Objective. To determine the effect of restraint use and seating position on injuries to children in motor vehicle crashes, with stratification by area of impact.
Methods. Children <15 years old involved in serious automobile crashes in Utah from 1992 through 1996 were identified from statewide motor vehicle crash records. Serious crashes are defined as those resulting in occupant injuries with broken bones or significant bleeding or property damage exceeding $750. Probabilistic methods were used to link these records with hospital records. Analysis used logistic regression controlling for age, restraint use, occupant seating position, and type of crash.
Results. We studied 5751 children and found 53% were rear seat passengers. More than 40% were unrestrained. Sitting in the rear seat offered a significant protective effect (adjusted odds ratio: 1.7; 95% confidence interval: 1.6–2.0), and restraint use enhanced this effect (adjusted odds ratio: 2.7; 95% confidence interval: 2.4–3.1). Mean hospital charges were significantly greater for front seat passengers.
Conclusions. Rear seat position during a motor vehicle crash provides a significant protective effect, restraint use furthers this effect, and usage rates of restraint devices are low. The rear seat protective effect is in addition to and independent of the protection offered from restraints.
Injuries suffered in motor vehicle crashes are the leading cause of death in children and adolescents in the United States.1 The effectiveness of seat belts and child car safety seats in preventing these injuries is well known.2,,3 Seating position in the vehicle during a crash also contributes to the risk for injury, with the location of impact relative to the occupant's seating position being an important factor. In frontal impact crashes, passengers in the rear seat have a reduced risk of serious injury and mortality.2,,4,5 The combined contribution of seating position and restraint use to injury prevention in automobile crashes has been reported.3,5–7 Available data describing the effect of restraint use and seating position for children in motor vehicle crashes are incomplete; studies variously had small numbers of children, low restraint use, or were not population-based. Because of these factors, we undertook the present study.
The purpose of our study was to determine the effect of seat belt or child restraint use and seating position on injuries to children in motor vehicle crashes. Specifically, we sought to determine whether sitting in the rear seat during a serious crash provided protection to the child occupant after allowing for the use of an age-appropriate restraint device. Additionally, we sought to determine whether differences in outcome were observed when controlling for the confounding variable of crash type. We used probabilistic methods8 to link statewide motor vehicle crash and hospital databases to address both these questions. The technique of probabilistic linkage allowed us to create a more complete database of specific crash events and patient outcomes that we could stratify by subject age, seating position, restraint use, and crash type.
Motor vehicle crash records for the years 1992 through 1996 were obtained from the Utah Department of Transportation. This file includes crashes reported on public roads in Utah, including all injury crashes and any crash resulting in property damage over $750. Excluded crashes were those not occurring on public roadways and those that did not result in any occupant injuries or significant property damage. These determinations and data collection were performed by local law enforcement officers. The Utah crash reports include all occupants in a vehicle, including noninjured occupants.
Records include extensive information about crash circumstances, including road, weather, and lighting conditions, vehicle damage descriptions, and driver actions. Occupant information includes age, gender, exact position within the vehicle, use of safety devices, and injury codes, as reported by the investigating police officer.
Statewide computerized hospital discharge records from 1992 through 1996 were obtained from the Office of Health Data Analysis, Utah Department of Health. This database provides information about length of stay, inpatient charges, discharge diagnosis, and discharge status (outcome). The database contains all Utah hospital discharges for the 5-year period (1 077 439 discharges).
We studied children from birth through 14 years old who were occupants in a serious automobile crash. We defined serious crashes as those resulting in a fatality, hospitalization, or a serious injury defined as injuries resulting in broken bones or significant bleeding to any of the automobile occupants. We stratified the crash victims into age groups as: young children 0 to 4 years old, school-aged children 5 to 11 years old, and older children defined as 12 to 14 years old. Vehicles were restricted to passenger cars, light trucks (including sport-utility vehicles), and vans. Children traveling in a nonpassenger area, such as a truck bed, were excluded from our analysis. As other investigators have done,2 restraint use was defined as optimal, suboptimal, or none. Optimal restraint included the use of a child safety seat from birth through 4 years old or use of a lap and shoulder belt for children from 5 through 14 years old. Suboptimal usage was defined as a lap belt, shoulder belt, or both in children 0 through 4 years old, or use of a lap belt or shoulder belt alone in the 5- through 14-year age groups. None means that no restraint device was used by the child.
The database for analysis was constructed by linking the crash database to the hospital file using probabilistic methods. These are described in detail by Jaro.8 Briefly, this technique allows linkage of information for a single crash event among disparate databases. For example, specific details for a single crash victim contained in the crash database can be linked to hospital data for the same crash victim. The ability to link specific crash events and parameters to hospital outcome data allows for a more complete analysis of the event in total. From this, we can make observations that would otherwise be impossible from an examination of either database individually.
The data were analyzed using logistic regression with the primary outcome being serious injury or death. The independent variables in our model were victim age, restraint use, type of crash, and front or rear seating position in the vehicle. Interaction covariates between seating position and crash type were also assessed. Performing multivariable logistic regression allowed us to calculate odds ratios (ORs) adjusted for all independent variables in our model as listed above. Total hospital charges stratified by restraint use and seating position were secondarily evaluated. Average hospital charges were calculated assigning a charge of $0 to passengers who were not admitted to the hospital. We defined a P value of <.05 to be statistically significant.
Probabilistic record linkage was performed using Automatch 4.1 (Matchware Technologies, Inc, Boston, MA). Multivariable logistic regression was done using SAS procedure (PROC LOGISTIC), and other statistical analyses were performed in SAS as well (SAS Institute, Inc, Cary, NC).
During the 5-year period from 1992 to 1996 the database included 284 550 crashes involving 464 128 vehicles and 573 772 occupants. Overall 5751 children were involved in serious crashes during the 5 years studied. Stratified by age group, this included 2016 children from birth through 4 years old, 2231 children from 5 through 11 years old, and 1504 children from 12 through 14 years old. There were 2383 children seriously injured, 134 children hospitalized, and 53 killed.
Seating Position and Restraint Usage
As might be expected, a larger percentage of older children sat in the front seat. A total of 2441 children were in the front seat and 3310 were in the rear seat at the time of the crash. We found 743 (37%) young children, 947 (42%) school-aged children, and 751 (50%) older children were riding in the front seat at the time of the crash. Restraint usage varied by age group and seating location in the vehicle (Fig 1). Of all children, 38% (2168 of 5751) were optimally restrained. Moreover, 27% of children 0 to 4 years old, 44% of children 5 to 11 years old, and 52% of children 12 to 14 years old were unrestrained in the crash. Suboptimal restraint use was noted in 34% of children birth to 4 years old versus 13% of children 12 to 14 years old (P < .01). The percentage of children optimally, suboptimally, or not restrained did not vary whether they were in the front or back seat (Table 1).
Controlling for age and restraint use, a child was 1.7 times more likely to suffer a serious injury or fatality when sitting in the front seat versus the rear seat during a crash. When age and seating position were controlled for, the benefit of restraints was clearly seen (Table 2). When restraints were optimally used, occupants were 2.7 times more likely to endure the crash without serious injury or fatality than unrestrained children, regardless of seating position. There was no significant difference between optimal and suboptimal restraint use in preventing serious injury or death (adjusted OR: 1.2; P = .94).
There were 53 deaths during this 4-year period yielding a fatality rate of .9% for children involved in serious crashes. Sixty percent of deaths occurred to front seat passengers versus 40% to those in the rear seat at the time of the crash. No form of restraint device was used in 70% of fatally injured passengers, 25% were optimally restrained, and 5% were suboptimally restrained.
Rear Seat Protection by Crash Type
We examined the odds of suffering serious or fatal injury, in relation to seating position, regardless of restraint use. Because the forces to which an occupant was exposed vary greatly depending on where they were sitting and where the primary impact occurred, we controlled for crash type. Rear seat passengers, compared with those in the front seat, were significantly less likely to be seriously injured or killed in frontal impact crashes (adjusted OR for injury: .73; 95% confidence interval [CI]: .60-.90). Rear seat occupants were also better protected against serious injury or death than those in the front seat during nonfrontal impact crashes (adjusted OR for injury: .57; 95% CI: .51-.64).
Hospital Admission Rates and Charges
Only 2.3% of crash victims required hospitalization (134 of 5751). By seating location, 2.6% (64 of 2441) of front seat passengers and 2.1% (70 of 3310) of rear seat passengers were admitted to the hospital. This difference was not significant (P = .2). Regardless of seating location, 1.2% (25 of 2168) of restrained and 3.3% (75 of 2301) of unrestrained victims required hospitalization (P < .001), underscoring the importance of restraints in preventing injury and decreasing the need for hospitalization. The uninjured occupant hospital charge was $0; the mean inpatient hospital charge for all front seat passengers was $284.18. This was significantly greater than that for rear seat passengers, which was $194.74 (P = .02). Unrestrained front seat passengers requiring admission incurred the highest hospital charges, averaging $11 190 (±$6930, 2 standard deviations), with a range from $540 to $139 470. Conversely, optimally restrained rear seat occupants incurred the lowest mean hospital charge averaging $6950 (±$3380, 2 standard deviations), 62% lower than those unrestrained in the front seat.
This study confirms that sitting in the rear seat during a motor vehicle crash offers a significant protective effect, restraint use enhances this effect, and usage rates of restraint devices are too low. The protective effect of the rear seat in a crash is in addition to the protection offered by restraints. Nationwide, 68% of automobile passengers used seatbelts in 1996, and a federal directive in January 1997 established a goal of 85% seatbelt usage with a goal of reducing child occupant fatalities 15% by the year 2000.9Nevertheless, our data show that many children remain unrestrained in cars at the time of a crash, significantly increasing their risk for serious injury or death.
Huelke and Compton4 reported the effect of seat belts on injury severity to front and rear seat passengers in the same frontal impact crash.4 They found that for adult victims of frontal impact crashes the rear seat was a safer location than the front seat. This observation also applied to rear-seated children 5 to 14 years old when they were in the same crash event. Only 8% of unrestrained rear-seated child passengers suffered a more severe injury, defined as one with a higher maximum abbreviated injury scale score, than an unbelted adult front seat passenger. When restraints were used, only 5% of children in the rear seat had a more severe injury than their front seat adult counterpart. Our data confirm that being in the rear seat during a crash is protective, with or without the use of restraints.
To expand on this observation, we examined the outcomes of rear seat passengers stratified by crash type. Because of the forces involved, frontal impact crashes are more injurious overall, and front seat passengers are 1.3 times more likely to be seriously injured or killed than those in the rear seat. We and others2,,7 have observed that rear seat pediatric passengers have decreased risks of morbidity and mortality when involved in serious automobile crashes. In our analysis, the protective effect of the rear seat persists, even in the more dangerous frontal impact crashes (adjusted OR: .73; 95%CI: .6-.9). An even greater protective effect is observed in rear seat passengers who are in nonfrontal impact crashes (adjusted OR: .57; 95% CI: .51-.64).
As children age, fewer are restrained when riding in an automobile. Twice as many children 12 to 14 years old are unbelted than are younger children 0 to 4 years old (Fig 1). This may be because older children are responsible for securing themselves without direct parental supervision or involvement. Furthermore, adolescents may view wearing a seat belt as uncool and may shun the seat belt for approval from their peers. Given their level of development, younger children must have an adult restrain them correctly and thus may be more likely to be restrained.
Suboptimal restraint usage decreases significantly as children age. Children 0 to 4 years old are 2.5 times more likely to be incorrectly restrained (which in this age group includes wearing a seat belt) at the time of a crash than children 12 to 14 years old. For young children, seat belt use carries risks because these restraint systems are engineered for the larger (>80-lb) body.9 Blunt abdominal trauma, a specific pattern of which known as the seat belt syndrome, is observed in small children held by seat belt systems designed for adults.10 Despite their potential for creating specific patterns of injury in the pediatric patient, seat belts are still proven to reduce mortality, injury severity, length of hospital stay, and chance of permanent impairment.11 Their correct use offers the greatest benefit.
The calculated mean hospital charges may seem low to the clinician. However, these reported charges are mean values of all crash victims including those not admitted to the hospital. We computed the mean values using a charge of $0 for nonhospitalized subjects to account for the large number of people who benefited from a restraint device or seating position and did not require hospitalization. This further illustrates the primary importance of restraint use as a preventive measure, reducing serious injury or mortality, and health care costs. When this selection bias is not accounted for, and mean charges are calculated based only on those patients admitted to the hospital, we could not detect any difference in hospital charges to those in the front versus rear seat, with or without restraints. This demonstrates selection bias in that those suffering injuries severe enough to require hospitalization, regardless of the circumstances, may incur similar charges. When examined in total, if all children were in the rear seat, there would have been 306 fewer injuries during this period. When multiplied by the average hospital charge per injured child, this equals a potential savings of $236 800 to the health care system in Utah alone.
Despite the known protective effect of rear seat position for children, few public health interventions have been made in this area. A recent exhaustive review identified studies evaluating interventions promoting pediatric rear seat use,12 3 of which demonstrated a significant increase in rear seat use after a public health intervention. The intervention was an educational effort in 1 case, and a legislative initiative in 2 cases. As the evidence for improved safety and better crash outcomes to rear-seated children becomes more conclusive, educational or legislative efforts to increase rear seat use will be increasingly important. Ongoing evaluation of the efficacy of these advocacy efforts will be crucial.
The limitations of this study are those that are inherent to research of this nature, primarily inaccuracy in crash reporting and data entry.13 Crashes in which no law enforcement official was notified, those occurring on private property (not on public roads), and those not meeting our previously stated inclusion criteria are not included in the database, and thus are not available for our analysis. Finally, use or nonuse of restraints is self-reported by the crash victims who may overreport their use. Selective overreporting is known to occur more among less severely injured passengers. This inflates the apparent effectiveness of seat belts.14 Given that our study population is children involved in crashes with a serious injury or fatality, these limitations should bias our results minimally.
In summary, sitting in the rear seat and using an age-appropriate restraint device significantly and independently decrease the risk of serious injury or death to children involved in serious automobile crashes. Optimal restraint usage rates were low in this study, <40% across all ages studied. Moreover, one third of the youngest children, 0 to 4 years old, were wearing an adult seat belt at the time of the crash when the appropriate restraint would have been a child safety seat. Nonuse or suboptimal restraint use places our children at unnecessarily high risk for serious injury or death. This should be considered when counseling patients and parents on car safety.
This research was supported by Cooperative Agreement Grant DTNH22-92-Y-57329 from the National Highway Traffic Safety Administration, United States Department of Transportation; and Cooperative Agreement Grant MCH-494003-03-0 from the Maternal and Child Bureau, Health Resources and Services Administration, Public Health Service, United States Department of Health and Human Services.
We thank the Utah Department of Transportation for providing research access to the crash files used in this study and the Office of Health Data Analysis, Utah Department of Health, for providing access to the inpatient hospital discharge database.
- Received June 25, 1999.
- Accepted December 10, 1999.
Reprint requests to (M.D.B.) Department of Pediatrics, Division of Critical Care, Primary Children's Medical Center, 100 N Medical Dr, Salt Lake City, UT 84113. E-mail:
- OR =
- odds ratio •
- CI =
- confidence interval
- Johnston C,
- Rivara F,
- Soderberg R
- Periodou E,
- Skalkidou A,
- Lescohier I,
- Trichopoulos D
- ↵National Highway Transportation Safety Administration. National Highway Transportation Safety Administration Data. Washington, DC: National Highway Transportation Safety Administration; 1999
- Osberg J,
- DiScala C
- Segui-Gomez M
- Agran P,
- Castillo D,
- Winn D
- Dean J,
- Reading J,
- Nechodom P
- Copyright © 2000 American Academy of Pediatrics