Association Between Hospitals Caring for a Disproportionately High Percentage of Minority Trauma Patients and Increased Mortality:  A Nationwide Analysis of 434 Hospitals FREE

Multiple studies^1- 3 have demonstrated racial disparities in mortality and functional outcomes after traumatic injury in the United States. Wong et al⁴ suggest that injuries are the third largest contributor to racial disparities in mortality in the United States. To develop intervention programs to mitigate these inequalities, it is imperative to elucidate the exact causes and mechanisms that lead to disparities in trauma outcome. Data from other medical specialties have implicated variations in quality of care and outcomes as a major contributor to this problem. According to 2 Institute of Medicine reports, Unequal Treatment⁵ and Crossing the Quality Chasm,⁶ health care in the United States often falls short of delivering its full potential benefits to patients and, in particular, to racial and ethnic minorities. Even when socioeconomic status and insurance coverage are controlled for, minorities receive poorer quality and less complete health care. For example, black patients receive less intensive hospital care, including fewer cardiac procedures, lung resections for cancer, kidney and bone marrow transplants, and less antiretroviral treatment for human immunodeficiency virus.⁷ In exploring the complex causes of these inequalities, patient characteristics such as comorbidities and provider characteristics such as treatment preferences and implicit biases have been considered.⁵

Institutional and health system–related factors that lead to differences in the quality of care between treatment facilities have also been posited as potentially major contributors to race-based differences in outcomes.^8- 9 Many hospitals, especially those located in economically depressed urban neighborhoods, serve a substantially higher proportion of minority patients. Such hospitals have been shown to have worse outcomes and suboptimal quality of care for a variety of diseases and surgical interventions.^9- 11 Patients who are treated for an acute myocardial infarction or undergoing procedures such as an esophagectomy, carotid endarterectomy, or coronary artery bypass graft at hospitals that disproportionately serve black patients have increased mortality.^12- 13

Based on these hospital-associated disparities among nontrauma patients, our investigation hypothesizes that the clustering of minority patients at hospitals with worse outcomes may substantially contribute to disparities in outcomes after trauma. Thus, the objective of our study is to use the largest available national trauma registry to determine whether patients treated at hospitals primarily serving minority patients with trauma have higher rates of in-hospital mortality. If this were found to be true, then these hospitals could be targeted for performance improvement initiatives that could help reduce disparities experienced by racial and ethnic minorities.

Our study analyzed the medical records of patients included in the National Trauma Data Bank (NTDB) between 2007 and 2008. The NTDB is maintained by the American College of Surgeons Committee on Trauma and contains approximately 1.3 million medical records from 592 participating hospitals that care for trauma patients. The NTDB is a convenience sample and the largest repository of trauma registry data in the United States. Patients from the years 2007 and 2008 were chosen owing to the NTDB's adoption in 2007 of the National Trauma Data Standard, which significantly improved the quality and fidelity of information present in the data bank.

Adult patients aged 18 to 64 years with an Injury Severity Score of 9 or greater who were white, black, or Hispanic were included (Figure 1). Patients with an Injury Severity Score of less than 9 were excluded owing to the very low incidence of the main outcome measure: mortality during index admission. Patients with unknown insurance status or from hospitals that treated fewer than 50 trauma patients per year were also excluded. Race was dichotomized into minority patients (blacks and Hispanics combined) and white patients, based on patient self-report. Demographic data on age, sex, and insurance status were assessed. For insurance status, patients were classified as those with private or commercial insurance, with government-sponsored insurance (such as Medicaid), or without insurance (ie, uninsured). Data on injury severity were also collected, including the Injury Severity Score, the presence of hypotension on arrival to the emergency department (systolic blood pressure <90 mm Hg),¹⁴ the presence of severe head and/or extremity injury (Abbreviated Injury Scale score ≥3), the type of injury (penetrating vs blunt), and the mechanism of injury.¹⁵ Hospitals were stratified into 3 groups based on the proportion of minority patients: predominantly majority hospitals (<25% of patients are minority patients), mixed hospitals (25%-50% of patients are minority patients), and predominantly minority hospitals (>50% of patients are minority patients). Patients at predominantly majority hospitals were used as the reference group. A bivariate analysis comparing patients in the different hospital strata was undertaken using the Pearson χ² test for categorical variables and the t test for continuous variables.

Adjusted odds of in-hospital mortality were compared between the 3 hospital strata by employing multivariate logistic regression. To statistically account for patient correlation among hospitals, generalized linear models (with a cluster-correlated robust estimate of variance) that adjust for within-hospital cluster correlation were used.¹⁶ The multivariate model controlled for patient characteristics known to predict mortality after trauma, including age, sex, insurance status, Injury Severity Score, presence of hypotension on arrival to the emergency department, presence of severe head and/or extremity injury, type of injury, and mechanism of injury. Owing to concern regarding the independent effect of trauma center status and type of injury, 2 subset analyses were also performed in which the data set was restricted to (1) patients treated at level 1 trauma centers and (2) patients with a blunt injury.

To determine the experience of individual races/ethnicities within each of the 3 hospital groups, a subset analysis by race was also performed. For this analysis, the adjusted mortality for minority patients (blacks and Hispanics) vs the adjusted mortality of white patients was calculated within each of the 3 hospital groups. Subsequently, the same analysis was also performed after stratifying the minority group of patients into blacks and Hispanics. Similarly, an analysis comparing outcomes relative to insurance status was conducted within the hospitals. Insured patients acted as the reference group, and the adjusted odds of mortality for uninsured and government-insured patients were estimated for each of the 3 hospital groups studied. Finally, to minimize the effect of confounding by comorbidities, a sensitivity analysis focusing on the 18- to 40-year-old age group was also performed. Statistical analyses were performed using Stata/MP statistical software version 11 (StataCorp, College Station, Texas). Statistical significance was set at P < .05.

There were 1 134 946 patients whose medical records were included in the NTDB for 2007 and 2008. Of these, 311 568 patients from 434 hospitals were included in our analysis (Figure 1). Of these, 215 176 (69%) were white, and nearly half of the patients had an Injury Severity Score between 9 and 15.Quiz Ref IDMore than half of the hospitals were categorized as predominantly majority hospitals, just over a quarter were mixed hospitals, and just 13% of the treating facilities were classified as predominantly minority hospitals serving a more than 50% minority population. The overall crude mortality rate was 5% (Table 1).

On bivariate analysis, predominantly majority hospitals had significantly lower mortality (P < .001), a higher proportion of female patients (P < .001), fewer patients with hypotension on arrival to the emergency department (P < .001), and fewer patients with penetrating injuries (P < .001) (Table 2). Insurance status closely tracked minority hospital status. Figure 2 shows that the proportion of privately insured patients in predominantly minority hospitals is less than half that of predominantly majority hospitals. Quiz Ref IDConversely, predominantly minority hospitals had nearly double the proportion of uninsured patients as did predominantly majority hospitals.

Table 3 outlines the hospital characteristics associated with each of the 3 minority categories. Interestingly, hospitals in the 2 categories with the highest proportions of minorities (ie, those with 25%-50% of patients who were minorities and those with >50% of patients who were minorities) were more likely to be level 1 trauma centers. In contrast, only 30% of the hospitals serving less than 25% of patients who were minorities had level 1 status. Mixed hospitals and predominantly minority hospitals were also more likely to be teaching hospitals and were somewhat larger than predominantly majority hospitals. However, the median number of core trauma surgeons was the same across all hospital categories.

Crude mortality was significantly lower at predominantly majority hospitals (Figure 3). After adjusting for age, sex, insurance status, injury severity, the presence of severe head and/or extremity injury, the presence of hypotension on arrival to the emergency department, and the type and mechanism of injury, the odds of mortality continued to increase with an increasing proportion of minority patients. As depicted in Figure 4, compared with the reference group of predominantly majority hospitals, patients treated at mixed hospitals had a 16% higher adjusted odds of death. Quiz Ref IDThis adjusted odds of death further increased to a 37% increased odds of death among patients treated at predominantly minority hospitals.

In a separate analysis of patients with a blunt injury only, trauma patients treated at mixed hospitals had an adjusted odds of death that was 18% higher than patients treated at predominantly majority hospitals, and patients at predominantly minority hospitals had an adjusted odds of death that was 45% higher than patients treated at predominantly majority hospitals. During stratification by trauma center, a similar effect was observed. When the data set was restricted to patients treated at level 1 trauma centers, patients had an adjusted odds of death that was 42% higher than patients treated at predominantly majority hospitals, if the trauma center was predominantly minority (adjusted odds ratio [AOR], 1.42 [95% confidence interval {CI}, 1.14-1.77]). However, there was no difference in mortality outcomes among patients in mixed or predominantly majority level 1 trauma centers (AOR, 1.11 [95% CI, 0.91-1.35]). Similarly, a subset analysis of 18- to 40-year-old patients showed an increased adjusted odds of death in the predominantly minority hospitals (AOR, 1.22 [95% CI, 1.01-1.48]). However, there was no difference in mortality between the mixed and predominantly majority hospitals in this subset (AOR, 0.99 [95% CI, 0.81-1.20]).

The experience of patients from each race/ethnicity was also investigated within each of the 3 hospital groups. There was no difference in adjusted mortality between whites and minorities within the same type of hospital. When the minority group was stratified into black patients and Hispanic patients, and when the analysis was repeated across the 3 hospital groups, the same was noted (Table 4). A similar analysis was performed assessing the effect of insurance status on patients within the 3 hospital categories. Uninsured patients were found to have significantly increased odds of mortality within all 3 hospital groups (Table 5).

Quiz Ref IDOur study demonstrates that a trauma patient's odds of in-hospital mortality are directly associated with the proportion of minority trauma patients treated at the facility where he or she is receiving care. Moderate to severely injured patients, treated at predominantly majority hospitals (<25% of patients were minorities), had the lowest adjusted odds of death, whereas those cared for at predominantly minority hospitals (>50% of patients were minorities) had the highest odds of death. These differences in mortality outcomes among trauma facilities serving higher proportions of minority patients may contribute significantly to the known racial disparities experienced by trauma patients in the United States.

The United States remains characterized by significant residential segregation by race and income levels,⁹ resulting in parallel segregation of patients among providers and hospitals.^9,17 Jha et al¹⁸ found that just 28% of all Veterans Affairs hospitals care for more than 75% of black veterans. A significant concentration of minority patients exists within the field of primary health care, with Bach et al¹⁷ determining that 80% of primary care visits by black patients occurred with just 22% of physicians.

Low-performing hospitals have been noted to serve higher proportions of minority, uninsured, and low-income patients.^8- 9 In our study, hospitals serving higher proportions of minority patients with trauma have a significantly disadvantaged payor mix, with nearly double the percentage of uninsured patients compared with hospitals with less than 25% of patients being minorities. This payor mix disparity may adversely influence the structure and process of care a hospital can bring to critically injured patients. For example, Werner et al¹⁹ found that hospitals caring for predominantly poor or underserved patients had a worse performance status than hospitals with more privately insured patients. It is thought that hospitals tasked with the difficult challenge of providing care to greater numbers of low-income and uninsured patients are not sufficiently financed to invest in quality-of-care improvement initiatives. In turn, suboptimal quality of care in these hospitals disproportionately affects minority patients, contributing to race-based disparities in health care.^8- 9

Significant variation in outcomes also exists between trauma centers, suggesting variations in quality of care.^20- 23 Although level 1 trauma centers have been shown to have lower mortality rates for severely injured trauma patients,^24- 25 not all similarly designated trauma centers achieve the same risk-adjusted outcomes.^20- 22 The reasons for these variations in outcome are unclear, but they may be related to the heterogeneity of patient populations by payor mix or injury mechanism or to inconsistent practice patterns.

Quiz Ref IDIn our analysis, hospitals serving higher proportions of minorities tended to be larger, more frequently level 1 trauma centers, and were more often university hospitals. However, the median numbers of core trauma surgeons, orthopedic surgeons, and neurosurgeons available were the same across all 3 categories of facilities. Despite this, inequalities in the quality of care may disproportionately affect minority patients, who tend to cluster at a relatively small number of low-performing institutions.⁸ Ly et al¹⁰ show that “black-serving” US hospitals have higher rates of potential safety events among both black and white patients than do other hospitals on 6 of 11 patient safety indicators. Similarly, it has been found that, with regard to acute myocardial infarction care, the lowest-performing hospitals in the United States serve a patient population that is older, more often black, and more likely to reside in lower income areas (based on each patient's zip code of residence).¹¹

Baicker et al⁹ suggest that differences in the quality of care observed between hospitals may originate in variations in practice patterns, technological capabilities, hospital capacity and supply of specialists, or patient characteristics. It may also be that the patients seen at predominantly minority hospitals have intrinsic differences in preexisting health status, injury patterns, or some of the social determinants of health. These patients may have more or undertreated comorbidities, which are poorly captured in existing trauma data sets.

Similarly, in our study, it is noted that predominantly minority hospitals and even the mixed hospitals had much higher rates of uninsured patients. This is particularly important given that uninsured patients had significantly increased odds of mortality within all 3 hospital groups. Lack of insurance is generally regarded as more than just the ability to pay a bill and suggests that there is something fundamentally different about the socioeconomic status of patients treated at the different types of facilities. The underinsured population with likely much less resources, which is seen at predominantly minority hospitals, may bring significant residual confounding that could not be controlled for. Issues such as treatment delay,²⁶ health illiteracy,²⁷ and differential rates of follow-up and access to rehabilitation services²⁸ have been implicated as potential reasons for the worse quality of care and worse outcomes among uninsured patients. Additional issues at public hospitals include nurse staffing shortages, constrained budgets, and lack of capital and technical support.^29- 30 Further research is needed to determine the underlying mechanisms that lead to worse outcomes at hospitals serving predominantly minority populations. If lack of resources in minority hospitals is a major driver of lower performance, policies and funds to strengthen and support these institutions will be an important step in reducing racial disparities in mortality after trauma.

Another important finding of our study is that minority patients, whether black or Hispanic, did not have worse outcomes at predominantly majority hospitals. Similarly, there was no difference in odds of mortality for whites, blacks, or Hispanics at predominantly minority hospitals; the poor outcomes at these hospitals were evenly spread across all 3 racial/ethnic groups. This important finding suggests that a large proportion of the racial disparities observed after trauma may be due to worse outcomes seen at predominantly minority hospitals.

The heavy concentration of minority patients at a relatively small number of hospitals presents a unique opportunity for interventions that could achieve an immediate and substantial impact in mitigating disparities. Hosking et al³¹ and Green et al³² have suggested that improving cultural competency, addressing health literacy, and implementing quality-of-care improvement initiatives focused on equity may reduce disparities in health care. Policies aimed at improving cultural competency and directing resources to centers serving a high proportion of minority patients can have a dual effect on worse outcomes caused by disparities in quality of care within and between hospitals among black patients with trauma.

There are several limitations to our study. The NTDB does not provide indicators of the quality of care received at each of these trauma centers. Therefore, in our analysis, it cannot be determined whether quality-of-care measures were worse among hospitals serving higher proportions of minorities. Second, there may be unmeasured confounders that contribute to the worse outcomes at hospitals serving predominantly minorities. For instance, our study was not able to control for comorbidities owing to a lack of available data in the NTDB. Another important confounder may have been increased risk of mortality due to trauma-patient recidivism.³³ Although the NTDB now uses the National Trauma Data Standard, approximately 5% of data were excluded owing to missing data on insurance status. However, given the very strong associations noted and the random manner in which data were missing, it is unlikely that this small amount of missing data would quantitatively change the results. Finally, outcomes after hospital discharge, including 30-day mortality, could not be assessed. Our study has several strengths as well, including the use of clustering for adjusted analyses and the use of the most recent versions of the NTDB to achieve a much lower preponderance of missing data.

Our large study of trauma patients demonstrates inequities in mortality rates associated with the proportion of minority patients treated at a facility. The exact mechanisms that lead to the higher mortality rates observed at hospitals with a disproportionately high percentage of minority patients need to be investigated further. Potential factors, such as prehospital transport variations, preexisting comorbidities, and other disparities that may affect survival, need careful examination. Additionally, an investigation of process measures and compliance with trauma protocols between hospitals may help explain the origins of these disparities. Given that hospitals serving predominantly minority patients also care for a high proportion of uninsured patients, initiatives to financially strengthen these institutions are especially important. Augmenting the assets of resource-poor institutions and implementing culturally competent quality-of-care improvement programs at hospitals that primarily serve minority populations may be an excellent first step toward reducing racial disparities in trauma outcomes and improving care for all patients.