ACADEMIC EMERGENCY MEDICINE 2010; 17:809–812 © 2010 by the Society for Academic Emergency Medicine
Objectives: This study sought to determine if insurance or race status affect trauma outcomes in pediatric trauma patients.
Methods: Using the National Trauma Data Bank (NTDB; v6.2), the following variables were extracted: age (0–17 years), payment type (insured, Medicaid/Medicare, or self-pay), race (white, Black/African American, or Hispanic), Injury Severity Score (ISS > 8), type of trauma (blunt or penetrating), and discharge status (alive or dead). Data were analyzed using logistic regression.
Results: Of the 70,781 patient visits analyzed, 67% were insured, 23% were Medicaid/Medicare, and 10% were self-pay. Self-pay patients had higher mortality (11%, compared to Medicaid/Medicare at 5% and insured at 4%; p < 0.001). African Americans and Hispanics also had higher mortality (7 and 6%) compared to whites (4%; p < 0.001). Self-pay patients more likely suffered penetrating trauma than insured patients (12% vs. 4%; p < 0.001), and mortality for penetrating trauma self-pay patients was 29%, compared to only 11% for penetrating trauma insured patients (p < 0.001). The mortality rate varied from a low of 3% for insured whites, to 18% for self-pay African Americans. Logistic regression (including race, insurance status, injury type, and ISS) revealed that African Americans and Hispanics both had an increased risk of death compared to whites (African American odds ratio [OR] = 1.37, Hispanic OR = 1.20). Medicaid/Medicare patients had a slightly increased risk of death with OR = 1.14, but self-pay patients were almost three times more likely to die (adjusted OR = 2.92).
Conclusions: After controlling for ISS and type of injury, mortality disparity exists for uninsured, African American, and Hispanic pediatric trauma patients. Although the reasons for this are unclear, efforts to decrease these disparities are needed.
Unfortunately, disparities based on race, ethnicity, and socioeconomic status continue to be challenges.1–4 While many social determinants such as disparity in median income, high school dropout rates, and residential segregation have improved in the past 50 years, health disparities have essentially remained unchanged.1 Multiple government reports have identified the glaring problem of health disparities.1–4
The 1985 Task Force Report on Black and Minority Health estimated that 60,000 excess deaths per year, in particular in blacks and African Americans, occurred because of health disparities. Satcher et al.5 estimated an excess 83,000 deaths per year from health disparity between blacks/African Americans and whites using 2002 data. The Institute of Medicine report Unequal Treatment: Confronting Racial and Ethnic Disparities in Health Care4 was a landmark study that became the primer for annual reports on disparities by the Agency for Healthcare Research and Quality. Most recently, Healthy People 2010 has made reducing racial disparities one of its two central national health priorities,6 and a recent IOM report7 establishing priorities for comparative-effectiveness research recommended a portfolio of 100 studies, 29 of which address racial and ethnic disparities. Clearly there is a need and desire to better understand the source of health care disparities.
While many studies and reports have documented the presence and extent of health disparities, there still remains a lack of understanding of the specific causes behind them. Historically, emphasis on health disparities and the majority of studies have centered on race. Several studies have found both race and socioeconomic status to be predictors of increased mortality, although one factor is usually used to account for the other. A study by Haider et al.8 found that in adult trauma patients, insurance status and race were both independent predictors of increased mortality, with lack of insurance being the stronger predictor. These findings are consistent with previous studies on patients with myocardial infarction that demonstrated that social class is a stronger predictor of morbidity and mortality than race.9
Injuries are the leading cause of death in our children.10 A recent study suggested that uninsured status and not race was a predictor for pediatric trauma mortality.11 To investigate this question further, and determine if race and insurance were both predictive of mortality in pediatric trauma, we analyzed pediatric data from the National Trauma Data Bank (NTDB). Our null hypothesis was that mortality from pediatric trauma was not related to race or insurance.
This study was a retrospective analysis of data from the NTDB (version 6.2). This study was approved by the St. John Hospital and Medical Center institutional review board via expedited review.
Study Setting and Population
The NTDB is maintained by the American College of Surgeons and represents the largest repository of trauma data on inpatients in the United States. All pediatric patients (ages 0–17 years) with an Injury Severity Score (ISS) of >8 were included in this study. The study population was limited to pediatric patients with ISS > 8 because these patients had suffered at least moderate injuries and would be at risk of dying. Mortality was the primary outcome measure. Burn patients were excluded as we intended to focus on blunt and penetrating trauma.
Patients were grouped by self-reported race (African American, white, Hispanic). Other races (i.e., Asian Americans, American Indians, Pacific Islanders) were excluded because of small sample sizes. “Uninsured” was defined as charity care, no charge, self-pay, and no insurance. Medicaid and Medicare were combined into one category, “Medicaid/Medicare.” Patients in all other categories were defined as “insured.” Data extracted included age, payment type, race, ISS, trauma type (blunt or penetrating), and discharge status.
Data were exported into Microsoft Access (Microsoft Corp., Redmond, WA) and analyzed using SPSS v. 17 (SPSS Inc., Chicago, IL). Student’s t-test or analysis of variance were used to compare continuous variables, and chi-square analysis was used for categorical variable comparison. Multivariate analysis was completed using logistic regression with a forward stepwise algorithm. For all analyses, a p-value of 0.05 or less was considered to indicate statistical significance.
There were 70,781 patients included in the analysis. There were 51,508 white (73%), 11,351 African American (16%), and 7,922 Hispanic (11%) patients. The majority of patients were insured (n = 47,211; 66%) or had Medicaid/Medicare (16,251; 23%), and only 10% (n = 7,319) were uninsured. The majority of patients suffered blunt trauma (66,611; 94%).
The overall mortality rate was 4.8%, with a range of 3% for insured whites to 18% for uninsured African Americans. The mortality was 3.7% for whites, 7.2% for African Americans, and 6.1% for Hispanics and 3.5% for insured, 4.7% for Medicaid/Medicare, and 11.1% for uninsured. The differences in mortality based on race were statistically significant (p < 0.0001), as were the differences in mortality based on insurance status (p < 0.0001).
Uninsured patients were more likely to suffer from penetrating injuries (12%) than insured patients (4%). Mortality for penetrating trauma was also higher in uninsured patients, at 29% compared to 11% in insured patients.
Logistic regression analysis (controlling for race, insurance, status, injury type, and ISS; Table 1) revealed that African American and Hispanic children both had increased mortality compared to the reference group of white children. Medicaid/Medicare children also had a slightly increased risk of death compared to insured (reference group). Uninsured patients had the highest risk of death compared to insured (p < 0.00001 in all comparisons).
Health care outcome disparities have been previously described in a variety of races and socioeconomic strata. While race and socioeconomic status together have often been used interchangeably to explain disparity, we found lack of insurance status and nonwhite race to be independent predictors of increased mortality, with lack of insurance being the stronger marker of increased mortality. Uninsured (or self-pay) pediatric trauma patients were nearly three times (adjusted odds ratio [OR] = 2.92) as likely to die compared to their insured counterparts, even after adjusting for age, ISS, type of trauma, and race. This finding is consistent with a similar pediatric trauma study that also found a high risk of trauma-related death in uninsured victims.11 In addition, our study also showed that African American and Hispanic pediatric trauma patients fares worse than their white counterparts (Table 1).
Even though our study clearly identifies a strong association between insurance status and mortality in pediatric trauma victims, it is unclear what exactly is the basis for this disparity. It is possible that differential trauma care provision exists between insured and uninsured children, with uninsured receiving a lower quality or timeliness of trauma care. Other factors related to lower socioeconomic status of uninsured children may also contribute to this disparity. For example, uninsured children may have less parental supervision leading to delays in seeking care after an injury. Additional potential socioeconomic factors leading to disparate care include greater risk-taking behaviors12 and potential delays in seeking medical attention in patients with lower socioeconomic status. It remains unclear whether the health insurance or the socioeconomic benefits that come along with the financial ability to obtain health insurance may contribute to this disparity. As in adults, this question remains a complex problem.
Unfortunately, race-based health care disparities continue to exist.5 Racial health care disparities have been described in numerous pediatric settings including appendicitis,13–15 asthma care,16–18 and vaccinations. A previous study using the NTDB cited only lack of insurance as a factor in outcome of pediatric trauma patients.11 Our study, which identified an adjusted OR of increased mortality of 1.37 for African American and 1.20 for Hispanic pediatric trauma patients, is different from this previous study as our study limits the analysis to children with ISS > 8. We focused on the analysis of this more severely injured patient population because we wanted to analyze children at risk of dying from trauma. This approach is similar to a published report by Haider et al.8 who also limited analysis to those with at least moderate injuries (ISS > 8) when evaluating the effect of race and insurance on traumatic mortality in adults. The differences between our study and the previous pediatric study may be related to the fact that any race-based disparities associated with mortality could potentially be diluted by large numbers of children with minor injuries in the previous study. Concentrating attention on patients with moderate to severe injuries (ISS > 8) allows the focus of the analysis to be on patients who are most at risk of death after a traumatic injury.
While we are unsure of the causes of increased mortality based on race, possible explanations include lower quality of care in the hospitals frequented by minority patients, patient preference regarding interventions, lack of trust in health care providers, and racial bias and stereotyping by providers. Bach et al.19 reported that a minority of physicians care for the majority of African American patients in the United States and that these physicians were more likely to report difficulty procuring adequate resources (subspecialty referrals, diagnostic imaging, and hospital admission for their patients). Another study found that the increased mortality of African Americans requiring cardiopulmonary resuscitation was attenuated when the quality of hospital (poor quality hospital defined as one with increased mortality of out-of-hospital cardiac arrest) was taken into account, suggesting that some difference in race mortality may be related to the quality of hospitals available to minorities.20 While some African American patients may lack trust in the medical establishment because of the Tuskegee experiment,21 obstetrics and gynecology experimentation of novel surgical procedures on non-consenting African American women,22 and other well-documented abuses of minorities in health care, such a trust deficit does not explain the race-based disparity among Hispanics who also experienced increased mortality. There is a lack of understanding of differential care among health care providers, as is demonstrated in a study that shows only a third of cardiologists who responded to a survey believed that “clinically similar patients receive different cardiovascular care based on what their race and ethnic background is,” despite evidence stating otherwise.23 More research is needed to better understand the basis of this issue and to tailor solutions to race-based disparities.
In summary, neither race nor insurance status alone can explain existing disparities in the outcome of pediatric trauma. Our research demonstrates the presence of health outcome disparities in pediatric trauma patients, with both insurance status and race being independent predictors of increased mortality. The underlying causes of these disparities remain poorly understood and will require more research centered on both insurance status and race. Prospective studies (including additional data such as comorbid conditions, past medical history, potential delays in care, potential differential treatment, and outcomes) may help identify potential solutions. The recent federal legislative efforts at universal insurance are a great starting point to address pediatric health care disparities. Unfortunately, the provision of health insurance alone may not remove all of the health care disparities in our children. Further studies are needed to be sure every American child is provided excellent health care.
This study was a retrospective observational analysis. No causal determinants for disparity can be inferred. Also, time to treatment and prehospital variables were not collected. Although the NTDB is large, it may not be nationally representative. However, comparing the uninsured rate in this study to nationally available statistics revealed close approximation (our study found 10% uninsured, and according to the U.S. Census Bureau in 2008, 9.9% of children were uninsured24). Additionally, the analysis focused on children at risk for death after trauma (ISS > 8) and did not include those with minor injuries. These conclusions may not be generalized to all children suffering trauma.
In this study of mortality in moderately to severely injured children, health care outcome disparities based on race and insurance status were found. Reasons for these disparities remain unanswered, and further research is needed to identify and implement solutions so that all American children are afforded the same survival opportunity after suffering traumatic injuries.
The authors thank Ruth Moore, PhD, for her assistance with this study.