Abdominal Injuries in Belt-Positioning Booster Seats

Abstract

Previous studies have demonstrated that booster seats reduce the risk of abdominal injuries by improving the fit of the seat belt on young children and encouraging better posture and compatibility with the vehicle seat. Recently, several studies have reported cases of abdominal injuries in booster seated children questioning the protective effects of these restraints. The objective of this study was to examine cases of abdominal injuries in booster seated children through parametric modeling to gain a thorough understanding of the injury causation scenarios. The Partners for Child Passenger Safety and CIREN in-depth crash investigation databases were queried to identify children in belt-positioning booster seats with abdominal injuries. The injury causation scenarios for these injuries were delineated using the CIREN Biotab method. The cases were modeled, using MADYMO with variations in key parameters, to determine the ranges of loads and loading rates for the abdomen and thorax. A parametric study was completed examining the influence of pretensioners and load limiters on the injury metrics obtained. Query of the two databases revealed three cases involving abdominal injuries to booster seated children. Children in two of the cases sustained a thoracic injury (AIS 3/AIS 4) in addition to their abdominal injuries (AIS 2) and review of these cases pointed to the role of shoulder belt loading in the injury causation. Modeling of these cases revealed chest compressions and accelerations of 30–53 mm and 41–89 g, respectively and abdominal deflection and velocity of 7.0–13.3 mm and 1.2–2.2 m/s, respectively. Parametric study suggested that coupling shoulder belt load limiting and lap belt buckle pretensioning resulted in improved chest and abdominal metrics while reducing head excursion, indicating that these technologies may provide injury reduction potential to pediatric rear seat occupants.

INTRODUCTION

The abdomen is the second most commonly injured body region in children using adult seat belts and injuries to this region can be associated with extended hospitalization and significant medical costs [Bergqvist, Hedelin, Lindblad et al., 1985, Durbin, Arbogast, Moll, 2001, Trosseille, Cassan, Schrooten, 2001, Tso, Beaver, Halter, 1993]. Reports of injuries to this region have been focused on seat belt induced abdominal injuries and lumbar spine fractures, the constellation of injuries known as “seat belt syndrome” [Garrett and Braunstein, 1962, Hoy and Cole, 1993, Kulowski and Rost, 1956, Lane, 1994]. The issue of seat belt syndrome, although identified decades ago, has received increased attention recently as focus has been placed on reducing the risks of seat belt syndrome in children. Children of all ages are at risk of developing seat belt syndrome, however the poor fit of the belt in children less than 8 years of age likely places them at higher risk than their older pre-teen and teenage counterparts.

Our previous work reviewed real-world cases of children restrained in adult seat belts and identified the mechanisms of abdominal injuries [Arbogast, Kent, Menon, et al., 2007]. Seat belt loading at the direct site of the injury was the most common injury causation scenario although the belt type and the relative position of the belt to the initial position of the child varied. Three scenarios were identified that resulted in seat belt syndrome injuries. These involved classic submarining, in which the child’s torso slides under the belt at the time of the crash, poor initial belt placement, in which the seat belt was placed across the abdomen before the crash event, and submarining with jackknifing, in which the child’s upper torso flexes over the belt while possibly submarining under the belt during the crash event. These injury causation scenarios may have been the result of several vehicle and child factors including seat belt geometry not ideal for children (e.g. a shallow lap belt angle), poor initial position of the child with the shoulder belt behind the back or slouched posture to position the knees over the edge of the seat. The use of a belt positioning booster seat improves these factors by improving the fit of the seat belt and encouraging better posture and compatibility with the vehicle seat itself.

Belt positioning booster seats are the recommended restraint for 4 to 8 year old children according the American Academy of Pediatrics and National Highway Traffic Safety Administration and reduce the risk of injury to children age 4–7 years old by 59% compared to similar age children in adult seat belts [Durbin, Elliott, Winston, 2003]. In this previous work, the reduction in injury risk was particularly evident in the abdomen, resulting in 0 injuries per 1,000 booster seat restrained children in crashes versus 4.4 per 1,000 for children using adult belts. This analysis, conducted on data from 1998 to 2002, was based primarily on children age 4 and 5 years of age due to the usage practices during that time period. Due in large part to the increased attention paid to the needs of children in motor vehicle safety beginning in the mid-1990’s, the period between 1999 and 2007 witnessed large increases in reported appropriate restraint use (including child safety seats, booster seats, and combination seats) by children from 4 to 8 years of age. From 1999 to 2007, there was a nearly threefold increase in child restraint use among 3 to 8 year old children in crashes. The largest relative increase in CRS use was for the oldest age group (6 to 8 years of age) [Partners for Child Passenger Safety, 2008]. As more children, in particular children 6 to 8 years of age, are appropriately restrained in booster seats, continued monitoring of their real-world experience is paramount.

Recently, the issue of abdominal injury prevention by booster seats has been questioned by other researchers. Several studies have reported the occurrence of these injuries in other field studies. Trosseille et al described findings from two crash investigation based studies in France [Trosseille, Chamourd, Tarriere, 1997]. They described abdominal injuries sustained by eleven booster-seated children who ranged in age from 3 to 7 years, but reported few crash or restraint use details such as impact type, severity, booster seat type or presence of misuse. More recently, Johannsen used data from the European CHILD (CHild Injury Led Design) project to reconstruct real-world crash events in an effort to validate newly designed abdominal sensors for the Q family of ATDs. In his study, he reconstructed four cases of frontal impacts involving abdominal injury in booster-seated children who ranged in age from 2½ to 4 years [Johannsen, Alonzo, Goubel, et al., 2005]. In a recent study of booster-seated children in Australia, Brown reported on 2 children, aged 2 years and aged 3 years, who sustained abdominal injuries, one as the result of a frontal impact and one as the result of a side impact [Brown and Bilston, 2006].

Due to the changing nature of the booster use landscape and these anecdotal reports of abdominal injuries in booster seat restrained children, this issue deserves further investigation. Real-world crash data involving child occupants represent the best means by which to understand the potential mechanism of these injuries and role of the booster seat in the injury causation pathway. The objectives of this study were two-fold. First, real-world cases of abdominal injuries in booster seated children were reviewed and modeled using MADYMO to understand the injury causation scenarios and injury mechanisms. Second, using an examplar crash scenario derived from the real-world cases, the effect of pre-tensioning and load limiting on the injury metrics was quantified through parametric study.

METHODS

Crash Reconstruction

In-depth crash investigations from two data sources were reviewed: Partners for Child Passenger Safety (PCPS) and the National Highway Traffic Safety Administration’s (NHTSA) Crash Injury Research and Engineering Network (CIREN). Each study was approved by the Institutional Review Boards of all participating hospitals and universities. Cases from PCPS and all CIREN Centers were reviewed and all cases involving a child restrained in a belt-positioning booster sustaining an AIS2+ abdominal injury were included in this analysis. There were no occupant age, height, or weight inclusion or exclusion criteria but rather inclusion was based only on restraint and injury. The case ascertainment process is described for each of the data sets followed by the in-depth investigation methods, which are similar between the two data sets.

For the cases obtained from PCPS dataset, data collected from December 1, 1998, to December 31, 2006 were used in this analysis. Detailed descriptions of the study population and methods involved in data collection and analysis have been previously published [Durbin, Bhatia, Holmes et al., 2001]. PCPS consists of a large scale, population based, child-specific crash surveillance system in which insurance claims from State Farm Insurance Co. (Bloomington, IL) serve as the source of subjects. Crashes qualifying for inclusion were those involving at least one child occupant less than16 years of age riding in a model year 1990 or newer State Farm-insured vehicle. Qualifying crashes were limited to those that occurred in fifteen states and the District of Columbia, representing three large regions of the United States (East: NY, NJ [until 11/01], PA, DE, MD, VA, WV, NC, DC; Midwest: OH, MI, IN, IL; West: CA, NV, AZ, TX [starting 6/03]). On a daily basis, data from qualifying and consenting claims were transferred electronically from all involved State Farm field offices to researchers at The Children’s Hospital of Philadelphia and University of Pennsylvania. Data in this initial transfer included contact information for the insured, the ages and genders of all child occupants, and a coded variable describing the medical treatment received by all child occupants. From these consenting claims, a series of cases were chosen for in-depth crash investigation based on manual review of claims files to identify children with moderate to severe injuries. Cases were screened via telephone with the policyholder to confirm the restraint status and medical details of the case. An in-depth investigation was then initiated following the methods described for the CIREN cases below.

For the cases obtained from CIREN dataset, Data collected from October 1, 1996 to December 31, 2006 were used in this analysis. NHTSA coordinates eight trauma centers that enroll cases in the CIREN system meeting several specific criteria. Children who present to the participating trauma center following a motor vehicle crash are screened using specific criteria such as vehicle and crash characteristics and the injury and restraint status of the child. Once a subject met the criteria for enrollment and consented to participation in the study, an in-depth investigation was initiated following the methods described below.

For each enrolled crash the following data were obtained: medical data on injured occupants, crash scene data, and vehicle damage information. Each crash scene and vehicle investigation used the format established by the National Automotive Sampling System (NASS). Scaled documentation of each crash site was performed including the collection of information on the roadway, traffic controls, road surface type, conditions, and road grade at pre- and post-impact locations. Physical evidence such as tire skid marks was used to determine the heading angle and post impact trajectory of the colliding vehicles. A scaled drawing of the impact and final rest positions of the vehicles was used to assist in the calculation of the speed and force of the impact.

Exterior vehicle inspections included detailed measurements of direct and indirect damage. Using a contour gauge, a damage crush profile was obtained for each vehicle from the front bumper or side plane and a specific collision deformation code (CDC), which incorporates the principal direction of force (PDOF), was assigned. These measurements were entered into a crash analysis program (Win SMASH, version 1.2.1, U.S. Department of Transportation) which estimates the change in velocity (delta-v) of the vehicle during impact and the energy absorbed during the crash event. Inspection of the interior of the vehicle was completed to determine the exact points of contact and restraint system use by the vehicle occupants. Contact points were identified using scuff patterns and skin transfers. The interior inspection also included an assessment of the integrity of the passenger compartment and measurement of component intrusion.

Injury data were compiled following review of individual hospital reports and radiographs. Injuries were coded using the 1998 AIS and an overall injury severity score (ISS) was calculated. Patient interviews were also performed to supplement crash and injury data.

Both CIREN and PCPS data sets were reviewed and three cases of abdominal injury in booster-seated children in frontal impacts were identified. Based on the crash investigations, all children were using the belt-positioning booster seat according to manufacturer’s instructions. There were other cases of abdominal injuries in booster seated children; however they were side impact collisions. Side impact crashes were excluded from the study, as the abdominal injury causation in side impact crashes likely involves the vehicle interior structure rather than the restraint system, which was the focus of this analysis. The following is a brief description of the three frontal impact cases studied:

1. The case occupant was a 4 year old male (104 cm, 18 kg), seated in a high back booster with a lap and shoulder belt in the right rear seating position, and was involved in a moderate frontal offset collision. He sustained a liver laceration and pancreatic contusion.

   • Delta-v: 21 km/h (13 mph)
   • PDOF: 340 degrees
   • Max Crush: 38 cm
   • Vehicle: 1996 Toyota SUV
2. The case occupant was a 4 year old female (109 cm, 22 kg), seated in a high back booster seat in the left rear seating position and restrained using a lap and shoulder belt, and was involved in a moderate frontal impact. She sustained bilateral lung contusions, a liver laceration, and rib fractures.

   • Delta-v: 50 km/h (28 mph)
   • PDOF: 0 degrees
   • Max Crush: 52 cm
   • Vehicle: 2000 Pontiac 4-door sedan
3. The case occupant was a 5 year old male (117 cm, 18 kg), seated in a backless booster with lap and shoulder belt in the third row right seating position, and was involved in a severe frontal offset collision. He sustained a lung contusion and splenic laceration.

   • Delta-v: 72 km/h (43 mph)
   • PDOF: 340 degrees
   • Max Crush: 69 cm
   • Vehicle: 1999 Chrysler minivan

All lap and shoulder belts were non-pretensioned and non-load limited belt systems. These cases were reviewed in-detail to determine the mechanisms of abdominal injury sustained by booster-seated children. The case review documented crash parameters of interest including crash severity, contact points, and restraint status. These parameters were examined in conjunction with detailed medical findings to assess the most probable source and mechanism of the injury.

Following review of the case, AIS 2+ injuries were coded using the CIREN BioTab method developed by L.W. Schneider [personal communication, Ann Arbor, MI 2007]. The BioTab approach to analyzing occupant injuries in a crash allows the researcher to attribute one or more Injury Causation Scenarios (ICS) to each injury, where each ICS includes the set of all factors that the researcher believes are essential for the injury to have occurred. Each ICS includes Involved Physical Components (IPC) (structures external to the occupant) that represent those structures that contacted the occupants and through the contact, caused the injury. Because it is not always possible to know for sure what caused an injury, the researcher must also assign confidence levels of Certain, Probable, and Possible to each ICS and to each involved physical component within each ICS.

Each case was reconstructed using computational simulations in order to fully understand the kinematics of and loading on the case occupant. For each case, the vehicle dynamics were reconstructed using HVE. This provided the overall timing and crash pulse associated with the crash events and defined the input into MADYMO, which allowed modeling of the complex nature of the seat belts and provided detailed information on the kinematics of the occupant. These MADYMO reconstructions provided an overall understanding of the occupant motion, belt loads and injury metrics in the given crash circumstances and allowed us to fully understand the interaction between the seat belt and the child’s abdomen and thorax.

For the HVE simulations the case vehicles were either selected from the vehicle database in HVE if available or a generic vehicle model was used when not available. The exterior vehicle specifications (front overhang, rear overhang, overall length and width, wheelbase and weight) of these vehicle models were updated based on the exterior specifications for the vehicles obtained from the crash investigation. The position of the vehicles and their initial velocities were assigned based on the crash investigation. To generate a valid crash pulse for the cases, Principal Direction of Force (PDOF), Collision Deformation Classification (CDC), Crush and Delta-V were matched between crash investigation and the HVE simulation by carrying out parametric variations of impact location, vehicle velocities, inter-vehicle friction, and post-impact motion. Care was taken to make sure that the post-impact motion obtained for both the case vehicle and the other vehicle in the HVE simulation were consistent with the information in crash investigation report. Once the specified parameters were matched between the HVE simulation and the crash investigation, the crash pulse was then obtained from HVE simulation and used for occupant simulation.

The detailed occupant kinematics were studied by conducting simulations using MADYMO software (version 6.2). For proper representation of the case occupant, the MADYMO Hybrid III 6-year-old ATD model was geometrically scaled to the size of the child as provided in the crash investigation using the inbuilt scaling algorithm available in the software. The MADYMO model consisted of the vehicle interior components i.e. occupant seat, front (driver or passenger) seat, side doors and the booster seat type used by the occupant based on the information in the crash investigation report. The properties for the interior vehicle structure and the contact characteristics between the occupant model and the vehicle interior were taken from the frontal impact application file available in MADYMO. To the extent possible, the booster seat type and belt configuration was matched to the crash investigation data. Particular attention was focused on modeling the proper seat and belt anchorage geometry. The booster seats used in the MADYMO modeling were ellipsoid models and were developed using the geometry of the booster seats from real-world cases. The material properties used for these seats were generic and were based on the data published in a previous study [Menon, Ghati, Jain, 2007].

In order to account for the variations in the outcome due to the uncertainty in the occupant interaction with the vehicle interiors during a crash, several parameters were varied in order to provide a realistic spectrum of occupant kinematics and injury metrics for a given crash scenario. (Table 1) Not all parameters were relevant for each case and thus not all were varied in the MADYMO model. For example, in case 1, the case occupant did not come in contact with the seat back in front of him and thus the seat back inclination was not relevant to vary. Slouching or other non-standard initial positions were not considered because each of the cases involved external bruises or abrasions that indicated the children were not grossly out of position. During these crash specific parametric analyses, care was taken to maintain the occupant-vehicle contacts listed in the crash investigation report. The following output variables were monitored: lap and shoulder belt forces, thoracic acceleration, compression and rate of compression, abdominal compression and rate of compression, and head excursion

Table 1.

Parameters varied and the values over which they were varied in the MADYMO models of the three cases.

Variable	Case 1	Case 2	Case 3
Shoulder Belt Slack/Pretension	15% slack	15% slack	15% slack
	No slack	No slack	No slack
	15% pretension	15% pretension	15% pretension
Front Seat Track position	n/a	Mid point	Mid point
	n/a	Rear most point	Rear most point
Shoulder Belt Anchor Position	60 cm	n/a, fixed in vehicle	n/a, fixed in vehicle
	65 cm	n/a, fixed in vehicle	n/a, fixed in vehicle
	70 cm	n/a, fixed in vehicle	n/a, fixed in vehicle
Front Seat back Inclination	n/a	Upright	n/a
	n/a	200inclination	n/a

n/a indicates a parameter not relevant for that particular case and thus one not varied in the MADYMO model

Parametric Study

An exemplar crash scenario was used for the parametric study by qualitatively examining the crash pulse for each of the cases along with their occupant kinematics. The exemplar scenario was parameterized in MADYMO in order to identify the sensitivity of output variables on specified input parameters. The model setup used the ECE R44 bench (Figure 1), which most closely matched the seat geometry from the reconstructed cases.

Figure 1.

Model set-up using the ECE R44 bench. Seat pan angle = 15±1°, seat back angle = 20±1°.

The estimation of crash pulses from real-world crash investigations has inherent limitations due to the assumptions made surrounding crash parameters such as impact location, vehicle velocities, post-impact motion, and inter-vehicle friction. In order to remove the effect of these assumptions from the parametric study of load limiters and pretensioners, an actual crash pulse directly measured during an NCAP test was used. The NCAP crash pulse from each of the three vehicles (or their sister/clone) was extracted from NHTSA’s database. The NCAP pulse for the vehicle in case 2 was chosen as it had a duration and magnitude similar to 2 of the three pulses from the HVE simulations (Case 1 and Case 2). Furthermore, the crash in case 2 was most similar to an NCAP test, as it was the only one of the three cases which was full frontal with minimal lateral component. The actual NCAP test used was a frontal, rigid barrier test of a 2000 Buick LeSabre (a sister/clone of the case vehicle) at 56.6 km/hr. Since it was not a focus of this study to examine the effect of pulse, only one NCAP pulse was used. The crash pulse and its comparison to the HVE-generated pulses from the three cases are seen in Figure 2.

Figure 2.

Parametric study crash pulse, generated from the NCAP test for Case 2, and the longitudinal crash pulses for each reconstruction from HVE.

The final simulation matrix used a common crash pulse (NCAP) and ATD model (Hybrid III 6 year old) and was restricted to backless and high back booster seats. Shoulder belt load limiters in this study were modeled at the 2500n, 3500n, and 4500n levels while lap belt buckle pretensioners were modeled with 25mm, 40mm, 50 mm and 75 mm strokes. The pretensioners were activated at 7 ms following initiation of the crash pulse and reached the full stroke at 13 ms. Combinations of load limiting and pretensioning were simulated with each load limiter and the 25 mm, 50 mm, and 75 mm pretensioners. In order to limit the study to understanding the effect of load limiting and pretensioners, other potential factors were held constant including initial position of the child occupant and position of the seat belt. For all simulations, the child was assumed to be seated forward facing in an upright position using the seat belt properly.

RESULTS

Each of the three cases was reviewed in detail and the injury causation scenarios were defined using the Biotab methodology. Table 2 shows an overview of the injuries and their injury causation scenarios. Two of the three cases had abdominal injuries with associated thoracic injuries. For all three cases, the case review team determined that the injuries to the abdomen and thorax were due to shoulder belt loading from properly positioned belts.

Table 2.

Injury Causation Scenarios

Injuries	Injury Causation Scenario	Confidence/Evidence
Case 1
Abdomen Liver laceration (AIS 2) Pancreatic contusion (AIS 2)	Shoulder belt loading to abdomen	Probable / chest abrasions and right clavicle fracture
Other AIS 2+ injuries	AIS 2 Right Clavicle Fracture
Case 2
Abdomen Liver laceration (AIS 2)	Shoulder belt loading to abdomen	Certain / chest abrasions and contusions
Chest Bilateral lung contusions(AIS 4) Right rib fractures #6–8 (AIS 3)	Shoulder belt loading to chest	Certain / chest abrasions and contusions
Other AIS 2+ injuries	None
Case 3
Abdomen Splenic laceration (AIS 2)	Shoulder belt loading to abdomen	Probable / chest abrasions and contusions
Chest Lung contusion (AIS 3)	Shoulder belt loading to chest	Certain / chest abrasions and contusions
Other AIS 2+ injuries	AIS 2 concussion

Table A.1 in the appendix shows a summary of the crash parameters from the HVE reconstructions compared with data from the crash investigations. The injury metrics as a result of the MADYMO modeling are shown in Table 3 below. The range of a particular metric for a specific case represents the range obtained over the combinations of parameters varied as indicated in Table 1.

Table 3.

Range of injury metrics from modeling of cases.

Measure		Case 1	Case 2	Case 3
Head	HIC (36)	65–75	420–428	1241–1315
	Head CG excursion (mm)	170–250	276–321	319–360
Chest	3ms (g)	27–35	41–45	78–89
	Compression (mm)	26–33	30–33	52–53
	Compression rate (m/s)	2.0–2.2	1.7–1.9	3.4–3.6
Abdomen	Compression (mm)	8.8–9.5	7.0–13.0	12.7–13.3
	Compression rate (m/s)	1.6–2.0	1.8–2.25	1.23–1.57
Belt loads	Lap belt (N)	1965–2647	2320–2630	6038–6173
	Shoulder belt (N)	2950–3522	5398–5703	12423–13065

Peak values of plots of interest from the parametric study are summarized in Table 4.

Table 4.

Peak values of injury metrics from the parametric study. Results are stratified by booster seat type and describe the effect of the addition of pretensioners and load limiters.

	Highback Booster				Backless Booster
	No LL	4500N	3500N	2500N	No LL	4500N	3500N	2500N
	Head Excursion (m)
No Pretens.	0.245	0.27	0.3	0.4	0.215	0.250	0.310	0.400
25mm	0.215	0.25	0.28	0.38	0.225	0.260	0.320	0.400
40mm	0.215				0.220
50mm	0.213	0.23	0.275	0.375	0.210	0.260	0.320	0.400
75mm	0.2	0.22	0.27	0.375	0.225	0.260	0.320	0.400
	Chest Acceleration (m/s2)
No Pretens.	375	350	300	250	325	325	300	260
25mm	360	280	390	258	320	270	260	360
40mm	310				320
50mm	330	290	325	260	300	290	280	340
75mm	480	540	540	460	740	750	420	300
	Chest compression (mm)
No Pretens.	38	34	32	27	42	38	35	31
25mm	36	33	32	26	40	35	31	27
40mm	36				39
50mm	36	35	32	26	39	35	32	28
75mm	36	34	33	28	42	37	32	28
	Chest Rate of Deflection (m/s)
No Pretens.	1.6	1.6	1.7	1.5	2.0	2.0	2.0	1.6
25mm	1.8	1.8	1.9	1.9	2.7	2.7	2.7	2.5
40mm	4.3				5.1
50mm	5.8	5.8	5.8	5.5	6.4	6.1	6.1	5.1
75mm	9.5	9.8	9	7.4	9.7	9.0	8	6.3
	Abdomen Deflection (mm)
No Pretens.	7	7.3	7.4	14.5	7.2	7.4	7	9.4
25mm	6.5	6.4	7.4	14	6.7	6.3	8	13
40mm	6.5				7.2
50mm	7	7.3	7.4	13	7.6	6.8	9	14
75mm	8	7.7	8.4	14.8	9.4	9.3	8.5	14
	Abdomen Rate of Deflection (m/s)
No Pretens.	0.7	0.6	0.6	0.6	0.9	0.94	0.9	0.75
25mm	0.5	0.5	0.51	0.8	0.42	0.5	0.5	0.65
40mm	1.2				0.4
50mm	2.0	2.1	2.1	1.7	0.9	0.8	0.9	1.0
75mm	3.2	3.3	3.3	2.7	4.0	3.2	2.6	1.9
	Shoulder Belt Load (N)
No Pretens.	6500	4500	3500	2500	6000	4500	3500	2500
25mm	6400	4500	3500	2500	6000	4500	3500	2500
40mm	6000				6500
50mm	6000	4500	3500	2500	6400	4500	3500	2500
75mm	6000	4500	3500	2500	6600	4500	3500	2500
	Lap Belt Load (N)
No Pretens.	2500	2580	2580	2600	2050	2150	2400	2750
25mm	2150	2200	2200	2200	1650	1700	1800	2350
40mm	2350				2100
50mm	1900	2100	2100	2800	3200	3000	3250	2800
75mm	2900	2750	2750	3400	8000	7200	6600	5200

DISCUSSION

This study in combination with existing epidemiological studies of abdominal injury risk for children seated in belt positioning booster seats clarifies recent data that highlights the occurrence of abdominal injuries in these restraints. Specifically, this analysis suggests that children restrained in booster seats sustain thoracic and abdominal injuries similar to a properly restrained adult in a moderate to high speed frontal crash [Rouhana, 2002]. The causation scenario of abdominal injuries in booster seated children delineated herein is different than that experienced by children who sustain abdominal injuries restrained by the seat belt alone. For children restrained using seat belts, injury is often due to the lap belt compressing the lower abdomen resulting in intestinal injuries [Arbogast et al., 2007] while in these booster seated children, the shoulder belt compresses the upper abdomen and thorax.

Children in two of the three cases sustained a thoracic injury: case 2 had bilateral lung contusions and rib fractures and case 3 had a unilateral lung contusion. Examination of the injury metrics from the modeling of these cases revealed chest compressions (for the rib fractures) and chest accelerations (for the lung contusions) of 30–53 mm and 41–89 g, respectively. Limited pediatric injury tolerance data exists for comparison. Ouyang et al [2006] applied blunt hub impacts to the thorax of nine pediatric post mortem human subjects. For their young cohort (2–4 years), peak chest deformation correlated with injury and they suggested a tolerance level of 45 mm, slightly higher than the chest compressions predicted by the MADYMO model of Case 2 but below those of Case 3. This relationship was not as evident in their older cohort (5–12 years). Of note, the injuries their subjects sustained were lung injuries not rib fractures. Their studies were conducted without intubating the lung or pressurizing the arterial system perhaps influencing the relationship between chest compression and lung injury.

For the abdomen, the children in the cases studied sustained two liver lacerations and a splenic laceration. The corresponding models reported abdominal deflection and velocity of 7.0–13.3 mm (a normalized abdominal deflection of 7–13% for an average 4–6 year old) and 1.2–2.2 m/s, respectively. However, there is limited comparative data available. Recently, Kent et al [2008] conducted a study using a juvenile porcine model and determined abdominal injury risk functions in response to belt loading. In their study, the 50% risk of injury corresponded to a normalized deflection of 0.32 (4.5 mm) and an abdominal velocity of 2 m/s. They stated that the relationship between abdominal velocity and injury was not a strong one. It is important to note that the modeling was conducted using the Hybrid III 6 year old ATD whose abdominal biofidelity has been questioned. Specifically, the abdomen of the physical ATD and its corresponding computational analog has been noted as being much stiffer than that of an actual child. This difference likely influences the deflection measures obtained from the model.

In the parametric study, introduction of load limiters increased the head excursion by 10% for the 4500 N load limiter and as much as 60% for the 2500N load limiter while pretensioning alone had a small minimizing effect on head excursion. When pretensioners were combined with load limiters, the pretensioners were able to offset or limit the increase in head excursion originally caused by the load limiters. For the backless booster, load limiting increased the head excursion by more than 80%, similar to the effect of load limiting in the highback booster case. However, contrary to the highback booster results, addition of pretensioning did not reduce the head excursion. It is important to note however, that the current head excursion limits in FMVSS 213 are 813mm (720mm with a tether) – none of the simulations exceeded 50% of this threshold. Of interest, the average horizontal distance from the centerline of the rear seat bight to the nearest point on the back of the seat forward of the subject occupant position in current vehicles was determined to be 712 mm (range 426–895mm) [Glass, 2002] suggesting that none of these head excursions would result in impact with front seat structures.

For the highback booster, chest acceleration was reduced by as much as 17% with the addition of the pretensioners, with one exception. Chest acceleration substantially increased for the simulations involving the 75 mm pretensioner, which is an artifact of the initial load application applied to the chest by this stiff pretensioner. Unlike head excursion, chest acceleration also improved with the introduction of load limiting. These trends were similarly seen in the backless booster simulations though the incremental benefit associated with lower load limiting was not as evident. The regulatory limit for chest acceleration for the Hybrid III 6 year old is 60g’s for FMVSS 208 and FMVSS 213. All parametric combinations tested were below this threshold except for some combinations of load limiters with the 75mm pretensioner. The highback boosters with the 75mm pretensioner were at the threshold while those in backless boosters with the 75 mm pretensioner sustained chest accelerations as high as 76g’s. Of note, the Injury Assessment Reference Value proposed for the chest acceleration for the Hybrid III 6 year old was 93g’s [Mertz, Irwin, Prasad, 2003]. All values of chest acceleration in the parametric studies were below this proposed reference value.

Pretensioners and load limiters independently reduced the chest compression for the highback booster. Combining load limiters and pretensioners resulted in an overall reduction of chest compression from the baseline but pretensioners did not necessarily have an additive effect on the load limiter benefits. The regulatory limits for chest compression (as measured by sternal deflection in the ATD) are 40 mm for the Hybrid III 6 year old in out-of-position air bag loading. No specific chest compression regulatory limits exist for belt or child restraint harness loading for this ATD. The IARV proposed by Mertz et al [2003] for seat belt loading of the Hybrid III 6 year old is 31mm. The authors state that the specific injury risk associated with this IARV is not known however it is likely that it corresponds to the lower bound of the AIS≥3 injury threshold. Virtually all of the parametric simulations exceed this threshold except for those with the 2500N load limiter. It is important to note however that this threshold was derived from scaling adult IARVs which correspond to the likelihood of rib fractures and may be inaccurate for assessing thoracic injury risk in children. The pediatric thoracic cage is more flexible suggesting that an IARV corresponding to soft tissue injury to the lung or the heart might be more appropriate. This is evidenced by the thoracic injury patterns seen in the cases studied herein. Studies based on impact of pediatric PMHS thoraces [Ouyang, Zhao, Xu et al., 2006] suggested a 45 mm threshold for injury. This threshold was derived not from injuries to the rib cage but rather to injuries to the thoracic contents. All parametric simulations in the current study were below this threshold.

The rate of chest compression increased by as much as 600% with the addition of pretensioners. These chest compression peak rate values were the result of the initial application of the pretensioner in the model. Addition of load limiters did not appreciably change rate of chest compression. The results were very similar across booster seat types. The IARV proposed by Mertz et al [2003] for sternal deflection rate for the Hybrid III 6 year old was 8.5 m/s corresponding to a 5 percent risk of AIS≥4 thoracic injury. Only the 75mm pretensioners in combination with no load limiter, the 4500 or 3500 load limiters exceeded this value. The highest value of rate of chest compression (9.8 m/s with 75 mm pretensioner and 4500 load limiter) corresponds to approximately a 20% risk of AIS≥4 thoracic injury according to the risk curves proposed by Mertz et al [2003].

Introduction of pretensioning had a slight positive effect on abdominal deflection, with reductions for 25mm and 40mm pretensioners in the highback booster cases. However, as pretensioning increased to 50mm and 75mm, the abdomen deflection increased to similar or higher levels than baseline. Load limiting had little effect on abdomen deflection, except for an approximate doubling of abdomen deflection for the 2500N load limiter. This increase is likely due to hyperflexion of the occupant torso over the lap belt as evidenced by the 180% increase in head excursion for this case. No regulatory limits exist for abdominal deflection. Mertz et al [2003] proposed an IARV of 24 mm for 6 year old abdominal deflection corresponding to the lower bound of AIS≥3 injury thresholds. None of the simulations were close to this threshold. For reasons discussed above, the Hybrid III 6 year old ATD abdomen is likely much stiffer than that of an actual child likely resulting in small values of abdominal deflection in the model. It is also important to realize that high magnitudes of abdominal deflection may not be of concern if the location of the abdominal deflection is below the Anterior Superior Iliac Spines (ASIS). If the deflection occurs above the ASIS, then the injury risk is likely proportional to the magnitude of deflection. In this study, abdominal deflection was measured between the center point of the abdominal insert ellipsoid and the lower lumbar spine ellipsoid which likely corresponds to slightly above the ASIS.

Pretensioning at 25 mm had a positive effect on the rate of abdomen deflection. The abdomen rate of deflection increased to as much as 450% of the baseline levels for the other levels of pretensioning. Again, no regulatory limits exist for abdominal deflection rate but an IARV of 8.5 m/s has been proposed [Mertz et al 2003] to correspond to the lower bounds of AIS≥4 injury thresholds. The abdominal deflection rates in this parametric study are much lower than this proposed threshold. Of note, the abdominal deflection rate for the 50mm pretensioner in the highback booster and the 75mm pretensioner in both booster types had abdominal deflection rates near or exceeding the threshold of 2 m/s described in Kent et al [2008].

Pretensioners had little effect on the shoulder belt loads while as expected load limiters reduced the shoulder belt loads to a great extent, by 30%, 45% and 60% for the 4500N, 3500N, and 2500N load limiters, respectively, in the highback booster cases.

Similar trends were seen for the backless booster seats. In the highback boosters, the combination of pretensioners and load limiters resulted in no clear pattern in resulting lap belt loads. Variations of 10–20% were seen. In contrast, for the backless booster seats, the 50 mm pretensioner increased lap belt loads by 50% and the 75 mm pretensioner by 100–300%.

Limitations

During the crash reconstructions and the parametric study, it was assumed that each of the child occupants were facing forward, properly restrained and seated upright. Although in the real-world cases there were clear witness marks which indicated that the children were not grossly out of position, the actual posture of the child during crash could be different from that modeled.

To generate a valid crash pulse for the reconstructed cases, PDOF, CDC, crush and delta-V were matched between the crash investigation and the HVE simulation by carrying out parametric variations with respect to crash parameters such as the impact location, vehicle velocities, and inter-vehicle friction. Care was also taken to ensure that the post-impact motion obtained for both the case and the non-case vehicle in the HVE simulation was consistent with the information provided in crash investigation. The damage profile from HVE was compared to the damage photos from crash investigation. Increasing emphasis was placed on matching the parameters in the following order: delta-v, crush, CDC, and PDOF

CONCLUSION

In-depth review of three real-world crashes involving abdominal injuries in booster-seated children pointed to the role of shoulder belt loading in the causation of abdominal injuries and associated chest injuries. Computational models quantified baseline injury metrics for these cases such as lap and shoulder belt forces, thoracic compression and rate of compression, abdominal compression and rate of compression, head excursion and HIC. Parametric study of these injury metrics revealed that baseline injury metrics such as chest compression could be improved through the use of shoulder belt load limiting at the expense of increases in head excursion. Coupling shoulder belt load limiting and lap belt buckle pretensioning resulted in improved chest and abdominal metrics while minimizing or reducing head excursion from the baseline, indicating that addition of these technologies to the rear seat environment may provide injury reduction potential.

Table A.1.

Comparison of crash parameters from the HVE reconstructions with those parameters from the real-world crash investigation.

Crash Parameter		Case 1	Case 2	Case 3
CDC	HVE	11FDEW3	12FDEW2	12FDEW4
	Crash Inv.	11FYEW4	12FDEW2	71FDEW3*
Delta-V	HVE	31 km/h	54 km/h	89 km/h
	Crash Inv.	21 km/h	50 km/h	72 km/h
Average Crush	HVE	24 cm	39 cm	49 cm
	Crash Inv.	17 cm	34 cm	44 cm
PDOF	HVE	341°	1°	346°
	Crash Inv.	340°	0°	340°

^*The PDOF was 11 o’clock; however due to the right lateral shift of the front part of the bumper at the axle, NASS/CIREN protocol requires an addition of 60 to the PDOF (i.e. 11+60=71). This protocol is followed whenever there is a shift in the axle perpendicular to impact greater than 4″ (10 cm).