Copyright 2007 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.2007
To calculate national estimates of motor vehicle crash (MVC)–related hospitalization and associated use of health care resources among patients 20 years and younger and to explore the effects of certain sociodemographic and health care system–related factors and injury severity on use of hospital resources and lengths of stay (LOSs) in the United States.
Data from the 2003 Healthcare Cost and Utilization Project Kids' Inpatient Database were used.
Pediatric inpatient care in 3438 hospitals in 36 US states.
Patients 20 years and younger hospitalized with MVC-related injuries.
Main Outcome Measures
National estimates of MVC-associated hospitalizations, rates, resource use, Injury Severity Scores, and demographics were calculated. Potentially significant covariate associations were studied using hospital charges and LOS.
Sixty-two thousand eight hundred eighty MVC-related hospitalizations occurred among patients 20 years and younger in the United States in 2003, resulting in more than $2.0 billion (SD = $1.2 million) in inpatient charges and 304 196 days (SD = 55 113 days) of hospitalization. Mean (SD) hospital charges and LOS were $33 440 ($55 113) and 4.8 (7.7) days, respectively. The mean (SD) Injury Severity Score was 10.3 (11.4). Adolescents aged 18 through 20 years had the highest hospitalization rates (197 cases per 100 000 children). Older age, being male, urban hospital location, mortality during hospitalization, higher injury severity, and longer LOS were significantly associated with higher total charges. Longer LOS was significantly associated with older age, urban hospital location, higher injury severity, and mortality.
Motor vehicle crash–related injuries among children burden health care resources, with nationwide charges exceeding $2 billion annually.
In 2000, total costs for motor vehicle crashes (MVCs) were $230.6 billion, with $32.6 billion in medical charges alone, making MVCs the most expensive cause of injury in the United States.1,2 Motor vehicle crashes are the leading cause of death due to unintentional injury from ages 1 to 64 years and are the second leading cause of unintentional injury-related deaths among infants.3 In the pediatric and adolescent populations, many factors, including improper use of child-restraint devices for young children and unsafe driving among older teenagers, contribute to these deaths.4,5
Motor vehicle crashes are also a major cause of pediatric morbidity.6 When compared with other injuries, children involved in MVCs generally incur more extensive and severe injuries.7,8 As a result, children's lengths of stay (LOSs) in hospitals are longer and the accrued medical charges, higher.9,10 The consequences of hospitalization (eg, billed charges, missed work/school days, posttraumatic stress) can have profound effects on the financial, physical, and social well-being of children and their families.11,12
Several studies have examined MVC-associated injuries and their consequences among the pediatric population in the United States,7,13,14 though few have attempted to make national estimates regarding resource use of such events.10,15 The objective of this study was to calculate national estimates of MVC-related hospitalization and associated use of health care resources for children and adolescents 20 years and younger involved in MVCs in the United States using the Kids' Inpatient Database (KID). The influence of demographic characteristics (eg, age, sex, and payer type), health system factors (eg, hospital size, location, and type), and injury severity on total hospital charges and LOS were also analyzed.
In this study, we performed a retrospective analysis of hospitalizations in the United States from January 1, 2003, to December 31, 2003, using data from the KID.16 The KID is a component of the Healthcare Cost and Utilization Project maintained by the Agency for Healthcare Research and Quality (AHRQ). The KID is the only hospital administrative data set designed to assess newborn, child, and adolescent use of hospital services. Since its implementation in 1997, the KID has undergone only minor changes in the recorded variables, types of participating facilities, and number of hospitals sampled. In 2003, the KID collected hospital discharge information on pediatric treatments and resource use from 3438 hospitals in 36 states. The 2003 KID includes nonfederal hospitals, short-term hospitals, academic medical centers, and specialty hospitals, such as obstetrics-gynecology, ear-nose-throat, orthopedic, and pediatric hospitals. Federal hospitals, short-term rehabilitation hospitals, long-term hospitals, psychiatric hospitals, and alcoholism/chemical dependency treatment centers were excluded from 2003 KID data. Hospitals were assigned to 6 strata for selection based on ownership/control, bed size, teaching status, rural/urban location, US region, and hospital type (pediatric vs other).17 All data are discharge level; therefore, individuals who were hospitalized multiple times will have multiple records in the KID. Both the government and private sector use total hospital charges and LOS as standards to monitor health care costs and quality of care. The 2003 KID, the most recent year of the KID with completed data, was used in this study.
States report total charges to the AHRQ. Professional KID coders edit the total charges reported to the AHRQ by each state in order to round charges to the nearest dollar, set zero charges to “missing,” and set excessively high (> $1 million) or excessively low (< $25) charges to “inconsistent.” Total charges include neither professional fees nor noncovered charges.
In the KID, hospital bed-size categories (small, medium, and large) are based on the number of short-term patient beds, geographical region of the hospital, and teaching status of the hospital. Hospital type follows the children's hospital categorization method used by the National Association of Children's Hospitals and Related Institutions.18 General children's hospitals are freestanding, general acute care children's hospitals. Children's units in general hospitals are nonfreestanding, general acute care hospitals, typically within larger hospitals. Specialty children's hospitals include orthopedic, rehabilitative, psychiatric, and other specialty-focused inpatient facilities. For the purpose of this analysis, specialty children's hospitals were combined with general children's hospitals. Nonchildren's hospitals admit patients of all ages and do not have a specific pediatric focus. According to the American Hospital Association's Annual Survey of Hospitals, metropolitan statistical area hospitals and non–metropolitan statistical area hospitals are characterized as urban and rural hospitals, respectively.19
A series of International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) diagnosis codes were used to identify MVC-related hospitalizations among children, consistent with the recommendation of the Centers for Disease Control and Prevention.20 Cases with an external cause-of-injury code (E-code) related to “any motor vehicle accident occurring on a public highway” (E810-E819) were selected.21 Since the KID lacks indicators for injury severity, ICDMAP software22 was used to assign an Injury Severity Score (ISS) to each MVC-related injury requiring hospitalization using ICD-9-CM diagnosis codes.
The KID provides discharge-level and hospital-level sampling weights to account for the complex survey design and sampling procedures. The KID was specifically designed to provide national estimates of hospitalizations. Data analyses were conducted using SAS23 and STATA SE24 statistical software to account for the weighting structures of the KID. SAS was used to calculate frequencies (unweighted sample and national estimates), 95% confidence intervals (CIs) for national estimates, percentages, means, medians, sums, and in analyses of unadjusted means. STATA SE was used to perform linear regression analyses and calculate associated 95% CIs. Some patients (<0.2%) had multiple MVC-related E-codes recorded for their injury. In these cases, the first recorded MVC-related E-code was used as the primary cause of injury.
The actual sample size is a statistically unweighted number and is noted when reported. All other frequencies, percentages, rates, means, medians, and sums in this analysis are national estimates calculated using the statistical weights. National estimates of MVC-related hospitalizations with 95% CIs for children were calculated by age, sex, race, primary expected payer, mortality status, hospital bed size, and hospital type. Median and mean ISSs for each demographic and health care–related subgroup were also calculated. Race was excluded from additional subanalyses because of the relatively high percentage of missing data (25.2%). Rates of MVC-related hospitalizations per 100 000 population by age group and sex were generated using publicly available US 2000 census data.25
Because of the skewed distributions, the 2 outcome variables (LOS and total charges) were log transformed to achieve normality; 0.1 was added to LOS before the log transformation to avoid numerical problems where LOS was equal to 0. Separate multivariate regression models were constructed for LOS and total hospital charges. To compare subgroups within each variable, a reference category was specified and assigned a value of 1 for each variable. All univariate regression models were analyzed and predictor variables (age, sex, expected primary payer, mortality status, hospital bed size, hospital type, hospital location, cause of injury, and injury severity) significant at the α level of .05 were included in a multivariable regression model. Clinically important 2-way interactions between predictor variables were then tested at the conservative α level of .01.
The institutional review board of the Columbus Children's Research Institute approved this study.
Of the 2 984 129 hospitalizations between January 1, 2003, and December 31, 2003, recorded in the KID, 38 532 actual cases involving children 20 years and younger were associated with an MVC E-code. Nationally, an estimated 62 880 (95% CI, 57 765-67 997) hospitalizations resulted from MVCs. The majority of children hospitalized because of MVCs were 15 years and older (66.5%). The most common expected primary payer was private insurance (59.8%), followed by Medicaid (22.8%). Overall, 1483 pediatric patients (2.4%) involved in MVCs died in the hospital (Table 1). Collision with a motor vehicle (E812) was the most common cause of hospitalization due to MVCs (34.3%), followed by loss of control (E816; 21.3%) (Table 2).
The hospitalization rate for all children involved in MVCs was 73 per 100 000. Adolescents 18 through 20 years of age (197 per 100 000) had the highest MVC-related hospitalization rate, while adolescents 15 through 17 years of age had the second highest hospitalization rate (139 per 100 000). Children 14 years and younger had a considerably lower MVC-related hospitalization rate (35 per 100 000). When compared with children 14 years and younger, adolescents 15 through 17 years had an incidence rate ratio of 3.95 (95% CI, 3.87-4.03) for MVC-related hospitalizations, while adolescents 18 through 20 years of age had an incidence rate ratio of 5.59 (95% CI, 5.49-5.69). Males had a higher hospitalization rate (86 per 100 000) than females (59 per 100 000) (Figure 1). Males had an incidence rate ratio of 1.46 (95% CI, 1.44-1.48) for MVC-related hospitalizations when compared with females.
Motor vehicle crash–associated hospitalizations in 2003 per 100 000 population by age and sex. The data source was the 2000 US census data25 and the Kids' Inpatient Database, 2003.16
Pediatric inpatients involved in MVCs accrued an estimated $2 042 484 046 (SD = $121 328 518) in hospital charges in 2003. A mean (SD) $33 440 ($55 113) was charged for each hospitalization resulting from an MVC, with a median of $17 571 and an interquartile range of $9217 to $34 534. A total of 304 196 (SD = 14 583) inpatient days were used for treatment of MVC-associated injuries in 2003. The mean (SD) LOS for pediatric MVC-associated hospitalizations was 4.8 (7.7) days. Length of stay ranged from less than 1 day to 160 days; the median was 3 days, and the interquartile range was 1 to 5 days (Table 3).
Pediatric inpatients involved in MVCs had a mean (SD) ISS of 10.32 (11.41) and a median (interquartile range) ISS of 8 (4-14) (Table 3). Adolescents 15 to 17 years of age had the highest ISS (mean, 11.1; median, 9), and adolescents 18 to 20 years of age had the second highest ISS (mean, 11.0; median, 9). Infants younger than 1 year had a mean ISS of 10.0 (median, 8), which was considerably higher than children 1 to 4 years of age (mean, 8.2; median, 5). Males involved in MVCs had an ISS of 10.6 (median, 9) while females had a lower ISS of 10.1 (median, 6). Children treated at large hospitals had a higher ISS (mean, 10.6; median, 8) than small and medium-sized hospitals. Children's units in general hospitals treated children with a higher ISS (mean, 10.9; median, 9) than other types of hospitals. The mean ISS of children treated in urban and rural hospitals was 10.5 (median, 8) and 8.4 (median, 5), respectively (Figure 2).
Mean and median Injury Severity Scores (ISSs) for motor vehicle crash–related hospitalizations according to characteristic, 2003. Injury Severity Scores were calculated from the Kids' Inpatient Database, 200316 using ICDMAP software.22
A multivariate analysis of total hospital charges was performed while controlling for potentially confounding variables (mortality status, age, sex, expected primary payer, hospital type, hospital size, hospital location, LOS, injury severity, and external cause of injury) (Table 4). Compared with all other age groups, adolescents 18 through 20 years of age had higher total charges. Females had 0.89 (95% CI, 0.85-0.93) lower hospital charges than males. Patients covered by Medicaid had higher total charges than patients who were privately insured or paid themselves. Total charges for urban hospitals were 1.72 (95% CI, 1.60-1.85) higher than for rural hospitals. Patients who died while hospitalized incurred more than 1.87 (95% CI, 1.71-2.04) greater charges than patients who survived their injuries. Longer LOSs were associated with 1.084 (95% CI, 1.076-1.092) higher hospital charges. Hospital charges increased by 1.014 (95% CI, 1.012-1.017) per ISS unit. Regarding external causes of injury, MVCs resulting from collisions resulting in loss of control (E816) had 1.08 (95% CI, 1.05-1.12) higher hospitalization charges when compared with MVCs involving another motor vehicle (E812) (Table 4). After analyzing interactions between variables of interest, only the interaction between sex and age was found to be significant. Older males had significantly higher hospital charges than younger males and than females of all age groups. The differences in hospital charges between hospital types were not statistically significant in the multivariate analyses.
In the multivariate analyses of LOS (Table 5), while controlling for the same potentially confounding variables as in the hospital charges model and comparing all age groups, adolescents 18 through 20 years of age had longer LOSs. Large hospitals were associated with 1.12 (95% CI, 1.00-1.26) longer LOSs than small hospitals. Urban hospitals had 1.16 (95% CI, 1.08-1.25) longer LOSs than rural hospitals. The LOSs of self-paying patients were 0.75 (95% CI, 0.71-0.80) shorter than LOSs of children covered by Medicaid. Patients who died during hospitalization had a 0.16 (95% CI, 0.14-0.19) shorter LOS than patients who survived. The LOS increased by 1.036 (95% CI, 1.032-1.040) per ISS unit. Sex was not a significant predictor of LOS. No significant 2-way interactions between variables existed in the LOS model (Table 5).
In the United States in 2003, an estimated 62 880 pediatric MVC-related hospitalizations occurred among children 20 years and younger, resulting in more than $2 billion in inpatient charges with 304 196 days of hospitalization. Mean (SD) hospital charges were $33 440 ($55 113) and mean (SD) LOS was 4.8 (7.7) days. The mean (SD) ISS for MVC-related hospitalizations was 10.32 (11.41). Increases in ISS were significantly associated with increases in both total hospital charges and LOS. The LOS was also a significant predictor of total hospital charges. In a study exploring trauma costs among individuals among all age groups, Lanzarotti and colleagues14 found mean (SD) hospital charges and mean (SD) LOS to be $31 000 ($52 000) and 4.8 (7.3) days, respectively. Haider and colleagues26 reported a mean ISS of 10.01 in pediatric injuries involving MVCs. Children hospitalized because of MVC-related injuries stayed approximately 1 day longer on average than children admitted for other pediatric illnesses. In addition, the average total hospital charges for MVC-related injury were more than 4 times greater than the average charges for other pediatric inpatient stays.27
Though the rate of MVC-related hospitalization continually increased with increasing age, there was a jump in the hospitalization rate among adolescents 15 years and older. Other researchers examining MVCs among children have observed that rates of MVC-related hospitalizations increased with the child's age.28- 31 Adolescents 18 through 20 years of age had the highest hospitalization rate (197 in 100 000), but adolescents 15 through 17 years of age had a higher MVC rate and a higher ISS.28,32 Government and private groups are working hard to lessen the burden of MVC-related injury in these age groups. The majority of US states have adopted graduated driver's licensing programs, which introduce driving privileges gradually to beginning drivers; these programs have seen promising reductions in MVC rates among these populations.33
Adolescents 15 years and older had higher charges per hospitalization than younger children, even after adjusting for LOS. Higher hospitalization charges have consistently been linked to increased injury severity.1,34,35 Higher injury rates and severity levels among this population are related to several factors, such as driving inexperience, risk-taking behavior, and greater risk exposure (wearing safety belts less, driving while intoxicated).36
Patients who died during hospitalization incurred significantly higher charges but had shorter LOSs when compared with patients who survived their injuries. Expensive intensive-care measures were most likely used in these situations, resulting in higher charges for this group of children. Other studies have also reported that MVC-related injuries resulting in death are associated with higher charges.9,15
In 2003, the Healthcare Cost and Utilization Project issued a report that included the most expensive diagnoses found in the KID for 2000. Currently, no such report is available for the 2003 KID. If our findings were applied to the report generated using the 2000 KID, MVC-related injury would be the most expensive cause of hospitalization due to injury or disease for children in the United States, costing more than cardiac and circulatory birth defects ($1.90 billion), prematurity, low birth weight and fetal growth retardation ($1.36 million), and pneumonia ($1.32 billion). Only charges related to the delivery and care of newborn infants would be greater ($17.2 billion).27
Certain limitations exist in this study. First, costs, reported in the KID as billed charges, may differ from the actual amount paid because of discounts, deductibles, copayments, and coinsurances.37 Second, as is the case with all large data sets, missing data and coding errors can potentially lead to bias in the data. Variables with large amounts of missing data (eg, race of patient, with 25.2% of values missing) were excluded from our analyses to minimize potential bias. Lastly, the true costs of these injuries are likely to be underestimated in this study, because certain health care data involving MVCs (eg, emergency department visits and deaths at the scene of the crash) are not included in the KID and subsequently are not accounted for in our analyses. Noncovered charges, professional fees, time lost from school and work by parents and other caregivers, and other societal costs are also not included in these estimates.
Motor vehicle crash–related injury in the United States results in high mortality, morbidity, and economic costs. In our study, inpatient charges totaled more than $2 billion for US children in 2003. This underestimates the true costs of these injuries because noncovered charges, professional fees, time lost from school and work by parents and other caregivers, and other societal costs are not included in this estimate. Healthy People 2010 has identified 7 initiatives to decrease MVC-related death and injury, 3 of which affect children and adolescents in particular: increase use of safety belts, increase use of child restraints, and increase the number of states that have graduated driver licensing laws.38 Achieving these objectives is critical to improve motor vehicle safety for all occupants, especially children. Special attention should be given to adolescents from 15 through 20 years of age, because this population is at highest risk.
Correspondence: Lara B. McKenzie, PhD, MA, Center for Injury Research and Policy, Columbus Children's Hospital, 700 Children's Dr, Columbus, OH 43205 (email@example.com).
Accepted for Publication: April 17, 2007.
Author Contributions: The authors have had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Gardner, Smith, Fernandez, and McKenzie. Acquisition of data: Gardner and Smith. Analysis and interpretation of data: Gardner, Smith, Chany, Fernandez, and McKenzie. Drafting of the manuscript: Gardner. Critical revision of the manuscript for important intellectual content: Gardner, Smith, Chany, Fernandez, and McKenzie. Statistical analysis: Gardner, Chany, and Fernandez. Obtained funding: Gardner, Smith, and McKenzie. Administrative, technical, and material support: Gardner, Smith, and McKenzie. Study supervision: Smith and McKenzie.
Financial Disclosure: None reported.
Funding/Support: The Samuel J. Roessler Memorial Scholarship Fund of The Ohio State University College of Medicine and Public Health provided Mr Gardner a 2-month stipend in support of his work on this study.
Additional Contributions: Brenda Shields, MS, assisted in the ISS analysis.
Gardner R, Smith GA, Chany AL, Fernandez SA, McKenzie LB. Factors Associated With Hospital Length of Stay and Hospital Charges of Motor Vehicle Crash–Related Hospitalizations Among Children in the United States. Arch Pediatr Adolesc Med. 2007;161(9):889-895. doi:10.1001/archpedi.161.9.889