood of injury. The likelihood of drivers of post-ADR69 vehicles sustaining AIS 2+ injuries was 1.5 times higher than for drivers of pre-ADR69 vehicles (0.64–3.51, p=0.3). The relative odds of sustaining AIS 2+ lower extremity injuries for drivers of vehicles with a frontal airbag deployment compared to drivers of vehicles without an airbag deployment was 0.60 (0.25–1.39, p=0.2). For drivers of post-ADR69 vehicles with an airbag deployment, the odds of sustaining a lower extremity injury was 0.90 than that for drivers of pre-ADR69 vehicles without an airbag system deployment, (0.39–2.10, p<0.8). The reduction in the likelihood of injury may be a consequence of improvements in seat belt design associated with frontal airbag systems, as well as other structural changes associated with the introduction of the ADR69 regulation.

HARM ASSOCIATED WITH ADR69 STATUS

As discussed in the method, HARM is a technique for costing injuries of varying severity for each body region. HARM analysis has the advantage of reflecting total injury outcome (and associated cost), whereas the logistic regression analysis indicates only the likelihood of sustaining an AIS 2+ injury.

Table 10 shows the mean HARM for each body region for drivers of pre-ADR69 and post-ADR69 manufactured vehicles, expressed in Australian dollars. Table 10 also provides the relative mean HARM estimate and the estimated cost of injury difference expressed as reduction (−) or increase (+) for drivers of post-ADR69 vehicles relative to pre-ADR69 manufactured vehicles. The regression model adjusts for EBS and vehicle market class. Due to the structure of the Poisson regression model it is not appropriate to sum the cost savings of individual body regions in Table 10 to derive an additive ADR69 benefit, for instance summing Head plus Face will provide an incorrect monetary estimate. The ‘Whole-of-body’ estimates includes all injuries sustained across all body regions, and involved a separate regression model.

Table 10.

Mean HARM (‘000) & adjusted cost estimates by body regions for drivers of post-ADR69 vehicles relative to drivers of pre-ADR69 vehicles, adjusted for EBS and market class

BODY REGION	Pre-ADR69 Mean (‘000)	Post-ADR69 Mean (‘000)	Estimate Post vs. Pre-ADR	95% CL (L/U)	P	Benefit (−) / Cost (+) $ Estimate (‘000)
Head	31.11	10.65	0.39	0.19–0.78	0.007	−18.97
Face	10.32	3.20	0.35	0.21–0.58	<0.001	−6.73
Neck	4.00	1.48	0.33	0.20–0.57	<0.001	−2.66
Chest	12.85	12.46	0.98	0.69–1.41	0.9	−0.23
Abdomen/pelvis	4.64	5.94	1.15	0.75–1.75	0.5	+0.68
Spine	1.65	4.31	1.78	0.99–3.20	0.05	+1.29
Upper Ex.	14.60	13.80	0.98	0.70–1.39	0.9	−0.26
Lower Ex.	25.91	33.38	1.42	0.98–2.05	0.06	+10.84
Whole-of-body	105.09	85.22	0.87	0.64–1.20	0.4	−13.24

By way of example, the mean cost of injury (HARM) associated with injuries of the head for drivers in the pre-ADR69 vehicle group was approximately AUD$31,000 compared to AUD$10,650 for drivers of post-ADR69 manufactured vehicles. Adjusting for EBS and vehicle market class differences between the ADR69 groups, the head HARM in the post-ADR69 group was 0.39 times (0.19–0.78, p=0.0007) that of drivers in the pre-ADR69 group, equating to a 61% reduction in HARM. By using the point estimate and the mean HARM associated with the pre-ADR69 group, these results translates to an average per case saving of AUD$18,970.

Table 10 also shows that the HARM sustained by drivers of post-ADR69 manufactured vehicles was lower for the face and neck, and higher for the spine and lower extremity than for drivers of pre-ADR69 manufactured vehicles. While the results are suggestive of an increased abdomen / pelvis injury cost associated with drivers of post-ADR69 vehicles, the result is ambiguous. The results also indicate no difference in the cost of chest and upper extremity injuries between the groups, but an indicative whole-of-body saving of approximately AUD$13,000 is noted.

DISCUSSION

The findings of this study demonstrated a reduction in the likelihood of injury risk, injury severity and reduced cost of injury for particular body regions for drivers of post-ADR69 vehicles and frontal airbag deployments involved in frontal impact tow-away and hospitalised crashes. The safety benefits associated with post-ADR69 vehicles and frontal airbag deployment were especially apparent for the head, face, and chest. The results do however show a slight increase in risk and cost of injury to the lower extremity and spine.

The analysis of ADR69 and airbag effectiveness in this paper was complicated by the fact that the ADR69 regulation, while not mandating the fitment of frontal airbag systems, provided the impetus for many manufacturers to do so. It was not possible in the context of this paper, therefore, to specify the precise level of ADR69 compliance and airbag effectiveness in mitigating injury risk absolutely. The results were however clear in delineating the combined effectiveness of the ADR69 regulation and airbags in reducing injury risk, particularly for the head.

The use of sophisticated regression models represents a significant advance from previous research examining the effectiveness of ADR69 and frontal airbags in Australia. When considering the findings of this paper it is imperative to note the strict entry criteria to the study, and the sample of crashes to which the results apply. Even so, the findings represent the most sophisticated and inclusive analysis carried out so far of ADR69 and airbags effectiveness using real-world frontal crashes in Australia.

An important issue is whether vehicles manufactured prior to the introduction of the ADR69 standard would have in fact met the ADR69 standard if tested. Indeed, some manufacturers elected to fit frontal driver airbags some years before the standard, and this is further complicated by the phased introduction of ADR69 from mid-1995. Precise information of the compliance status of these vehicles was not available, however it is noteworthy that some pre-ADR69 vehicles would likely have met the standard [Seyer, 1992, 1993]. Despite this, the changes in injury risk reported here between the ADR69 groups is more than noteworthy, and is likely to be driven more by the widespread introduction of frontal airbag systems rather than the introduction of ADR69 in mid-1995 per se.

There were a number of limitations associated with the study. The inability to separate clearly the impact of airbags from the ADR69 regulation is a limitation, but is explained by the overlap in their introduction. The results of the study may have some degree of selection bias, as participation in any of the three databases was voluntary. The direction of this sampling bias is at this stage unclear and requires clarification. Sample weights were not available in the datasets and are required in order to provide unbiased and representative results applicable to the entire fleet, particularly as the datasets spanned separate time periods. The results are therefore best viewed as being sample-specific rather than representative of the entire Australian vehicle fleet. A further limitation of the study was a degree of self-reporting of injuries for drivers not admitted to hospital, and the accuracy of this method is unknown. Finally, the cost of injury figures used in this paper are dated and require updating in order to provide accurate, current day savings associated with ADR69 and airbag systems in Australia.

CONCLUSION

Despite these limitations, this paper demonstrates significant benefits associated with the introduction of ADR69 as well as the parallel introduction of frontal airbags in Australia, while pointing the way for improvements in design standards for the further protection of the abdomen and pelvis, the spine, and the lower extremity. Future studies would be best placed to use a larger sample with a range of injury severities so that the influence of variables such as height and age on injury risk could be explored more fully than was possible within the context of this paper.

Finally, manufacturers and regulatory bodies responsible for vehicle standards could perhaps work with the goal of optimising the relationship between the seat belt and the airbag in order to further reduce injuries, particularly those to the abdomen and pelvis, while improved design and a modified standard may reduce injury risk associated with the extremities.