Analysis of crash parameters and driver characteristics associated with lower limb injury

Accident Analysis and Prevention 83 (2015) 37–46

Contents lists available at ScienceDirect

Accident Analysis and Prevention journal homepage: www.elsevier.com/locate/aap

Analysis of crash parameters and driver characteristics associated with lower limb injury Xin Yea,* ,1, Gerald Poplina,1, Dipan Bosea,1, Aaron Forbesb,2 , Shepard Hurwitzb,2 , Greg Shawa,1, Jeff Crandalla,1 a b

Center for Applied Biomechanics, University of Virginia, Charlottesville, VA, USA School of Medicine, University of North Carolina, Chapel Hill, NC, USA

A R T I C L E I N F O

A B S T R A C T

Article history: Received 15 September 2014 Received in revised form 22 May 2015 Accepted 30 June 2015 Available online 18 July 2015

This study aims to investigate changes in frequency, risk, and patterns of lower limb injuries due to vehicle and occupant parameters as a function of vehicle model year. From the National Automotive Sampling System-Crashworthiness Data System, 10,988 observations were sampled and analyzed, representing 4.7 million belted drivers involved in frontal crashes for the years 1998–2010. A logistic regression model was developed to understand the association of sustaining knee and below knee lower limb injuries of moderate or greater severity with motor vehicle crash characteristics such as vehicle type and model years, toepan and instrument panel intrusions in addition to the occupant’s age, gender, height and weight. Toepan intrusion greater than 2 cm was significantly associated with an increased likelihood of injury (odds ratio: 9.10, 95% confidence interval 1.82–45.42). Females sustained a higher likelihood of distal lower limb injuries (OR: 6.83, 1.56–29.93) as compared to males. Increased mass of the driver was also found to have a positive association with injury (OR: 1.04, 1.02–1.06), while age and height were not associated with injury likelihood. Relative to passenger cars, vans exhibited a protective effect against sustaining lower limb injury (OR: 0.24, 0.07–0.78), whereas no association was shown for light trucks (OR: 1.31, 0.69–2.49) or SUVs (OR: 0.76, 0.28–2.02). To examine whether current crash testing results are representative of real-world NASS-CDS findings, data from frontal offset crash tests performed by the Insurance Institute for Highway Safety (IIHS) were examined. IIHS data indicated a decreasing trend in vehicle foot well and toepan intrusion, foot accelerations, tibia axial forces and tibia index in relation to increasing vehicle model year between the year 1995 and 2013. Over 90% of vehicles received the highest IIHS rating, with steady improvement from the upper and lower tibia index, tibia axial force and the resultant foot acceleration considering both left and right extremities. Passenger cars received the highest rating followed by SUVs and light trucks, while vans attained the lowest rating. These results demonstrate that while there has been steady improvement in vehicle crash test performance, below-knee lower extremity injuries remain the most common AIS 2+ injury in real-world frontal crashes. ã 2015 Elsevier Ltd. All rights reserved.

Keywords: Lower extremity injury NASS-CDS Logistic regression

1. Introduction The most common injuries sustained in frontal motor vehicle crashes involve the lower extremities. The socioeconomic burden

* Corresponding author. Fax: +1 4342963453. E-mail addresses: [email protected] (X. Ye), [email protected] (G. Poplin), [email protected] (D. Bose), [email protected] (A. Forbes), [email protected] (S. Hurwitz), [email protected] (G. Shaw), [email protected] (J. Crandall). 1 Permanent address: 4040 Lewis and Clark Drive, Charlottesville, VA, 22911, USA. 2 Permanent address: 400 Silver Cedar Court, Chapel Hill, NC, 27514, USA. http://dx.doi.org/10.1016/j.aap.2015.06.013 0001-4575/ ã 2015 Elsevier Ltd. All rights reserved.

of these injuries is high, due to protracted recovery time and related health care costs. Several studies have examined potential occupant and vehicle factors for increasing the risk of lower extremity injuries. Lower extremity injuries were observed more frequently in frontal collisions compared to other crash modes, and seatbelts were not effective restraints in preventing these injuries (Dischinger, 1996; Dischinger et al., 2005). In addition, sex was noted to affect lower limb injury risk, with female drivers having a higher likelihood of ankle and foot fractures than male drivers when height was explicitly considered as a covariate, as height has also been shown to correlate with injury risk (Crandall et al., 1996). In a field study using Folksam Insurance data over the years 1985–

38

X. Ye et al. / Accident Analysis and Prevention 83 (2015) 37–46

1991, age was found to be less significant of a predictor for Abbreviated Injury Scale (AIS) 2 and 3+ foot/ankle injuries than seating position and impact location (Parenteau et al., 1996). Rudd (2009) investigated lower limb injury risk and associated factors, and compared the risk and severity of injury based on crash characteristics and vehicle specifications from NASS-CDS data 1994–2007. Results indicated that the incidence of foot and ankle injury increased despite structural improvements to vehicles. Austin et al. (2012) examined NASS-CDS data from 1997 to 2009 to assess whether intrusion causes lower extremity injuries. For Maximum AIS (MAIS) 2 and greater lower limb injuries, 68.0% of the injury sources originated from the floor including the toepan, while foot controls including brakes accounted for 25.2%, instrument panel and knee bolster for 5.7%, and other injury sources resulted in the remaining 1.1% of injuries. While toepan intrusion is often considered to be a significant contributor in lower extremity injury risk, an updated analysis of the influence of recent vehicle changes to lower limb injuries has yet to have been established. This study aims to investigate the prevalence and patterns of lower limb injuries in relation to vehicle and driver parameters. In addition, this study evaluates and contextualizes results from realworld lower limb injuries against improvements made in standardized crash testing outcomes over a similar period of time. 2. Methodology 2.1. NASS-CDS data analysis The National Automotive Sampling System-Crashworthiness Data System (NASS-CDS) is a nationally representative database maintained by the U.S. National Highway Traffic Safety Administration for the assessment of traffic safety and evaluation of crashresultant injuries. In the current study, the NASS-CDS database was queried to obtain sampled data on the following inclusion criteria: frontal crashes during the calendar years 1998–2010 with the principal direction of force (PDOF) between 11 and 1 o’clock (PDOF = 30 ), and no resulting rollover event, ejection or fire. All case occupants were drivers with age greater than 16 years and who were properly belted. Further classification of frontal crashes was performed based on Collision Deformation Code (CDC), and vehicle class. Vehicle model years were restricted to 1998–2011, since significant changes in airbag systems occurred in 1998 with depowering of inflator and reduction of volume (Kahane, 2006). For survey analysis and survey frequency, weighting factors were

assigned a range between zero and one million. Cases with survey weighting factors greater than one million were considered outliers and excluded from the analysis. In the examination of demographic data, height, weight and sex were considered as potential injury predictors. Body Mass Index (BMI) was calculated and categorized based on standard ranges: underweight (BMI  18.5), normal weight (18.6  BMI  24.9), overweight (25  BMI  29.9) and obese (BMI  30) (James et al., 2001). Vehicles were categorized into four body types, namely passenger cars, SUVs, light vans, and pickup trucks (<5000 kg curb weight). Injury outcomes were coded based on the AIS (AAAM, 1990). The occupant’s body was divided into seven anatomical regions based on the AIS coding standards: (1) head/face, (2) neck, (3) thorax, (4) abdomen, (5) spine, (6) upper extremity, and (7) lower extremity. Pelvis contents including soft tissues, vessels, etc. were included in the abdominal region, while musculoskeletal structures (e.g., sacroilium, pubic symphysis, etc.) were included in the lower extremity category. For each body region, selection criteria limited analysis to moderate or more severe injuries (i.e., AIS 2+). Maximum AIS was documented for each anatomical region by counting the most severe injury and excluding concurrent injuries of the same AIS level. The classification of lower extremity region was applied using a functional approach similar to that of Klinich and Schneider (2003), with the lower limb been categorized into five sub-regions: hip, femur shaft, knee, tibia/fibula shaft, and ankle/foot (Table 1). The functional classification was then mapped with the AIS and queried from the NASS-CDS database. Lower extremity injury codes pertaining to skin, blood vessels, or the nerves were excluded in this anatomical categorization. A multi-variate logistic regression model was developed for analyzing the associated odds ratios of knee and below knee lower limb injuries in the specified frontal crash conditions(SAS, 2004). NASS-CDS measured the driver toepan and instrument panel intrusion in magnitude ranges (i.e., 2, 3–8, 9–15, 16–30, 31–46, 46–61, 61 cm). A preliminary analysis showed the majority of the intrusions occur at lower level with 97.14% of the vehicle’s toepan intrusion and 97.56% of vehicle’s instrument panel intrusion occurring at less than or equal to 2 cm. As a result, very few observations were included at levels of large intrusion. Therefore, a dichotomous variable was defined to quantify the intrusion level, with less or equal to 2 cm as little/no intrusion or larger than 2 cm as moderate/severe intrusion. Delta-V was included in the regression model as a continuous variable for the measurement

Table 1 Categorization of AIS 2+ lower limb injuries. Category

AIS 2+ lower limb injuries

Hip

Femoral head/neck fractures Hip dislocation NFS

Hip NFS laceration into joint

Femur shaft

Femur fractures Femoral vessel injury

Femur fracture Femur supracondylar

Knee (fractures, soft tissue injuries)

Femoral condyle fracture Collateral/cruciate ligament laceration Patellar tendon Patella fracture Popliteal vein NFS

Tibia condyle fracture Joint laceration Dislocation Sprain Popliteal artery NFS

Tibia/fibula shaft

Tibia fracture Fibula fracture

Tibia open/comminuted fracture

Ankle/foot

Dislocation Foot fracture (talus, calcaneus, medial mal., posterior mal.,toe)

Fibula lateral malleolus fracture Achilles tendon Ligament laceration

X. Ye et al. / Accident Analysis and Prevention 83 (2015) 37–46

2.2. IIHS test data review A previous study examining the German In-Depth Accident Study database (GIDAS) indicated that there was a higher risk of foot and ankle injuries existed for pre-Euro NCAP (model year 1997 and older) vehicles than for vehicles produced after the introduction of Euro NCAP testing (Schubert et al., 2010). In current study, data from moderate frontal offset crash tests performed by the Insurance Institute for Highway Safety (IIHS) were queried to assess how well current crash test results represent the real-world lower extremity injures exhibited from the NASS-CDS analysis. Vehicle body types were divided into four categories: passenger cars, SUVs, pickup trucks and minivans. Vehicle model years from the IIHS test data ranged from 1995 to 2013. Vehicle body category was defined based on IIHS vehicle classification scheme, and the final query resulted in the inclusion of 289 passenger cars, 136 SUVs, 29 pickup trucks and 30 Mini-vans. Model year was extrapolated in time to reflect the vehicle fleet of each calendar year as an approach to increase the amount of usable data, as well as to trace backward and to predict forward the vehicle performance of the same model (IIHS, 2012). Measurement data from the vehicle included intrusion of instrument panel, toepan, footrest and brake pedal. Axial force and bending moment at lower and upper, left and right tibia were also included in the query, as well as resultant acceleration from both left and right feet as measurement from the tests dummy for the systematic examination of the crash tests. Tibia Index was calculated as an injury tolerance criterion for combined bending and axial compressive loading of the tibia during IIHS tests. A correction formula was applied to calculate the upper tibia bending moment, since the curvature of the Hybrid III dummy leg would result in overestimation of the upper tibia bending moments (Zuby et al., 2001). IIHS applied a categorized rating to evaluate crash test performance,

with tibia index <0.8 referring to good; 0.8–1.0 acceptable; 1.0– 1.2 marginal; and >1.2 referring to poor. Higher tibia index was associated with a higher risk of sustaining tibia shaft injuries (Kuppa et al., 2001). In addition, we developed an overall rating for each vehicle, which was based on the lowest composite score obtained from the values among the upper and lower tibia index, tibia axial force and the resultant foot acceleration in both left and right extremities. 3. Results 3.1. NASS-CDS data analysis Based on the NASS-CDS query, 10,988 observations (i.e., drivers) for the years 1998–2010 were queried to represent a weighted number of 4,684,596 belted drivers involved in frontal crashes. The queried 10,988 drivers represented a total number of 10,256 crash cases with 10,988 vehicles involved. Of the 10,988 drivers, 2747 drivers (weighted: 371,272) sustained at least one AIS 2+ whole-body injury. 907 sampled observations, representing a weighted number of 171,958 drivers, sustained at least one AIS 2+ lower limb injury (Table 2). The distribution of injuries by body regions for drivers who sustained at least one AIS 2+ injury (N = 2747,weighted: 371,272) in frontal crashes revealed that lower extremity injuries were the most frequent injured body region, accounting for 46.32% of all the injuries (Fig. 1). Upper extremity injuries accounted for 17.88% of the total distribution, followed by thorax (13.24%), head and face (11.94%), spine (6.03%), abdomen (4.52%) and neck (0.07%). Coinciding with improvements to vehicle frontal crashworthiness, the likelihood of AIS 2+ whole-body injuries has declined over the past decade, compared with the data from previous study (Kuppa et al., 2001). Despite the overall reductions in injury, AIS 2+ injuries to the lower extremity remain the most frequently injured body region (3.67%) for all drivers included in this study, followed by head/face (0.95%) (Fig. 2). The vehicle distribution of Delta-V as an indicator of crash severity was fully examined (Fig. 3). Of the 10,988 vehicles (which also represents the same number of drivers), 7839 had known

50 45 Distribuon Percentage (%)

of crash severity. Delta-V was defined as the change of vehicle velocity during approaching period, which is the time between initial contact and occurrence of maximum crush at the interface (National Automotive Sampling System-Crashworthiness Data System, 2010). Vehicle model years ranged from 1998 to 2011, and were assessed as a categorical term with 1998 as the reference year to account for any chronological changes in vehicle structure. Vehicle type was included as a categorical variable to assess whether SUVs, vans, light trucks would be safer for drivers in frontal crashes compared to passenger cars. Age, gender, height and weight were also considered as continuous and independent variables. BMI was excluded from the analyses as individual height and weight were deemed more appropriate independent variables to adjust for, given the focus on lower extremity injuries. PDOF was also excluded in the regression model as it was already embedded as a case selection criterion. A correlation check was computed between Delta-V, toepan intrusion and instrument panel intrusion to assure no co-linearity between the variables and each variable could be assessed independently. While statistically significant, the Pearson correlation coefficients were relatively small in magnitude, suggesting that these three parameters could be considered mutually independent (Pearson correlation coefficients: 0.21, 0.25 and 0.29, respectively).

39

40 35 30 25 20 15 10 5 0 head/face

neck

thorax

abdomen

spine

upper lower extremity extremity

Fig. 1. Distribution of injured body regions for drivers with at least one AIS 2+ injury in frontal crashes.

Table 2 Data included from NASS-CDS dataset. Frontal Crashes

Raw number

Weighted frequency

Drivers Drivers with lower extremity AIS 2+ injury Drivers with knee and below knee AIS 2+ injury Drivers with hind foot or ankle AIS 2+ injury

10,988 907 747 312

4,684,596 171,958 162,420 49,028

40

X. Ye et al. / Accident Analysis and Prevention 83 (2015) 37–46

907 drivers involved in frontal crashes, representing 171,958 drivers when accounting for sampling weights. When restricted to the knee and below, 747 drivers (weighted N = 162,420) exhibited AIS 2 + injuries in the region, and 312 drivers (weighted N = 49,028) sustained AIS 2+ hind foot injuries when further restricted to the lower limb. The majority of the crashes occurred at relatively low Delta-V ranges (i.e., below 30 km/h). For all frontal crashes, 85.7% occurred within the 0–30 km/h range, while approximately 62.4% of AIS 2+ lower extremity injuries occurred in crashes with vehicle Delta-V below 30 km/h (Fig. 3). Compared to all frontal crashes, there was a substantial increase in the median Delta-V when the analysis was limited to crashes resulting in moderate or greater hind-foot injuries. The median Delta-V for all frontal crashes and crashes with hind-foot AIS 2+ injuries were 18.14 km/h and 27.79 km/h, respectively. Few hindfoot injury cases occurred in crashes of 0–15 km/h, while 51.7% of the total occurred within a Delta-V range of 16–30 km/h and 30.7% of injuries within the relatively low Delta-V range of 31–45 km/h. Driver information and vehicle attributes were queried for all drivers, including those with and without AIS 2+ lower limb injuries (Table 3). The bodyweight of all queried drivers followed a trend similar to a normal distribution, with overweight and obese consisting of 31.73% and 17.23%, respectively. A higher proportion of obesity in BMI category was noticed for drivers with lower limb injuries. To better describe injuries pertaining to the lower extremity, a breakdown of AIS 2+ injury distribution was conducted among the hip, femur shaft, knee, tibia/fibula shaft and ankle/foot (Fig. 4). The total number of 907 drivers with one or multiple AIS 2+ lower extremity injuries was included in the query, representing 171,958 weighted drivers in frontal crashes. The proportion of each lower extremity injury showed that ankle/foot injuries were the leading lower extremity injury region. Ankle/foot AIS 2+ injuries accounted for 60.22% of all AIS 2+ lower extremity injuries, regardless of other concomitant injuries that may occur. Of all the

5.0

Distribuon Percentage (%)

4.5 4.0 3.5 3.0 2.5 2.0 1.5 1.0 0.5 0.0 head/face

neck

thorax

abdomen

spine

upper lower extremity extremity

Fig. 2. Distribution percentage for AIS 2+ injuries of all drivers in frontal crashes.

Fig. 3. Distribution of drivers by vehicle Delta-V.

Delta-V. At least one AIS 2+ lower limb injury was documented in

Table 3 Driver and vehicle descriptive statistics summary of NASS-CDS dataset, 1998–2010. Total Sample numbers Weighted frequency

With lower limb AIS 2+ injuries

Without lower limb AIS 2+ injuries

10,988 907 4,684,596 171,958 Mean or percentage (95% confidence interval) 37 (36.69–38.20) 42 (39.76–44.01)

10,081 4,512,638

48.85 (42.76–54.95) 51.15 (45.05–57.24) 77.24 (75.94–78.55) 171 (171.30–71.96)

23.46 (1.10–45.83) 76.54 (54.17–98.90) 84.43 (80.29–88.58) 167 (162.60–171.43)

49.82 (44.92–54.72) 50.18 (45.28–55.08) 76.95 (75.58–78.32) 172 (171.40–172.23)

14.96 (12.76–17.15) 36.09 (31.99–40.18) 31.73 (28.36–35.10) 17.23 (15.39–19.06)

5.69 (0.07–11.32) 19.46 (2.79–36.12) 35.89 (13.41–58.37) 38.96 (33.85–44.07)

15.38 (12.77–17.99) 36.90 (33.36–40.43) 31.25 (28.31–34.20) 16.47 (14.29–18.65)

63.58 (61.02–66.14) 16.99 (14.71–19.28) 6.53 (4.86–8.20) 12.90 (9.98–15.82) 3.71 (3.53–3.89) Weighted frequency (percentage)

60.97 (49.16–72.78) 21.69 (16.84–26.53) 3.60 (0.00–7.27) 13.74 (2.77–24.71) 3.37 (2.19–4.56)

63.68 (61.39–65.97) 16.81 (14.51–19.12) 6.64 (5.08–8.19) 12.86 (10.19–15.54) 3.72 (3.55–3.90)

Toepan intrusion 2 cm >2 cm

4,550,480 134,116

125,623 (2.76%) 46,335 (34.55%)

4,424,857 (97.24%) 87,781 (65.45%)

Instrument panel intrusion 2 cm >2 cm

4,570,136 114,460

146,696 (3.21%) 25,262 (22.07%)

4,423,440 (96.79%) 89,198 (77.93%)

Age (year) Gender Male (%) Female (%) Weight (kg) Height (cm) BMI (%) Underweight (18.5) Normal (18.6–24.9) Overweight (25–29.9) Obese (30) Vehicle type (%) Passenger cars SUVs Vans Trucks, pickups Vehicle age (years)

37 (36.41–38.14)

X. Ye et al. / Accident Analysis and Prevention 83 (2015) 37–46

toepan intrusion throughout vehicle model years from 1995 to 2013. From the calculation of tibia index, all vehicle types reached good or acceptable ratings by the year 2012, with tibia index ranging from 0.2 to 0.5, with less than 2% probability of AIS 2+ leg shaft fracture. Over 90% of vehicles reached the highest IIHS rating (i.e., good), which is based on the lowest composite score obtained among the upper and lower tibia index, tibia axial force and the resultant foot acceleration in both left and right extremities. As for ratings of vehicle types, passenger cars achieved the highest rating (i.e. good) followed by SUVs and light trucks, while vans attained the lowest score.

80.0 70.0 60.0 Percentage(%)

41

50.0 40.0 30.0 20.0 10.0 0.0 hip

femur sha

knee

bia/fibula sha

ankle/foot

Fig. 4. Distribution of lower limb body regions for drivers sustaining at least one AIS 2+ lower limb injury in frontal crashes.

drivers who sustained at least one lower limb injury, 36.62% suffered knee injuries as the second most frequently injured lower limb region. In addition, the results of percentage distribution of AIS 2+ lower limb injuries for all drivers indicated that ankle/foot injuries still remained the predominant lower limb injury in frontal crashes (Fig. 5). Results from the regression analysis indicated that toepan intrusion greater than 2 cm was significantly associated with AIS 2 + knee and below knee injury (OR: 9.10, CI 1.82–45.42). In addition, females had a higher likelihood of sustaining lower limb injuries (OR: 6.83, CI 1.56–29.93) compared to males, and increased body weight was found to be related to injury (OR: 1.04, CI 1.02–1.06). Occupant age and height were not significantly associated with injury, with ORs of 1.02 (CI 1.00–1.06) and 0.99 (CI 0.96–1.01), respectively. Relative to passenger cars, vans exhibited a significant decrease in sustaining lower limb injury (OR: 0.24, CI 0.07–0.78), followed by SUVs though the result was non-significant (OR: 0.76, CI 0.28–2.02), whereas light trucks showed no such protective association (OR: 1.31, CI 0.69–2.49) (Table 4). 3.2. IIHS test data review From the IIHS tests data, a significant decrease in intrusion was shown in vehicle foot well and toepan intrusion, foot accelerations, and tibia axial forces in relation to increasing vehicle model year. Fig. 6 demonstrates an example of the decreasing trend in left

3.50 3.00

Percentage (%)

2.50 2.00 1.50 1.00 0.50 0.00 hip

femur sha

knee

bia/fibula sha ankle/foot

Fig. 5. Percentage distribution of AIS 2+ injury for lower extremity regions.

4. Discussion The distribution of injuries by body regions of drivers who sustained at least one AIS 2+ injuries was compared with previous results. With a comparable selection criterion, these results indicated an increased proportion of lower extremity injuries (46.3%) as compared to the 28.2% of lower extremity reported by Kuppa et al. (2001), which assessed the same crash types from years 1993 to 1999. While the current study indicated similar distribution of injury frequency and proportion; the likelihood of AIS2+ whole-body injuries has declined correspondingly over the past decade. The percentage for head/face injuries dropped from previous 3.58% to current 0.95%, while lower extremity remained the most frequent injured body region with a percentage distribution of 3.67%, compared to the previous 4.37%. Current query was also expanded to frontal occupants (i.e., drivers and frontal seat passengers) as to examine whether any change of injury distribution or incidence percentage might occur. A total number of 13,387 frontal occupants were included, which represented 11,176 vehicles involved in a total number of 10,741 cases. Marginal change occurred in the results when the target population was expanded to frontal occupants, but the relative ranking of injured body regions remained consistent. Logistic regression results from this study demonstrated that gender (i.e., females), increased body weight, and toepan intrusion were the significant factors that increase the likelihood of moderate or greater lower limb injury. Consistent with previous findings (Thomas and Bradford, 1995; Crandall et al., 1996; Austin, 2012), results from the present study indicated female drivers had a significantly higher likelihood of sustaining lower extremity injuries than males. Adjusting for other vehicle, crash and occupant characteristics, injured drivers (i.e., knee and below knee AIS 2+) were 6.83 times more likely to be female. Several reasons were postulated for the increased odds ratio of lower extremity fractures in women. Female drivers may be seated in a more forward track position than males, and therefore more likely to be affected by instrument panel or toepan intrusions. Crandall et al. (1996) suggested that the increased injury risk for shorter drivers was associated with the increased gap between ankle heels and floor pan. Footwear may also contribute to the gender difference of lower extremity injury, as 20% of variation in leg loads was dependent on the style of shoe and an increased likelihood of ankle instability was associated with women’s high heel shoes. The height difference between male and female drivers could also result in various pedal interactions, which is also a contributing factor associated with lower limb injuries(Thomas et al., 1995). Body weight was another statistically significant factor positively related to lower limb injury. The results indicated that an increase of 1 kilogram in mass was associated with a 4% increase in odds of lower limb AIS 2+ injury. Previous study also indicated that obesity was associated with the increase in lower limb injury severity (i.e., MAIS) (Arbabi et al., 2003). The association of occupant age with lower extremity injury was also examined.

42

X. Ye et al. / Accident Analysis and Prevention 83 (2015) 37–46

Table 4 Results from multi-variable logistic regression model. Independent variables

OR estimates

95% Wald confidence limits

Occupant age (per year) Gender Male Female Height (per cm) Body weight (per kg) Model Year (per year) Toepan intrusion 2 cm >2 cm Instrument panel intrusion 2 cm >2 cm Delta-V (per km/h) Vehicle type Passenger car SUV Van Light truck

1.02

1.00

1.06

0.0602

Reference 6.83 0.99 1.04 0.96

1.56 0.96 1.02 0.87

29.93 1.01 1.06 1.06

0.0108* 0.3095 0.0002* 0.4369

Reference 9.10

1.82

45.42

0.0071*

Reference 1.06 1.03

0.61 0.98

1.83 1.10

0.8429 0.1840

Reference 0.76 0.24 1.31

0.28 0.07 0.69

2.02 0.78 2.49

0.5788 0.0176* 0.4158

P-value

Raw observations: 10,988; weighted frequency: 4,684,596. An asterisk (*) indicated statistical significance in p-value.

While the odds ratio resulting from the multivariate regression suggested a positive association with age (OR 1.02), the result was only marginally significant (p-value = 0.0602). This finding was consistent with Parenteau et al. (1996) who found that age was not as significant a factor as seating position and crash position; however several related studies have demonstrated significant increased likelihood of lower extremity injury with increasing age (Crandall and Martin, 1997; Austin, 2012). The difference in significance was likely due to the fact that age distribution of current data followed a left skewed distribution, and the increased injury risk of elder generation due to osteoporosis and other factors may be underestimated. The results based on logistic regression model in current study suggested that height (OR: 0.99, CI: 0.96–1.01) was not associated with knee and below knee injury for drivers in frontal crashes, even though a slight decrease in odds ratio was observed with increasing height. Dischinger et al. (2005) has examined trauma data and police crash reports and found that there was an inverse relation between driver height and incidence of lower limb fractures. Crandall et al. (1996) also found that foot and ankle injury risk decreased with increasing of driver height. However, Rudd (2009) found contradictory results and suggested that

shorter occupants sustained more foot/ankle injuries than their taller counterparts based on overall injury distribution. Intrusion of the toepan was found to be a significant predictor of injury. Austin et al. (2012) found that the odds of sustaining lower limb injury with the occurrence of floor or toe pan intrusion were twice the odds of experiencing a similar injury without intrusion. However, reducing intrusion may only be a partial solution for reductions in lower extremity injuries, since 50% of drivers with AIS 2+ lower limb injury had virtually no residual intrusion. Ye et al. (2014) found axial loading to be the predominant injury mechanism of hind-foot injuries, regardless of the intrusion magnitude, which may suggest a more causal relationship with the deceleration of the vehicle components than the magnitude of deformation. Given large odds ratio for intrusion greater than 2 cm (OR: 9.10, CI 1.82–45.42), intrusion is clearly a sufficient mechanism of injury but not a necessary one given the large number of cases without large intrusion levels. With the improvement of vehicle stiffness, many crashes produce little or no intrusion. This is supported by the fact that the majority of drivers (73%) with lower limb AIS 2+ injuries were involved in crashes with less than 2 cm of intrusion. Similarly, 85.3% of drivers with AIS 2+ injuries included in this study had little or no intrusion to the instrument

Fig. 6. Left toepan intrusion (resultant) by vehicle category and model year from IIHS crash test data.

X. Ye et al. / Accident Analysis and Prevention 83 (2015) 37–46

panel. In addition to intrusion effects, braking can affect the foot/ ankle kinematics since pedal interactions have been shown to increase the likelihood of foot injuries (Pilkey et al., 1994; Crandall et al., 1996). In the current study, 53.3% of the queried drivers were braking as an attempted avoidance maneuver (derived from the NASS-CDS variable “maneuver”). Previous studies have found that the driver’s right pedal foot has the highest risk (Crandall et al., 1996; Assal et al., 2002). Given the scope of current study and the limitations within NASS-CDS data collection, the difference in injury risk between left and right aspects of the lower extremity was not feasible. Given the nature of NASS-CDS with the overwhelming majority of crashes exists at lower speeds (85.7% 30 km/h) and low levels of intrusion (97.1% 2 cm), both toepan and instrument panel intrusion were defined as dichotomous terms. Despite the potential lack of precision in the 9.10 odds ratio relating to toepan intrusion and injury likelihood (as indicated by large confidence intervals), there continues to be an increased likelihood of lower limb injury with increased toepan intrusion (i.e., >2 cm). More severe crashes resulted in a higher proportion of AIS 2+ injuries relative to the distribution of intrusion levels, significantly increasing the injury risk, as measured by the odds ratio. An analysis of lower limb injury in the NASS-CDS data by vehicle type revealed that crashes with vans and SUVs were less likely to cause a lower limb injury relative to passenger cars, while crashes of pickup trucks had a higher likelihood sustaining lower limb injuries. While crashworthiness by vehicle type may play a role, a more detailed characterization in terms of occupant compartment geometry could be considered in future work. The results from the IIHS tests demonstrated a steady improvement in vehicle crash test performance, most likely attributable to the increase in vehicle stiffness. The average dynamic stiffness of vehicles has increased by 34%, beginning at 584 N/mm in the early 1980s to 781 N/mm for vehicles in 2003 (Swanson et al., 2003). The apparent injury prediction for the lower extremity as a function of vehicle type for IIHS crash tests differed from the NASS-CDS findings. From the IIHS tests data, passenger cars received the highest rating followed by SUVs and light trucks, while vans attained the lowest score. In contrast, we found that front seat occupants in vans had the lowest risk of lower limb injury according to the NASS-CDS data. This difference in vehicle type ranking may be due to a combination of the weighting factors used in analysis of specific injury predictors and differences in occupant position across the vehicle types. In addition, the composite metric of injury predictors used in this study provided equal weighting among the components. Furthermore, injury predictor for the tibia index was dominated by the moment whereas real-world lower limb injuries predominantly result from axial loading (Ye et al., 2014). One reason for this disparity is that tibia index was formulated at the mid-diaphysis tolerance whereas most leg fractures of occupants occur in the distal-third (Ivarsson et al., 2008). This study elucidated the disconnect between crash tests results and real-world crash injury potentials, as frontal crash tests were performed at a high speed (i.e., 64 km/h), while the vast majority of lower extremity injuries occur in crashes at much lower speeds. There are some limitations in the present study. First, vehicle model year was restricted to 1998 and newer to coincide with changes in airbag technologies. Had vehicle model year restrictions been relaxed to include vehicles of 10 years from the crash date, the magnitude of some risk estimates would be reduced; however, the direction and interpretation of results should remain unchanged. Another potential limitation involved the lack of specificity in categorizing vehicle type. Utilizing NASS-CDS

43

criteria, vehicles were classified into one of four general categories: SUV, truck, van, or passenger vehicles. NASS-CDS does permit additional descriptions to generate more detailed classifications, which could provide more precise rankings for vehicle types. In addition, inherent limitations exist with any survey-based data system like NASS-CDS. While each observation within NASS-CDS is provided a weighting factor for improved generalization at a national level, selection criteria may produce an uneven distribution of weights although the present study did ensure no cases with outlying weights. Another limitation is that Delta-V information is not available from NASS-CDS for a substantial number of cases, creating the potential for selection bias as less severe crashes, as described by median injury severity and intrusion levels, may be more likely to be missing Delta-V. In cases where Delta-V was known, 4.6% of drivers sustained AIS 2+ injuries, whereas 2% of drivers with missing Delta-V data had sustained an AIS 2+ injury. An assessment of the crash, vehicle and occupant characteristics between cases with and without documented Delta-V revealed that occupant and vehicle attributes were largely similar, and the current logistic regression model was deemed appropriate. 5. Conclusion This study presents an in-depth investigation of the prevalence, patterns and potential for lower extremity injuries in frontal crash modes. With improvements in vehicle design and stiffness, over ten thousand sampled drivers from the last twelve years in NASSCDS were examined and analyzed to evaluate the likelihood of lower extremity injury as a function of vehicle attributes and driver characteristics. The analysis concluded with regard to lower extremity injuries sustained in frontal crashes:  The percentage of whole-body AIS 2+ injuries for drivers in frontal crashes has decreased, but lower extremity remained the most frequently injured region.  Toepan intrusion greatly increased the odds of sustaining AIS 2+ lower limb injuries. Crashes with increased instrument panel intrusion and higher Delta-V increased the likelihood of lower limb injuries as well.  Female drivers had a significantly higher likelihood of lower extremity injuries than males, as did individuals with increased body weight. Vans, followed by SUVs, were less likely to cause lower limb injury in comparison to passenger cars, with light trucks being the worst for sustaining lower limb injuries. As results from this NASSCDS analysis were contextualized against the improvements in vehicle structure and design, it is clear that lower extremity injuries among vehicle occupants remains a prominent issue of concern. Current study provides valuable information to manufacturers for continued efforts focused on the mitigation and prevention of lower extremity injuries, through all available technologies including computational modeling, experimental tests and improved methods for reducing the gap with real world injury risk. Acknowledgments The National Highway Traffic Safety Administration provided both technical and financial support via Cooperative Agreement No. DTNH22-10-H-00293. Acknowledgment also goes to Raul Arbelaez and Andrew Brethwaite for providing the IIHS frontal offset crash test data. Note that the views expressed by the authors do not necessarily represent those of the sponsors.

44

Appendix A

X. Ye et al. / Accident Analysis and Prevention 83 (2015) 37–46

X. Ye et al. / Accident Analysis and Prevention 83 (2015) 37–46

45

46

X. Ye et al. / Accident Analysis and Prevention 83 (2015) 37–46

References AAAM, 1990. The Abbreviated Injury Scale—1990 Revision. Published Report. Association for the Advancement of Automotive Medicine. Arbabi, S., Wahl, W.L., Hemmila, M.R., Kohoyda Inglis, C., Taheri, P.A., Wang, S.C., 2003. The cushion effect. J. Trauma 54 (6), 1090–1093. Assal, M., Huber, P., Tencer, A., Rohr, E., Mock, C., Kaufman, R., 2002. Are drivers more likely to injure their right or left foot in a frontal car crash: a crash and biomechanical investigation. 46th Annual Proceedings for the Association for the Advancement of Automotive Medicine . Austin, R.A., 2012. Lower Extremity Injuries and Intrusion in Frontal Crashes. Report No. DOT HS 811578. National Highway Traffic Safety Administration, Washington, DC. Crandall, J.R., Martin, P.G., 1997. Lower limb injuries sustained in crashes and corresponding biomechanical research. International Symposium on Real World Crash Injury Research, Leicestershire, UK. Crandall, J.R., Martin, P.G., Bass, C.R., Pilkey, W.D., Dischinger, P.C., Burgess O’Quinn, A.R., Schmidhauser, T.D.C.B., 1996. Foot and ankle injury: the roles of driver anthropometry, footwear, and pedal controls. 40th Annual Proceedings of the Association for the Advancement of Automotive Medicine 1–18. Dischinger, P.C., 1996. The risk of ankle/foot fractures among women drivers, Proceedings of the 2nd National Conference on Women’s Travel Issues, Baltimore, MDChapter 28 , pp. 531–536. Dischinger, P.C., Read, K.M., Kufera, J.A., Kerns, T.J., Ho, S.M., Burch, C.A., Jawed, N., Burgess, A.R., 2005. CIREN report: consequences and costs of lower-extremity injuries. DOT HS 809871. IIHS, 2012. Predicted availability of safety features on registered vehicles, 26 Highway Loss Data Institute Bulletin, 28. Insurance Institute for Highway Safety, Arlington, VA April 2012. Ivarsson, B.J., Manaswi, A., Genovese, D., Crandall, J.R., Hurwitz, S.R., Burke, C., Fakhry, S., 2008. Site, type, and local mechanism of tibial shaft fracture in drivers in frontal automobile crashes. Forensic Sci. Int. 175 (2), 186–192. James, P.T., Leach, R., Kalamara, E., Shayeghi, M., 2001. The worldwide obesity epidemic. Obes Res. 9 (4), 228S–233S. Kahane, C.J., 2006. An Evaluation of the 1998–1999 Redesign of Frontal Air Bags. NHTSA Technical Report DOS HS 810685. National Highway Traffic Safety Administration, U.S. Department of Transportation.

Klinich, K.D., Schneider, L., 2003. Pedestrian Lower Extremity Injuries: A Preliminary Summary of Injury Patterns, Biomechanical Data, Currently Proposed Surrogates and Test Procedures, and Related Issues. Alliance for Automobile Manufacturers. Kuppa, S., Wang, J., Haffner, M., Eppinger, R., 2001. Lower extremity injuries and associated injury criteria. 17th Conference on the Enhanced Safety of Vehicles, Paper No. 457 . National Automotive Sampling System-Crashworthiness Data System, 2010. NASSCDS Coding and Editing Manual. U.S. Department of Transportation. Parenteau, C.S., Viano, D.C., Lovsund, P., Tingvall, C., 1996. Foot–ankle injuries: influence of crash location, seating position and age. Accid. Anal. Prev. 28 (5), 607–617. Pilkey, W.D., Sieveka, E.M., Crandall, J.R., Klopp, G., 1994. The influence of foot placement and vehicular intrusion on occupant lower limb injury in full-frontal and frontal-offset crashes. Paper No. 94 S4 W31. International Technical Conference on the Enhanced Safety of Vehicles . Rudd, R.W., 2009. Updated analysis of lower extremity injury risk in frontal crashes in the United States. 21st International Technical Conference on the Enhanced Safety of Vehicles, Stuttgart, Germany Paper No. 09-0556. SAS/STAT, 2004. 9.1 Users Guide. SAS Institute Inc., Cary, North Carolina, USA. Schubert, R., Junge, M., Zwipp, H., 2010. Injuries of foot and ankle in front seat car occupants. 4th International Conference on Expert Symposium on Accident Research 230–238. Swanson, J., Rockwell, T., Beuse, N., Summers, L., Summers, S., Park, B., 2003. Evaluation of stiffness measures from the U.S. new car assessment program. 18th Conference on the Enhanced Safety of Vehicles Paper No. 527. Thomas, P., Bradford, M., 1995. A logistic regression analysis of lower limb injury risk in frontal collisions. 39th Annual Proceedings of the Association for the Advancement of Automotive Medicine, Chicago, Illinois. Thomas, P., Charles, J., Fay, P., 1995. Lower Limb Injuries—The Effect of Intrusion, Crash Severity and the Pedals on Injury Risk and Injury Type in Frontal Collisions. SAE Paper 952728. Ye, X., Crandall, J., Forbes, A., Bose, D., Hurwitz, S., Funk, J., Shaw, G., Sochor, M., Poplin, G., Freeth, R., Rizzo, A., Rudd, R., Scarboro, M., 2014. Case series analysis of hind-foot injuries sustained by drivers in motor vehicle crashes. Proceedings of International Conference of Crashworthiness, Kuching, Malaysia. Zuby, D.S., Nolan, J.M., Sherwood, C.P., 2001. Effect of Hybrid III Leg Geometry on Upper Tibia Bending Moments. SAE Paper 2001-01-0169, Detroit, Michigan.