- Email: [email protected]
Effect of weight, height and BMI on injury outcome in side impact crashes without airbag deployment
Accident Analysis and Prevention 72 (2014) 193–209
Contents lists available at ScienceDirect
Accident Analysis and Prevention journal homepage: www.elsevier.com/locate/aap
Effect of weight, height and BMI on injury outcome in side impact crashes without airbag deployment Chinmoy Pal a , Okabe Tomosaburo a , K. Vimalathithan b , M. Jeyabharath b,∗ , M. Muthukumar b , N. Satheesh b , S. Narahari b a b
Nissan Motor Company Ltd., Japan Renault Nissan Technology Business Centre India, Chennai, India
a r t i c l e
i n f o
Article history: Received 29 June 2013 Received in revised form 7 May 2014 Accepted 18 June 2014 Keywords: BMI Side impact Lower extremity injury Thoracic injury Abdomen injury Head injury
a b s t r a c t A comprehensive analysis is performed to evaluate the effect of weight, height and body mass index (BMI) of occupants on side impact injuries at different body regions. The accident dataset for this study is based on the National Automotive Sampling System-Crashworthiness Data System (NASS-CDS) for accident year 2000–08. The mean BMI values for driver and front passenger are estimated from all types of crashes using NASS database, which clearly indicates that mean BMI has been increasing over the years in the USA. To study the effect of BMI in side impact injuries, BMI was split into three groups namely (1) thin (BMI < 21), (2) normal (BMI 24–27), (3) obese (BMI > 30). For more clear identification of the effect of BMI in side impact injuries, a minimum gap of three BMI is set in between each adjacent BMI groups. Car model years from MY1995–1999 to MY2000–2008 are chosen in order to identify the degree of influence of older and newer generation of cars in side impact injuries. Impact locations particularly side-front (F), side-center (P) and side-distributed (Y) are chosen for this analysis. Direction of force (DOF) considered for both near side and far side occupants are 8 o’clock, 9 o’clock, 10 o’clock and 2 o’clock, 3 o’clock and 4 o’clock respectively. Age <60 years is also one of the constraints imposed on data selection to minimize the effect of bone strength on the occurrence of occupant injuries. AIS2+ and AIS3+ injury risk in all body regions have been plotted for the selected three BMI groups of occupant, delta-V 0–60 kmph, two sets (old and new) of car model years. The analysis is carried with three approaches: (a) injury risk percentage based on simple graphical method with respect to a single variable, (b) injury distribution method where the injuries are marked on the respective anatomical locations and (c) logistic regression, a statistical method, considers all the related variables together. Lower extremity injury risk appears to be high for thin BMI group. It is found that BMI does not have much influence on head injuries but it is influenced more by the height of the occupant. Results of logistic analysis suggest that BMI, height and weight may have significant contribution towards side impact injuries across different body regions. © 2014 Elsevier Ltd. All rights reserved.
1. Introduction Body mass index (BMI) of occupants has been one of the factors causing changes in the pattern and the distribution of serious injuries in car crashes. Mock et al. (2002) found that an increase in BMI would increase the risk of severe injuries in automobile
Abbreviations: BMI, body mass index; MY, model year; AY, accident year; CY, calendar year; AIS, abbreviated injury scale; MAIS, maximum abbreviated injury scale; NCAP, new car assessment program; DOF, direction of force; NASS CDS, National Automotive Sampling System Crashworthiness Data System; ROC, receiver operator characteristics. ∗ Corresponding author. Tel.: +91 4467448224.. E-mail address: [email protected] (M. Jeyabharath). http://dx.doi.org/10.1016/j.aap.2014.06.020 0001-4575/© 2014 Elsevier Ltd. All rights reserved.
accidents. Arbabi et al. (2003) identified that obesity would cause a negative effect due to increase in mass and momentum but have positive effect due to cushioning of extra tissue. Shankuan et al. (2006) found that underweight men have higher risk of injury than the normal ones but for women, BMI have no significant effect on change in risk of injury. The chest and pelvis are the most possible regions of the body to be injured in side impacts accidents and from NASS data it is observed that dominant source of contact causing pelvis and chest injuries is the side interior surface especially the door and its surrounding structures (Samaha and Elliott, 2003). Automobile accidents are also the major causes of serious thoracic and abdomen injuries in particular for lateral crashes, which contributes to thoracoabdominal injury (Siegel et al., 1993). Obese men had a broadly higher risk of injury, especially serious injury, to the upper body regions including head, face, thorax, and spine
194
C. Pal et al. / Accident Analysis and Prevention 72 (2014) 193–209
Table 1 Number of drivers and front passengers in the US NASS data CY2000–2008.
Table 2 Side impact dummies in NCAP.
Accident year
No. of drivers
No. of front passengers
Dummy
Height
Weight
BMI
2000–2008
50507
14166
ES2-male 50th %ile SID-IIs-female 5th %ile World SID-male 50th %ile
177.8
72.0
22.8
152.0
50.0
21.6
175.3
77.3
25.2
than normal weight men during frontal coliision (Shankuan et al., 2010). Side impact crashes are the second most important crashes with high frequency in real world accidents, resulting in one third of occupant fatalities (NHTSA, 2005). Compared to far-side (passenger) impacts, near side (driver) impacts are 2.3 times more common (Haland et al., 1993). Impact direction, impact location, vehicle speed before crash and change of impact velocity (delta-V) are the various crash related factors that could influence the severity of injuries (Terrel et al., 2003). Impact location (P, Y) has strong effects on the injury outcomes of nearside drivers. Serious thoracic injury is found to be more associated with impact location than impact direction (Xinghua et al., 2012). Oblique impact directions (10 o’clock and 2 o’clock) were found to be the most common injury crashes (Augenstein et al., 2003). In side impact, accidents leading to fatal injuries are slightly higher (32%) for 10 o’clock when compared to 9 o’clock impacts (29%) (Rattenbury et al., 2001). From our previous work (Pal et al., 2014), it was found that gender has significant contribution on pedestrian pelvis injuries. From the survey of past literature, it is found that only few studies have determined the influence of BMI in various car crash modes. Hence, an in-depth and a systematic study of the influence of BMI in side impact injuries is the key focus of the present study. 2. Methods 2.1. Trend of BMI in the USA In order to estimate the effect of BMI in side impact injuries, mean BMI is calculated for all occupants in all type of crashes reported in NASS-CDS from calendar year (CY) 2000–08. Number of drivers and front seat passengers involved are tabulated in Table 1. Mean BMI is calculated for drivers and front passengers plus drivers using NASS CDS data base for calendar year 2000–2008. From Fig. 1, it is observed that mean BMI of drivers and front passengers plus drivers are increasing every year. This agrees with the findings from prior studies (Jakobsson and Lindman, 2005) that US NASS data set shows an increase of mean BMI value from calendar year 1993 to 2003, which incorporates 67,000 occupants in all type of car accidents. In this present study, mean BMI of drivers are increased from 25.97 in CY2000 to 26.71 in CY2008. In the similar way, the mean BMI of front passengers plus drivers are increased
Fig. 1. Variation of mean BMI for driver and front passenger estimated from US NASS data.
from 25.65 in CY2000 to 26.45 in CY2008. This agrees well with trend of National Health Statistics in the USA, where continuous increase in BMI is presently a national concern for rising medical expenditure. It is to be noted that when front passenger is also included in mean BMI calculation, the mean BMI of front passengers plus drivers are less than that of mean BMI of drivers population. This is because children are also included in front seat passengers population. BMI of side impact dummies (ES2 and World SID) currently used in NCAP are tabulated in Table 2. To identify the effect of BMI in side impact injuries, the range of BMI values are categorized into three distinct groups (thin: <21, normal: 24–27 and obese: >30). Here, a gap of three points of BMI (21–24 and 27–30) is maintained between the selected BMI groups to extract the exact effect of BMI. Population percentage for drivers (marked in blue) and front passengers plus driver (marked in pink) involved in all type of crashes for three different BMI groups are shown in Fig. 2. It is seen from Fig. 1 that World SID 50th percentile male BMI lies within the selected normal BMI group and SIDII 5th percentile female BMI is just above the thin BMI group. It is also noted that BMI of World SID dummy is still less than mean BMI of the US driving population. It is observed from Fig. 2 that there are 16.6% of thin people, 23.6% of normal people and 19.9% of obese people population exist in total driver populations. Similarly, 14.6% of thin people, 24.2% of normal people and 20.8% of obese people population exist in total front passengers plus drivers populations. The groups of others include data between BMI 21–24 and BMI 27–30. None of the present side impact dummies currently used in NCAP are representing obese people group whose BMI values are above 30.
Fig. 2. Percentage distribution of different BMI groups in the US NASS data for front seat occupants. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
C. Pal et al. / Accident Analysis and Prevention 72 (2014) 193–209
195
Table 3 Driver side (DS) and passenger side (PS). Variable
Description
Code
508
Direction of force—DS
8, 9, 10, 28,29, 30, 48, 49, 50, 68, 69, 70, 88, 89,90 2, 3, 4, 22, 23, 24, 42, 43, 44, 62, 63, 64, 82, 83, 84 F, P, Y
Direction of force—PS 510 812
807 364
Fig. 3. Accident analysis variable for side impact front seat passengers including drivers.
2.2. Accident database Accident data used in this analysis are taken from the NASS-CDS which is a nationwide crash data collection program sponsored by the U.S Department of Transportation (DOT). About 5000 crashes are collected for light vehicles each year on the US roadways that are towed away from the accident scene due to heavy damage. The NASS-CDS data can be extrapolated to national estimates by using inflation factor (RATWGT) to represent a broad range of all police reported motor vehicle crashes. 2.3. Variable selection Accident data selection criteria of this analysis for drivers and front passengers are indicated in Fig. 3. These variables are also tabulated in Table 3 in Appendix A. Direction of force (DOF) considered for both near side and far side are 8 o’clock, 9 o’clock, 10 o’clock and 2 o’clock, 3 o’clock and 4 o’clock respectively. Impact locations namely F, P, Y corresponding to the locations of side-front, sidecenter, side distributed respectively are considered for this analysis. Table 4 Driver and front passengers without airbag cases delta-V (0–60 kmph).
809 810 107 306 Driver side (DS) and passenger side (PS)
Longitudinal and lateral location Occupant’s seat position—DS Occupant’s seat position—PS Age <60 Lateral component. delta-V—DS (0–60 kmph) Lateral component. delta-V—PS (−60–0 kmph) Height Weight Accidental year Model year
11 13
999 (Unknown removed) 999 (Unknown removed) 2000–2008 1995–2008
Lateral delta-V (0–60 kmph) constraint is imposed to nullify the effect of very high delta-V influence. Age range 0–60 years is considered to avoid very old people. This will remove the influence of bone strength in injuries because older people have softer bones than normal people. AIS2+ and AIS3+ injuries are considered in body regions such as head, thorax, abdomen, lower extremity, spine and upper extremity. Older cars with model year (MY)1995–1999 and newer cars with MY 2000–08 is considered to find the car model year influence on injuries. Table 4 and Table 5. 2.4. Approach To study the effect of BMI on side impact injuries, three approaches were adopted in the current study.
196
C. Pal et al. / Accident Analysis and Prevention 72 (2014) 193–209
Table 5 Raw and weighted data for MAIS2+ and MAIS3+ injuries. Injury Scale
AIS2+ injuries
AIS3+ injuries
Raw data
Weighted data
Raw data
Weighted data
BMI
<21
24–27
>30
<21
24–27
>30
<21
24–27
>30
<21
24–27
>30
No. of occupants Head Face Neck Thorax Abdomen Spine Upper extremity Lower extremity
96 93 9 1 63 35 27 37 104
90 110 4 0 85 52 20 42 85
96 100 5 0 109 68 29 60 74
7924 4511 442 22 2879 1456 230 2299 11355
8680 6576 275 0 4686 1740 1139 3962 6098
7858 5395 246 0 6696 3123 1233 4763 5962
65 66 1 0 57 6 4 10 44
62 83 0 0 74 14 4 14 34
74 76 0 0 87 20 6 18 24
4965 2682 140 0 2687 433 90 358 4157
4149 2408 0 0 3899 493 323 874 2374
5662 3374 0 0 5217 755 342 1970 2022
Fig. 6. AIS2+ injury risk in abdomen region for different model years; older and newer cars.
Fig. 4. AIS2+ injury risk in head region for different model years; older and newer cars.
Fig. 5. AIS2+ injury risk in thorax region for different model years; older and newer cars.
Fig. 7. AIS2+ injury risk in lower extremity region for different model years; older and newer cars.
C. Pal et al. / Accident Analysis and Prevention 72 (2014) 193–209
Fig. 8. AIS2+ injury risk in spine region for different model years; older and newer cars.
(a) Injury risk: Accident cases involving MAIS2+ cases were considered for this study. Within those MAIS2+ cases, AIS2+ injury risk is calculated by dividing the number of AIS2+ injuries within the BMI group by the number of AIS 1+ injuries within the same BMI group. for
AIS 2+Injury risk
A
BMI group
For B body region
Number of AIS 2+Injuries
[For B body region A BMI group] = Number of MAIS 2 + AIS 1 + Injuries [For all body region A BMI group]
Fig. 9. AIS2+ injury risk in upper extremity region for different model years; older and newer cars.
197
Fig. 10. AIS3+ injury risk in head region for different model years; older and newer cars.
(b) Injury distribution: Injuries for each case are plotted on the respective anatomic locations. By this approach, dominant injury locations and symmetrical conditions for individual BMI group can be identified. (c) Logistic regression: It models the relationship between a dependent and one or more independent variables and allows us to look at the fit of the model as well as at the significance of the relationships between dependent and independent variables. Logistic regression can be binomial or multinomial. Binomial or binary logistic regression deals with situations in which the observed outcome for a dependent variable can have only two possible types (for example, “dead” vs. “alive”). Multinomial logistic regression deals with situations where the
Fig. 11. AIS3+ injury risk in thorax region for different model years; older and newer cars.
198
C. Pal et al. / Accident Analysis and Prevention 72 (2014) 193–209
Fig. 12. AIS3+ injury risk in abdomen region for different model years; older and newer cars.
Fig. 14. AIS3+ injury risk in spine region for different model years; older and newer cars.
outcome can have three or more possible types (for example, “minor” vs. “serious” vs. “fatal” injuries). Binary logistic regression was used in this study to identify the significance of different variables for AIS2+ injury occurrence.
MY1995–1999 (old cars) and MY2000–2008 (new cars) and deltaV range of 0–60 kmph respectively. Figs. 4–9 gives the injury risk of AIS2+ injuries for drivers and front passengers involved in car accidents without airbag cases.
3. Results and discussion: 3.1. Injury risk (AIS2+ injuries) AIS2+ injury risk for different body regions such as head, thorax, abdomen, lower extremity, spine and upper extremity are plotted for three different selected BMI groups, two sets of car models
Fig. 13. AIS3+ injury risk in lower extremity region for different model years; older and newer cars.
3.1.1. Head Comparison of AIS2+ injury risk in head region for older cars and newer cars are shown in Fig. 4. This comparison will help us to understand the difference in injury trend between new and old car model years. It is seen from Fig. 4 that the head injury risk is slightly higher for newer cars compared to those of older ones. There is not much difference observed in injury risk percentage among three selected BMI groups.
Fig. 15. AIS3+ injury risk in upper extremity region for different model years; older and newer cars.
C. Pal et al. / Accident Analysis and Prevention 72 (2014) 193–209 Table 6 Selection criteria for injury distribution–pie chart (based on NASS-CDS Codebook). Injuries based on NASS-CDS codebook
Lower extremity region OIC BODY REGION 16 Pelvis 20 Thigh 11 Knee 12 Lower leg 17 Foot 25 Lower limb Abdomen region OIC SYSTEM ORGAN 17 Spleen 12 Liver 11 Kidney Others Spine region Specific anatomic structure Lumbar 6 4 Thoracic 2 Cervical
3.1.2. Thorax Comparison of AIS2+ injury risk in thorax region for older and newer cars are shown in Fig. 5. It is observed that the increasing trend of injury risk percentage from thin to obese people group. Injury risk percentage is always less for thin people group irrespective of car model years.
3.1.3. Abdomen Comparison of AIS2+ injury risk in abdomen region for older cars and newer cars are shown in Fig. 6. Injury risk percentage in abdomen region is less for thin group in newer car models.
199
Increasing trend of injury percentage is observed from thin to obese people group for newer cars. 3.1.4. Lower extremity Comparison of AIS2+ injury risk in lower extremity region for older cars and newer cars are shown in Fig. 7. It is observed that decreasing trend of injury risk percentage from thin to obese people group and it is more prominent for older cars. Injury risk percentage is always high for thin people group irrespective of car model years. It is also seen from Fig. 7 that the lower extremity injury risk is higher for older cars compared to newer cars. 3.1.5. Spine Comparison of AIS2+ injury risk in spine region for older cars and newer cars are shown in Fig. 8. Since the number of data is less, it is difficult to make any concrete conclusion. Further investigation needed by observing the trend with new data sets say for calendar year CY2009–2012. 3.1.6. Upper extremity Comparison of AIS2+ injury risk in upper extremity region for both older cars and newer cars are shown in Fig. 9. It is observed that injury risk percentage is almost same for both older and newer cars. 3.2. AIS3+ injury risk Within MAIS3+ cases, AIS3+ injury risk is calculated by dividing the number of AIS3+ injuries within the BMI group by the number of AIS 1+ injuries within the same BMI group. So far, the trend of AIS2+ injury risk for different body regions was discussed. Similarly, AIS3+ injury risk for all body regions are
Table 7 Selection criteria for injury distribution–pie chart (based on AIS-98 codebook). Injuries based on AIS-98 codebook
Thorax region Ribs Sternum Lungs Heart + inter ventricular septum Diaphragm Pneumothorax Vessels Head region Cerebellum + brain stem Cerebrum Skull-Base,Vault Unconsius
Type of anatomical structure 5 4
2 Type of anatomical structure 4 5 6
Specific anatomical structure 2 8 14 10, 13 6 22 2 Specific anatomical structure 2, 4 6 2,4 2,4,6,8,10
Table 8 Selection criteria for brain injury distribution–pie chart (AIS-98 Codebook based). Injuries based on AIS-98 codebook
Head Injury region
Brain stem Cerebellum
Cerebrum
Diffuse axonal injury Diffuse axonal injury Epidural Hematoma Intra cerebellar Hematoma Subdural Hematoma Sub arachnoid Hemorrhage Diffuse axonal injury Epidural Hematoma Intra cerebral Hematoma Subdural Hematoma Intra ventricular Hemorrhage Sub arachnoid Hemorrhage
Type of anatomical structure 4
Specific anatomical structure 2 4
6
Level of injury 6 6 14,18,22 26,30,34 38,42,46 66 28 30,32,34,36 38,40,42,44,46,48 50,52,54,56 78 84
200
C. Pal et al. / Accident Analysis and Prevention 72 (2014) 193–209
Fig. 16. AIS2+ lower extremity injury distribution across different BMI groups. (a)–(c) AIS2+ pelvis anterior region injury distribution. (d)–(f) AIS2+ pelvis posterior region injury distribution. (g)–(i) AIS2+ leg injury distribution. (j)–(l) Injury distribution percentage within lower extremity anatomical structure. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
plotted in Figs. 10–15. It is observed that injury risk trend of AIS3+ injuries are almost same as that of AIS2+ injuries. 3.3. Injury distribution AIS2+ injuries are plotted on their respective anatomical location and pie charts are used to highlight the injured segment or entity within each body region. Injury distribution study is carried out for different body regions namely lower extremity, thorax, abdomen, spine and head. This study aims to highlight the
proportion of injured body segment within each body region and symmetrical injury pattern variations with respect to BMI. Hence, only driver injury without airbag cases was considered for this section of study. Selection criteria for injured segment within each body region (used in pie chart) are tabulated in Tables 6–8 in appendix A 3.3.1. Lower extremity Lower extremity injuries for drivers across different BMI groups are plotted as points with different colors on the respective
C. Pal et al. / Accident Analysis and Prevention 72 (2014) 193–209
201
Fig. 17. AIS2+ injury proportions on gender basis. (a) All body region injuries and (b) lower extremity injuries.
anatomical locations as shown in Fig. 16. Lower extremity comprises of five major parts namely: pelvis, thigh, tibia, ankle and foot. For convenience, lower extremity injuries are classified into two groups (a) pelvis injuries and (b) leg injuries (thigh, tibia, ankle and foot). As shown in Fig. 16, the injury pattern is bilateral (both left and right side) for thin people group whereas injury pattern is left side dominant for obese people group. Pelvic injuries are dominant across all BMI groups, but decreasing trend in injury percentage of pelvis is observed with increase of BMI from thin to obese group. Thin people relatively have more chance to sustain pelvic injury (87%) possibly due to less fat tissue around the waist and needs further investigation. Leg injury percentage increases with the increase in BMI (13% in thin to 48% in obese BMI group), may be because the legs of obese group occupy more legroom space. Since anatomical profile and dimensions of pelvis is not same for male and female, gender can be an important factor. 3.3.2. Lower extremity—gender based AIS2+ injuries for male and female drivers for all BMI are plotted in Fig. 17. For all body injuries, both male and female share almost same level of injury percentage. However, for lower extremity injuries female injury percentage is dominant. AIS2+ injuries for males and females are plotted for different BMI groups as shown in Fig. 18. Pelvis injuries appear to be more dominant among male drivers. For female drivers, pelvis injuries appear to be more sensitive to the change in BMI. A decreasing trend in pelvis injury percentage is observed for females moving from thin to obese BMI group. Since the cases are very less, it is not possible to make any strong conclusion. 3.3.3. Thorax Thorax injuries for drivers across different BMI groups are plotted as points with different colors on the respective anatomical locations as shown in Fig. 19. Each point with different color indicates the set of injuries corresponding particular accident cases. Thorax region comprises of 12 pairs of thoracic ribs and organs like lungs, diaphragm, heart, etc. In Fig. 19 number of rib injury locations is high compared to other regions. As, multiple rib fracture is considered as single injury for pie and but marked in multiple locations, for additional details refer Table 9 in appendix A. Injury pattern is bilateral for thin people and left side dominant for normal people. An increasing trend of rib-injury percentage is observed from thin to obese BMI group. Rib injury may occur due to intrusion, for thin BMI group rib injury is 27% and lungs injury is 41%,
Fig. 18. AIS2+ injury proportions on gender and BMI basis. (a), (c) and (e) AIS2+ injury proportion for male across different BMI groups. (b), (d) and (f) AIS2+ injury proportion for female across different BMI groups.
which is of different pattern in obese group. Obese BMI group has 52% rib injury, 3% heart injury, 5% sternum injury, etc., shows more intrusion based injury and higher depth of injury comparing with thin BMI group. 3.3.4. Abdomen Abdomen injuries for drivers across different BMI groups are plotted as points with different colors on the respective anatomical locations as shown in Fig. 20. Each point with different color indicates the set of injuries corresponding particular accident cases. It is seen from Fig. 20 that spleen injury is always dominant across all BMI groups and this injury is highest for thin people group (75%) but the cases involved are very less. Kidney and liver injuries are higher in obese people group when compared with thin and normal BMI groups. The other injuries in abdomen region are retroperitoneum, mesentery, stomach, jejunum–ileum, bladder for all BMI groups. It also shows that injury pattern is distributed on both sides for obese people. This change in distribution of abdomen injury pattern is probably due to two possible factors (a) the gap between the abdomen and the surrounding structures and (b) the other could be the weight of the occupant. 3.3.5. Spine Spine injuries for drivers across different BMI groups are plotted as points with different colors on the respective anatomical locations as shown in Fig. 21. It is observed that there is no thoracic
202
C. Pal et al. / Accident Analysis and Prevention 72 (2014) 193–209
Fig. 19. AIS2+ thoracic injury distribution across different BMI groups. (a)–(c) AIS2+ ribs injury distribution. (d)–(f) AIS2+ lungs and diaphragm injury distribution. (g)–(i) AIS2+ heart, aorta and other thoracic cavity injury distribution. (j)–(l) Injury distribution percentage within thoracic anatomical structure. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
vertebrae injury for thin people. Lumbar injury is all most same pattern for all BMI group. Cervical vertebrae injury is higher for thin people group when compared to other higher BMI group. As number of spine injured cases reported in NASS CDS is less, it is difficult to arrive at any concrete conclusion. 3.3.6. Head Head injury distribution for driver with respect to BMI is shown in Fig. 22. It is seen that the injury percentage for all three BMI groups are almost same. From this observation, no concrete
conclusion could be inferred. Since BMI is the ratio of weight and square of height, height can also be a deciding factor to determine the injury trends, which will be discussed in the following section. To carry out height based study, the data has been segregated into three groups namely short (150–165 cm), medium (165–175 cm), tall (greater than 175 cm). Head injury distribution for driver with respect to height is shown in Fig. 23. It is observed that cerebrum injuries are always dominant for all height groups and this injury is highest for the short people group, which is 79%. The other two height groups (medium
Table 9 Sample case for explaining rib fracture injury distribution. Accident year = 2005, PSU:45, Case id: 109 Vehicle 1 > Occupant 1 > Injury codes Injury number
Source of data
AIS code
Injury description
Aspect
2
Post-ER medical record
4502324
Rib cage fracture >3 ribs on one side and ≤3 ribs on other side, stable chest or NFS = with hemo-/pneumothorax
L Rib 4, L Rib 5, L Rib 6, L Rib 7, L Rib 8, L Rib 9, L Rib 10, L Rib 12
Number of rib injury = 1 (used for Pie chart) Number of rib injured location = 8 (used for anatomical markings)
C. Pal et al. / Accident Analysis and Prevention 72 (2014) 193–209
203
Fig. 20. AIS2+ abdomen injury distribution across different BMI groups. (a)–(c) AIS2+ spleen, liver, kidney and other abdomen injury distribution. (d)–(f) Injury distribution percentage within abdomen anatomical structure. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
and tall) the cerebrum injury is almost same 56%. The skull injury percentage is higher for tall people group (27%) when compared to short people group (12%). Head of the taller people is close to the roof side rail so there is a higher probability of contact injury and may need further investigation. Jillian et al. (2012) considered the following injuries occurring in the brain to understand the brain injury mechanism as shown in Fig. 24. These are as follows: (1) Sub-dural hematoma: bleeding inside the sub-dural space. (2) Intracerebral hematoma: bleeding inside the brain. (3) Epidural hematoma: bleeding between the skull and the dura mater. (4) Intraventricular hemorrhage: bleeding into the brains ventricular system. (5) Sub-arachnoid hemorrhage: bleeding into the sub-arachnoid space. Brain injury type for driver with respect to height is shown in Fig. 25. It is seen that subarachnoid type of injury is higher in tall people group (50%) when compared to the short people group (29%). Since in this type of injury bleeding is in the subarachnoid space, the depth of injury could be high. Hence, the depth of injury for taller people could be higher, when compared to short people. The intraventricular hemorrhage and epidural hematoma type of injury is higher in small people group when compared to the tall people group. The height of the occupant influences the seat track position for the driver with respect to the B-pillar position. Hence, the degree of interaction of the occupant with the intruding vehicle door and the structures around it changes with height of
the occupant. Therefore, tall people may have more possibility to hit B-pillar or roof side rail and their depth of injury can be more. These observations need further investigation before making any concrete statement.
3.4. Comparative study for “with and without weighting factor” cases “With weighting factor” data are those, injury counts are multiplied by their respective inflation factor. “Without weighting factor” is a raw data without multiplying by any weighting factor. Fig. 26 shows the AIS2+ injury risk percentage within BMI group for individual body regions “with and without weighting factor” for 0–60 kmph delta-V for MY1995–2008 incorporating driver plus front passengers. Decreasing trend of injury risk percentage is observed moving from thin to obese BMI group for lower extremity, is common for both “with and without weighting factor”. Increasing trend of injury risk percentage is observed from thin to obese BMI group for thorax, is common for both “with and without weighting factor”. Increasing trend of injury risk percentage is observed from thin to obese BMI group for abdomen, is same for both “with and without weighting factor”. No definite trend is observed for head in both “with and without weighting factor” for thin and obese BMI group. Raw and weighted data for MAIS2+ and MAIS3+ injuries are tabulated in Tables 4 and 5 in Appendix A.
204
C. Pal et al. / Accident Analysis and Prevention 72 (2014) 193–209
Fig. 21. AIS2+ spine injury distribution across different BMI Groups. (a)–(c) AIS2+ lumbar, cervical and thoracic spine injury distribution. (d)–(f) Injury distribution percentage within spine anatomical structure. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
Fig. 22. AIS2+ injury distribution for head across different BMI groups (a) AIS2+ head injury distribution for BMI < 21 group. (b) AIS2+ head injury distribution for BMI: 24–27 group. (c) AIS2+ head injury distribution for BMI > 30 group.
C. Pal et al. / Accident Analysis and Prevention 72 (2014) 193–209
205
Fig. 23. AIS2+ injury distribution for head across different height groups (a) AIS2+ head injury distribution for height 150–165 cm. (b) AIS2+ head injury distribution for height 165–175 cm. (c) AIS2+ head injury distribution for height > 175 cm.
3.5. Comparative study for Impact locations F, P and Y and F, P, Y, Z and D So far, the trend of AIS2+ and AIS3+ injury risk for different body regions was discussed based on impact locations F, P and Y and other variable mentioned in the previous sections. It is also noteworthy to find out the effect of impact locations D and Z in side impact injuries. Fig. 27 shows the AIS2+ injury risk percentage for MY 1995–2008 including driver plus front passengers within respective BMI groups for different impact locations. It is observed that injury risk trend is almost same for all body regions. Fig. 28 shows the accident analysis variable incorporating impact locations Z and D in addition to F, P and Y.
3.6. Comparative study for “with and without side airbag” cases “With airbag” are those cases when either one of the side airbags (seat-back or roof-side rail or door-panel mounted) is in deployed conditions and “without airbag” are those cases without any side airbag available or not deployed at the time of crash. Total number of “with airbag” cases are very less, almost one-tenth of “without airbag” cases. Hence, it is not possible to make any definite conclusion by “with airbag” results.
Fig. 24. Commonly occurring injuries in the brain.
Fig. 29 shows the AIS2+ injury risk percentage within BMI group including “with and without airbag” cases across different body regions plotted for drivers considering MY1995–08 and Delta-V 0–60 kmph. Overall there is no appreciable difference in the injury risk trend for “with and without airbag” cases with small exception in lower extremity.
Fig. 25. AIS2+ injury distribution for head across different height groups. (a)–(c) Different type of injuries distributed in brain for different height groups.
206
C. Pal et al. / Accident Analysis and Prevention 72 (2014) 193–209
Fig. 26. AIS2+ injury risk % within respective BMI group with and without weighting factor. AIS2+ injury risk % (a) and (b) for lower extremity (c) and (d) for thorax (e) and (f) for abdomen (g) and (h) for head.
3.7. Logistic regression In this section, logistic regression was carried out to identify the significance of various parameters in predicting AIS2+ injury for different body regions. Contributing rate of parameters changes with the type of dataset chosen. All the vehicles will have a driver and hence, the number of driver cases reported in NASS-CDS database is much more than other occupants in the vehicle. Hence, a dataset was chosen only with drivers. Height of an occupant will affect
the seat track position with respect to B-pillar. Therefore, belt path through thorax and abdomen region of occupant and seat track position with respect to seat location (belted driver track position with steering wheel and belted front passenger track position) may change with the height of the occupant. Hence, a dataset was also chosen involving belted occupant. The number of “with side airbag deployed” cases are very less and they are omitted for this study. Driver without airbag-deployed dataset (includes belted and unbelted drivers) and belted driver plus front passenger without
C. Pal et al. / Accident Analysis and Prevention 72 (2014) 193–209
207
Fig. 27. AIS2+ injury risk % within respective BMI groups for different impact locations. AIS2+ injury risk % (a) and (b) for lower extremity (c) and (d) for thorax (e) and (f) for abdomen (g) and (h) for head.
airbag-deployed dataset were chosen for this study. Variables considered for the logistic regression analysis are tabulated in Table 10 of Appendix A. The above mentioned datasets were evaluated with “BMI” and “height and weight” separately and the results are tabulated in Tables 11 and 12 in Appendix A. BMI, height and weight were not split into any individual groups and they were treated as continuous variable. Based on the observation from Tables 11 and 12 in Appendix A: Head: Age and lateral delta-V may have more influence on the head injuries. Height of the occupant may have some influence
on the head injuries and this agrees with the injury distribution mentioned in Section 3.3.6. Thorax: Age and lateral delta-V may have more influence on the thorax injuries. Height and gender of the occupant may have some influence on the thorax injuries. Compared to the degree of influence of age, delta-V, height and gender on thorax injuries, those of BMI and model year on thorax injuries appear to be less. These results are contradicting from Sections 3.1.2 and 3.3.3, which need further investigations. Abdomen: Lateral delta-V may have more influence on the abdomen injuries. BMI of the occupant may have some influence on
208
C. Pal et al. / Accident Analysis and Prevention 72 (2014) 193–209 Table 10 Logistic regression variables. Variable
Description
510 812 807 364 809 810 306 854 324 309 821 110 808
Longitudinal and lateral location Occupant’s seat position—driver and passenger Age <60 Lateral component delta-V (0–60 kmph) Height Weight Model year Seat track Vehicle curb weight Body type Manual seat belt use Multi impact Sex BMI Theta (angle between lateral and longitudinal Delta V) Side airbag deployed during crash
Fig. 28. Accident analysis variable incorporating impact locations Z and D in addition to F, P and Y.
the abdomen injuries. These match well with the results of Sections 3.1.3 and 3.3.4. Lower extremity: Lateral delta-V may have more influence on the lower extremity injuries. BMI and gender of the occupant may have some influence on the lower extremity injuries and this supports
the statements of the previous Sections 3.1.4 and 3.3.1. Nevertheless, compared to the degree of influences of other factors, model year influence on lower extremity injuries appears to be low.
Table 11 Logistic regression result based on BMI.
Head Thorax Abdomen Lower extremity Spine Upper extremity
WoAB-driver WoAB-belted WoAB-driver WoAB-belted WoAB-driver WoAB-belted WoAB-driver WoAB-belted WoAB-driver WoAB-belted WoAB-driver WoAB-belted
MY
Age
0.52 0.50 0.16 0.91 0.87 0.14 0.81 0.69 0.33 0.01 0.21 0.37
0.00 <0.0001 0.00 <0.0001 0.23 0.63 0.77 0.73 0.88 0.75 0.01 0.01
LAT Del V 0.01 0.03 0.00 0.00 <0.0001 <0.0001 0.00 0.00 0.09 0.00 0.71 0.82
Gender 0.85 0.54 0.40 0.13 0.80 0.40 0.05 0.30 0.00 <0.0001 0.36 0.91
BMI
Theta
ROC
N
0.59 0.18 0.66 0.63 0.12 0.02 0.04 0.03 0.76 0.61 0.47 0.83
0.12 0.11 0.84 0.98 0.93 0.16 0.76 0.13 0.44 0.75 0.77 0.18
0.74 0.74 0.72 0.70 0.74 0.75 0.75 0.73 0.70 0.76 0.67 0.67
338 312 338 312 338 312 338 312 338 312 338 312
Table 12 Logistic regression result based on height and weight. MY Head Thorax Abdomen Lower extremity Spine Upper extremity
Head Thorax Abdomen Lower extremity Spine Upper extremity
WoAB-Driver WoAB-Belted WoAB-Driver WoAB-Belted WoAB-Driver WoAB-Belted WoAB-Driver WoAB-Belted WoAB-Driver WoAB-Belted WoAB-Driver WoAB-Belted
WoAB-driver WoAB-belted WoAB-driver WoAB-belted WoAB-driver WoAB-belted WoAB-driver WoAB-belted WoAB-driver WoAB-belted WoAB-driver WoAB-belted
0.46 0.46 0.17 0.97 0.84 0.14 0.81 0.66 0.30 0.01 0.18 0.36
Age
LAT Del V
0.00 <0.0001 <0.0001 <0.0001 0.20 0.55 0.74 0.73 0.89 0.76 0.01 0.01
0.00 0.02 0.00 0.00 <0.0001 <0.0001 0.00 0.00 0.09 0.00 0.71 0.83
Gender 0.53 0.10 0.04 0.04 0.27 0.76 0.12 0.97 0.00 <0.0001 0.06 0.64
Height
Weight
Theta
ROC
N
0.19 0.01 0.03 0.12 0.08 0.16 0.48 0.69 0.08 0.00 0.14 0.52
0.65 0.17 0.92 0.75 0.21 0.03 0.06 0.03 0.77 0.24 0.30 0.58
0.13 0.15 0.74 0.88 0.94 0.19 0.76 0.10 0.58 0.51 0.65 0.15
0.74 0.75 0.73 0.71 0.74 0.75 0.75 0.73 0.71 0.79 0.68 0.68
338 312 338 312 338 312 338 312 338 312 338 312
MY
Age
LAT Del V
Gender
BMI
Theta
ROC
N
0.52 0.50 0.16 0.91 0.87 0.14 0.81 0.69 0.33 0.01 0.21 0.37
0.00 <0.0001 0.00 <0.0001 0.23 0.63 0.77 0.73 0.88 0.75 0.01 0.01
0.01 0.03 0.00 0.00 <0.0001 <0.0001 0.00 0.00 0.09 0.00 0.71 0.82
0.85 0.54 0.40 0.13 0.80 0.40 0.05 0.30 0.00 <0.0001 0.36 0.91
0.59 0.18 0.66 0.63 0.12 0.02 0.04 0.03 0.76 0.61 0.47 0.83
0.12 0.11 0.84 0.98 0.93 0.16 0.76 0.13 0.44 0.75 0.77 0.18
0.74 0.74 0.72 0.70 0.74 0.75 0.75 0.73 0.70 0.76 0.67 0.67
338 312 338 312 338 312 338 312 338 312 338 312
C. Pal et al. / Accident Analysis and Prevention 72 (2014) 193–209
209
4. Conclusion Based on NASS-CDS database, the mean BMI of the USA population increases every year. Hence, it will be important to study to effect of BMI on injuries during vehicle crashes. BMI of an occupant has significant influence on the occurrence of lower extremity injuries. Lower extremity injury risk appears to be high for thin people group. Gender of an occupant may have some influence on the lower extremity injuries. BMI of an occupant has significant influence on the occurrence of abdomen injuries. Abdomen injury risk appears to be low for thin people group. Age of an occupant has significant influence on the occurrence of head, thorax and upper extremity injuries. Height of an occupant may have some influence on the occurrence of head injuries. Results of logistic analysis suggest that, for better understanding BMI’s influence, it is necessary to investigate the effect of weight and height. Appendix A. See Tables 3–12. References
Fig. 29. AIS2+ injury risk % within BMI group across different body regions (a) with airbag (b) without airbag (c) sum of with and without airbag. AIS2+ injury risk % within BMI group across different body regions (a) with airbag (b) without airbag (c) sum of with and without airbag.
Spine: Lateral delta-V and gender may have more influence on the spine injuries. Height of the occupant appears to have some influence on the spine injuries. These results are different from those of Sections 3.1.5 and 3.3.5, which needs further investigations. However, one should note that the total number of spine injuries is very less and coming to any concrete conclusion is not appropriate. Upper extremity: Age of the occupant may have more influence on the upper extremity injuries. Other parameters like gender, height, model year and BMI do not have significant influence on the upper extremity injuries. These results support the findings from Section 3.1.6. Overall-all body region: Height of the occupant has influence on side impact injuries for head, thorax and spine body regions and weight of the occupant has influence on side impact injuries for abdomen and lower extremity body regions.
Arbabi, S., Wahl, W.l., Hemmila, M.R., Kohoyda-Inglis, C., Taheri, P.A., Wang, S.C., 2003. The vushion effect. J. Trauma 54 (6), 1090. Augenstein, J., Perdeck, E., Stratton, J., Digges, K., Bahouth, G., 2003. Characteristics of crashes that increase the risk of serious injuries. Annu. Proc. Assoc. Adv. Automot. Med. 47, 561–576. Haland, Y., Lovsund, P., Nygren, A., 1993. Life-threatening and disabling injuries in car-to-car side affect—implications for development of protective systems. Accid. Anal. Prev. 25 (2), 199–205. Jillian E. Urban, Sarah Lynch, Christopher T., Whitlow, et al., Sep 2012. Understanding brain injury mechanism: integrating real-world lesions, ATD response and finite element modeling. CIREN Public Meeting. Jakobsson, Lotta, Lindman, Magdalena, 2005. Does BMI Influence the occupant injury risk pattern in car crashes. IRCOBI Conference—Prague (Czech Republic). Mock, C.N., Grossman, D.C., Kaufman, R.P., Mack, C.D., Rivara, F.P., 2002. The relationship between body weight and risk of death and serious injury in motor vehicle crashes. AAP 31, 221–228. National Highway Traffic Safety Administration, 2005. Safety Facts. NHTSA, US DOT, Washington, DC, Report #HS 810 6312005. Pal, C., Okabe, T., Vimalathithan, K., Manoharan, J. et al. Estimation of Pelvis Injuries and Head Impact Time using Different Pedestrian Human FE Models, SAE Technical Paper 2014-01-0522, 2014, doi:10.4271/2014-01-0522. Rattenbury, S.J., Gloyns, P.F., Nolan, J.M., 2001. Vehicle deformation in real-world side impact crashes and regulatory crash tests. In: Proceedings of 17th International Technical Conference on the ESV, Amsterdam, Netherlands, Paper 356. Samaha, R.R., Elliott, D.S., 2003. NHTSA side impact research: motivation for upgraded test procedures. In: Proceedings of the 18th Conference on the Enhanced Safety of Vehicles, Paper 492, Nagoya, Japan. Siegel, J.H., Mason-Gonzalez, S., Dischinger, P., Cushing, B., Read, K.R., Robinson, R., Smialek, J., Heatfield, B., Hill, W., Bents, F., Jackson, J., Livingston, D., Clark, C.C., 1993. Safety belt restraints and compartment intrusions in frontal and lateral motor vehicle crashes: mechanisms of injuries, complications, and acute care costs. J. Trauma 5, 736–759. Shankuan, Zhu, Jong-Eun, Kim, Xiaoguang, Ma, 2010. BMI and risk of serious upper body injury following motor vehicle crashes: concordance of realworld and computer-simulated observations. PLoS Med. 7 (3), e1000250, http://dx.doi.org/10.1371/journal.pmed.1000250. Shankuan, Zhu, Layde, P.M., Guse, C.E., Laud, P.W., Pintar, F., Nirula, R., Hargarten, S., 2006. Obesity and risk for death due to motor vehicle crashes. Am. J. Publ. Health 96 (4), 734–739, http://dx.doi.org/10.2105/AJPH.2004.058156 (Impact Factor: 3.93). Terrel, M., Newland, C., Seyer, K., 2003. The dependence of side impact injury risk on MDB configuration. In: Proceedings of the 18th Conference on the ESV, Nagoya, Japan. Paper no. 399. Lai, Xinghua, Ma, Chunsheng, Hub, Jingwen, Zhoua, Qing, 2012. Impact direction effect on serious-to-fatal injuries among driver in near-side collisions according to impact location: focus on thoracic injuries. Accid. Anal. Prev. 48, 442–450.
Contents lists available at ScienceDirect
Accident Analysis and Prevention journal homepage: www.elsevier.com/locate/aap
Effect of weight, height and BMI on injury outcome in side impact crashes without airbag deployment Chinmoy Pal a , Okabe Tomosaburo a , K. Vimalathithan b , M. Jeyabharath b,∗ , M. Muthukumar b , N. Satheesh b , S. Narahari b a b
Nissan Motor Company Ltd., Japan Renault Nissan Technology Business Centre India, Chennai, India
a r t i c l e
i n f o
Article history: Received 29 June 2013 Received in revised form 7 May 2014 Accepted 18 June 2014 Keywords: BMI Side impact Lower extremity injury Thoracic injury Abdomen injury Head injury
a b s t r a c t A comprehensive analysis is performed to evaluate the effect of weight, height and body mass index (BMI) of occupants on side impact injuries at different body regions. The accident dataset for this study is based on the National Automotive Sampling System-Crashworthiness Data System (NASS-CDS) for accident year 2000–08. The mean BMI values for driver and front passenger are estimated from all types of crashes using NASS database, which clearly indicates that mean BMI has been increasing over the years in the USA. To study the effect of BMI in side impact injuries, BMI was split into three groups namely (1) thin (BMI < 21), (2) normal (BMI 24–27), (3) obese (BMI > 30). For more clear identification of the effect of BMI in side impact injuries, a minimum gap of three BMI is set in between each adjacent BMI groups. Car model years from MY1995–1999 to MY2000–2008 are chosen in order to identify the degree of influence of older and newer generation of cars in side impact injuries. Impact locations particularly side-front (F), side-center (P) and side-distributed (Y) are chosen for this analysis. Direction of force (DOF) considered for both near side and far side occupants are 8 o’clock, 9 o’clock, 10 o’clock and 2 o’clock, 3 o’clock and 4 o’clock respectively. Age <60 years is also one of the constraints imposed on data selection to minimize the effect of bone strength on the occurrence of occupant injuries. AIS2+ and AIS3+ injury risk in all body regions have been plotted for the selected three BMI groups of occupant, delta-V 0–60 kmph, two sets (old and new) of car model years. The analysis is carried with three approaches: (a) injury risk percentage based on simple graphical method with respect to a single variable, (b) injury distribution method where the injuries are marked on the respective anatomical locations and (c) logistic regression, a statistical method, considers all the related variables together. Lower extremity injury risk appears to be high for thin BMI group. It is found that BMI does not have much influence on head injuries but it is influenced more by the height of the occupant. Results of logistic analysis suggest that BMI, height and weight may have significant contribution towards side impact injuries across different body regions. © 2014 Elsevier Ltd. All rights reserved.
1. Introduction Body mass index (BMI) of occupants has been one of the factors causing changes in the pattern and the distribution of serious injuries in car crashes. Mock et al. (2002) found that an increase in BMI would increase the risk of severe injuries in automobile
Abbreviations: BMI, body mass index; MY, model year; AY, accident year; CY, calendar year; AIS, abbreviated injury scale; MAIS, maximum abbreviated injury scale; NCAP, new car assessment program; DOF, direction of force; NASS CDS, National Automotive Sampling System Crashworthiness Data System; ROC, receiver operator characteristics. ∗ Corresponding author. Tel.: +91 4467448224.. E-mail address: [email protected] (M. Jeyabharath). http://dx.doi.org/10.1016/j.aap.2014.06.020 0001-4575/© 2014 Elsevier Ltd. All rights reserved.
accidents. Arbabi et al. (2003) identified that obesity would cause a negative effect due to increase in mass and momentum but have positive effect due to cushioning of extra tissue. Shankuan et al. (2006) found that underweight men have higher risk of injury than the normal ones but for women, BMI have no significant effect on change in risk of injury. The chest and pelvis are the most possible regions of the body to be injured in side impacts accidents and from NASS data it is observed that dominant source of contact causing pelvis and chest injuries is the side interior surface especially the door and its surrounding structures (Samaha and Elliott, 2003). Automobile accidents are also the major causes of serious thoracic and abdomen injuries in particular for lateral crashes, which contributes to thoracoabdominal injury (Siegel et al., 1993). Obese men had a broadly higher risk of injury, especially serious injury, to the upper body regions including head, face, thorax, and spine
194
C. Pal et al. / Accident Analysis and Prevention 72 (2014) 193–209
Table 1 Number of drivers and front passengers in the US NASS data CY2000–2008.
Table 2 Side impact dummies in NCAP.
Accident year
No. of drivers
No. of front passengers
Dummy
Height
Weight
BMI
2000–2008
50507
14166
ES2-male 50th %ile SID-IIs-female 5th %ile World SID-male 50th %ile
177.8
72.0
22.8
152.0
50.0
21.6
175.3
77.3
25.2
than normal weight men during frontal coliision (Shankuan et al., 2010). Side impact crashes are the second most important crashes with high frequency in real world accidents, resulting in one third of occupant fatalities (NHTSA, 2005). Compared to far-side (passenger) impacts, near side (driver) impacts are 2.3 times more common (Haland et al., 1993). Impact direction, impact location, vehicle speed before crash and change of impact velocity (delta-V) are the various crash related factors that could influence the severity of injuries (Terrel et al., 2003). Impact location (P, Y) has strong effects on the injury outcomes of nearside drivers. Serious thoracic injury is found to be more associated with impact location than impact direction (Xinghua et al., 2012). Oblique impact directions (10 o’clock and 2 o’clock) were found to be the most common injury crashes (Augenstein et al., 2003). In side impact, accidents leading to fatal injuries are slightly higher (32%) for 10 o’clock when compared to 9 o’clock impacts (29%) (Rattenbury et al., 2001). From our previous work (Pal et al., 2014), it was found that gender has significant contribution on pedestrian pelvis injuries. From the survey of past literature, it is found that only few studies have determined the influence of BMI in various car crash modes. Hence, an in-depth and a systematic study of the influence of BMI in side impact injuries is the key focus of the present study. 2. Methods 2.1. Trend of BMI in the USA In order to estimate the effect of BMI in side impact injuries, mean BMI is calculated for all occupants in all type of crashes reported in NASS-CDS from calendar year (CY) 2000–08. Number of drivers and front seat passengers involved are tabulated in Table 1. Mean BMI is calculated for drivers and front passengers plus drivers using NASS CDS data base for calendar year 2000–2008. From Fig. 1, it is observed that mean BMI of drivers and front passengers plus drivers are increasing every year. This agrees with the findings from prior studies (Jakobsson and Lindman, 2005) that US NASS data set shows an increase of mean BMI value from calendar year 1993 to 2003, which incorporates 67,000 occupants in all type of car accidents. In this present study, mean BMI of drivers are increased from 25.97 in CY2000 to 26.71 in CY2008. In the similar way, the mean BMI of front passengers plus drivers are increased
Fig. 1. Variation of mean BMI for driver and front passenger estimated from US NASS data.
from 25.65 in CY2000 to 26.45 in CY2008. This agrees well with trend of National Health Statistics in the USA, where continuous increase in BMI is presently a national concern for rising medical expenditure. It is to be noted that when front passenger is also included in mean BMI calculation, the mean BMI of front passengers plus drivers are less than that of mean BMI of drivers population. This is because children are also included in front seat passengers population. BMI of side impact dummies (ES2 and World SID) currently used in NCAP are tabulated in Table 2. To identify the effect of BMI in side impact injuries, the range of BMI values are categorized into three distinct groups (thin: <21, normal: 24–27 and obese: >30). Here, a gap of three points of BMI (21–24 and 27–30) is maintained between the selected BMI groups to extract the exact effect of BMI. Population percentage for drivers (marked in blue) and front passengers plus driver (marked in pink) involved in all type of crashes for three different BMI groups are shown in Fig. 2. It is seen from Fig. 1 that World SID 50th percentile male BMI lies within the selected normal BMI group and SIDII 5th percentile female BMI is just above the thin BMI group. It is also noted that BMI of World SID dummy is still less than mean BMI of the US driving population. It is observed from Fig. 2 that there are 16.6% of thin people, 23.6% of normal people and 19.9% of obese people population exist in total driver populations. Similarly, 14.6% of thin people, 24.2% of normal people and 20.8% of obese people population exist in total front passengers plus drivers populations. The groups of others include data between BMI 21–24 and BMI 27–30. None of the present side impact dummies currently used in NCAP are representing obese people group whose BMI values are above 30.
Fig. 2. Percentage distribution of different BMI groups in the US NASS data for front seat occupants. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
C. Pal et al. / Accident Analysis and Prevention 72 (2014) 193–209
195
Table 3 Driver side (DS) and passenger side (PS). Variable
Description
Code
508
Direction of force—DS
8, 9, 10, 28,29, 30, 48, 49, 50, 68, 69, 70, 88, 89,90 2, 3, 4, 22, 23, 24, 42, 43, 44, 62, 63, 64, 82, 83, 84 F, P, Y
Direction of force—PS 510 812
807 364
Fig. 3. Accident analysis variable for side impact front seat passengers including drivers.
2.2. Accident database Accident data used in this analysis are taken from the NASS-CDS which is a nationwide crash data collection program sponsored by the U.S Department of Transportation (DOT). About 5000 crashes are collected for light vehicles each year on the US roadways that are towed away from the accident scene due to heavy damage. The NASS-CDS data can be extrapolated to national estimates by using inflation factor (RATWGT) to represent a broad range of all police reported motor vehicle crashes. 2.3. Variable selection Accident data selection criteria of this analysis for drivers and front passengers are indicated in Fig. 3. These variables are also tabulated in Table 3 in Appendix A. Direction of force (DOF) considered for both near side and far side are 8 o’clock, 9 o’clock, 10 o’clock and 2 o’clock, 3 o’clock and 4 o’clock respectively. Impact locations namely F, P, Y corresponding to the locations of side-front, sidecenter, side distributed respectively are considered for this analysis. Table 4 Driver and front passengers without airbag cases delta-V (0–60 kmph).
809 810 107 306 Driver side (DS) and passenger side (PS)
Longitudinal and lateral location Occupant’s seat position—DS Occupant’s seat position—PS Age <60 Lateral component. delta-V—DS (0–60 kmph) Lateral component. delta-V—PS (−60–0 kmph) Height Weight Accidental year Model year
11 13
999 (Unknown removed) 999 (Unknown removed) 2000–2008 1995–2008
Lateral delta-V (0–60 kmph) constraint is imposed to nullify the effect of very high delta-V influence. Age range 0–60 years is considered to avoid very old people. This will remove the influence of bone strength in injuries because older people have softer bones than normal people. AIS2+ and AIS3+ injuries are considered in body regions such as head, thorax, abdomen, lower extremity, spine and upper extremity. Older cars with model year (MY)1995–1999 and newer cars with MY 2000–08 is considered to find the car model year influence on injuries. Table 4 and Table 5. 2.4. Approach To study the effect of BMI on side impact injuries, three approaches were adopted in the current study.
196
C. Pal et al. / Accident Analysis and Prevention 72 (2014) 193–209
Table 5 Raw and weighted data for MAIS2+ and MAIS3+ injuries. Injury Scale
AIS2+ injuries
AIS3+ injuries
Raw data
Weighted data
Raw data
Weighted data
BMI
<21
24–27
>30
<21
24–27
>30
<21
24–27
>30
<21
24–27
>30
No. of occupants Head Face Neck Thorax Abdomen Spine Upper extremity Lower extremity
96 93 9 1 63 35 27 37 104
90 110 4 0 85 52 20 42 85
96 100 5 0 109 68 29 60 74
7924 4511 442 22 2879 1456 230 2299 11355
8680 6576 275 0 4686 1740 1139 3962 6098
7858 5395 246 0 6696 3123 1233 4763 5962
65 66 1 0 57 6 4 10 44
62 83 0 0 74 14 4 14 34
74 76 0 0 87 20 6 18 24
4965 2682 140 0 2687 433 90 358 4157
4149 2408 0 0 3899 493 323 874 2374
5662 3374 0 0 5217 755 342 1970 2022
Fig. 6. AIS2+ injury risk in abdomen region for different model years; older and newer cars.
Fig. 4. AIS2+ injury risk in head region for different model years; older and newer cars.
Fig. 5. AIS2+ injury risk in thorax region for different model years; older and newer cars.
Fig. 7. AIS2+ injury risk in lower extremity region for different model years; older and newer cars.
C. Pal et al. / Accident Analysis and Prevention 72 (2014) 193–209
Fig. 8. AIS2+ injury risk in spine region for different model years; older and newer cars.
(a) Injury risk: Accident cases involving MAIS2+ cases were considered for this study. Within those MAIS2+ cases, AIS2+ injury risk is calculated by dividing the number of AIS2+ injuries within the BMI group by the number of AIS 1+ injuries within the same BMI group. for
AIS 2+Injury risk
A
BMI group
For B body region
Number of AIS 2+Injuries
[For B body region A BMI group] = Number of MAIS 2 + AIS 1 + Injuries [For all body region A BMI group]
Fig. 9. AIS2+ injury risk in upper extremity region for different model years; older and newer cars.
197
Fig. 10. AIS3+ injury risk in head region for different model years; older and newer cars.
(b) Injury distribution: Injuries for each case are plotted on the respective anatomic locations. By this approach, dominant injury locations and symmetrical conditions for individual BMI group can be identified. (c) Logistic regression: It models the relationship between a dependent and one or more independent variables and allows us to look at the fit of the model as well as at the significance of the relationships between dependent and independent variables. Logistic regression can be binomial or multinomial. Binomial or binary logistic regression deals with situations in which the observed outcome for a dependent variable can have only two possible types (for example, “dead” vs. “alive”). Multinomial logistic regression deals with situations where the
Fig. 11. AIS3+ injury risk in thorax region for different model years; older and newer cars.
198
C. Pal et al. / Accident Analysis and Prevention 72 (2014) 193–209
Fig. 12. AIS3+ injury risk in abdomen region for different model years; older and newer cars.
Fig. 14. AIS3+ injury risk in spine region for different model years; older and newer cars.
outcome can have three or more possible types (for example, “minor” vs. “serious” vs. “fatal” injuries). Binary logistic regression was used in this study to identify the significance of different variables for AIS2+ injury occurrence.
MY1995–1999 (old cars) and MY2000–2008 (new cars) and deltaV range of 0–60 kmph respectively. Figs. 4–9 gives the injury risk of AIS2+ injuries for drivers and front passengers involved in car accidents without airbag cases.
3. Results and discussion: 3.1. Injury risk (AIS2+ injuries) AIS2+ injury risk for different body regions such as head, thorax, abdomen, lower extremity, spine and upper extremity are plotted for three different selected BMI groups, two sets of car models
Fig. 13. AIS3+ injury risk in lower extremity region for different model years; older and newer cars.
3.1.1. Head Comparison of AIS2+ injury risk in head region for older cars and newer cars are shown in Fig. 4. This comparison will help us to understand the difference in injury trend between new and old car model years. It is seen from Fig. 4 that the head injury risk is slightly higher for newer cars compared to those of older ones. There is not much difference observed in injury risk percentage among three selected BMI groups.
Fig. 15. AIS3+ injury risk in upper extremity region for different model years; older and newer cars.
C. Pal et al. / Accident Analysis and Prevention 72 (2014) 193–209 Table 6 Selection criteria for injury distribution–pie chart (based on NASS-CDS Codebook). Injuries based on NASS-CDS codebook
Lower extremity region OIC BODY REGION 16 Pelvis 20 Thigh 11 Knee 12 Lower leg 17 Foot 25 Lower limb Abdomen region OIC SYSTEM ORGAN 17 Spleen 12 Liver 11 Kidney Others Spine region Specific anatomic structure Lumbar 6 4 Thoracic 2 Cervical
3.1.2. Thorax Comparison of AIS2+ injury risk in thorax region for older and newer cars are shown in Fig. 5. It is observed that the increasing trend of injury risk percentage from thin to obese people group. Injury risk percentage is always less for thin people group irrespective of car model years.
3.1.3. Abdomen Comparison of AIS2+ injury risk in abdomen region for older cars and newer cars are shown in Fig. 6. Injury risk percentage in abdomen region is less for thin group in newer car models.
199
Increasing trend of injury percentage is observed from thin to obese people group for newer cars. 3.1.4. Lower extremity Comparison of AIS2+ injury risk in lower extremity region for older cars and newer cars are shown in Fig. 7. It is observed that decreasing trend of injury risk percentage from thin to obese people group and it is more prominent for older cars. Injury risk percentage is always high for thin people group irrespective of car model years. It is also seen from Fig. 7 that the lower extremity injury risk is higher for older cars compared to newer cars. 3.1.5. Spine Comparison of AIS2+ injury risk in spine region for older cars and newer cars are shown in Fig. 8. Since the number of data is less, it is difficult to make any concrete conclusion. Further investigation needed by observing the trend with new data sets say for calendar year CY2009–2012. 3.1.6. Upper extremity Comparison of AIS2+ injury risk in upper extremity region for both older cars and newer cars are shown in Fig. 9. It is observed that injury risk percentage is almost same for both older and newer cars. 3.2. AIS3+ injury risk Within MAIS3+ cases, AIS3+ injury risk is calculated by dividing the number of AIS3+ injuries within the BMI group by the number of AIS 1+ injuries within the same BMI group. So far, the trend of AIS2+ injury risk for different body regions was discussed. Similarly, AIS3+ injury risk for all body regions are
Table 7 Selection criteria for injury distribution–pie chart (based on AIS-98 codebook). Injuries based on AIS-98 codebook
Thorax region Ribs Sternum Lungs Heart + inter ventricular septum Diaphragm Pneumothorax Vessels Head region Cerebellum + brain stem Cerebrum Skull-Base,Vault Unconsius
Type of anatomical structure 5 4
2 Type of anatomical structure 4 5 6
Specific anatomical structure 2 8 14 10, 13 6 22 2 Specific anatomical structure 2, 4 6 2,4 2,4,6,8,10
Table 8 Selection criteria for brain injury distribution–pie chart (AIS-98 Codebook based). Injuries based on AIS-98 codebook
Head Injury region
Brain stem Cerebellum
Cerebrum
Diffuse axonal injury Diffuse axonal injury Epidural Hematoma Intra cerebellar Hematoma Subdural Hematoma Sub arachnoid Hemorrhage Diffuse axonal injury Epidural Hematoma Intra cerebral Hematoma Subdural Hematoma Intra ventricular Hemorrhage Sub arachnoid Hemorrhage
Type of anatomical structure 4
Specific anatomical structure 2 4
6
Level of injury 6 6 14,18,22 26,30,34 38,42,46 66 28 30,32,34,36 38,40,42,44,46,48 50,52,54,56 78 84
200
C. Pal et al. / Accident Analysis and Prevention 72 (2014) 193–209
Fig. 16. AIS2+ lower extremity injury distribution across different BMI groups. (a)–(c) AIS2+ pelvis anterior region injury distribution. (d)–(f) AIS2+ pelvis posterior region injury distribution. (g)–(i) AIS2+ leg injury distribution. (j)–(l) Injury distribution percentage within lower extremity anatomical structure. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
plotted in Figs. 10–15. It is observed that injury risk trend of AIS3+ injuries are almost same as that of AIS2+ injuries. 3.3. Injury distribution AIS2+ injuries are plotted on their respective anatomical location and pie charts are used to highlight the injured segment or entity within each body region. Injury distribution study is carried out for different body regions namely lower extremity, thorax, abdomen, spine and head. This study aims to highlight the
proportion of injured body segment within each body region and symmetrical injury pattern variations with respect to BMI. Hence, only driver injury without airbag cases was considered for this section of study. Selection criteria for injured segment within each body region (used in pie chart) are tabulated in Tables 6–8 in appendix A 3.3.1. Lower extremity Lower extremity injuries for drivers across different BMI groups are plotted as points with different colors on the respective
C. Pal et al. / Accident Analysis and Prevention 72 (2014) 193–209
201
Fig. 17. AIS2+ injury proportions on gender basis. (a) All body region injuries and (b) lower extremity injuries.
anatomical locations as shown in Fig. 16. Lower extremity comprises of five major parts namely: pelvis, thigh, tibia, ankle and foot. For convenience, lower extremity injuries are classified into two groups (a) pelvis injuries and (b) leg injuries (thigh, tibia, ankle and foot). As shown in Fig. 16, the injury pattern is bilateral (both left and right side) for thin people group whereas injury pattern is left side dominant for obese people group. Pelvic injuries are dominant across all BMI groups, but decreasing trend in injury percentage of pelvis is observed with increase of BMI from thin to obese group. Thin people relatively have more chance to sustain pelvic injury (87%) possibly due to less fat tissue around the waist and needs further investigation. Leg injury percentage increases with the increase in BMI (13% in thin to 48% in obese BMI group), may be because the legs of obese group occupy more legroom space. Since anatomical profile and dimensions of pelvis is not same for male and female, gender can be an important factor. 3.3.2. Lower extremity—gender based AIS2+ injuries for male and female drivers for all BMI are plotted in Fig. 17. For all body injuries, both male and female share almost same level of injury percentage. However, for lower extremity injuries female injury percentage is dominant. AIS2+ injuries for males and females are plotted for different BMI groups as shown in Fig. 18. Pelvis injuries appear to be more dominant among male drivers. For female drivers, pelvis injuries appear to be more sensitive to the change in BMI. A decreasing trend in pelvis injury percentage is observed for females moving from thin to obese BMI group. Since the cases are very less, it is not possible to make any strong conclusion. 3.3.3. Thorax Thorax injuries for drivers across different BMI groups are plotted as points with different colors on the respective anatomical locations as shown in Fig. 19. Each point with different color indicates the set of injuries corresponding particular accident cases. Thorax region comprises of 12 pairs of thoracic ribs and organs like lungs, diaphragm, heart, etc. In Fig. 19 number of rib injury locations is high compared to other regions. As, multiple rib fracture is considered as single injury for pie and but marked in multiple locations, for additional details refer Table 9 in appendix A. Injury pattern is bilateral for thin people and left side dominant for normal people. An increasing trend of rib-injury percentage is observed from thin to obese BMI group. Rib injury may occur due to intrusion, for thin BMI group rib injury is 27% and lungs injury is 41%,
Fig. 18. AIS2+ injury proportions on gender and BMI basis. (a), (c) and (e) AIS2+ injury proportion for male across different BMI groups. (b), (d) and (f) AIS2+ injury proportion for female across different BMI groups.
which is of different pattern in obese group. Obese BMI group has 52% rib injury, 3% heart injury, 5% sternum injury, etc., shows more intrusion based injury and higher depth of injury comparing with thin BMI group. 3.3.4. Abdomen Abdomen injuries for drivers across different BMI groups are plotted as points with different colors on the respective anatomical locations as shown in Fig. 20. Each point with different color indicates the set of injuries corresponding particular accident cases. It is seen from Fig. 20 that spleen injury is always dominant across all BMI groups and this injury is highest for thin people group (75%) but the cases involved are very less. Kidney and liver injuries are higher in obese people group when compared with thin and normal BMI groups. The other injuries in abdomen region are retroperitoneum, mesentery, stomach, jejunum–ileum, bladder for all BMI groups. It also shows that injury pattern is distributed on both sides for obese people. This change in distribution of abdomen injury pattern is probably due to two possible factors (a) the gap between the abdomen and the surrounding structures and (b) the other could be the weight of the occupant. 3.3.5. Spine Spine injuries for drivers across different BMI groups are plotted as points with different colors on the respective anatomical locations as shown in Fig. 21. It is observed that there is no thoracic
202
C. Pal et al. / Accident Analysis and Prevention 72 (2014) 193–209
Fig. 19. AIS2+ thoracic injury distribution across different BMI groups. (a)–(c) AIS2+ ribs injury distribution. (d)–(f) AIS2+ lungs and diaphragm injury distribution. (g)–(i) AIS2+ heart, aorta and other thoracic cavity injury distribution. (j)–(l) Injury distribution percentage within thoracic anatomical structure. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
vertebrae injury for thin people. Lumbar injury is all most same pattern for all BMI group. Cervical vertebrae injury is higher for thin people group when compared to other higher BMI group. As number of spine injured cases reported in NASS CDS is less, it is difficult to arrive at any concrete conclusion. 3.3.6. Head Head injury distribution for driver with respect to BMI is shown in Fig. 22. It is seen that the injury percentage for all three BMI groups are almost same. From this observation, no concrete
conclusion could be inferred. Since BMI is the ratio of weight and square of height, height can also be a deciding factor to determine the injury trends, which will be discussed in the following section. To carry out height based study, the data has been segregated into three groups namely short (150–165 cm), medium (165–175 cm), tall (greater than 175 cm). Head injury distribution for driver with respect to height is shown in Fig. 23. It is observed that cerebrum injuries are always dominant for all height groups and this injury is highest for the short people group, which is 79%. The other two height groups (medium
Table 9 Sample case for explaining rib fracture injury distribution. Accident year = 2005, PSU:45, Case id: 109 Vehicle 1 > Occupant 1 > Injury codes Injury number
Source of data
AIS code
Injury description
Aspect
2
Post-ER medical record
4502324
Rib cage fracture >3 ribs on one side and ≤3 ribs on other side, stable chest or NFS = with hemo-/pneumothorax
L Rib 4, L Rib 5, L Rib 6, L Rib 7, L Rib 8, L Rib 9, L Rib 10, L Rib 12
Number of rib injury = 1 (used for Pie chart) Number of rib injured location = 8 (used for anatomical markings)
C. Pal et al. / Accident Analysis and Prevention 72 (2014) 193–209
203
Fig. 20. AIS2+ abdomen injury distribution across different BMI groups. (a)–(c) AIS2+ spleen, liver, kidney and other abdomen injury distribution. (d)–(f) Injury distribution percentage within abdomen anatomical structure. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
and tall) the cerebrum injury is almost same 56%. The skull injury percentage is higher for tall people group (27%) when compared to short people group (12%). Head of the taller people is close to the roof side rail so there is a higher probability of contact injury and may need further investigation. Jillian et al. (2012) considered the following injuries occurring in the brain to understand the brain injury mechanism as shown in Fig. 24. These are as follows: (1) Sub-dural hematoma: bleeding inside the sub-dural space. (2) Intracerebral hematoma: bleeding inside the brain. (3) Epidural hematoma: bleeding between the skull and the dura mater. (4) Intraventricular hemorrhage: bleeding into the brains ventricular system. (5) Sub-arachnoid hemorrhage: bleeding into the sub-arachnoid space. Brain injury type for driver with respect to height is shown in Fig. 25. It is seen that subarachnoid type of injury is higher in tall people group (50%) when compared to the short people group (29%). Since in this type of injury bleeding is in the subarachnoid space, the depth of injury could be high. Hence, the depth of injury for taller people could be higher, when compared to short people. The intraventricular hemorrhage and epidural hematoma type of injury is higher in small people group when compared to the tall people group. The height of the occupant influences the seat track position for the driver with respect to the B-pillar position. Hence, the degree of interaction of the occupant with the intruding vehicle door and the structures around it changes with height of
the occupant. Therefore, tall people may have more possibility to hit B-pillar or roof side rail and their depth of injury can be more. These observations need further investigation before making any concrete statement.
3.4. Comparative study for “with and without weighting factor” cases “With weighting factor” data are those, injury counts are multiplied by their respective inflation factor. “Without weighting factor” is a raw data without multiplying by any weighting factor. Fig. 26 shows the AIS2+ injury risk percentage within BMI group for individual body regions “with and without weighting factor” for 0–60 kmph delta-V for MY1995–2008 incorporating driver plus front passengers. Decreasing trend of injury risk percentage is observed moving from thin to obese BMI group for lower extremity, is common for both “with and without weighting factor”. Increasing trend of injury risk percentage is observed from thin to obese BMI group for thorax, is common for both “with and without weighting factor”. Increasing trend of injury risk percentage is observed from thin to obese BMI group for abdomen, is same for both “with and without weighting factor”. No definite trend is observed for head in both “with and without weighting factor” for thin and obese BMI group. Raw and weighted data for MAIS2+ and MAIS3+ injuries are tabulated in Tables 4 and 5 in Appendix A.
204
C. Pal et al. / Accident Analysis and Prevention 72 (2014) 193–209
Fig. 21. AIS2+ spine injury distribution across different BMI Groups. (a)–(c) AIS2+ lumbar, cervical and thoracic spine injury distribution. (d)–(f) Injury distribution percentage within spine anatomical structure. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
Fig. 22. AIS2+ injury distribution for head across different BMI groups (a) AIS2+ head injury distribution for BMI < 21 group. (b) AIS2+ head injury distribution for BMI: 24–27 group. (c) AIS2+ head injury distribution for BMI > 30 group.
C. Pal et al. / Accident Analysis and Prevention 72 (2014) 193–209
205
Fig. 23. AIS2+ injury distribution for head across different height groups (a) AIS2+ head injury distribution for height 150–165 cm. (b) AIS2+ head injury distribution for height 165–175 cm. (c) AIS2+ head injury distribution for height > 175 cm.
3.5. Comparative study for Impact locations F, P and Y and F, P, Y, Z and D So far, the trend of AIS2+ and AIS3+ injury risk for different body regions was discussed based on impact locations F, P and Y and other variable mentioned in the previous sections. It is also noteworthy to find out the effect of impact locations D and Z in side impact injuries. Fig. 27 shows the AIS2+ injury risk percentage for MY 1995–2008 including driver plus front passengers within respective BMI groups for different impact locations. It is observed that injury risk trend is almost same for all body regions. Fig. 28 shows the accident analysis variable incorporating impact locations Z and D in addition to F, P and Y.
3.6. Comparative study for “with and without side airbag” cases “With airbag” are those cases when either one of the side airbags (seat-back or roof-side rail or door-panel mounted) is in deployed conditions and “without airbag” are those cases without any side airbag available or not deployed at the time of crash. Total number of “with airbag” cases are very less, almost one-tenth of “without airbag” cases. Hence, it is not possible to make any definite conclusion by “with airbag” results.
Fig. 24. Commonly occurring injuries in the brain.
Fig. 29 shows the AIS2+ injury risk percentage within BMI group including “with and without airbag” cases across different body regions plotted for drivers considering MY1995–08 and Delta-V 0–60 kmph. Overall there is no appreciable difference in the injury risk trend for “with and without airbag” cases with small exception in lower extremity.
Fig. 25. AIS2+ injury distribution for head across different height groups. (a)–(c) Different type of injuries distributed in brain for different height groups.
206
C. Pal et al. / Accident Analysis and Prevention 72 (2014) 193–209
Fig. 26. AIS2+ injury risk % within respective BMI group with and without weighting factor. AIS2+ injury risk % (a) and (b) for lower extremity (c) and (d) for thorax (e) and (f) for abdomen (g) and (h) for head.
3.7. Logistic regression In this section, logistic regression was carried out to identify the significance of various parameters in predicting AIS2+ injury for different body regions. Contributing rate of parameters changes with the type of dataset chosen. All the vehicles will have a driver and hence, the number of driver cases reported in NASS-CDS database is much more than other occupants in the vehicle. Hence, a dataset was chosen only with drivers. Height of an occupant will affect
the seat track position with respect to B-pillar. Therefore, belt path through thorax and abdomen region of occupant and seat track position with respect to seat location (belted driver track position with steering wheel and belted front passenger track position) may change with the height of the occupant. Hence, a dataset was also chosen involving belted occupant. The number of “with side airbag deployed” cases are very less and they are omitted for this study. Driver without airbag-deployed dataset (includes belted and unbelted drivers) and belted driver plus front passenger without
C. Pal et al. / Accident Analysis and Prevention 72 (2014) 193–209
207
Fig. 27. AIS2+ injury risk % within respective BMI groups for different impact locations. AIS2+ injury risk % (a) and (b) for lower extremity (c) and (d) for thorax (e) and (f) for abdomen (g) and (h) for head.
airbag-deployed dataset were chosen for this study. Variables considered for the logistic regression analysis are tabulated in Table 10 of Appendix A. The above mentioned datasets were evaluated with “BMI” and “height and weight” separately and the results are tabulated in Tables 11 and 12 in Appendix A. BMI, height and weight were not split into any individual groups and they were treated as continuous variable. Based on the observation from Tables 11 and 12 in Appendix A: Head: Age and lateral delta-V may have more influence on the head injuries. Height of the occupant may have some influence
on the head injuries and this agrees with the injury distribution mentioned in Section 3.3.6. Thorax: Age and lateral delta-V may have more influence on the thorax injuries. Height and gender of the occupant may have some influence on the thorax injuries. Compared to the degree of influence of age, delta-V, height and gender on thorax injuries, those of BMI and model year on thorax injuries appear to be less. These results are contradicting from Sections 3.1.2 and 3.3.3, which need further investigations. Abdomen: Lateral delta-V may have more influence on the abdomen injuries. BMI of the occupant may have some influence on
208
C. Pal et al. / Accident Analysis and Prevention 72 (2014) 193–209 Table 10 Logistic regression variables. Variable
Description
510 812 807 364 809 810 306 854 324 309 821 110 808
Longitudinal and lateral location Occupant’s seat position—driver and passenger Age <60 Lateral component delta-V (0–60 kmph) Height Weight Model year Seat track Vehicle curb weight Body type Manual seat belt use Multi impact Sex BMI Theta (angle between lateral and longitudinal Delta V) Side airbag deployed during crash
Fig. 28. Accident analysis variable incorporating impact locations Z and D in addition to F, P and Y.
the abdomen injuries. These match well with the results of Sections 3.1.3 and 3.3.4. Lower extremity: Lateral delta-V may have more influence on the lower extremity injuries. BMI and gender of the occupant may have some influence on the lower extremity injuries and this supports
the statements of the previous Sections 3.1.4 and 3.3.1. Nevertheless, compared to the degree of influences of other factors, model year influence on lower extremity injuries appears to be low.
Table 11 Logistic regression result based on BMI.
Head Thorax Abdomen Lower extremity Spine Upper extremity
WoAB-driver WoAB-belted WoAB-driver WoAB-belted WoAB-driver WoAB-belted WoAB-driver WoAB-belted WoAB-driver WoAB-belted WoAB-driver WoAB-belted
MY
Age
0.52 0.50 0.16 0.91 0.87 0.14 0.81 0.69 0.33 0.01 0.21 0.37
0.00 <0.0001 0.00 <0.0001 0.23 0.63 0.77 0.73 0.88 0.75 0.01 0.01
LAT Del V 0.01 0.03 0.00 0.00 <0.0001 <0.0001 0.00 0.00 0.09 0.00 0.71 0.82
Gender 0.85 0.54 0.40 0.13 0.80 0.40 0.05 0.30 0.00 <0.0001 0.36 0.91
BMI
Theta
ROC
N
0.59 0.18 0.66 0.63 0.12 0.02 0.04 0.03 0.76 0.61 0.47 0.83
0.12 0.11 0.84 0.98 0.93 0.16 0.76 0.13 0.44 0.75 0.77 0.18
0.74 0.74 0.72 0.70 0.74 0.75 0.75 0.73 0.70 0.76 0.67 0.67
338 312 338 312 338 312 338 312 338 312 338 312
Table 12 Logistic regression result based on height and weight. MY Head Thorax Abdomen Lower extremity Spine Upper extremity
Head Thorax Abdomen Lower extremity Spine Upper extremity
WoAB-Driver WoAB-Belted WoAB-Driver WoAB-Belted WoAB-Driver WoAB-Belted WoAB-Driver WoAB-Belted WoAB-Driver WoAB-Belted WoAB-Driver WoAB-Belted
WoAB-driver WoAB-belted WoAB-driver WoAB-belted WoAB-driver WoAB-belted WoAB-driver WoAB-belted WoAB-driver WoAB-belted WoAB-driver WoAB-belted
0.46 0.46 0.17 0.97 0.84 0.14 0.81 0.66 0.30 0.01 0.18 0.36
Age
LAT Del V
0.00 <0.0001 <0.0001 <0.0001 0.20 0.55 0.74 0.73 0.89 0.76 0.01 0.01
0.00 0.02 0.00 0.00 <0.0001 <0.0001 0.00 0.00 0.09 0.00 0.71 0.83
Gender 0.53 0.10 0.04 0.04 0.27 0.76 0.12 0.97 0.00 <0.0001 0.06 0.64
Height
Weight
Theta
ROC
N
0.19 0.01 0.03 0.12 0.08 0.16 0.48 0.69 0.08 0.00 0.14 0.52
0.65 0.17 0.92 0.75 0.21 0.03 0.06 0.03 0.77 0.24 0.30 0.58
0.13 0.15 0.74 0.88 0.94 0.19 0.76 0.10 0.58 0.51 0.65 0.15
0.74 0.75 0.73 0.71 0.74 0.75 0.75 0.73 0.71 0.79 0.68 0.68
338 312 338 312 338 312 338 312 338 312 338 312
MY
Age
LAT Del V
Gender
BMI
Theta
ROC
N
0.52 0.50 0.16 0.91 0.87 0.14 0.81 0.69 0.33 0.01 0.21 0.37
0.00 <0.0001 0.00 <0.0001 0.23 0.63 0.77 0.73 0.88 0.75 0.01 0.01
0.01 0.03 0.00 0.00 <0.0001 <0.0001 0.00 0.00 0.09 0.00 0.71 0.82
0.85 0.54 0.40 0.13 0.80 0.40 0.05 0.30 0.00 <0.0001 0.36 0.91
0.59 0.18 0.66 0.63 0.12 0.02 0.04 0.03 0.76 0.61 0.47 0.83
0.12 0.11 0.84 0.98 0.93 0.16 0.76 0.13 0.44 0.75 0.77 0.18
0.74 0.74 0.72 0.70 0.74 0.75 0.75 0.73 0.70 0.76 0.67 0.67
338 312 338 312 338 312 338 312 338 312 338 312
C. Pal et al. / Accident Analysis and Prevention 72 (2014) 193–209
209
4. Conclusion Based on NASS-CDS database, the mean BMI of the USA population increases every year. Hence, it will be important to study to effect of BMI on injuries during vehicle crashes. BMI of an occupant has significant influence on the occurrence of lower extremity injuries. Lower extremity injury risk appears to be high for thin people group. Gender of an occupant may have some influence on the lower extremity injuries. BMI of an occupant has significant influence on the occurrence of abdomen injuries. Abdomen injury risk appears to be low for thin people group. Age of an occupant has significant influence on the occurrence of head, thorax and upper extremity injuries. Height of an occupant may have some influence on the occurrence of head injuries. Results of logistic analysis suggest that, for better understanding BMI’s influence, it is necessary to investigate the effect of weight and height. Appendix A. See Tables 3–12. References
Fig. 29. AIS2+ injury risk % within BMI group across different body regions (a) with airbag (b) without airbag (c) sum of with and without airbag. AIS2+ injury risk % within BMI group across different body regions (a) with airbag (b) without airbag (c) sum of with and without airbag.
Spine: Lateral delta-V and gender may have more influence on the spine injuries. Height of the occupant appears to have some influence on the spine injuries. These results are different from those of Sections 3.1.5 and 3.3.5, which needs further investigations. However, one should note that the total number of spine injuries is very less and coming to any concrete conclusion is not appropriate. Upper extremity: Age of the occupant may have more influence on the upper extremity injuries. Other parameters like gender, height, model year and BMI do not have significant influence on the upper extremity injuries. These results support the findings from Section 3.1.6. Overall-all body region: Height of the occupant has influence on side impact injuries for head, thorax and spine body regions and weight of the occupant has influence on side impact injuries for abdomen and lower extremity body regions.
Arbabi, S., Wahl, W.l., Hemmila, M.R., Kohoyda-Inglis, C., Taheri, P.A., Wang, S.C., 2003. The vushion effect. J. Trauma 54 (6), 1090. Augenstein, J., Perdeck, E., Stratton, J., Digges, K., Bahouth, G., 2003. Characteristics of crashes that increase the risk of serious injuries. Annu. Proc. Assoc. Adv. Automot. Med. 47, 561–576. Haland, Y., Lovsund, P., Nygren, A., 1993. Life-threatening and disabling injuries in car-to-car side affect—implications for development of protective systems. Accid. Anal. Prev. 25 (2), 199–205. Jillian E. Urban, Sarah Lynch, Christopher T., Whitlow, et al., Sep 2012. Understanding brain injury mechanism: integrating real-world lesions, ATD response and finite element modeling. CIREN Public Meeting. Jakobsson, Lotta, Lindman, Magdalena, 2005. Does BMI Influence the occupant injury risk pattern in car crashes. IRCOBI Conference—Prague (Czech Republic). Mock, C.N., Grossman, D.C., Kaufman, R.P., Mack, C.D., Rivara, F.P., 2002. The relationship between body weight and risk of death and serious injury in motor vehicle crashes. AAP 31, 221–228. National Highway Traffic Safety Administration, 2005. Safety Facts. NHTSA, US DOT, Washington, DC, Report #HS 810 6312005. Pal, C., Okabe, T., Vimalathithan, K., Manoharan, J. et al. Estimation of Pelvis Injuries and Head Impact Time using Different Pedestrian Human FE Models, SAE Technical Paper 2014-01-0522, 2014, doi:10.4271/2014-01-0522. Rattenbury, S.J., Gloyns, P.F., Nolan, J.M., 2001. Vehicle deformation in real-world side impact crashes and regulatory crash tests. In: Proceedings of 17th International Technical Conference on the ESV, Amsterdam, Netherlands, Paper 356. Samaha, R.R., Elliott, D.S., 2003. NHTSA side impact research: motivation for upgraded test procedures. In: Proceedings of the 18th Conference on the Enhanced Safety of Vehicles, Paper 492, Nagoya, Japan. Siegel, J.H., Mason-Gonzalez, S., Dischinger, P., Cushing, B., Read, K.R., Robinson, R., Smialek, J., Heatfield, B., Hill, W., Bents, F., Jackson, J., Livingston, D., Clark, C.C., 1993. Safety belt restraints and compartment intrusions in frontal and lateral motor vehicle crashes: mechanisms of injuries, complications, and acute care costs. J. Trauma 5, 736–759. Shankuan, Zhu, Jong-Eun, Kim, Xiaoguang, Ma, 2010. BMI and risk of serious upper body injury following motor vehicle crashes: concordance of realworld and computer-simulated observations. PLoS Med. 7 (3), e1000250, http://dx.doi.org/10.1371/journal.pmed.1000250. Shankuan, Zhu, Layde, P.M., Guse, C.E., Laud, P.W., Pintar, F., Nirula, R., Hargarten, S., 2006. Obesity and risk for death due to motor vehicle crashes. Am. J. Publ. Health 96 (4), 734–739, http://dx.doi.org/10.2105/AJPH.2004.058156 (Impact Factor: 3.93). Terrel, M., Newland, C., Seyer, K., 2003. The dependence of side impact injury risk on MDB configuration. In: Proceedings of the 18th Conference on the ESV, Nagoya, Japan. Paper no. 399. Lai, Xinghua, Ma, Chunsheng, Hub, Jingwen, Zhoua, Qing, 2012. Impact direction effect on serious-to-fatal injuries among driver in near-side collisions according to impact location: focus on thoracic injuries. Accid. Anal. Prev. 48, 442–450.