Objective. To determine the risk of significant injury associated with premature graduation of young (2- to 5-year-old) children to seat belts from child restraint systems (CRS).
Background. Advocates recommend use of child safety seats for children younger than age 4 and booster seats for children age 4 and older. Despite these recommendations, many children are prematurely taken out of these child restraints and placed in seat belts. Although data exist to support the use of child restraints over nonrestraint, no real-world data exist to evaluate the risk of significant injury associated with premature use of seat belts.
Design/Methods. Partners for Child Passenger Safety includes a child-focused crash surveillance system based on a representative sample of children ages 0 to 15 years in crashes involving 1990 and newer vehicles reported to State Farm Insurance Companies in 15 states and the District of Columbia. Driver reports of crash circumstances and parent reports of child occupant injury were collected via telephone interview using validated surveys. Results were weighted based on sampling frequencies to represent the entire population.
Results. Between December 1, 1998, and November 30, 1999, 2077 children aged 2 to 5 years were included and were weighted to represent 13 853 children. Among these young children, 98% were restrained, but nearly 40% of these children were restrained in seat belts.
Compared with children in CRS, children in seat belts were more likely to suffer a significant injury (relative risk: 3.5; 95% confidence interval: [2.4, 5.2]). Children in seat belts were at particular risk of significant head injuries (relative risk: 4.2; 95% confidence interval: [2.6, 6.7]) when compared with children in CRS.
Conclusions. Premature graduation of young children from CRS to seat belts puts them at greatly increased risk of injury in crashes. A major benefit of CRS is a reduction in head injuries, potentially attributable to a reduction in the amount of head excursion in a crash.
Motor vehicle crashes (MVCs) are the leading cause of death and acquired disability for children older than age 1.1 In 1998 alone, 697 children younger than age 6 years died as occupants in MVCs and nearly 100 000 were injured.2 Advocates recommend the use of appropriate child restraint systems (CRS) to protect children in crashes. These recommendations include infant and convertible child safety seats (CSS) for children younger than age 4 and booster seats for children from age 4 until they fit properly in the vehicle seat belt (usually age 9).3
Despite these recommendations, many preschoolers are prematurely taken out of these child restraints and placed in seat belts.4Although data exist to support the use of child restraints over nonrestraint,5 no real-world data exist to evaluate the risk of clinically significant injury associated with premature use of seat belts. The purpose of this study was to quantify the nature and risk of significant injury associated with premature graduation to seat belts in preschool-aged children.
Data were collected as part of Partners for Child Passenger Safety (PCPS), a child-focused crash surveillance system based on a representative sample of children ages 0 to 15 years in crashes involving 1990 and newer vehicles reported to State Farm Insurance Companies in 15 states and the District of Columbia. Driver reports of crash circumstances and parent reports of child occupant injury were collected via telephone interview using validated survey instruments. Results (from December 1, 1998 through November 30, 1999) were weighted based on sampling frequencies to represent the entire eligible population.
Case Identification by State Farm Automobile Insurance Company
State Farm Automobile Insurance Company is the largest insurer of automobiles in the United States, with >38 million vehicles under coverage, representing approximately 20% of the automobile insurance market in the country. The electronic insurance claims database was used as the mechanism to identify subjects for inclusion in the PCPS surveillance system. This database is updated continuously at the time of the initial report of a crash to a State Farm claim representative. Claims qualifying for inclusion were those reporting a crash including at least 1 child occupant ≤15 years of age riding in a model year 1990 or newer insured vehicle. Only children riding as occupants (nondrivers) in the insured vehicle were eligible for inclusion. Qualifying claims were also limited to crashes that occurred in 15 states (Delaware, Maryland, North Carolina, New Jersey, New York, Pennsylvania, Virginia, West Virginia, Illinois, Indiana, Michigan, Ohio, Arizona, California, and Nevada) and the District of Columbia, representing 3 large regions of the United States. These regions account for >50% of State Farm's claims volume, and were chosen to provide a mixture of tort and no fault states because this was presumed to influence the reporting of crashes to an insurance company. In addition, these states provided a representative sample of most vehicle types and driving conditions in the United States. Crashes resulting in bodily injury, as well as those resulting only in property damage, were eligible for inclusion. Excluded were claims occurring outside the designated geographic areas, or those involving rental or other fleet vehicles. Child occupants of non-State Farm insured vehicles (for crashes involving 2 or more vehicles) were also excluded from the surveillance system. Members of the Strategic Resources Department at State Farm Insurance Companies maintain summary information about all eligible claims to calculate the capture rate (the percentage of eligible claims for which claim representatives approached the policyholders for consent) and consent rate (among those claims captured, the percentage of policyholders who consented to be part of the study). This summary information includes vehicle model year, ages of the child occupants, and treatment status of the child (no treatment, ambulatory medical treatment including emergency department visits, and hospital admission).
Approximately 7750 claim representatives from 365 field offices throughout the 3 study regions were trained to obtain a minimum amount of standardized data on all qualifying claims and to obtain consent from the insured for inclusion in the surveillance system. On a daily basis, data from consenting claims were transferred electronically from all field offices to State Farm corporate headquarters in Bloomington, Illinois. After several quality assurance checks at State Farm headquarters, the data were then forwarded via e-mail to researchers at The Children's Hospital of Philadelphia/University of Pennsylvania (CHOP/Penn) on a daily basis, 6 days per week (no transmissions on Sundays).
Subject Selection Via Electronic Sampling
After passing through several additional quality assurance procedures, the data were then subjected to an automated sampling algorithm to select claims for in-depth telephone interviews to obtain information about the crash and all child occupants. Each claim was classified based on the medical treatment received by child occupants after the crash. Crashes were then sampled with known probabilities according to the highest treatment status of its child occupants. Specifically, crashes involving children who were treated in emergency departments, physician's offices, or admitted to the hospital were oversampled to ensure the capture of all injured children while maintaining a representative sample of all crashes.
Contact information from sampled claims was then transferred electronically on a daily basis, on the same day that the data were originally received from State Farm, to Response Analysis Corporation (RAC), a telephone survey firm based in Princeton, New Jersey. Within 24 hours of receipt of the data, RAC initiated telephone contact with the insured. The firm conducted a 25-minute survey with the driver of the vehicle and parent(s) of the involved children. After completion of the interview, data were transferred electronically back to CHOP/Penn on a daily basis for inclusion in the surveillance system.
Survey Design and Validation
A survey instrument was developed for this project. Among other information, the instrument was designed to ascertain the seating position (eg, front row, right passenger position) restraint use (restrained versus unrestrained), and restraint type (child safety seat, booster seat, lap only seat belt, lap/shoulder seat belt) for each child occupant. Questions regarding the description of restraint type and use were validated on a separate population by comparing parent responses to direct observation of the restraint system by trained child passenger safety technicians.6 Further, restraint questions were validated by on-site professional investigations of crashes involving 101 children on whom interviews were conducted and demonstrated a high degree of agreement (89%) between parent report and the investigator's report of restraint use (unpublished data).
Questions regarding injuries to children were designed to provide responses that were classified by body region and severity based on the Abbreviated Injury Scale (AIS) system.7 The AIS is the most widely used severity scoring system based on anatomic injury data. The AIS rates the severity of an injury from 1 (eg, contusions and minor lacerations) to 6 (uniformly fatal) for each of 7 body regions. Significant injuries were defined as AIS 2 or more, and included concussion and more severe brain injuries, internal organ injuries, and most fractures. These questions were previously validated to differentiate AIS 2 or more from minor injuries.8
Reliance on electronic transfers of data and automated quality assurance and sampling procedures has resulted in a system capable of collecting and analyzing interview data and initiating a crash investigation within 72 hours of the crash. Data encryption techniques were used to ensure maximum security of the data during all transfers among entities. The institutional review boards of both The Children's Hospital of Philadelphia and the University of Pennsylvania School of Medicine approved the study protocol. In addition, the project is in compliance with all relevant state insurance laws in the involved regions.
The study sample was weighted, based on the sampling proportions corresponding to the treatment status of each claim, to represent the total eligible population of claims consenting for participation in the surveillance system. All analyses were then conducted on the weighted population. Simple descriptive statistics were calculated, including frequencies for categorical variables and mean, standard deviation, median, and range for continuous variables. The χ2 or Fisher's exact test, when indicated, were used to examine the association between restraint type and significant injury, both overall, and significant head injury, in particular. Relative risks (RR) with 95% confidence intervals (CI) were calculated.
Between December 1, 1998, and November 30, 1999, 56 053 crash claims meeting inclusion criteria were reported to State Farm Insurance Companies. From among all eligible claims, 38 557 claims involving 59 643 child occupants (1.5 children per claim) were transferred to CHOP/Penn for inclusion in the surveillance system and represented a capture rate of 87%. The consent rate among policyholders asked to participate in the project was 81%. Among the consenting claims received by CHOP/Penn, 11 123 (29%) were sampled for a telephone interview. The average time to first contact with the policyholder by the telephone survey firm was 16 hours from receipt of the data. The consent rate for completion of the telephone survey was 99%.
Complete interview data were obtained on 8334 children who were then weighted to represent 48 108 child occupants 0 to 15 years of age. Overall restraint usage among all children was high at 95%. However, as noted in Fig 1, compliance with current recommendations9,,10 regarding the proper type of restraint device varied widely with age. In general, the majority of children older than 8 years were in compliance with current recommendations (use of a seat belt). Similarly, the majority of children younger than age 3 were in compliance with current recommendations (use of a child safety seat, rear-facing for infants). However, few children between 4 and 8 years of age were properly restrained for their age, because of their failure to use a booster seat. As noted in Fig 2, seat belt use began at age 2 and was the most common form of restraint by age 4. Booster seat use in our population peaked at age 3 (29% of 3-year-olds) and declined dramatically with each succeeding year so that <1% of children older than age 5 were restrained in booster seats.
To address the aim of risk of seat belts to preschoolers, the weighted sample of children ages 2 to 5 formed the basis of this study. These 13 853 children accounted for 28.7% of all children included in the surveillance system. There were 7036 (51%) boys and 6817 (49%) girls. See Table 1 for representative descriptive data on the crashes included in the study sample. A broad distribution of vehicle model year was represented, as well as a broad distribution of principal impact area and crash severity, as represented by the driveable status of the vehicle. All states in the study regions contributed crashes to the surveillance system with California, Illinois, Michigan, and Pennsylvania contributing approximately half of the cases.
Among 2- to 5-year-old children, restraint usage was 98%. In this age group, seat belt usage overall was 38.5% but varied widely with 5.5% of 2-year-olds, 16.3% of 3-year-olds, 54.8% of 4-year-olds and 81.3% of 5-year-olds in seat belts. Among children in seat belts, 19% were using only the lap belt. Among children restrained in boosters, 50% were using shield boosters and 50% were using belt-positioning boosters. Overall, 1331 young children (9.6%) suffered some type of injury, with 135 children (1% of all children) suffering significant injuries. Significant injuries occurred to all body regions with head injuries the most common (58% of all significant injuries).
Young children in seat belts were more likely to suffer a significant injury (RR: 3.5; 95% CI: [2.4, 5.2]) than young children in CRS. Young children in seat belts were at particular risk of significant head injuries (RR: 4.2; 95% CI: [2.6, 6.7]) when compared with children in CRS. Children in CRS suffered significant injuries primarily to the head (51% of significant injuries), with additional injuries to the face (13%), and extremities (16%). Of note, young children in seat belts suffered the only significant abdominal injuries (n = 8) in the entire population. These 8 children were all restrained in lap/shoulder belts, however, 3 of the children had placed the shoulder portion of the belt behind their backs.
The risk of significant injury from premature graduation to seat belts was somewhat greater for 2- to 3-year-olds (RR: 4.0; 95% CI: [2.0, 7.9]) than for 4- to 5-year-olds (RR: 2.4; 95% CI: [1.4, 4.3]). The increased risk of injury to children in seat belts was similar when compared separately with children in child safety seats (RR: 3.4; 95% CI: [2.2, 5.2]) and children in booster seats (RR: 4.0; 95% CI [2.0, 7.9]). There was no difference (P = .23) in risk of injury for children restrained in lap only versus lap/shoulder belts.
This study provides the first real-world evidence for an increased risk of injury in preschool-aged (2- to 5-year-old) children attributable to inappropriate restraint in vehicle seat belts rather than CRS. Despite high restraint usage, more than one-third of children in this age range were inappropriately restrained by vehicle seat belts. This inappropriate restraint resulted in a 3.5-fold increased risk of significant injury and a more than fourfold increased risk of significant head injury. These data bolster the recent emphasis on the proper restraint of children to prevent injuries in crashes, in particular the use of booster seats for children 4 years and older.
A vehicle seat belt fits correctly when the lap portion of the belt rides low over the hips and is held in place by mature anterior superior iliac spines.11 A well-fit shoulder portion of the belt crosses the sternum and shoulder. Correct seat belt fit is not usually achieved until a child is 9 years old, the age at which the child's femur length is long enough for the child to sit against the back of the seat, the anterior superior iliac spines are sufficiently developed to anchor the belt, and the child's sitting height is sufficient for the shoulder belt to fit properly over the shoulder and sternum. By these guidelines, virtually no child younger than the age of 6 is large enough to be properly restrained in a seat belt.12
When a child is prematurely graduated to a seat belt from a CSS, the lap portion of the belt rides up over the abdomen and the shoulder portion crosses the neck or face. This places the child at risk for submarining or sliding out of the lap belt during a crash. Rapid, jack-knife bending about a poorly positioned vehicle seat belt increases the risk of intraabdominal and spinal cord injuries, also known as seat belt syndrome, and brain injury resulting from the impact of the head with the child's knees or the vehicle interior.9,,13,14 The data presented in this study point to the high risk of head injuries in young children restrained by vehicle seat belts, likely attributable to increased head excursion. In addition, although abdominal injuries were not common, they only occurred in children in seat belts not those in CRS. Of note, the risk of injury was similar for children in lap belts and lap/shoulder belts, suggesting that the addition of a poorly fitting shoulder portion of the belt offers no added protection for young children.
In this study, many 2- and 3-year-old children were restrained in booster seats; a younger age than current guidelines recommend.9,,10 Limited sample sizes precluded meaningful analysis of the risk of injury associated with shield versus belt-positioning booster seats. A shield booster has a shield that crosses in front of the child to restrain the child's hips instead of using the vehicle seat belt. A belt- positioning booster, either with or without a high back, raises the child up to improve the fit of both the lap and shoulder portions of the seat belt. Of note, for children aged 2 and 3 years, the risk of injury associated with booster seat use was similar to the risk of injury for children in CSS. This result may be attributable to a high degree of CSS misuse as reported in previous studies,15 which may serve to reduce the effectiveness of CSS. This misuse frequently involves loose fit of the child in the CSS harness and loose fit of the CSS in the vehicle.3 Both of these forms of misuse result in excessive excursion of the child and would result in increased head excursion with the potential for resultant head injury. Reductions in CSS misuse might further improve the effectiveness of CSS in preventing head injuries. Shield boosters are no longer recommended for use because of risk of submarining and ejection9 and these accounted for half of the booster seats used in this study. Further reductions in injury might be achieved by the specific promotion of belt-positioning boosters rather than shield boosters.
Before PCPS, sources of child crash injury data were inadequate for estimating the exposure of children to specific crash circumstances, particularly children who are uninjured or who receive only minor injuries.16 Although the Fatality Analysis Reporting System provides data regarding mechanisms of child occupant injury, this data source only includes fatal crashes and cannot provide estimates for child exposure to nonfatal or noninjury crashes. Similarly, trauma center-based investigations can elucidate mechanisms of injury but cannot provide any estimates of exposure because of the nature of trauma system triage in which more seriously injured children are cared for in these centers. The National Automotive Sampling System (NASS), a third data source, has the potential for providing exposure estimates but as described by Newgard and Jolly,17 is fraught with missing data elements and, as a population-based sampling system, includes relatively few children. PCPS was created to overcome some of the deficiencies in current sources of child occupant injury data by including large numbers of injured and noninjured children in a broadly representative exposure-based surveillance system.
The estimate of overall restraint usage in this study was higher than current national estimates of child restraint usage.2 This difference was attributable, in part, to the contemporary nature of PCPS results that were based on 1999 data. In a recent study using NASS data from 1988 through 1995, Edwards18 found that child restraint use was significantly related to driver restraint use; children of restrained drivers were more than twice as likely to be restrained than children of unrestrained drivers. Our results on age-specific child restraint usage are, in fact, strikingly similar to those for restrained drivers in the Edwards study: 97% of preschool-aged children were restrained when the driver was restrained. This suggests that, although estimates of restraint use from PCPS may not be generalizable to children in all crashes, they are likely generalizable to children in crashes with restrained drivers. Currently, national driver restraint use is estimated at 69%, with California reporting restraint use of 88% in the general population.5
Information regarding restraint use and type were obtained via telephone interview with the driver/parent of the child. Previous studies have questioned the validity of self-reported restraint use.19,,20 We have attempted to reduce this potential source of bias in several ways. Rather than simply asking whether or not the child was restrained, questions regarding restraint use in the telephone survey were designed to require the respondent to describe specific characteristics of the restraint system and the way it was used. In addition, responses to multiple questions were analyzed to identify inconsistencies that would suggest incorrect responses. Finally, as noted above, preliminary evaluation of the agreement between parent-reported restraint use and the results of on-site crash investigations suggest that the vast majority of respondents provide valid and accurate assessments of restraint use and type.
PCPS has the potential for limitation because of its reliance on >7000 insurance claim representatives to identify qualifying cases and to obtain initial consent for inclusion in the study. Mechanisms to ensure continuous updating and training of these individuals have been created by State Farm to maintain high rates of capture of qualifying claims. As a result, this system with reports of >1000 crashes involving children per week provides the largest child-focused crash surveillance system in the world. The surveillance system is limited, however, to children occupying model year 1990 and newer vehicles insured in 15 states and the District of Columbia. Results of this study may, therefore, not be generalizable to children occupying older or uninsured vehicles, or to children residing in nonstudy states. Finally, the results of this study are not generalizable to children older than 5 years. As demonstrated in Fig 1, few children above the age of 5 were restrained in booster seats and, as such, the risk of injury associated with the use of seat belts as compared with that associated with booster seat use could not be assessed.
Premature graduation of young children from CRS to seat belts puts them at greatly increased risk of significant injury in crashes. A major benefit of CRS is a reduction in head injuries, potentially attributable to a reduction in the amount of head excursion in a crash. Results of this study support public health efforts directed toward ensuring appropriate restraint of children, particularly the use of belt-positioning booster seats by children who have outgrown CSS. To reduce the risk of injury, children should remain in CSS until they are at least 4 years old and weigh 40 pounds (18 kg), at which point children should be placed in belt-positioning booster seats. Children should remain in booster seats until they are the appropriate height and weight for seat belts.
This work was funded by State Farm Insurance Companies.
We thank State Farm Insurance Companies for their financial support of this work through the Partners for Child Passenger Safety project. In addition, the authors would like to thank the many dedicated claim representatives and personnel from State Farm, the research team on the Partners project, and Response Analysis Corporation who devoted countless hours to this study, and the parents who generously agreed to participate in the study.
- Received February 4, 2000.
- Accepted March 17, 2000.
Reprint requests to (F.K.W.) The Children's Hospital of Philadelphia, 34th St and Civic Center Blvd, 3535 TraumaLink, 10th Floor, Philadelphia, PA 19104. E-mail:
- MVC =
- motor vehicle crash •
- CRS =
- child restraint systems •
- CSS =
- child safety seats •
- PCPS =
- Partners for Child Passenger Safety •
- CHOP/Penn =
- The Children's Hospital of Philadelphia/University of Pennsylvania •
- RAC =
- Response Analysis Corporation •
- AIS =
- Abbreviated Injury Scale •
- RR =
- relative risk •
- CI =
- 95% confidence interval •
- NASS =
- National Automotive Sampling System
- ↵Committee on Injury Prevention and Control. Reducing the Burden of Injury: Advancing Prevention and Treatment. Bonnie R, Fulco C, Liverman C, eds. Washington, DC: National Academy Press; 1999
- ↵http://www.nhtsa.dot.gov/people/ncsa/tsf97about.htmlTraffic Safety Facts 1997. Washington, DC: National Highway Traffic Safety Administration; 1998
- ↵Winston FK, Durbin DR, Bhatia E, Werner J, Sorenson W. Patterns of inappropriate restraint for children in crashes. 43rd Annual Proceedings of the Association for the Advancement of Automotive Medicine. Barcelona, Spain: Association for the Advancement of Automotive Medicine; 1999:59–69
- ↵http://www.nhtsa.dot.gov/people/ncsa/pdf/OccPrt98.pdfTraffic Safety Facts 1998. Washington, DC: National Highway Traffic Safety Administration; 1999
- ↵Arbogast KB, Durbin DR, Morris SD, Winston FK. Assessing child restraint misuse by parental survey. Inj Prev. In press
- ↵The Abbreviated Injury Scale, 1990 Revision. Des Plaines, IL: Association for the Advancement of Automotive Medicine; 1990
- ↵http://www.aap.org/family/01352.htmSelecting and Using the Most Appropriate Car Safety Seats for Growing Children. Guidelines for Counseling Parents (RE9618). Elk Grove Village, IL: American Academy of Pediatrics; 1996
- ↵http://www.nhtsa.dot.gov/people/injury/childps/RxFlyer/RxFlyer4 html. Protect Your Kids in the Car. Washington, DC: National Highway Traffic Safety Administration, Emergency Nurses Association, American College of Emergency Physicians; 1997
- ↵http://www.safekids.org/home.htmlFrequently Asked Questions: What Is the Best Car Seat to Buy for My Child? Washington, DC: SafeKids; 1999
- Hamill PVV,
- Drizd TA,
- Johnson CL,
- Reed RB,
- Roche AF,
- Moore WM
- ↵Gotschall C, Better A, Blaus D, Eichelberger M, Bents F, Warner M. Injuries to children restraint in 2- and 3-point belts. 42nd Annual Proceedings of the Association for the Advancement of Automotive Medicine. Charlottesville, VA: Association for the Advancement of Automotive Medicine; 1998:29–43
- Kulowski K,
- Rost W
- ↵Gotschall C, Dougherty D, Eichelberger M, Bents FD. Traffic-related injuries to children: lessons from real world crashes. 42nd Annual Proceedings of the Association for the Advancement of Automotive Medicine. Charlottesville, VA: Association for the Advancement of Automotive Medicine; 1998
- ↵Newgard C, Jolly BT. A descriptive study of pediatric injury patterns from the National Automotive Sampling System. 42nd Annual Proceedings of the Association for the Advancement of Automotive Medicine. Charlottesville, VA: Association for the Advancement of Automotive Medicine; 1998
- ↵Edwards J, Sullivan K. Where are all the children seated and when are they restrained? Child Occupant Protection Second Symposium. Orlando, FL: Society of Automotive Engineers; 1997:35–42
- Nelson DE
- Copyright © 2000 American Academy of Pediatrics