Objective: To determine the relationship between body mass index (BMI) and injury risk among US children in motor vehicle crashes.
Design: Cross-sectional study using data from the Partners for Child Passenger Safety study, a child-focused crash surveillance system.
Participants: A probability sample of children, 9–15 years of age, involved in crashes in parent-operated vehicles between 1 December 2000 and 31 December 2006.
Main outcome measure: The odds ratio of Abbreviated Injury Severity (AIS) 2+ injuries (overall and body region specific) by BMI category: underweight, normal, overweight, and obese.
Results: The study sample included 3232 children in 2873 vehicles, representing a population estimate of 54 616 children in 49 037 vehicles. Approximately 15% (n = 502) sustained an AIS 2+ injury to any body region; 34% of the children were overweight or obese. There was no overall increase in injury risk by BMI; however, body region differences were found. In multivariate logistic regression, compared with normal weight children, the odds of sustaining an AIS 2+ injury to the extremities for overweight and obese children was 2.64 (95% CI 1.64 to 4.77) and 2.54 (95% CI 1.15 to 5.59), respectively.
Conclusions: Although overweight and obese children are not at increased overall risk of injury, they are at increased risk of injury to the lower and upper extremities. This increased risk may be due to a combination of physiology, biomechanical forces, and vehicle design.
Statistics from Altmetric.com
Notable contributions have been made to the identification of risk factors for pediatric injuries resulting from motor vehicle crashes, the leading cause of death and a leading cause of non-fatal injuries for youth.1–5 One potential risk factor that has not received significant attention is obesity. The high prevalence of obesity among youth warrants its exploration as a potential risk factor for injury during transport. Empirical studies on adult occupants of motor vehicles involved in a crash has revealed that obese occupants are at greater risk of injury and mortality and sustain different patterns of injury from lean occupants.6–10 Although research in this area is limited for children, the association seems plausible. Research on some types of pediatric injury has shown an increased risk for overweight or obese children.1112 In addition, one study estimated that a substantial number of US children are at a weight that may not allow them to safely use currently available car seats, therefore possibly increasing their risk of injury.13
The objective of this study was to determine the relationship between body mass index (BMI) and the risk of injury resulting from motor vehicle crashes in older children in the USA. We used data from the Partners for Child Passenger Safety (PCPS) study, a large child-focused crash surveillance system. We hypothesized that, after factors related to the crash and restraint system are taken into account, child BMI would be significantly associated with the risk of injury. On the basis of the adult data on this topic, we also hypothesized that the pattern of body region injured would vary by BMI, such that injuries to the extremities would be more prevalent among heavier children.
The study sample comprised children enrolled in the PCPS study. A thorough description of the development of the PCPS study has been published previously.14 The project consists of a large-scale, child-specific crash surveillance system; insurance claims from State Farm Insurance Co (Bloomington, Illinois, USA) function as the source of subjects, with telephone survey and on-site crash investigations serving as the primary sources of data. Vehicles qualifying for inclusion were State Farm-insured, model year 1990 or newer, and involved in a crash with at least one child occupant <16 years of age. Qualifying crashes were limited to those that occurred in 15 states and the District of Columbia, representing four regions of the USA (East, Midwest, South, and West).
A stratified cluster sample was designed to select vehicles (sampling unit). If a vehicle was sampled, all child occupants in that vehicle were included in the survey. Drivers of sampled vehicles were contacted by phone, and, if a passenger had received medical treatment, were screened via an abbreviated survey to verify the presence of at least one child occupant with an injury. All vehicles with at least one child who sustained an injury and a 10% random sample of vehicles in which child occupants received medical treatment but did not sustain an injury were selected for a full interview; a 2.5% sample of crashes where no medical treatment was received was also selected. This sampling scheme was designed to select the majority of injured children. Sampling weight, which was proportional to the inverse of the sampling probability, was applied to all cases to maintain the representativeness of the overall population.
Crash investigation teams went to the crash scenes within 24 h of notification to document the crash environment, damage to involved vehicles, and occupant contact points. Information collected from on-scene investigations were supplemented by reports from occupants, witnesses, medical personnel, police reports, and medical records. The study protocol was reviewed and approved by the Institutional Review Boards of both The Children’s Hospital of Philadelphia and the University of Pennsylvania School of Medicine.
This analysis used PCPS data collected between 1 December 2000 and 31 December 2006. Our sample included vehicles with older children, aged 9–15 years, whose driver was their parent or guardian. We included only the children riding in parent-operated vehicles because child height and weight reported by non-parent drivers tended to be inaccurate. To exclude children who would be optimally restrained in a child safety seat or booster seat, we included only children at least 5 feet tall, who were not riding in booster seats. Although 4 foot 9 inches is the suggested height for children to transition out of a booster seat, to account for variability in parent recall and rounding of their child’s height, we used a cut-off of 5 feet. Cases with missing weight data (2.9% of the 9–15 year old children riding with parents) were excluded from this analysis. After application of these exclusions, the sample was 3232 children in 2873 vehicles, which represented population estimates of 54 616 children in 49 037 vehicles.
Definitions and validity of variables
Child obesity was defined by BMI, which is calculated from a child’s weight and height. To determine if a child is obese, his/her BMI centile is compared with his/her age (months) and a gender-specific BMI distribution developed by the US Centers for Disease Control and Prevention (CDC). The CDC defines a child as underweight, normal, at-risk for overweight, or overweight if his/her BMI is <5th, between 5th and 85th, between 85th and 95th, or >95th centile, respectively, for his/her age. We classified a child’s BMI using the CDC method; however, in the PCPS data, parents often reported their child’s age in years rather than months. Therefore, we used the midpoint between a child’s reported age and next birthday to categorize age in months. The degree of misclassification using this method was assessed and found to be very low (<3%). Although the CDC uses the term “at-risk for overweight”, recently published studies classify these children as overweight, and overweight children according to the CDC definition, as obese.1516 Therefore, we classified children as underweight, normal, overweight, or obese.
Child weight and height in the PCPS data were parent-reported. The validity of parent-reported measures was explored by reassessing height and weight for the subset of cases in which a crash investigation occurred. Among children aged 9–15, the correlation coefficient between the survey and the crash investigation was 0.90 for the two reported heights and 0.81 for the two weights. In the survey, when parents reported a height and weight, they were asked if the values were recently measured or their best guess. The correlation coefficients were 0.98 and 0.79 for recently measured and best-guessed child height, respectively. The correlation coefficients were 0.90 and 0.92 for recently measured and best-guessed child weight, respectively.
Injuries were also parent-reported. Survey questions on injuries to children were designed to provide responses that were classified by body region and severity based on the Abbreviated Injury Scale (AIS) score.17 The ability of parents to accurately distinguish AIS 2 or greater injuries from less severe ones has been previously validated for all body regions of injury.14 A child passenger was considered injured if he/she had an AIS 2+ injury at any body region. These included: AIS2+ head injury (concussion, intracranial hemorrhage, skull fracture); AIS2+ face injury (facial bone fractures); AIS2+ chest injury (rib or other chest fractures, injuries to organs such as heart, lung, trachea, and esophagus); AIS2+ abdominal injury (injuries to internal organs such as liver, spleen, pancreas, kidneys, stomach/intestine, bladder, and genitals); AIS2+ neck/spine/back injury (spinal cord injuries); and AIS2+ extremities injuries (fractures at extremities except those at fingers or toes).
A number of important covariates were also included in this study. Vehicle model years were categorized according to typical airbag system characteristics: no air bag (1990–1993); first-generation air bag (all pre-model year 1998 vehicles and 1998 vehicles without redesigned air bags); second-generation air bag systems (all 1998 vehicles with redesigned systems and all 1999–2001 vehicles); and advanced airbags (2002 and newer vehicles). Information on the redesign date was determined from data obtained from the US Department of Transportation National Highway Traffic Safety Administration’s Special Crash Investigation Division; they obtained the information from the manufacturers as part of their evaluation of the air bag regulation. Restraint status of children and seating position at the time of the crash were determined by the validated telephone survey. Ongoing comparison of driver-reported child occupant restraint use and seating position with evidence from the crash investigations has shown a high degree of agreement (κ statistic = 0.99 for seat row, 0.74 for restraint use).
Logistic regression modeling was used to compute a child passenger’s odds ratio of sustaining an AIS2+ injury. The primary independent variable was BMI, which was modeled according to the previously noted CDC-defined weight categories for children. Injuries were analyzed to any body part and, consistent with study hypotheses, separately for head and extremity injuries. Adjustment was made for the following confounders: age and gender of child, direction of initial impact, vehicle type, child restraint status, seating position of child, age of driver, exposure of child to passenger airbag, and crash severity (any intrusion into the passenger compartment and tow away status of the vehicle). The final model was selected using a backward selection method. Potential confounders with p>0.1 were excluded. Results of logistic regression modeling were expressed as adjusted odds ratios with corresponding 95% CI.
Previous literature on the association between driver BMI and risk of motor vehicle crash-related injury and death suggests a non-linear relationship between injury risk and BMI.89 Therefore, we explored the curve of the relationship by dividing BMI into eight equal width categories and calculated the logit of injury risk of each category. A quadratic term of BMI was also added to models to test its significance. Both methods suggested a goodness-of-fit for logistic regression model with a linear term of BMI. A sensitivity analysis was also conducted to evaluate the effect of bias in parent-reported height and weight, by including only those children in the model whose height and weight were reported to be “recently measured”. The final model and the effects of BMI and all other confounders were essentially unchanged.
As sampling was based on the likelihood of an injury, subjects least likely to be injured were under-represented in the study sample in a manner potentially associated with the predictors of interest.18 To account for this potential bias, analytical methods were used to account for sampling strategy. To compute p values and 95% CIs to account for the stratification of subjects by medical treatment, clustering of subjects by vehicle, and the disproportional probability of selection, Taylor Series linearization estimates of the logistic regression parameter variance were calculated using SAS-callable SUDAAN: Software for the Statistical Analysis of Correlated Data, Version 9.0 (Research Triangle Institute, Research Triangle Park, NC, USA, 2004). Sample values (unweighted) and population estimates (weighted on the basis of the intricate sample method) are both presented.
During the study period, a total of 3232 children age 9–15 years were involved in crashes from 2873 vehicles. This reflected a population estimate of 54 616 children in 49 037 vehicles (table 1). Approximately 15% (n = 502) of the children sustained an AIS 2+ injury. Approximately 62% of the children involved in crashes had a BMI in the normal range; 18%, 16%, and 5% were overweight, obese, and underweight, respectively. The sample was nearly equally distributed by gender, and ∼60% of the sample was between 13 and 15 years of age.
The largest group of vehicles involved in crashes was passenger cars (39%), followed by minivans (26%) or sport utility vehicles (26%). Over half of the children were seated in the front row of the vehicle. Ninety-six percent of the children were restrained at the time of the crash, with most using a joint lap/shoulder belt system. There was no intrusion in over 90% of the crashes. The initial direction of impact was mostly to the front (43%) or rear (32%) of the vehicle. In at least 93% of the crashes, there was also no deployment of the driver or passenger airbag and no vehicle rollover.
Risk of injury in relation to BMI
Overall 1.78% of children sustained an AIS2+ injury to any body part (table 2). This value ranged from 1.20% for the underweight children to 2.06% for the overweight children. A larger percentage of injuries occurred to overweight and obese children than to underweight and normal sized children for injuries to any body region, the abdomen, and the extremities. This pattern was most evident for injuries to the upper and lower extremity.
There was no statistically significant increase in overall AIS2+ injury risk associated with increased BMI (table 3). After adjustment for potential confounders, however, analyses by body region revealed a significant risk of extremity injury associated with increased BMI. Compared with normal weight children, the odds of sustaining an AIS 2+ injury for overweight and obese children was 2.64 (95% CI 1.64 to 4.77) and 2.54 (95% CI 1.15 to 5.59), respectively. Review of the lower and upper extremity fractures showed that overweight and obese children sustained injury to the foot, ankle, elbow, and forearm, whereas leaner children sustained injury to the femur, pelvis, and clavicle (figs 1 and 2).
Table 4 gives detailed results for the final model on the upper and lower extremity injuries. The multivariate model identified crash severity (ie, intrusion and tow away status) and driver airbag deployment as the strongest risk factors for injury (p<0.001). Children were more likely to sustain an extremity fracture in a crash if they were riding in an older car, sitting in the front row, unrestrained, and were involved in a more severe crash as indicated by the presence of a rollover or driver airbag deployment. BMI remained an additional risk factor for extremity injury in motor vehicle crashes.
This research demonstrated no significant increase in overall AIS2+ injury risk as BMI increased. However, we found that the risk of sustaining an AIS 2+ injury to the extremities was over two and a half times as great for overweight and obese children as for normal weight children. This relationship was still present even after statistical modeling accounted for the presence of other important risk factors for child injury, including severity, seating position, etc. It is worth noting that this analysis also confirmed the overwhelming effects of previously identified risk factors for pediatric motor vehicle crash-related injuries: riding in an older car, sitting in the front row, being unrestrained, and being in a severe crash.231920
Our finding of no overall increased risk of injury, but increased risk of injury to the extremities was consistent with similar research in adults.6 Prior studies of adult motor vehicle crash injuries have also shown that higher BMI is associated with injuries to the extremities.6710 Although conclusive evidence has yet to identify the cause of this increased risk, the impact of body mass on the transfer of energy to the extremities during the crash was hypothesized.
For children, we suspect that the differences in body part injured by BMI may be due to a combination of physiology and biomechanical factors. Prior research has shown that bone strength is inversely related to fractures (primary cause of extremity injuries) and that obesity may influence the risk of fractures.21–23 Therefore, it seems likely that overweight and obese children may be more susceptible to fractures at equivalent levels of crash severity than children of normal BMI.
Motor vehicle crashes are the leading cause of death and a leading cause of non-fatal injuries for youth.
Research has found an association between childhood obesity and certain pediatric injuries.
Although there are some adult data on the association between obesity and motor vehicle crash-related injury, limited research has explored this issue for children.
Overweight and obese children are not at increased overall risk of injury; however, they are at increased risk of injury to the lower and upper extremities.
The increased risk of injury to the extremities for children in crashes may be due to a combination of physiology, biomechanical forces, and vehicle design.
Our analysis of fracture location showed that overweight and obese children sustained injury to more distal aspects of the extremity: these children sustained injuries to the foot, ankle, elbow, and forearm, whereas leaner children sustained injuries to the femur, pelvis, and clavicle. We hypothesize that, for a given crash acceleration, the increased mass of the overweight and obese children may result in increased force on impact with interior vehicle structures—often with distal aspects of the body—and an increased likelihood of fracture. These distal injuries may also simply be evidence that larger children are in fact closer to interior vehicle structures and thus more likely to hit them with increased velocity. An absence of injuries to the boney structures upon which the seat belt lies—the pelvis and the clavicle—for overweight or obese children may indicate a relative sparing of these structures, due in part to a layer of energy-absorbing body fat between the seat belt and the boney skeleton. Future biomechanical exploration of these hypotheses may highlight potential modifications to vehicle design that could address the ever-increasing population of overweight and obese motor vehicle occupants.
Some limitations must be considered in the interpretation of our results. Our study sample included older children, in insured vehicles of model year 1990 and newer, in 15 states and the District of Columbia. Thus, to the extent that the effect of BMI on injuries of uninsured drivers during crashes differ substantially between vehicles of model year before 1990, our analysis may be not be generalizable to children, 9–15 years of age, riding in those vehicles. Many variables were from passenger accounts of the crash and therefore are potentially subject to recall bias. Also, the reliance on parent-reported weight, height, and injury may be subject to misclassification biases. Research exploring the validity of parent-reported child height and weight have revealed biases; however, parent-reported measures are reasonably valid for classifying children as obese or non-obese in large epidemiological studies.24 The ability of parents to accurately distinguish AIS 2+ injuries from those that are less severe has previously been validated for all body regions of injury.14 Although use of parent-reported measures remains a limitation of our study, for reasons stated above, we believe that they had a limited effect on the results.
IMPLICATIONS FOR PREVENTION
Our findings document another potential risk associated with overweight and obesity in children. These results may have important implications for future studies on the interaction of physiological, biomechanical, and vehicle factors during motor vehicle crashes. The association of obesity with injury should be considered as part of ongoing efforts to reduce the burden on young people of morbidity and mortality related to motor vehicle crashes. While future studies continue to explore this association, to reduce the risk of injury during transport, all children should use a restraint appropriate for their age and size, with children under the age of 13 seated in the rear of the vehicle.
We acknowledge the following individuals for their insightful comments on this research and article: Dr Flaura Winston and Dr Mark Zonfrillo of the Children’s Hospital of Pennsylvania and Professor Susan P Baker of the Center for Injury Research and Policy at the Johns Hopkins Bloomberg School of Public Health.
Funding: We acknowledge the commitment and financial support of the State Farm Mutual Automobile Insurance Company for the creation and ongoing maintenance of the Partners for Child Passenger Safety (PCPS) program, the source of data for this study. We also thank the State Farm policyholders who consented to participate in PCPS. This publication was also supported in part by Grant Number 5R49CE3000198 from the Centers for Disease Control and Prevention (CDC). Its contents are solely the responsibility of the authors and do not necessarily represent the official views of the CDC.
Competing interests: None.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.