A Descriptive Study of Pediatric Injury Patterns from the National Automotive Sampling System

Abstract

This study describes information from the National Automotive Sampling System for injury mechanisms in the pediatric age group (age 0–16). The total number of pediatric cases in the NASS database for this three year sampling period is 2141(weighted 591,084). No restraint use was identified in 23–43% of the children. For age < 1yr, 60% of patients suffer a facial injury. Head injuries make up only 10% of the total injuries, but are severe. For those age 1–4 yrs abdominal injuries and lower extremity injuries begin to appear. For those age 5–10 yrs, the predominant change over younger occupants is the proportion of spinal injuries. By age 11–16, injuries to the spine, upper extremities, and lower extremities outnumber injuries to the face and head. However, in this population, the greatest proportions of AIS 3–5 injuries still occur to the head and abdomen.

Trauma is recognized as the leading cause of death in the pediatric population over one year of age [Hall, 1994]. Motor vehicle occupant injury in children encompasses 12–18% of all pediatric injuries in the U.S. [Agran, Castillo, Winn, 1990, Newman, Bowman, Eichelberger, et al 1990]. When all motor vehicle-related injuries are considered, these injuries account for the number one cause of death and disability among the pediatric population in this country [Hazinski, Francessuti, Lapidus et al 1993]. Although crash occupants do not account for the most trauma-related emergency department visits or the most hospital admissions in children, they are recognized as more severe injuries than others by many measures. [Guyer, Ellers, 1990, Peclet, Newman, Eichelberger et al 1990]

Many have recognized that preventing injury with restraint systems is effective at reducing death and disability [Osberg, Di Scala 1992]. While some statistics show an improving trend in pediatric restraint use [Mazurek, 1994], other studies indicate significant problems with inappropriate installation and serious misuse of childhood restraint devices[Margolis, Wagenaar, Molnar 1988, Hazinski et al 1996]. There are also large regional differences in restraint use, documented as low as 9% in pre-adolescent children in certain rural areas [Hazinski et al 1996]. Data from 1990 shows that 70% of the fatalities in pediatric motor vehicle occupants under 5 years of age were unrestrained [Mazurek, 1994].

Although there is a body of literature detailing injury patterns associated with seat belt use in the pediatric population, as well as numerous case reports of injuries associated with restraint use and misuse, the relation between the larger constellation of variables encountered in the biomechanics of MVC (seating position, type of restraint used, combination of restraint systems) remains to be elucidated for this age group. While injury patterns related to collision type and the accompanying biomechanics have been established in the adult population (Dischinger, Cushing, Kerns 1993, Siegel, Mason-Gonzalez, Dischinger 1993, Hill 1992], it is unknown whether these same patterns apply to the pediatric population. We describe the information available in a large national database that could be used to assess the relationship between mechanism of MVC, seat position, restraint use, and injury patterns in the pediatric population.

METHODS

This study is a retrospective analysis of 2141 pediatric occupants, aged 0–16 years, injured in motor vehicle collisions covered in the National Accident Sampling System (NASS) database for a 3 year period from January 1993 through December 1995.

The NASS database is operated and maintained by the National Highway and Traffic Safety Administration (NASS, 1996). To qualify for entry into NASS, the MVC must have a police report, must be reported to the state, must involve a “harmful event” defined as property damage and/or personal injury, and must have occurred as a result of an “accident” defined as at least one harmful event produced by an unstabilized situation. The collision must involve a motor vehicle in-transport and the collision must have occurred on a public trafficway. At each of 24 sampling sites the research team investigates a sample of the respective police reported collisions. The investigation consists of a detailed review of Police Accident Reports, hospital records, prehospital care records, photographs of the vehicles and the vehicles themselves. This sampling system and the NASS database were designed to compute national estimates and insure the validity of the data without having to investigate every MVC in the country. Each individual case within the NASS database is assigned a weighted number based on the sampling mechanism to represent the nationwide incidence of similar cases. We present the unweighted numbers for our sample of 2141 patients as well as the corresponding weighted numbers in Table 1. The descriptive analysis relies on weighted numbers.

Table 1.

Restraint Use by Age (weighted)

Age	Belted	Child Seat	Unrestrained	Unknown
< 1 year	733	6561	5592	224
1–4 years	25687	20604	14572	4149
5–10 years	112869	0	65400	4427
11–16 years	175683	0	143147	11437

In this study, the patient population is stratified by age. The age categories (<1, 1–4, 5–10, 11–16) were chosen to represent the different anatomic developmental stages in children. The <1 year-old group encompasses the general range for the rear-facing child safety seat, while the 1–4 year old group generally corresponds to the range for the front-facing child safety seat. The 5–10 year old age groups represent children too large for the child seat, but often too small to be appropriately restrained by the belt systems. The 11–16 year group corresponds to the population large enough to fit the typical lap-shoulder belt system, but still with anatomic differences from the adult.

For each age range, the groups are further subdivided by occupant seat position, then by restraint use (yes or no) and if restraints were used, the groups are further divided by the type and combination of restraints used (infant seat-rear facing, toddler seat-front facing, child seat with unknown orientation, and seat belt). Booster seats and air bag deployments were not identified as individual categories due to lack of sufficient number of patients. The belted occupants were not separated by belt type. Each occupant has a corresponding set of injury data, including: Abbreviated Injury Score (AIS-90), maximum AIS (MAIS), Injury Severity Score (ISS), and body region of injury.

RESULTS

The total number of pediatric cases in the NASS database for this three year sampling period is 2141(weighted 591,084). Included are 55(13109) children <1 year old, 355(65012) ages 1–4 years, 588(182696) ages 5–10 years, and 1143(330267) ages 11–16 years. No restraint use was identified in 43% of those < 1 yr old, 22% of those 1–4 yrs old, 36% of those 5–10 yrs old, and 43% of those 11–16 yrs old.(Table 1) In 78% percent of the cases in which infants were in safety seats, the direction in which the seat was facing was unknown.

Because of low unweighted numbers, one must view some of this data with great caution. This caution is particularly necessary for those < 1 year old, who represent only 55 total cases. Estimates made under these circumstances may be regarded as unstable.

Among those < 1 yr old, the rear passenger side was the most common seating position (32%), although for 40% the seating position was unknown. For those age 1–4 the rear driver’s side was most common (28%), closely followed by the front passenger seat (25%). From age 5–16, the most common position is the front passenger seat (50%). Information on impact type, ΔV, and compartment intrusion were unknown to such a degree as to make comparisons unrevealing.

Reported injury distributions (MAIS per body region) vary among the age groups. For age < 1yr, 60% of patients suffer a facial injury, although all of these are AIS 1 injuries. (Fig 1) Head injuries make up only 10% of the total injuries, but 614/1621 of the head injuries are AIS 4 or 5. For those age 1–4 yrs, the pattern of head and facial injuries remains unchanged, but abdominal injuries (AIS 1–3 - 6.7% of total injuries) and lower extremity injuries (AIS 1–3 - 15% of total injuries) appear. (Fig 2) For those age 5–10 yrs, the predominant change over younger occupants is the proportion of spinal injuries. (Fig 3) Spinal injuries make up 12% of total injuries in this group, with all of them being AIS 1 or 2. By age 11–16, injuries to the spine, upper extremities, and lower extremities outnumber injuries to the face and head. (Fig 4) In this older population, the greatest proportions of AIS 3–5 injuries still occur to the head and abdomen. Of 50756 head injuries, 2599 are AIS 3–5. Of 12088 abdominal injuries, 1261 are AIS 3–5. Injury Severity Scores of less than 16 were much less common than those greater than 16, although ISS was not calculated in the majority of cases. (Table 2) Given the low unweighted n in the < 1 yr old group (55), estimates may be unstable.

Figure 1.

MAIS Injury Distribution < 1 yr olds

Figure 2.

MAIS Injury Distribution - 1–4 yr olds

Figure 3.

MAIS Injury Distribution - 5–10 yr olds

Figure 4.

MAIS Injury Distribution -11–16 yr olds

Table 2.

Injury Severity Score by Age (percent within age group)

Age	ISS < 16	ISS > 16	ISS Unknown
< 1 year	5093 (39)	0 (0)	8016 (61)
1–4 years	25741 (40)	57 (<1)	39214 (60)
5–10 years	45591 (25)	309 (<1)	136796 (75)
11–16 years	91273 (28)	1949 (<1)	237045 (72)
Total	167698 (28.4)	2315 (0.39)	421071 (71.2)

The amount of unknown data was concerning. The database requires that vehicle-specific data come from police crash reports and that injury-specific data come from hospital medical records. Only 29% of cases had injury data detailed enough for calculation of ISS.

DISCUSSION

Injury patterns and the associated biomechanics in motor vehicle collisions involving children are complex. This complexity arises from the many different variables affecting children, as opposed to adults, in vehicular crashes. Not only are there different types and severities of collisions, but children use a variety of restraint types, different seating locations, and are a much more heterogeneous population of passengers. Although comorbidities in children are not as much of a concern in children as compared to adults, the pediatric population encompasses vastly different body mechanics, anatomical relations, and sizes. Each one of these factors can potentially change the biomechanics and injury patterns in collisions.

In one study which separated belted pediatric occupants by age (0–3, 4–9, and 10–14 years), the head was the most commonly injured part of the body in all ages 0–14 years [Agran, Dunkle, Winn 1987]. In the 0–3 year old group, most other injuries were relatively infrequent, with chest and abdominal injuries as well as spinal strains being the least frequently injured. In the 4–9 year old group, there was a rise in extremity injuries (25%) and an increase in abdominal injuries. It was proposed that this increase in abdominal injuries in the 4–9 year olds, the highest incidence of the three groups, was due to positioning an inappropriately sized passenger in a belt system designed for a much larger person. With the belting system designed to ride over the anterior iliac crests, providing for relatively stable anchor points in the adult, the anterior iliac crests in children are much smaller and more underdeveloped, allowing the lap portion of the belt system to encroach on the relatively unprotected abdomen. [Agran, Winn, Dunkle, 1989] Although an inappropriately placed lap belt may also leave an adult at risk for abdominal injury, the child occupant is unique in this scenario. There is a smaller and less mature bony pelvis and thoracic cage, less developed abdominal musculature, decreased abdominal adipose tissue, relatively larger solid viscera, and a more compact torso that is larger in comparison with the extremities than the adult, all resulting in an increased potential for abdominal injuries from a high-riding lap belt. [Agran et al 1989, Schafermeyer, 1993, Pieper, 1994]. Finally, in this same study, the 10–14 year old group had a marked increase in spinal strains coincident with a rise in the incidence of “whiplash” as the mechanism of injury (33% for each). Combined with a decrease in the incidence of head injury, the injury pattern in the 10–14 age group was ascribed to better seat belt positioning (secondary to more appropriate body size). The result was an improved ability to restrain forward movement with a resultant transfer of energy to the cervical spine and neck musculature [Agran et al 1987].

While some studies have reported the head and face to be the most frequently injured body regions in children, even among restrained passengers [Agran et al 1987, Agran et al 1989], other studies have found this trend to be true only with unrestrained children [Osberg et al 1992, Stylianos, Harris 1990]. The discrepancy is likely due to differences in restraint type (lap belt versus lap and shoulder belt) as well as improper use of restraints, as children who have outgrown child safety seats are still too small to appropriately fit the belt system designed for an adult and may disregard the shoulder. The lack of a shoulder belt combined with an ill-fitting lap belt and the body mechanics of a child may indeed fail to protect the head and face from contact with structures within the passenger compartment [Agran et al 1989]. The child’s sitting height is less than that of an average adult which may increase proximity to compartment structures, and when combined with a higher center of gravity, a greater proportion of body mass above the belt, a greater head-to-body ratio, and underdeveloped neck muscles, there may be increased forward motion and increased risk of head and facial impact with the vehicle interior [Agran et al 1989, Mazurek 1994].

The anatomical and pathophysiological differences in cranial and spinal structures in children can also affect the accompanying injury patterns and severity of injuries. Not only is there a greater head-to-body ratio, but the cranial bones are thinner, providing less protection, and the brain cells are less myelinated and consequently more vulnerable to injury. In addition, increased intracranial pressure and global injury (diffuse cerebral swelling) has been shown to occur in a greater percentage of children than adults with intracranial head injury [Mazurek, 1994]. While the cranial structures may be at particular risk in child occupants of collisions, the majority of these injuries are reportedly minor [Agran et al 1990, Agran et al 1987, Agran et al 1989, Mazurek 1994].

Spinal injury in children is relatively uncommon owing to greater physiologic tissue mobility in the neck and spinal structures, however, motor vehicle collisions have been reported as the most frequent mechanism of injury in pediatric spinal trauma (Buckley, Gotschall, Robertson 1994, Pang, Pollack 1989, Hadley, Zabranski, Browner 1988, Schafermeyer 1993]. There have been numerous case reports of spinal trauma in pediatric passengers, including inappropriately restrained (Steele, Aks 1995, Conry, Hall 1987, Bodenham, Swindells, Newman 1992, Fuchs, Barthel, Flannery 1989] and appropriately restrained [Diekema, Allen 1988, Fuchs et al 1989] occupants. The increased mobility seen in the immature spine is a result of ligamentous laxity, underdeveloped paraspinous musculature, wedge-shaped vertebral bodies, and a shallow, horizontal orientation of the facet joints [Hadley et al 1988, Pang et al 1989, Fuchs, et al 1989, Schafermeyer 1993]. These anatomic differences allow for greater subluxation at the expense of stability [Fuchs et al 1989, Pang et al 1989]. When combined with the greater head-to-body size, there is higher resultant torque placed on the cervical spine and a higher susceptibility to flexion-extension injuries [Fuchs et al 1989, Pang et al 1989]. As a result of these characteristics, the fulcrum of cervical spine movement is higher in young children (C2-3) when compared to adults (C5-6) [Fuchs et al 1989, Pang et al 1989, Bonadio 1993], resulting in a proportionately larger percentage of high cervical spine injuries in children [Hadley et al 1988, Fuchs et al 1989, Pang et al 1989]. The failure of the odontoid synchondroses to fuse until later childhood may also contribute to the increased incidence of high cervical spine injuries [Schafermeyer 1993]. This hypermobility is further demonstrated in documented cases of spinal cord injury without radiographic abnormality in children (the SCIWORA syndrome), in which one study reported most of the children with severe injury were under 8 years of age [Pang et al 1989]. While these structural differences may offer protection to injury in minor trauma, the same characteristics place the immature spine at risk when more force is encountered [Hadley et al 1988]. Generally by eight years of age, the horizontal facets have converted to more stable upright position and much of the ligamentous laxity and underdeveloped musculature is beginning to mature [Fuchs et al 1989, Schafermeyer 1993], which may translate into different spinal injury patterns for children older than eight years.

Cervical spine injuries are not the only spinal injuries described in pediatric passengers. The combination of lumbar spine injuries (in particular, flexion-distraction injuries) and injury to the intestinal viscera in connection to wearing a lap belt has been termed the “seat belt syndrome”. Although this injury pattern is not unique to the pediatric population, children are at risk due to the frequency of poorly fitted restraints and the relatively common occurrence of children that either do not use the shoulder harness as a result of it riding high across the neck and face or who are seated in back seat of a vehicle sold before lap-shoulder systems were implemented in the rear seats [Newman et al 1990]. The mechanism has been described as the passenger sliding beneath an improperly placed lap belt, shifting the axis of rotation to the level of the umbilicus [Reid, Letts, Black 1990]. There is hyperflexion of the lumbar spine and with the little abdominal protection offered by the pediatric anatomy, an accompanying intestinal injury is often seen. Some authors report a unique association with frontal collisions [Newman et al 1990].

Extremity trauma, which is fairly common in children involved in motor vehicle collisions [Osberg et al 1992, Agran et al 1987, Buckley et al 1994], also does not seem to be affected by restraint use (Agran, Castillo, Winn 1992, Osberg et al 1992]. The incidence of extremity injury in pediatric occupants has been shown to increase in incidence among older age groups of children, amounting to 27% of the patients in the 10–14 year age group [Agran et al 1987], possibly secondary to the increased limb length compared to younger children, placing the extremities in closer proximity to compartment structures.

With regard to restraint use, different patterns have emerged. Numerous studies have validated the use of seat belts in reducing severity of injury, morbidity and mortality in the pediatric population [Osberg et al 1992, Agran et al 1992, Decker, Dewey, Hutcheson et al 1984]. Other studies which have focused on the young pediatric population (0–4 years) to compare the protective efficacy of the child safety seat versus the seat belt versus no restraint have shown the safety seat to offer the most protection [Agran, Dunkle, Winn 1985]. Yet even in this young population, the seat belt did offer more protection from serious injury than no restraint at all [Agran et al 1985]. However with the protection that restraint systems offer comes a separate injury pattern specific to those using those types of restraints. Agran et al. reported a significant increase in spinal strains in belted adolescents (as previously detailed) and showed a surge in intracranial, facial, and soft tissue injuries in unrestrained adolescents [1992]. Many authors have reported an association between seat belt use and intra-abdominal injuries [Osberg et al 1992, Stylianos et al 1990, Newman et al 1990, Reid et al 1990, Tso, Beaver, Haller 1993]. Osberg et al. showed that although there was a higher incidence of severe abdominal injuries in belted children, the severity of overall injury (as indicated by the ISS) of the belted passengers was less, indicating there were fewer multiple injury patients wearing seat belts [1992].

There have been few studies to link specific injuries to the impact type and the seat location in the pediatric age group. One group noted an increase in spinal injuries to front seat occupants involved in rear collisions. In the same study there was a significant increase in the number of “lower torso” injuries for both front and back seat passengers in frontal collisions, however most of these injuries were minor contusions and abrasions [Agran et al 1989]. An additional finding in this study was a higher mean maximum AIS (i.e. more severe) injury as well as a possible association with head and facial injuries in same-side passengers involved in lateral collisions. There were no differences in extremity injuries by seat location or impact site. Concerning seat position, Decker et al. reported an increased “risk of visible injury” to child occupants < 4 years in the front seat (odds ratio 1.6), and the greatest risk to occupants of the front center seat (odds ratio 1.9) [1984]. Both of these findings were significant regardless of restraint use. Decker also noted a greater risk for the rear center seat compared to the rear side seats, although this trend did not reach statistical significance. In a separate study which involved the 4–9 year population, the restrained passenger in the back seat was shown to have the lowest mean ISS, however the mean ISS between restrained and unrestrained front seat passengers and between passengers in all seat locations was not significant [Agran et al 1992].

LIMITATIONS

The main weaknesses of this study center on the study design (retrospective) and the data base. The retrospective nature of the project allows only a descriptive analysis of the sample population and prevents one from controlling the many possible variables at play: data abstraction inaccuracies; diagnostic and charting differences between institutions and between individuals; restraint misuse and extent of misuse if present; biomechanical and environmental issues at the scene that may have altered the injury patterns; size and design differences between vehicles; and inherent differences in anthropomorphic characteristics between subjects that are not eliminated by stratifying by age and weight. Perhaps most importantly, significant areas of data, notably ISS, ΔV, PDOF, and intrusion had large percentages of “unknown” values. These “unknown” values weaken the overall quality of the data but strengthen the argument for development of prospective data gathering systems.

Use of the NASS database dilutes the value of the correlations drawn from the study due to the requirement for statistical manipulation of the unweighted data to produce national estimates for collisions and related injuries. However, NASS was specifically designed to allow national estimates to be made from the available resources. Analyses of this magnitude have not been possible otherwise. Not only have great efforts been taken to ensure the statistical validity of the NASS-based national estimates (NASS manual, 1996; Appendix B, 1993), but comparisons made between portions of the NASS data and that of a single state’s database have supported the validity of NASS (Agran, AJDC 1990). Although the limitations of studies using NASS data must be recognized, the benefits of such a large estimate and spectrum of collisions and injuries may help to demonstrate some of the associations between MVC variables and certain injury patterns that have not been possible with smaller unweighted sample sizes.

Similar databases have also been criticized for excluding fatalities and thus providing an accordant underrepresentation of higher severity injuries (Peclet, 1990). While there is some truth in this assertion, the motor vehicle trauma cases severe enough to cause fatality represent a very small portion of MVC patients presenting to emergency departments (Agran, AJDC 1990). Furthermore, there is not an absolute need to include fatally injured patients in order to study the mechanisms and injury patterns in severely injured patients. ISS has been well recognized as a representative indicator of injury severity (Baker, 1974), and the population with an ISS≥16 has been shown to represent severely injured patients, including children (Eichelberger, 1988).

Another limitation concerns the exclusion of uninjured patients from the study. Our results do not make inferences on the likelihood of sustaining injury as a pediatric occupant of a MVC. Rather, our data describes specific injury patterns among differing severities of injured children involved in MVCs.

The original intent of this work was to go beyond the description of the circumstances of crashes involving various age and weight groups and to associate specific injury patterns with specific mechanisms. The reliability of conclusions in this area with data of such small absolute numbers and large areas of unknown variables would be highly questionable. We choose to describe what can be reasonably concluded about the pediatric population from the largest crash database available.

CONCLUSION

The usefulness of this study lies in describing the what is known about the pediatric population involved in motor vehicle crashes. Large percentages of children are unrestrained, head and facial injuries are common, and torso and extremity injuries become more common in older age groups. More detailed study will be required to determine the relationship between specific crash mechanisms and circumstances and specific injury patterns. This important analysis may require more accurate abstraction of data in retrospective databases and creation of prospective data collection systems designed to answer the many important questions that remain to be answered about children in cars.