A Field Data Analysis of Risk Factors Affecting the Injury Risks in Vehicle-To-Pedestrian Crashes

Abstract

The head, torso, and lower extremity are the most commonly injured body regions during vehicle-to-pedestrian crashes. A total of 312 cases were selected from the National Automotive Sampling System (NASS) Pedestrian Crash Data Study (PCDS) database to investigate factors affecting the likelihood of sustaining MAIS 3+, AIS 3+ head, AIS 3+ torso, and AIS 2+ lower extremity injuries during vehicle-to-pedestrian frontal crashes. The inclusion criteria were pedestrians: (a) aged 14 years or older, (b) with a height of 1.5 m and taller, and (c) who were injured in an upright standing position via vehicle frontal collision. The injury odds ratios (ORs) calculated from logistic regression analyses were used to evaluate the association between selected injury predictors and the odds of sustaining pedestrian head, torso, and lower extremity injuries. These predictors included a crash factor (impact speed), pedestrian factors (age, gender, height, and weight), and vehicle factors (front bumper central height, front bumper lead, ground to front/top transition point height (FTTPH), and rear hood opening distance (RHOD)). Results showed that impact speed was a statistically significant predictor for head, torso, and lower extremity injury odds, as expected. Comparison of people 65 years of age and older to young adults aged 14 to 64 showed that age was also a significant predictor for torso (p<0.001, OR=23.8) and lower extremity (p=0.020, OR=2.44) injury odds, but not for head injuries (p=0.661). Vehicles with higher FTTPH and more vertical frontal structures were aggressive to pedestrians, especially regarding injuries to the torso. A very short RHOD would be more likely to lead the pedestrian to impact the windshield and windshield frame, thus increasing the head injury risk.

INTRODUCTION

Pedestrians are the most vulnerable road users because they are unprotected in traffic crashes. About 65 percent of 1.17 million annual traffic fatalities world-wide are pedestrians [World Bank, 2007]. In 2006, 4,784 pedestrians were killed, and 61,000 pedestrians were injured during traffic crashes in the United States, representing 11 percent of traffic fatalities and 2 percent of traffic related injuries [NHTSA, 2006]. In China, about 25,000 pedestrians were killed in traffic crashes in 2006, which was 25 percent of all traffic related fatalities [TAMPS, 2005]. In Japan, 2,450 pedestrians were killed in traffic crashes in 1999, accounting for approximately 28 percent of all traffic collision associated deaths [Maki et al., 2003]. In the European Union, more than 7,000 pedestrians are killed every year [EEVC, 1998].

The head (including face), torso (including thorax, abdomen, and spine), and lower extremity (including pelvis) are the most frequently injured body regions for pedestrians [Longhitano et al., 2005a; Longhitano et al., 2005b]. The head is the most commonly injured body region in fatal vehicle-to-pedestrian crashes [Maki et al., 2003; Mizuno et al., 1999], while lower extremity injuries often resulted in long term disability. Furthermore, Lefler et al. (2004) reported that the number of light trucks and vans increased in the United States between 1980 and 1999, and accounted for nearly 50 percent of new light passenger vehicle sales. The study also revealed that the likelihood of sustaining serious pedestrian thorax injury was substantially greater when colliding with light trucks or vans than passenger cars. It was reported that the torso was the second most commonly injured body region in pedestrian to light truck or van crashes [Longhitano et al., 2005a]. Therefore, head, torso, and lower extremity injuries are of major concern in vehicle-to-pedestrian crashes.

For the past few years, numerous field data analyses have been conducted using frequency or logistic regression statistical methods to investigate the injury mechanisms and identify the main factors affecting injury risks during vehicle-to-pedestrian crashes. Pedestrian factors, such as age and gender, have been reported to be associated with pedestrian injury risks. An increase in age was shown to lead to an increase in the injury severity [Miles-Doan, 1996].

Pedestrians older than 65 years were more likely to suffer severe injuries or death than young adults [Holubowycz, 1995; Lee et al., 2005; Sze et al., 2007; Zajac et al., 2003], while the adult mortality rate was higher than that for children in vehicle-to-pedestrian crashes [Roudsari et al., 2004]. Holubowycz (1995) found that a large proportion of the seriously or fatally injured pedestrians were males. Conversely, Lee et al. (2005) indicated that females were more likely to be severely injured, while Pitt et al. (1990) and Stone et al. (2003) reported that injury or fatality rate was not significantly different between males and females. To date, studies using pedestrian height and/or weight to predict pedestrian injury risks have been scarce.

Vehicle factors, such as vehicle body type, vehicle weight, and front structure characteristics, have also been studied previously. Atkins et al. (1988) found that the severity of injuries sustained by pedestrians increased with an increase in striking vehicle weight. However, Mizuno et al. (1999) reported that it was not the mass but the geometrical incompatibility between sports utility vehicles (SUV) and passenger cars that induced the high fatality rate of pedestrians. Al-Ghamdi (2002) reported that the relationship between pedestrian injury severity and vehicle body type was not statistically significant, while the studies from Lefler et al. (2004), Lee et al. (2005), Ballesteros et al. (2003), Longhitano et al. (2005a), and Roudsari et al. (2004) indicated that pedestrians struck by passenger cars sustained less severe injuries compared to those struck by light truck vehicles (LTV) or vans. The major differences among all vehicle body types that are related to frontal impact pedestrian safety are vehicle frontal structural parameters, such as front bumper height, front bumper lead, ground to front/top transition point height, and so on. Therefore, it would be more beneficial if the relationship between these frontal structural parameters (in contrast to the vehicle body type) and the associated risks of pedestrian injuries were studied.

In the literature, among crash factors investigated, impact speed is the only non-controversial parameter reported for predicting pedestrian injury severity in vehicle-to-pedestrian crashes. Higher impact speeds always resulted in more severe injuries to pedestrians [Ballesteros et al., 2004; Jensen, 1999; Lee et al., 2005; Lefler et al., 2004; Miles-Doan, 1996; Pitt et al., 1990; Roudsari et al., 2004; Sze et al., 2007].

In vehicle-to-pedestrian crashes, pedestrians are injured by contacting the vehicle (primary impact), striking the ground (secondary impact), smashing into other objects in the environment, and/or even from a source such as heat, flame, or chemical not associated with impact. Otte et al. (2001) analyzed 293 traffic crashes from 1985 to 1999 involving adult pedestrians who were taller than 1.5 m, came into contact with frontal plane of cars, and with collision speeds between 20 and 70 km/h. They concluded that injuries due to secondary impacts were not as frequent or severe compared to injuries caused by primary impacts.

Based on the short literature review outlined above, there are four aspects of field data analysis that could be improved when studying pedestrian injuries:

1. Data selection: Most previous field data analyses used injured pedestrian data from both adults and children. Children are generally shorter than adults, and thus, the impact position for children is very different from adults during vehicle-to-pedestrian crashes. Consequently, the injury sources coded in traffic related injury databases are, in general, different from those documented for adult pedestrian injuries. Moreover, the tolerances and responses of children are different from those of adults during impacts. Therefore, there is a need to treat adults and children separately when investigating pedestrian injuries.

2. Pedestrian height and weight: To the best of the authors’ knowledge, the height and weight of pedestrians have not been investigated in previous field data analyses. Similar to the differences found in coded injury sources between children and adults, pedestrian height also affects the position of contact on the vehicle frontal components. In turn, pedestrians of different heights may sustain different injuries in vehicle-to-pedestrian crashes. Intuitively, greater pedestrian weight could possess greater pedestrian inertia and impact forces. Therefore, pedestrian height and weight were analyzed in this study.

3. Vehicle parameters: In most previous analyses, the vehicle body type was considered as the sole vehicle parameter to predict pedestrian injury risk. However, the same vehicle body type could be designed with different frontal structural parameters, such as the front bumper center height, front bumper lead, ground to front/top transition point height, and so on. The risk of pedestrian injuries could be predicted more accurately using these sub-parameters of vehicle.

4. Pedestrian injuries: Pedestrian injury severity was evaluated by the maximum AIS (MAIS) of the entire body in most of previous studies. Each region of the human body, such as head, torso, and lower extremity, has different injury mechanisms. In order to understand detailed injury risks of various body regions of the pedestrian and subsequent investigation of the individual injury mechanisms, the head, torso, and lower extremity injury odds in vehicle-to-pedestrian crashes should be individually studied.

The objective of this study was to predict the head, torso, and lower extremity injury risks of standing pedestrians impacted via frontal components of vehicles. Logistic regressions were performed in conjunction with a crash factor, pedestrian factors, and vehicle factors.

METHODS

To analyze the effect of impact speed, pedestrian factors, and vehicle factors on head, torso and lower extremity injury odds in vehicle-to-pedestrian crashes, the National Accident Sampling System (NASS) Pedestrian Crash Data Study (PCDS), an in-depth accident database publicly available in the United States, was used.

NASS-PCDS

The PCDS database, which included a total of 552 vehicle-to-pedestrian crashes with 4,500 injuries, was collected from 1994 to 1998 in six cities: Buffalo, Chicago, Dallas, Fort Lauderdale, San Antonio, and Seattle. A pedestrian was defined as one person who was on a roadway, a sidewalk, a path contiguous with a traffic way, or on private property. Persons in or on a non-motorist conveyance are not pedestrians and were excluded from the study. Each case must meet five criteria as follows: (1) the vehicle should be a late model year vehicle made in or after 1990 and should be moving forward when the collision occurred, (2) the pedestrian should not be lying or sitting prior to the crash, (3) the striking portion of the vehicle structure should be original manufactured parts without previous damage and/or parts removed in the impact area, (4) the pedestrian impacts should be the vehicle’s only impacts, (5) the first point of contact between the vehicle and the pedestrian should be forward of the top of the A-pillar [NASS-PCDS, 1996].

Case selection

Because the objective of this study was to investigate the injury risk associated with an upright standing pedestrian, a total of 103 (18.7%) cases were excluded from further study due to these reasons: (1) one case in which the severity of the injury was not recorded, (2) one case in which the pedestrian was crouching, (3) two cases in which the pedestrians were bending at waist, and (4) 99 cases in which the pedestrians were impacted by the left side or right side of the striking vehicles.

Children are generally much shorter than adults, and their tissue properties are very different from adults as well [Roudsari et al., 2004], particularly in terms of comparing the developing child with the elderly adult. These differences could result in different kinematics and injury tolerances between children and adults [Longhitano et al., 2005b]. Hence, pedestrians aged 13 or younger were excluded from this study. For adult pedestrians with stature shorter than 1.50 m, the contact areas on the pedestrians’ bodies and the striking vehicles, as well as the kinematics during vehicle-to-pedestrian crashes could be different from taller adults, so these cases were also excluded, similar to the study by Matsui (2005). Table 1 shows the distribution of pedestrians with age 14 and older and height 1.5 m and taller among the remaining 449 pedestrians who were upright when hit by a vehicle. Combining these two parameters, a total of 312 vehicle-to-pedestrian cases were investigated in this study.

Table 1.

Age and height distribution of the pedestrians reported in PCDS. Of the remaining 449 cases, 312 pedestrians with an age of 14 and older and a height of 1.5 m and taller were included in statistical analyses.

Total	Items	No. of Cases	Percent
449 cases	Age 14 and older	366	81.5%
	Height 1.5 m and taller	334	74.4%
	Age ≥ 14 & Height ≥ 1.5m	312	69.5%

Injury selection

There were 2,903 injuries sustained by the 312 pedestrians selected from the PCDS database. These injuries were divided into nine categories based on body region: head, face, neck, thorax, abdomen, spine, upper extremity, lower extremity, and unspecified. Among the 2,903 injuries, the coded injury source for 2,319 (79.9%) injuries was some part of the vehicle compared to only 503 (17.3%) injuries in which the ground was coded as the injury source (Table 2).

Table 2.

Distribution of coded injury sources and injury severity with each injury source

Injury Source	No. of Cases (Percent)	AIS	No. of Cases (Percent)
Vehicle	2,319 (79.9%)	1	1410 (60.8%)
		2+	905 (39.0%)
		3+	521 (22.5%)
Ground	503 (17.3%)	1	441 (87.7%)
		2+	62 (12.3%)
		3+	37 (7.4%)
Object in environment	60 (2.1%)	1	48 (80.0%)
		2+	12 (20.0%)
		3+	5 (8.3%)
Non-contact injury	17 (0.6%)	1	9 (52.9%)
		2+	8 (47.1%)
		3+	4 (23.5%)
Unknown	4 (0.1%)	1	4 (100.0%)
		2+	0
		3+	0
Total	2,903 (100%)	-	-

Of the 503 pedestrian injuries with the ground coded as the source of injury, 441 injuries were AIS 1 injuries, accounting for 87.7% of all the ground induced injuries (Table 2). Only 37 (7.4%) injuries were AIS 3+ injuries. Therefore, pedestrians were coded to be not as frequently and severely injured by the ground during vehicle-to-pedestrian crashes. The same conclusion was reported by Otte et al. (2001). Consequently, injuries resulting from ground impacts were not considered in this study in order to investigate the pedestrian injury risks caused directly by vehicles in vehicle-to-pedestrian crashes. Furthermore, the sample size of injuries with the ground coded as the injury source was insufficient to identify any meaningful association between vehicle factors and pedestrian injury risk resulting from ground impact.

Pedestrians considered in this study had at least one: (a) MAIS 3 to 6 (MAIS 3+) injury, (b) AIS 3 to 6 (AIS 3+) injury to the head, (c) AIS 3+ injury to the torso, or (c) AIS 2 to 6 (AIS 2+) lower extremity injury. The nature of AIS is to indicate the likelihood of fatality. As a result, injuries to the head and torso are in general more “serious” (AIS 3) than injuries to the lower extremity. Most published studies reported pedestrians AIS 3+ head/torso injuries and AIS 2+ lower extremity injuries. The same considerations were used in the current study to allow direct comparisons between previous and current analyses. Table 3 shows the distribution of AIS 1+, AIS 2+, and AIS 3+ injuries to different body regions in which the vehicle was coded as the injury source. Table 4 represents the distribution of MAIS 3+, AIS 3+ head, AIS 3+ torso, and AIS 2+ lower extremity injuries coded as induced by vehicles.

Table 3.

Distribution of AIS 1+, AIS 2+, and AIS 3+ injuries to different body regions in which the vehicle was coded as the injury source

Body Region	AIS 1+ No. of Cases (Percent)	AIS 2+ No. of Cases (Percent)	AIS 3+ No. of Cases (Percent)
Head	700 (30.2%)	280 (30.9%)	226 (43.4%)
Torso	405 (17.5%)	256 (28.3%)	134 (25.7%)
Lower extremity	822 (35.4%)	310 (34.3%)	138 (26.5%)
Upper extremity	382 (16.5%)	59 (6.5%)	23 (4.4%)
Neck	9 (0.4%)	0	0
Unspecified	1 (0.0%)	0	0
Total	2319 (100.0%)	905 (100.0%)	521 (100.0%)

Table 4.

Distribution of MAIS 3+, AIS 3+ head, AIS 3+ torso, and AIS 2+ lower extremity injuries coded as induced by vehicle contact among the 312 cases studied

Human Body Injury	No. of Cases	Percent
MAIS 3+	118	37.8%
Head/Face AIS 3+	67	21.5%
Torso AIS 3+	62	19.9%
Lower extremity AIS 2+	119	38.1%

Injury Predictors

Injury predictors used in this study for estimating pedestrian injury risks were: crash factor (impact speed in km/h), pedestrian factors, and vehicle factors. Pedestrian factors include age (year), gender, height (cm), and weight (kg). Vehicle factors consist of front Bumper Central Height (BCH), front Bumper Lead (BL), ground to Front/Top Transition Point Height (FTTPH), and Rear Hood Opening Distance (RHOD). An illustration of vehicle factors is shown in Figure 1.

Figure 1.

A schematic diagram of vehicle frontal structure (a) and vehicle factors (b)

BCH: front Bumper Central Height (cm)

BL: front Bumper Lead (cm)

FTTPH: ground to Front/Top Transition Point Height (cm)

RHOD: Rear Hood Opening Distance (cm)

More specifically, BCH is the average height of the top and bottom of the bumper, and BL is the longitudinal width of the front bumper measured from the leading edge of the bumper to where the horizontal structure ends and the vertical structure begins [NASS-PCDS, 1996].

The front/top transition point is the location on the front of the vehicle where the top of the hood plane transitions into the frontal structures. This could be the point at which the front of the hood curves downward, the hood edge, or the top edge of the upper grille panel. FTTPH is the vertical height from the ground level to the front/top transition point, along the vehicle’s centerline. The RHOD is a continuous distance following the contour of the vehicle along the vehicle’s centerline, from the ground level to the rear hood opening [NASS-PCDS, 1996].

Table 5 shows the various predictors and different categories assigned to these predictors. Vehicles involved in vehicle-to-pedestrian crashes are categorized as passenger cars (PCs), vans, and light truck vehicles (LTVs) including utility vehicles and light conventional trucks. Because 65 years old was reported as the frontier of the elderly in the published literature [Atkins et al., 1988; Lee et al., 2005; Sze et al., 2007; Zajac et al., 2003], we chose to define pedestrians aged 65 and higher as the elderly group.

Table 5.

Injury predictor categories

Predictor	Reference (no. of case)	Category 1 (no. of case.)	Category 2 (no. of case.)	No. of cases with missing values
Speed (km/h)	0–24 (145)	25–55 (95)	56+ (34)	38
Age (year)	14–64(257)	65+ (55)	-	0
Gender	Female(153)	Male(159)	-	0
Height (cm)	161–175 (170)	176+ (84)	150–160 (58)	0
Weight (kg)	61–90 (185)	91+ (47)	0–60 (77)	3
BCH (cm)	0–42.5 (58)	43–50 (185)	50.5+ (69)	0
BL (cm)	0–6 (38)	7–12 (245)	13+ (28)	1
FTTPH (cm)	0–75 (87)	76–100 (155)	101+ (70)	0
RHOD (cm)	171–200 (153)	201+ (71)	0–170 (88)	0

Statistical methods

Logistic regression analysis was employed to evaluate the relationship between the selected factors and the odds of pedestrian injuries. Because it is important to adjust or account for the confounding effects between variables, logistic regression was conducted with all the selected variables (including crash factor, pedestrian factors, and vehicle factors) as predictors or covariates in a model to investigate the independent contribution of each factor. Therefore, the final odds ratio (OR) presented for each factor was the OR after controlling/adjusting for all the other factors. For example, the OR predicted for one of the vehicle factors was the OR after adjusting the crash factor, pedestrian factors, and all the other vehicle factors. All analyses were conducted using SPSS 15.0. A significance level of 0.05 was introduced to test statistical significance.

RESULTS

The final study sample consisted of 312 cases involving 312 pedestrians and 312 vehicles. The odds of sustaining MAIS 3+, AIS 3+ head, AIS 3+ torso, and AIS 2+ lower extremity injuries for the crash factor, pedestrian factors, and vehicle factors in vehicle-to-pedestrian crashes were investigated. All injuries caused by vehicle-to-pedestrian impacts are listed in the following tables while injuries resulting from sources other than the vehicle structure were not considered.

Pedestrian injury ORs for crash factor

Table 6 shows the effect of impact speed on the risk of pedestrian injury after adjusting for the other factors. The speed was divided into three categories, with the 0–24 km/h group set as the reference value. As seen in Table 6, impact speed was a statistically significant (p < 0.001) predictor of a pedestrian sustaining the injuries included in this study. Pedestrians impacted by a 25 to 55 km/h car had 3.9 to 82.0 times the risk of sustaining the injuries investigated compared to those struck by a car traveling at 0 to 24 km/h. When the impact speed was 56 km/h and higher, the injury odds ranged from 21.2 to 4342.1 times those of a 0 to 24 km/h impact. Although statistical significance was found for all body regions studied, the lower extremity was found to be not as sensitive as the head and torso to impact speed as indicated by the lower elevation of injury ORs.

Table 6.

Logistic regression analysis to evaluate the effect of impact speed on the risk of pedestrian injury after adjusting for all the other factors

Predictor: Impact speed (reference: 0–24 km/h)		25–55 km/h versus reference (0–24 km/h)		56 km/h and higher versus reference (0–24 km/h)
		p	OR (95% CI)	p	OR (95% CI)
Whole body	MAIS 3+	0.000	8.590 (4.174–17.681)	0.000	334.672 (38.340–2921.356)
Head	AIS 3+	0.000	6.951 (2.391–20.206)	0.000	230.316 (48.220–1100.066)
Torso	AIS 3+	0.000	81.955 (9.253–725.869)	0.000	4342.136 (300.154–62814.848)
Lower extremity	AIS 2+	0.000	3.937 (2.093–7.403)	0.000	21.243 (7.286–61.936)

Pedestrian injury ORs for pedestrian factors

The ORs for one of pedestrian factors (age, gender, height, and weight) were based on results after adjusting for the crash factor (impact speed), vehicle factors and the other pedestrian factors. For instance, the ORs predicted for pedestrian age were based on results after adjusting the impact speed, vehicle factors (BCH, BL, FTTPH and RHOD) and pedestrian gender, height, and weight.

Pedestrian age

Table 7 shows the injury ORs related to the effect of age on pedestrian injury. Pedestrian age was a statistically significant predictor for the injury odds of pedestrians aged 65 years and older with AIS 2+ lower extremity injuries (p=0.020) and AIS 3+ torso injuries (p<0.001) compared to young adults 14 to 64 years of age. However, it was not a statistically significant predictor for injury odds of pedestrians with AIS 3+ head injuries (p=0.661) and MAIS 3+ injuries (p=0.069). The AIS 3+ torso injury odds were 23.8 times (95% CI 4.98–114.15) greater and the AIS 2+ lower extremity injury odds were 2.44 times (95% CI 1.15–5.18) greater for the elderly (>=65) as compared to young adults.

Table 7.

Logistic regression analysis to evaluate the effect of age on the risk of pedestrian injury after adjusting for all the other factors

Bold face indicates statistically significant

Predictor: Age		Elderly (65 and more) versus Younger (14–64)
		p	OR (95% CI)
Whole body	MAIS 3+	0.069	2.243 (0.938–5.360)
Head	AIS 3+	0.661	1.290 (0.414–4.022)
Torso	AIS 3+	0.000	23.842 (4.980–114.150)
Lower extremity	AIS 2+	0.020	2.439 (1.148–5.181)

Pedestrian gender

Gender was not a statistically significant predictor of MAIS 3+, AIS 3+ head, AIS 3+ torso, and AIS 2+ lower extremity injury odds (p>0.7) as seen in Table 8.

Table 8.

Logistic regression analysis to evaluate the effect of gender on the risk of pedestrian injury after adjusting for all the other factors

Predictor: Gender		Male versus female
		p	OR (95% CI)
Whole body	MAIS 3+	0.758	1.137 (0.502–2.574)
Head	AIS 3+	0.706	1.219 (0.435–3.412)
Torso	AIS 3+	0.997	0.997 (0.264–3.774)
Lower extremity	AIS 2+	0.810	0.915 (0.445–1.883)

Pedestrian height

Table 9 shows that pedestrian height was not a statistically significant predictor of MAIS 3+, AIS 3+ torso, and AIS 2+ lower extremity injury odds. However, pedestrian height was a statistically significant predictor of AIS 3+ head injury odds (p=0.031, OR=0.193) for shorter pedestrians compared to medium height pedestrians (161 to 175 cm).

Table 9.

Logistic regression analysis to evaluate the effect of height on the risk of pedestrian injury after adjusting for all the other factors

Bold face indicates statistically significant

Predictor: Height (reference: 161–175 cm)		Taller (176 cm and higher) versus reference (161–175 cm)		Shorter (150–160 cm) versus Reference (161–175 cm)
		p	OR (95% CI)	p	OR (95% CI)
Whole body	MAIS 3+	0.077	0.406 (0.150–1.102)	0.448	0.700 (0.278–1.760)
Head	AIS 3+	0.323	0.548 (0.167–1.803)	0.031*	0.193 (0.043–0.860)
Torso	AIS 3+	0.400	0.511 (0.107–2.438)	0.384**	0.465 (0.083–2.609)
Lower extremity	AIS 2+	0.519	0.763 (0.335–1.738)	0.267	1.574 (0.707–3.506)

^*The AIS 3+ injured head sample size was only 5.

^**The AIS 3+ injured torso sample size was only 5.

Pedestrian weight

The odds of heavy (91 kg and heavier) pedestrians sustaining MAIS 3+ injuries were 4.26 times (95% CI 1.49–12.2) those of medium weight (61 to 90 kg) pedestrians, and the odds of heavy pedestrians sustaining AIS 2+ lower extremity injuries were 3.03 times (95% CI 1.24–7.41) greater. The AIS 3+ head injury ORs of heavier to medium weight pedestrians were 2.32 times (95% CI 0.70–7.74) and the AIS 3+ torso injury ORs for the same group were 1.76 times (95% CI 0.39–7.91) greater, but neither of these values was statistically significant. It was found that pedestrian weight was not a statistically significant predictor of pedestrian injury when comparing pedestrians weighing less than 61 kg to those weighing 61 to 90 kg (Table 10).

Table 10.

Logistic regression analysis to evaluate the effect of weight on the risk of pedestrian injury after adjusting for all the other factors

Bold face indicates statistically significant

Predictor: Weight (reference: 61–90 kg)		Heavier (91 kg and heavier) versus medium (61–90 kg)		Less weight (0–60 kg) versus medium (61–90 kg)
		p	OR (95% CI)	p	OR (95% CI)
Whole body	MAIS 3+	0.007	4.257 (1.486–12.194)	0.471	1.363 (0.587–3.166)
Head	AIS 3+	0.169	2.324 (0.698–7.740)	0.214	2.016 (0.667–6.095)
Torso	AIS 3+	0.460	1.762 (0.392–7.910)	0.199*	0.397 (0.097–1.625)
Lower extremity	AIS 2+	0.015	3.034 (1.243–7.405)	0.722	0.874 (0.418–1.830)

^*The AIS 3+ injured torso sample size was only 8.

Pedestrian injury ORs for vehicle factors

The ORs calculated for one of vehicle factors (BCH, BL, FTTPH, and RHOD) were based on results after adjusting for crash factor, pedestrian factors and the other vehicle factors. For instance, the ORs predicted for BCH were based on results after adjusting for the impact speed, pedestrian age, gender, height, and weight, and BL, FTTPH, and RHOD.

BCH

Table 11 shows pedestrian injury ORs evaluated for the effect of BCH after statistically adjusting all the other factors. The BCH was not a statistically significant predictor for pedestrian injury odds.

Table 11.

Logistic regression analysis to evaluate the effect of BCH on the risk of pedestrian injury after adjusting for all the other factors

Predictor: BCH (reference: 0–42.5 cm)		Medium (43–50 cm) versus lower (0–42.5 cm)		Higher (50.5 cm and higher) versus lower (0–42.5 cm)
		p	OR (95% CI)	p	OR (95% CI)
Whole body	MAIS 3+	0.863	1.088 (0.418–2.830)	0.744	1.255 (0.321–4.907)
Head	AIS 3+	0.250	0.510 (0.162–1.604)	0.513	0.562 (0.100–3.162)
Torso	AIS 3+	0.421	0.534 (0.116–2.462)	0.058	0.115 (0.012–1.077)
Lower extremity	AIS 2+	0.504	1.324 (0.581–3.021)	0.483	1.532 (0.466–5.044)

BL

Table 12 shows that the BL was a significant predictor of pedestrians with AIS 3+ head (p=0.045) injuries, after adjusting for all the other factors. The AIS 2+ lower extremity injury odds with a longer BL (more than 13 cm) were 2.98 times (95% CI 0.75–11.78) those with shorter BL (0 to 6 cm), but it was not statistically significant (p=0.120). The BL was not a statistically significant predictor of AIS 3+ torso and AIS 2+ lower extremity injury odds (p>0.1) either.

Table 12.

Logistic regression analysis to evaluate the effect of BL on the risk of pedestrian injury after adjusting for all the other factors

Bold face indicates statistically significant

Predictor: BL (reference: 0–6 cm)		Medium (7–12 cm) versus shorter (0–6 cm)		Longer (13 cm and longer) versus shorter (0–6 cm)
		p	OR (95% CI)	p	OR (95% CI)
Whole body	MAIS3+	0.933	0.959 (0.359–2.558)	0.062	4.550 (0.928–22.299)
Head	AIS3+	0.212	2.957 (0.539–16.208)	0.045	7.761 (1.045–57.623)
Torso	AIS3+	0.519	1.784 (0.307–10.377)	0.426	2.442 (0.271–22.007)
Lower extremity	AIS2+	0.521	0.752 (0.314–1.799)	0.120	2.977 (0.752–11.780)

FTTPH

Table 13 shows the pedestrian injury ORs to evaluate the effect of FTTPH after adjusting for all the other factors in vehicle-to-pedestrian crashes. It was found that pedestrian AIS 3+ torso injury odds with higher FTTPH (101 cm and higher) were 20.8 times (95% CI 2.30–187.9) those with lower FTTPH (75 cm and lower). However, the FTTPH was not a statistically significant predictor of pedestrians with MAIS 3+, AIS 3+ head, and AIS 2+ lower extremity injury odds.

Table 13.

Logistic regression analysis to evaluate the effect of FTTPH on the risk of pedestrian injury after adjusting for all the other factors

Bold face indicates statistically significant

Predictor: FTTPH (reference: 0–75 cm)		Medium (76–100 cm) versus lower (0–75 cm)		Higher (101 cm and higher) versus lower (0–75 cm)
		p	OR (95% CI)	p	OR (95% CI)
Whole body	MAIS3+	0.325	0.638 (0.261–1.562)	0.974	0.978 (0.256–3.741)
Head	AIS3+	0.646	0.773 (0.257–2.321)	0.742	1.323 (0.250–7.003)
Torso	AIS3+	0.497	1.716 (0.361–8.163)	0.007	20.782 (2.299–187.851)
Lower extremity	AIS2+	0.314	0.673 (0.312–1.454)	0.265	0.505 (0.152–1.678)

RHOD

Table 14 shows the pedestrian injury ORs to evaluate the effect of RHOD after statistically adjusting for all the other factors in vehicle-to-pedestrian crashes. It was found that the RHOD was a statistically significant predictor of pedestrians with AIS 3+ head injury odds, but not significant for MAIS 3+, AIS 3+ torso, and AIS 2+ lower extremity injuries. The AIS 3+ head injury ORs when comparing shorter RHOD to medium RHOD were 4.30 times (95% CI 1.34–13.8) greater.

Table 14.

Logistic regression analysis to evaluate the effect of RHOD on the risk of pedestrian injury after adjusting for all the other factors

Bold face indicates statistically significant

Predictor: RHOD (reference: 171–200 cm)		Longer (201 cm and longer) versus medium (171–200 cm)		Shorter (0–170 cm) versus medium (171–200 cm)
		p	OR (95% CI)	p	OR (95% CI)
Whole body	MAIS3+	0.160	2.019 (0.758–5.377)	0.239	1.669 (0.712–3.914)
Head	AIS3+	0.857	1.116 (0.338–3.677)	0.014	4.297 (1.339–13.785)
Torso	AIS3+	0.276	2.220 (0.529–9.322)	0.849	1.160 (0.252–5.334)
Lower extremity	AIS2+	0.195	1.751 (0.751–4.084)	0.978	1.010 (0.483–2.115)

DISCUSSION

The purpose of this study was to determine factors affecting the odds of upright standing pedestrian sustaining MAIS 3+, AIS 3+ head, AIS 3+ torso, or AIS 2+ lower extremity injuries during vehicle-to-pedestrian crashes, using logistic regression methods after adjusting covariants. It is hoped that vehicle factors identified as significantly affecting injury odds could be used to guide further experimental and numerical investigations to explore mechanisms of head, torso, and lower extremity injuries due to vehicle-to-pedestrian crashes.

Crash factor

Impact speed

Numerous studies have concluded that impact speed was a significant predictor of pedestrian serious injury or mortality odds using logistic regression or frequency methods. With each increasing impact speed level, the injury severity or mortality increased [Ballesteros et al., 2004; Jensen, 1999; Lee et al., 2005; Miles-Doan, 1996; Pitt et al., 1990; Roudsari et al., 2004; Sze et al., 2007]. However, results found in these studies were limited to the study of MAIS. In the current study, not only the pedestrian MAIS 3+ injury odds, but also AIS 3+ head, AIS 3+ torso, and AIS 2+ lower extremity injury odds, were analyzed using NASS-PCDS database. It was found that the impact speed was not only a statistically significant predictor for MAIS 3+ injury odds, but also AIS 3+ head, AIS 3+ torso, and AIS 2+ lower extremity injury odds. For this reason, it is critical to adjust injury outcomes for impact speed when calculating the injury ORs for other factors.

Pedestrian factors

Age

Several previous studies concluded that pedestrians older than 65 years were more frequently involved in severe or fatal crashes compared to younger victims [Holubowycz, 1995; Lee et al., 2005; Sze et al., 2007; Zajac et al., 2003]. In the current study, it was found that elderly pedestrians were more likely to sustain AIS 3+ torso and AIS 2+ lower extremity injuries, but not AIS 3+ head injuries. One can speculate that elderly people are more prone to having osteoporosis in bones compared to younger persons, but that the difference between the mechanical properties of head for elderly and younger people is not as significant. Consequently, more torso and lower extremity injures were observed. Our results suggest that there is a need to identify the mechanical properties, such as Young’s and shear moduli, of long bone instead of using age alone to determine if the strength of long bone is a significant factor affecting injury outcomes.

Gender

In the present study, gender was found to be a statistically insignificant (p>0.7) predictor of MAIS 3+ injury odds sustained by pedestrians after adjusting for all the other factors. This finding was consistent with that reported by Pitt et al. (1990) and Stone et al. (2003). Furthermore, gender was not a statistically significant predictor of AIS 3+ head, AIS 3+ torso, and AIS2+ lower extremity injury odds. The authors believe that the effect of gender was overshadowed by the fact that the effects of weight and height were removed prior to calculating the injury odds due to gender difference.

Height

Although statistical significance was found in this study for shorter stature pedestrian sustaining AIS 3+ head injury as compared to medium height pedestrian, the sample size of AIS 3+ head injuries was only 5, which was insufficient to identify any meaningful association between pedestrian height and pedestrian injury risk. Hence, more studies are needed to explore the effect of height on pedestrian injuries.

Weight

Because heavier pedestrians possess greater inertia, larger forces will occur between the vehicle and the pedestrian, which may result in more serious injuries to pedestrians. The authors found significant differences in overall (MAIS 3+) and AIS 2+ lower extremity injuries for heavier pedestrians compared to medium weight pedestrians but not in AIS 3+ head and torso injuries. Factors other than weight may contribute more to head and torso injuries and require further research.

Pedestrian factors, such as age, gender, height, and weight, are useful information when designing experiments to study pedestrian responses and injury tolerances. However, pedestrian-friendly vehicles cannot be designed for only the elderly, females, shorter, or heavier pedestrians. Therefore, vehicle factors must be studied while simultaneously adjusting for impact speed, pedestrian age, gender, height, and weight before determining which vehicle parameters are friendlier to the impacted pedestrian.

Vehicle factors

In vehicle-to-pedestrian frontal crashes, the first contact generally occurs between the front bumper and pedestrian leg or knee joint, followed by thigh-to-hood (bonnet) edge contact. Consequently, the pelvis, torso, and head are impacted by the hood leading edge, hood, and hood/windshield, respectively [Pritz, 1983; Yang, 1997]. Based on data reported in these studies, the vehicle frontal factors BCH and BL correspond to pedestrian lower extremity injury, FTTPH corresponds to pedestrian torso injury, and RHOD corresponds to pedestrian head injury.

Roudsari et al. (2004) concluded that LTVs and vans were more likely to result in severe pedestrian injuries, using logistic regression analysis with the PCDS database. In the current study, detailed vehicle frontal parameters, instead of vehicle type, were investigated to predict pedestrian injury odds to recognize the wide variations even within the same vehicle type.

BCH

Matsui (2005) found that lower BCH corresponded to tibia and knee ligament injuries, while higher BCH corresponded to femur injury, based on an investigation of the PCDS database. Using frequency analysis with the Medical University of Hanover database, Snedeker et al. (2003) also found that shorter pedestrians were more likely to sustain upper leg (including pelvis) fractures compared to taller pedestrians. In the current study, we did not distinguish different types of lower extremity injury. Therefore, it was not surprising that BCH or pedestrian height was not found to be a statistically significant parameter for predicting pedestrian lower extremity AIS 2+ injury odds.

BL

The BL is considered an important parameter in the EEVC lower extremity test protocol. However, we did not find this parameter to be significantly associated with injury odds to lower extremity.

FTTPH

Generally speaking, pedestrians impacted by a more vertically orientated frontal structure (e.g. the hood leading edge and windshield of SUV in Figure 2) would sustain higher impact forces for a shorter duration. In contrast, a more horizontally orientated frontal structure (such as the hood of passenger car in Figure 2) would allow some sliding on the hood surface to prolong the impact duration and reduce the impact force. Therefore, one can speculate that the more horizontally orientated hood area is less injurious than the vertically orientated hood leading edge and windshield to the torso and head of pedestrians.

Figure 2.

Lateral view of different vehicle body type

Vehicles with higher FTTPH are generally more vertically orientated and hence more likely to impact the pedestrian torso with greater force, especially for shorter pedestrians (Figure 3). Our data also showed that more AIS 3+ torso injuries were significantly associated with higher FTTPH. Therefore, reducing the FTTPH may be an important approach for future vehicle design when attempting to mitigate pedestrian torso injuries.

Figure 3.

Ford Explorer 2008 model with pedestrian scaled to 5^th percentile female (left) and 50^th percentile male (right) [Chinese Art World, 2008; Ford, 2008]

RHOD

For vehicles with medium RHOD (171–200 cm), the pedestrian head is more likely to impact the central area of the hood. In contrast, the head is more likely to strike the windshield and its boundary in vehicles with shorter RHOD. As described earlier, a more vertically orientated structure is more likely to generate larger contact force. Therefore, the windshield may be more hazardous to the pedestrian head than the hood central area. This assertion is supported by our finding that vehicles with shorter RHOD had significantly higher AIS 3+ head injury odds compared to those with mediuma RHOD. Kerrigan et al. (2005a) conducted full-scale pedestrian impacts on two PMHS and three Polar-II dummy tests using a late-model SUV at 40 km/h. The same group also conducted similar impacts on three PMHS and three Polar-II dummy tests using a late-model sedan at the same impact speed (Kerrigan et al. 2005b). They concluded that head resultant velocity at the time of strike was higher in impact with the sedan compared to impact with the SUV. Because RHOD was generally shorter in sedans than SUVs and our study indicates that a shorter RHOD had higher head injury odds, results from this study seem to be consistent with the studies reported by Kerrigan et al. (2005a and b).

Injury predictors for head, torso, and lower extremity injuries

Significant predictors for pedestrian MAIS 3+, AIS 3+ head, AIS 3+ torso, and AIS 2+ lower extremity injury odds are shown in Table 15. The authors believe these findings would be very useful for future vehicle design to mitigate pedestrian injuries.

Table 15.

Significant predictors for injuries in different body regions

Predictor	MAIS 3+ injury	AIS 3+ head injury	AIS 3+ torso injury	AIS 2+ lower extremity injury
Impact speed	S	S	S	S
Age	-	-	S	S
Gender	-	-	-	-
Height	-	S	-	-
Weight	S	-	-	S
BCH	-	-	-	-
BL	-	S	-	-
FTTPH	-	-	S	-
RHOD	-	S	-	-

S indicates statistically significant.

Limitations and future work

The PCDS database is 10 years old and the fleet has changed since then. At present, no data are publicly available to examine the effect of these fleet changes. Nevertheless, our study concludes that pedestrian head injury risks are generally higher when impacted by a sedan compared to an SUV at the same impact severity. This is consistent with results reported by Kerrigan et al. (2005a and b) who conducted PMHS and dummy tests using an SUV and a sedan of late model year. Therefore, we believe that even though the fleet has changed, the basic trend obtained is still valid. Furthermore, one of the purposes of this study was to obtain general directions for designing pedestrian friendly vehicles, but not to design a specific vehicle. In other words, we are interested in determining which direction the safety engineers should follow to improve pedestrian safety. Still, until a new database based on late-model vehicles is available, effects related to the fleet change can only be speculated upon.

Before changing vehicle factors (such as FTTPH and RHOD) according to those suggested in this study, readers must note that any design alteration will produce some unintended consequences. Retrospective examination of the association between vehicle factors and pedestrian injuries do not convey the injury etiology and hence cannot be used directly to improve pedestrian safety. For example, changing FTTPH and RHOD may alter the visibility of the vehicle and hence could affect the incidence of pedestrian crashes not possible to detect using the current methods. More studies need to be conducted in the future to fully prove the validity of the suggested changes in vehicle factors.

As stated earlier, this study did not distinguish between different types of injuries (such as skull fracture vs. brain injury for the head), nor did we differentiate different injury types within the same body region in detail (such as tibia vs. femur fractures for the lower extremity). For future studies, it would be more beneficial to study each injury type separately if sufficient number of cases can be identified to investigate the injury mechanisms in more detail. However, the current study was only focused on finding significant predictors for pedestrian injury odds, giving a basis for more detailed injury mechanism studies in later days.

Ground induced injuries were not considered in the current study, since our results showed that pedestrians were not as frequently and severely injured by impacts to the ground as compared to vehicle impacts during vehicle-to-pedestrian crashes. However, one should be aware that the coded injury sources in the PCDS database are assessed by investigators based on available physical evidence. As a result, it is rather subjective and difficult to ensure its accuracy. Otte et al. (2001) reported that 65% of pedestrians in vehicle-to-pedestrian crashes sustained at least one injury induced by the ground. More studies need to be conducted to investigate whether vehicle is the major injury source for pedestrians.

In the current study, vehicle parameters, such as FTTPH and RHOD, were found to play an important role in determining the pedestrian injury risk. In future studies, finite element models could be used to investigate their effects in greater detail to assess the effects of vehicle parameters on pedestrian injury risk.

CONCLUSIONS

In this study, the NASS PCDS database was used to investigate upright standing pedestrian MAIS 3+, AIS 3+ head, AIS 3+ torso, and AIS 2+ lower extremity injury ORs for a crash factor (impact speed), pedestrian factors (age, gender, height and weight), and vehicle factors (BCH, BL, FTTPH, RHOD) in vehicle-to-pedestrian frontal crashes using a logistic regression method. For each regression analysis, all the other factors were adjusted to evaluate each factor more objectively and without bias.

The results show that impact speed is a statistically significant predictor for AIS 3+ head, AIS 3+ torso, and AIS 2+ lower extremity injury odds. Pedestrian age is also significant for AIS 3+ torso and AIS 2+ lower extremity injury odds, but not for AIS 3+ head injury odds. FTTPH is a statistically significant predictor for pedestrian AIS 3+ torso injury odds. Vehicles with higher FTTPH and more vertical frontal structures are aggressive to pedestrians, especially regarding injuries to the torso. A very short RHOD would be more likely to lead the pedestrian to impact the windshield and windshield frame, thus increasing the head injury risk.