Abstract
Background
Race-related health disparities have been well documented in the United States. In some settings, Black patients have better outcomes in hospitals that serve high proportions of Black patients. We hypothesized that Black trauma patients would have lower mortality in high Black-serving hospitals.
Methods
We identified all adult patients with Black or White race and with an injury severity score (ISS) of ≥4 from the 2017 National Inpatient Sample. We collected hospital identifier, mechanism, age, sex, comorbidities, urban-rural location, insurance, zip code income quartile, and injury severity calculated from ICD-10 codes. We used a previously published method to group hospitals by proportion of Black patients served: high Black-serving (H-BS, top 5%), medium (M-BS, 5% to 25%), and low (L-BS, bottom 75%). Adjusted logistic regression using an interaction variable between race and hospital service rank (reference: White patients in H-BS) was used to identify factors associated with mortality.
Results
We analyzed 184,080 trauma patients (median age 72 [IQR 55-84], ISS 9 [4-10]), of whom 11.7% were Black. Overall mortality was 4%. Of 2,376 hospitals, 126 (5.3%) were H-BS and 469 (19.7%) M-BS. 29.8% of Black and 3.6% of White patients were treated at H-BS hospitals, while 71.7% of White and 23.6% of Black patients were treated at L-BS hospitals (p<0.001). Black patients had the lowest mortality at H-BS hospitals (OR 0.76 [0.64-0.92]) and the highest mortality (OR 1.43 [1.13-1.80]) at L-BS hospitals. White patients had the lowest mortality at L-BS hospitals (OR 0.76 [0.64-0.92]).
Conclusions
After adjusting for patient and hospital factors, disparities exist such that Black and White patients have the best outcomes in hospitals that treat those patients most frequently, suggesting potential for racial bias at the institutional level. Further efforts must be made to promote equitable treatment at all hospitals and reduce these disparities.
Level of Evidence
III
Study Type
Epidemiological
Keywords: Trauma, disparities, mortality, race
Background
Trauma is the leading cause of death in the United States up to the age of 45 and the third overall among all ages.1 However, outcomes after trauma are not evenly distributed, particularly across racial groups. Investigations of racial disparities in trauma surgical management show that patients from underrepresented racial and ethnic groups frequently face increased odds of mortality after injury.2-4 This mirrors findings across non-trauma surgical specialties that have consistently demonstrated that Black patients are more likely to die after surgery than White patients, after adjusting for other clinical, social, and demographic confounders.5,6
Hospital characteristics, including populations served and the quality of care provided, are increasingly being investigated for their contributions to racial disparities in hospital-based care.7-9 Previous studies have shown that the majority of elderly Black patients receive care in a small concentration of hospitals, Black patients are more likely to undergo complex surgeries at hospitals with lower case volumes, and Black mothers are more likely to die after childbirth.7,10,11 In trauma, it is known that socioeconomic and demographic factors are associated with higher mortality.4 Prior studies have shown conflicting results on the effect of race and mortality in trauma, but a recent meta-analysis suggested that non-White race was associated with higher odds of death. Mechanisms for these effects are not definitively understood.12
We sought to examine the effect of hospital population served on trauma mortality in the United States to better understand how hospital characteristics may relate to racial disparities in trauma care. We hypothesized that, after adjusting for injury-related and hospital-specific factors, Black trauma patients would have lower mortality in hospitals that treat higher proportions of Black patients and, conversely, higher mortality in hospitals that treat lower proportions of Black patients.
Materials and Methods
Data Source
The 2017 iteration of the National (Nationwide) Inpatient Sample (NIS) of the Healthcare Cost and Utilization Project (HCUP), as supported by the Agency for Healthcare Research and Quality (AHRQ), was utilized for this study.13 The NIS only includes data from inpatient stays rather than individuals, so patient events and diagnoses unrelated to the stay are not available and not included in this analysis.
Patient Selection
Inpatients aged 18 years or older in the NIS with Black or White race were included in the analysis if they had an injury severity score (ISS) of 4 or more. A previously published open-source program using the International Classification of Diseases (ICD) Programs for Injury Categorization using R statistical software (ICDPIC-R) was utilized to map injury diagnosis codes to injury severity codes, including the maximum abbreviated injury scores for each body region and a summary injury severity score (ISS) for each admission.14 External causes of injuries were also mapped using ICDPIC-R.
Variables extracted from the NIS database included demographic variables (hospital identifier, age, sex, urban-rural location, insurance, zip code income quartile), injury characteristics (penetrating mechanism and ISS), and comorbidities. Hospital location was described as urban or rural. Health insurance type was classified as Medicare, Medicaid, private insurance, self-pay, no charge, or other, which included worker’s compensation and government programs not previously described. Zip code income quartile reflects the estimated median household income of residents in each patient’s home zip code as reported by AHRQ. The Elixhauser measure, which is a categorization of medical comorbidities into thirty-one conditions, was used to determine comorbidities.15 These variables were extracted from ICD-10 diagnosis codes using the STATA “Elixhauser” module for ICD-10 codes.
Hospital Ranking
A methodology previously published by Ly et al was used to stratify hospitals according to proportion of Black trauma patients served.16 Trauma admissions were grouped by hospital using the NIS hospital identifier. The percent of Black and White trauma patients served by each hospital, as identified by the NIS hospital identifier, was calculated and the hospitals were subsequently ranked by proportion of Black patients served overall. Cut-offs for groups were established, such that hospitals were identified as high Black-serving (H-BS, top 5th percentile), medium (M-BS, between 5th and 25th percentile), and low (L-BS, below the 25th percentile, or the lowest 75 percent).
Outcome Variable
Primary outcome of interest was in-hospital mortality, as defined by the inpatient discharge status field in NIS.
Analysis
Patient characteristics for all included subjects were described, and bivariate analysis was performed for patients with White race compared to patients with Black race. Variables examined included patient demographics, total Elixhauser comorbidity count, injury severity, hospital characteristics, and mortality. Demographic and injury-related clinical variables were also compared across the three hospital stratifications (H-BS, M-BS, and L-BS) as well as across the four injury severity categories (minor injury [ISS 4-9], moderate injury [ISS 10-15], severe injury [ISS 16-24], and profound injury [ISS 25-75]). Comparisons between groups was performed using the Chi-square test for categorial variables and a Kruskal-Wallis equality of populations rank test for continuous variables.
Adjusted logistic regression using an interaction variable between race and hospital service rank was performed to identify factors significantly associated with mortality. The reference group was defined as White patients in H-BS hospitals. The regression was adjusted for penetrating trauma, injury severity score, gender, age, a summary score of Elixhauser comorbidities, payer, zip code income quartile, and rural hospital status. Odds ratios, 95% confidence intervals, and p-values are reported.
Additional adjusted logistic regression models were performed to compare odds of mortality across individual injury severity categories (minor, moderate, severe, and profound) and individual traumatic mechanisms (penetrating injury and non-penetrating injury). For the injury severity regression models, the regression was adjusted for mechanism, gender, age, Elixhauser comorbidity summary, payer, zip code income quartile, and rural hospital status. For the traumatic mechanism regression models, the regression was adjusted for injury severity, gender, age, Elixhauser comorbidity summary, payer, zip code income quartile, and rural hospital status. As for the primary logistic regression, odds ratios, 95% confidence intervals, and p-values are reported.
All p-values and confidence intervals are presented throughout the manuscript. ICDPIC-R was applied to NIS data using R version 3.6.3 and RStudio v1.2.5033. Statistical analysis was performed using Stata MP, version 16.0 statistical software (Statacorp, College Station, TX).
This study was deemed exempt from review by our Institutional Review Board given its use of deidentified data. Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines were utilized in the drafting of this manuscript.
Results
Population Demographics
We identified 184,080 adult trauma patients who met inclusion criteria, as reflected in Table 1. Of these, 88.3% were White and 11.7% were Black. Median age was 72 years (IQR 55-84) for the total population, 74 years (IQR 58-85) for White patients, and 54 years (IQR 33-70) for Black patients (p < 0.001). Males accounted for nearly 49% of all patients and, more granularly, 47% of White patients and 63% of Black patients (p < 0.001). Aside from being younger and predominantly male compared to their White counterparts, Black patients were more likely to present with penetrating trauma than White patients (14% vs. 2%, p < 0.001). Median ISS was 9 regardless of race, with interquartile ranges varying slightly between Black and White patients. Median number of Elixhauser comorbidities was three, with interquartile ranges again varying slightly by patient race. Medicare and Medicaid patients represented 62% and 10%, respectively, of all study subjects. Among White subjects, 65% were Medicare and 8% were Medicaid patients, compared with Black subjects, among whom 36% were Medicare and 27% were Medicaid patients. Over 10% of Black patients were categorized as “self-pay,” compared to just over 3% of White patients.
Table 1:
Study population stratified by race
| Total | White | Black | p-value | |
|---|---|---|---|---|
| n = 184,080 | n = 162,600 | n = 21,480 | ||
| Age, years , median [IQR] | 72 [55-84] | 74 [58-85] | 54 [33-70] | < 0.001 |
| Sex , n (%) | < 0.001 | |||
| Male | 90,090 (48.9%) | 76,527 (47.1%) | 13,563 (63.2%) | |
| Female | 93,978 (51.1%) | 86,064 (52.9%) | 7,914 (36.9%) | |
| Mechanism of Trauma , n (%) | < 0.001 | |||
| Penetrating | 6,063 (3.3%) | 3,044 (1.9%) | 3,019 (14.1%) | |
| Other | 178,017 (96.7%) | 159,556 (98.1%) | 18,461 (86.0%) | |
| ISS , median [IQR] | 9 [4-10] | 9 [4-10] | 9 [4-16] | < 0.001 |
| Elixhauser comorbidities , median [IQR] | 3 [2-5] | 3 [2-5] | 3 [1-4] | < 0.001 |
| Insurance status , n (%) | < 0.001 | |||
| Medicare | 114,268 (62.2%) | 106,302 (65.5%) | 7,966 (36.3%) | |
| Medicaid | 19,158 (10.4%) | 13,362 (8.2%) | 5,796 (27.1%) | |
| Private insurance | 34,604 (18.9%) | 30,506 (18.8%) | 4,098 (19.2%) | |
| Self-pay | 7,798 (4.3%) | 5,577 (3.4%) | 2,221 (10.4%) | |
| No charge | 632 (0.3%) | 443 (0.3%) | 189 (0.9%) | |
| Other | 7,145 (3.9%) | 6,027 (3.7%) | 1,118 (5.2%) | |
| Median household income for patient’s zip code , n (%) | < 0.001 | |||
| 1: $1-$43,999 | 49,778 (27.6%) | 37,945 (23.8%) | 11,833 (56.3%) | |
| 2: $44,000-$55,999 | 47,601 (26.4%) | 43,497 (27.2%) | 4,204 (20.0%) | |
| 3: $56,000-$73,999 | 44,041 (24.4%) | 40,973 (25.7%) | 3,068 (14.6%) | |
| 4: $74,000+ | 39,133 (21.7%) | 37,229 (23.3%) | 1,904 (9.1%) | |
| Hospital location type , n (%) | < 0.001 | |||
| Urban | 136,610 (74.7%) | 118,307 (73.2%) | 18,303 (86.1%) | |
| Rural | 46,307 (25.3%) | 43,356 (26.8%) | 2,951 (13.9%) | |
| Died during hospitalization , n (%) | 7,264 (4.0%) | 6,381 (3.9%) | 883 (4.1%) | 0.187 |
Median and interquartile range presented for continuous variables and n (%) presented for categorical variables. Comparisons between groups was performed using the Chi-square test for categorial variables and a Kruskal-Wallis equality of populations rank test for continuous variables. IQR=Interquartile range.
Estimated median household income according to the patient’s home zip code was classified by quartile. The distribution among the entire study population was fairly even: 27% in the first or poorest quartile, 26% in the second quartile, 24% in the third quartile, and almost 22% in the fourth or wealthiest quartile. The distribution among specifically White patients largely mirrored the overall study population distribution. Among Black patients, however, the distribution favored the lower quartiles: 56% of Black patients were in the first or poorest quartile, 20% in the second quartile, 14% in the third quartile, and 9% in the fourth or wealthiest quartile. Black patients were more likely than White patients to receive care at an urban hospital (86% vs. 73%, p < 0.001). In unadjusted bivariate analysis, Black patients had similar mortality to White patients (4.11% vs. 3.92%, p = 0.187).
Hospital Stratification and Populations Served
A total of 2,376 hospitals were included in the analysis. After stratifying by proportion of Black patients served, 126 hospitals (5.3%) were designated H-BS, 469 hospitals (19.7%) were designated M-BS, and the remaining 1,781 hospitals (75.0%) were designated L-BS. The 5% of hospitals designated as H-BS treated 6.6% of the total trauma patients, M-BS hospitals treated 27.3%, and L-BS hospitals, which accounted for 75% of all included hospitals, treated 66.6% of patients. Nearly half of the patients treated at H-BS hospitals were White (5,774 of 12,177 H-BS patients, 47.4%). Only a small proportion of patients treated at L-BS hospitals were Black (5,065 of 121,708, 4.2%). Of all Black trauma patients in the sample, more than three-quarters (76.4%) were treated in just 25% of the hospitals.
Patient characteristics by hospital stratification – H-BS, M-BS, and L-BS – are presented in Table 2. Median age of patients admitted to H-BS hospitals was 57 years (IQR 37-74) compared to 68 years (51-82) at M-BS and 75 years (59-85) at L-BS hospitals (p < 0.001). Patients admitted to H-BS hospitals were more likely to be younger, male (61% at H-BS vs. 52% at M-BS vs. 46% at L-BS, p < 0.001), and to present after a penetrating mechanism of trauma (11% at H-BS vs. 4.5% at M-BS vs. nearly 2% at L-BS, p < 0.001). Median ISS was 9 across all three hospital stratifications. The median number of Elixhauser comorbidities in patients admitted to H-BS hospitals was 2, compared to 3 at both M-BS and L-BS hospitals (p < 0.001).
Table 2:
Characteristics of study population stratified by hospital proportion of Black patients served
| Admissions to a High Black- Serving Hospital |
Admissions to a Medium Black- Serving Hospital |
Admissions to a Low Black- Serving Hospital |
p-value | |
|---|---|---|---|---|
| n = 12,175 | n = 50,192 | n = 121,706 | ||
| Race, n (%) | < 0.001 | |||
| White | 5,774 (47.4%) | 40,183 (80.1%) | 116,643 (95.8%) | |
| Black | 6,403 (52.6%) | 10,012 (19.9%) | 5,065 (4.2%) | |
| Age , median [IQR] | 57 [37-74] | 68 [51-82] | 75 [59-85] | < 0.001 |
| Sex , n (%) | < 0.001 | |||
| Male | 7,502 (61.6%) | 26,164 (52.1%) | 56.424 (46.4%) | |
| Female | 4,673 (38.4%) | 24,023 (47.9%) | 65,282 (53.6%) | |
| Mechanism of Trauma , n (%) | < 0.001 | |||
| Penetrating | 1,395 (11.5%) | 2,263 (4.5%) | 2,405 (2.0%) | |
| Not Penetrating | 10,782 (88.5%) | 47,932 (95.5%) | 119,303 (98.0%) | |
| ISS , median [IQR] | 9 [4-16] | 9 [4-13] | 9 [4-16] | < 0.001 |
| Elixhauser comorbidities , median [IQR] | 2 [1-4] | 3 [1-5] | 3 [2-5] | < 0.001 |
| Insurance status , n (%) | < 0.001 | |||
| Medicare | 4,993 (41.2%) | 27,896 (55.7%) | 81,379 (67.0%) | |
| Medicaid | 2,608 (21.5%) | 6,128 (12.2%) | 10,422 (8.6%) | |
| Private insurance | 2,572 (21.2%) | 10,607 (21.3%) | 21,425 (12.7%) | |
| Self-pay | 1,194 (9.9%) | 2,993 (6.0%) | 3,611 (3.0%) | |
| No charge | 141 (1.2%) | 233 (0.5%) | 258 (0.2%) | |
| Other | 599 (5.0%) | 2,235 (4.5%) | 4,311 (3.6%) | |
| Median household income for patient’s zip code , n (%) | < 0.001 | |||
| 1: $1-$43,999 | 6,304 (52.5%) | 16,635 (33.8%) | 26839 (22.5%) | |
| 2: $44,000-$55,999 | 2,141 (17.8%) | 12,801 (26.0%) | 32,659 (27.4%) | |
| 3: $56,000-$73,999 | 2,036 (17.0%) | 10,937 (22.2%) | 31,068 (26.0%) | |
| 4: $74,000+ | 1,527 (12.7%) | 8,839 (18.0%) | 28,767 (24.1%) | |
| Hospital location type , n (%) | < 0.001 | |||
| Urban | 10,024 (82.8%) | 39,532 (79.4%) | 87,054 (71.9%) | |
| Rural | 2,084 (17.2%) | 10,273 (20.6%) | 33,950 (28.1%) | |
| Died during hospitalization , n (%) | 567 (4.7%) | 2,137 (4.3%) | 4,560 (3.8%) | < 0.001 |
Median and interquartile range presented for continuous variables and n (%) presented for categorical variables. Comparisons between groups was performed using the Chi-square test for categorial variables and a Kruskal-Wallis equality of populations rank test for continuous variables. IQR=Interquartile range.
Social factors varied significantly between hospital groups. H-BS hospitals were the least likely to treat Medicare patients (41.2% of patients) but the most likely to treat Medicaid patients (21.5%). Conversely, L-BS hospitals were the most likely to treat Medicare patients (67.0%) and the least likely to treat Medicaid patients (8.6%) (p < 0.001). Similarly, when examining household income, H-BS hospitals were more likely to admit and manage patients in the lower or poorer estimated median household income quartile (52.5% of patients) compared to L-BS hospitals (22.5%), whereas L-BS hospitals were more likely to treat patients in the fourth or wealthiest quartile (24.1%) compared to H-BS hospitals (12.7%) (p < 0.001). In summary, H-BS hospitals were more likely to care for poorer patients while L-BS hospitals were more likely to care for wealthier patients, based on the estimated median household income of all residents in each patient’s home zip code.
Patient Stratification Along Injury Severity Category
Study subjects were further grouped into their corresponding injury severity score categories, as presented in Table 3. Of the 184,080 adult trauma patients included in this study, 134,609 (73.1%) patients were minorly injured (ISS 4-9), 15,907 (8.6%) were moderately injured (ISS 10-15), 24,868 (13.5%) were severely injured (ISS 16-24), and 8,696 (4.7%) were profoundly injured (ISS 25-75).
Table 3:
Characteristics of study population stratified by Injury Severity Score category
| Minor Injury (ISS 4-9) |
Moderate Injury (ISS 10-15) |
Severe Injury (ISS 16-24) |
Profound Injury (ISS 25-75) |
p-value | |
|---|---|---|---|---|---|
| n = 134,609 | n = 15,907 | n = 24,868 | n = 8,696 | ||
| Race, n (%) | < 0.001 | ||||
| White | 120,690 (89.7%) | 13,763 (86.5%) | 21,363 (85.9%) | 6,784 (78/0%) | |
| Black | 13,919 (10.3%) | 2,144 (13.5%) | 3,505 (14.1%) | 1,912 (22.0%) | |
| Age , median [IQR] | 74 [58-85] | 63 [42-80] | 70 [54-82] | 54 [32-72] | < 0.001 |
| Sex , n (%) | < 0.001 | ||||
| Male | 60,895 (45.2%) | 9,036 (56.8%) | 14,261 (57.3%) | 5,898 (67.9%) | |
| Female | 73,708 (54.8%) | 6,871 (43.2%) | 10,604 (42.7%) | 2,795 (32.3%) | |
| Mechanism of Trauma , n (%) | < 0.001 | ||||
| Penetrating | 2,968 (2.2%) | 778 (4.9%) | 1,005 (4.0%) | 1,312 (15.1%) | |
| Not Penetrating | 131,641 (97.8%) | 15,129 (95.1%) | 23,863 (96.0%) | 7,384 (84.9%) | |
| Elixhauser comorbidities , median [IQR] | 3 [2-5] | 2 [1-4] | 3 [2-5] | 2 [1-4] | < 0.001 |
| Hospital Stratification , n (%) | < 0.001 | ||||
| High Black-Serving | 7,561 (5.6%) | 1,443 (9.1%) | 2,054 (8.3%) | 1,119 (12.9%) | |
| Medium Black-Serving | 34,884 (25.9%) | 4,799 (30.2%) | 7,553 (30.4%) | 2,959 (34.0%) | |
| Low Black-Serving | 92,164 (68.5%) | 9,665 (60.8%) | 15,261 (61.4%) | 4,618 (53.1%) | |
| Died during hospitalization , n (%) | 3,636 (2.7%) | 500 (3.1%) | 1,709 (6.9%) | 1,419 (16.3%) | < 0.001 |
The racial distribution of study subjects varied by ISS grouping, with the proportion of Black patients increasing as injury severity worsened. Among those with minor injuries, 89.7% of patients were White and 10.3% were Black, compared to 86.5% White and 13.5% Black moderately injured patients, 85.9% White and 14.1% Black severely injured patients, and 78% White and 22% Black profoundly injured (p < 0.001).
Among minorly injured patients, median age was 74 [IQR 58-85] and 45.2% were male. Among profoundly injured patients, median age was two decades younger at 54 [IQR 32-72] and 67.9% males (p < 0.001). Only 2.2% of minorly injured patients sustained penetrating trauma compared to 4.9% of moderately injured patients, 4.0% of severely injured patients, and 15.1% of maximally injured patients (p < 0.001). Number of Elixhauser comorbidities was similar at 2 or 3 across all four injury severity groups.
Although L-BS hospitals admitted the greatest absolute number of patients, study subjects with higher ISS were significantly more likely to receive care at H-BS hospitals. Just 5.6% of patients with ISS 4-9 were treated at H-BS hospitals whereas 12.9% of patients with ISS 25-75 were treated at H-BS hospitals. Correspondingly, the proportion of patients served at L-BS hospitals decreased as injuries worsened (p < 0.001).
Mortality also significantly increased in accordance with worsened injury severity grouping.
Mortality and Logistic Regression Results
Unadjusted mortality rates showed that H-BS hospitals had higher crude rates of death, with mortality in 4.7% of patients served at these hospitals compared to M-BS hospitals (4.3%) and L-BS hospitals (3.8%) (p < 0.001; Table 2). Logistic regression with the outcome of mortality, using an interaction term between race and hospital category, is presented in Table 4. After adjusting for other factors and accounting for race and white-black hospital rank, the odds of mortality differ significantly by race and hospital category.
Table 4:
Mortality by race and hospital stratification
| Odds of Mortality* |
|||
|---|---|---|---|
| Odds Ratio | p-value | 95% Conf. Int. | |
| White Patients | |||
| High Black-Serving Hospital | 1.00 (Ref) | (Ref) | (Ref) |
| Medium Black-Serving Hospital | 0.84 | 0.010 | 0.73 – 0.96 |
| Low Black-Serving Hospital | 0.75 | < 0.001 | 0.66 – 0.85 |
| Black Patients | |||
| High Black-Serving Hospital | 0.76 | 0.004 | 0.64 – 0.92 |
| Medium Black-Serving Hospital | 1.20 | 0.092 | 0.97 – 1.48 |
| Low Black-Serving Hospital | 1.43 | 0.003 | 1.13 – 1.80 |
Regression adjusted for injury mechanism (blunt or penetrating), injury severity score, gender, age, Elixhauser comorbidity count, payer, zip code income quartile, and rural hospital status.
Compared to the reference group of White patients at H-BS hospitals, mortality for White patients decreased as the hospital proportion of Black patients served decreased. White patients at L-BS hospitals had a lower mortality (OR 0.75 [0.66 – 0.85], p < 0.001) than White patients at M-BS hospitals (OR 0.84 [0.73 – 0.96], p = 0.010). The opposite was true for Black patients, who had the lowest mortality at H-BS hospitals (OR 0.76 [0.64 – 0.92], p = 0.004) and the highest mortality at L-BS hospitals (OR 1.43 [1.13 – 1.80], p = 0.003).
Two additional sets of adjusted logistic regression models were performed. To be consistent with the previous logistic regressions, White patients at H-BS hospitals remained the reference group in both sets of subgroup logistic regression modeling. The first set compared mortality odds across the four individual injury severity groupings. For mildly injured patients (ISS 4-9), White patients demonstrated significantly decreased mortality at M-BS (OR 0.66 [0.54 - 0.80], p < 0.001) and L-BS (OR 0.62 [0.52 – 0.75], p < 0.001) hospitals. Conversely, mildly injured Black patients demonstrated a significant increase in mortality in M-BS hospitals (OR 1.39 [1.01 – 1.91], p = 0.042).Among patients with profound injuries (ISS 25-75), Black patients demonstrated significantly decreased mortality odds at H-BS hospitals (OR 0.59 [0.42 – 0.84], p = 0.003). For moderately (ISS 10-15) and severely (ISS 16-24) injured patients, logistic regression modeling did not reveal any significant differences between White and Black patients in terms of mortality odds.
The second set of logistic regression models was performed for subgroups of patients with penetrating and non-penetrating injuries. The only significant finding was that White patients with non-penetrating traumatic injuries demonstrated improved odds of mortality in M-BS (OR 0.83 [0.72 – 0.95], p = 0.006) and L-BS (OR 0.74 [0.65 – 0.85], p < 0.001) hospitals.
Discussion
Racial disparities in medical and surgical care have been widely reported.17 However, the extent and mechanisms by which race impacts trauma-related mortality is not fully understood. The degree to which hospital characteristics in particular impact racial disparities in trauma care is even less understood. After adjusting for injury and hospital factors, our study demonstrated that Black trauma patients had the lowest mortality rates at hospitals serving the highest proportion of Black patients, despite these hospitals having the highest unadjusted overall mortality rates. Conversely, our study also demonstrated that Black trauma patients had the highest mortality rates at hospitals that served the lowest proportion of Black patients. These findings reaffirm the need to identify and address the underlying etiologies of healthcare inequity.
A variety of studies have examined the relationship between race and trauma mortality, with inconsistent findings.2,3,18-20 A meta-analysis by Haider et al pooled data from seven studies and found that Black patients had worse trauma mortality rates than White patients.4 A more recent study, however, performed a population-based adjusted analysis using NIS data and concluded that Black race was not associated with higher mortality.12 We also elected to analyze data provided by the NIS, a national longitudinal hospital inpatient database, to better understand whether or not hospital-level factors were associated with racial disparities in trauma. The large, nationally-representative population eliminated regional variances and allowed our analysis to focus specifically on trends related to trauma-receiving hospitals. Although NIS data has previously been examined for race-related trends in mortality, our study differs as we examine a possible mechanism that links race and mortality at the hospital-level.
Our study demonstrated that trauma is not a disease of equals. For Black Americans, trauma afflicts predominantly the young men from socioeconomically disadvantaged locales, and they are more likely to have sustained serious penetrating injury.21 Here, trauma is a disease of the underserved and socioeconomically high-risk, with a clear gender disparity. White trauma patients, on the other hand, are older, both male and female, and represent a near-even distribution of wealth, suggesting that age is more of a driving factor than socioeconomical status among White Americans. Our analysis also showed that while most trauma patients receive care at urban medical centers, Black patients are significantly more likely to do so. Given that outcomes substantially diverge across hospitals depending on the proportion of Black patients served, these findings suggest severe and complex underlying health inequities.
Previous studies have suggested that hospitals serving high proportions of Black patients provide lower quality of care.5,9,10,16,22 Our findings challenge this assertion. In our analysis, H-BS hospitals demonstrated overall higher unadjusted mortality compared to L-BS and M-BS hospitals. After accounting for patient demographics, injury characteristics, and hospital characteristics, we found that hospitals that treated higher proportions of Black patients had lower adjusted rates of mortality for Black trauma patients. This is most evident among Black patients with profound injury severity (ISS 25-75) whose odds of mortality decreased by nearly half when treated at H-BS hospitals. This suggests that the proportion of Black patients served is not an adequate proxy for hospital performance. Although these findings apply strictly to trauma-related outcomes, they challenge earlier conclusions made across various subspecialties and patient safety indicators that predominantly Black-serving hospitals provide inferior care. In our analysis, Black trauma patients have improved survivorship in H-BS hospitals, notably if they present with severe injuries. On the other side of the same coin, White trauma patients had the lowest mortality at L-BS hospitals, which also manifested in subgroup analysis of non-penetrating patients.
While we identified a clear disparity in trauma care at the hospital level, the underlying reasons for these differences are likely multifactorial. One possible mechanism is societal racism, which not only contributes to the prehospital factors affecting the initial traumatic injury but also may contribute to patient outcomes. Another possible explanation is the volume effect. Black patients were proportionally more likely to present with penetrating trauma to urban hospitals, and previous studies have shown that patients with penetrating injuries have improved survival at centers that see high volumes of these types of injuries, which are also predominantly urban centers.23-25 It may be that H-BS hospitals, which see a greater number of younger Black patients with penetrating trauma, are better equipped to manage these types of traumatic injuries. Hospitals characterized as L-BS have higher volumes of older patients and may, thus, be more prepared or skilled in the management of geriatric trauma, which includes more White patients. Another possible explanation deals more directly with race: earlier research has suggested that Black men receive better care from Black clinicians.26 This may be related to beneficial cultural competency, improved rapport-building behavior, and the maintenance of positive or trusted hospital reputations within specific communities.27
The divergence in survivorship among Black and White trauma patients is also closely tied to various socioeconomic factors such as insurance status. Consistent with previous research investigating the relationship between insurance status and overall outcomes, our study finds that Black trauma patients, when adjusted for injury and hospital characteristics, face significant disadvantages in insurance coverage compared to White patients.19 Black patients are more likely to be uninsured, and those with government insurance are more likely to require Medicaid and less likely to be enrolled in Medicare. This is largely a function of the nearly twenty-year gap in median age between White and Black patients such that Black patients are admitted for trauma before they reach Medicare eligibility. Nevertheless, numerous studies report an independent association between insurance status and mortality, with uninsured patients faring worse.20 Similarly, a study of surgical patients treated at Department of Defense facilities revealed that in a system with equal and ubiquitous access to healthcare (i.e. no disparity in coverage), Black race is not associated with increased risk of death when compared to the general population.28 Our findings support the conclusion that type or lack of insurance coverage directly relates to or is a proxy for other socioeconomic disparities that affect trauma mortality.
Our analysis demonstrated that H-BS hospitals are also more likely than L-BS hospitals to exist in urban centers. This is consistent with earlier studies showing that Black patients are more likely to seek care at large, urban, and academic medical centers.11 Much like the previous assertion tying quality of care to proportion of Black patients served, the implication here is that urban hospitals provide worse care, which subsequently leads to the increased mortality risk reported among Black and other minority patients. Kahn et al reported that while Black patients overall receive significantly worse quality of care and are discharged in greater states of instability, this is partly offset by the fact that they are more likely to seek or receive care at urban hospitals where quality of care has been shown to be better.29,30 Additionally, three-quarters of trauma centers in our analysis exist in urban centers, meaning White trauma patients are also more likely to receive care in an urban setting, albeit not quite as likely as Black trauma patients. In summary, the divergent outcomes among Black and White patients that we uncovered is less likely a function of hospital location and may instead be better understood as a direct consequence of who is affected by trauma and how.
Our study has several important limitations. As is the case with large administrative retrospective databases, the individual entries are unverifiable, and unmeasured or otherwise inaccurate variables may impact the accuracy of our findings. The substantial cohort size should, however, mitigate the effect of these potential inaccuracies in data input or recording. In addition, our study focused on patients who were labeled by an imperfect classification system as White or Black, and we cannot comment on similar effects for any other race or ethnicity. Another concern is reflected in the low median ISS of 9 of qualifying study subjects, for which mortality might be attributable less to the traumatic injury itself and more to baseline medical comorbidity burden. For this reason, we performed subgroup analyses for patients in each of the four individual injury severity groupings as well as for patients with penetrating versus non-penetrating traumatic injuries to identify more granular additional dimensions to our overall findings. Due to the administrative nature of the database, we were also unable to identify non-comorbidity risk factors that relate to age, such as frailty and functional status. We were also unable to examine factors related to patient access to healthcare in these data, which may also contribute to disparity in care.
Moreover, this analysis of inpatient discharges does not consider patients who died at the scene or were discharged from the emergency department. Data from other sources – such as the National Violent Death Reporting System (NVDRS) and the National Death Index (NDI), both maintained by the Centers for Disease Control and Prevention (CDC) – that capture patients who die at the scene or are treated and released from emergency departments should supplement NIS data in future iterations of this analysis to more accurately reflect the trauma population and racial disparities faced therein.31,32 For example, disparities in deaths-at-scene, time to response, resource allocation, and triage decisions may be uncovered with this data. Finally, conclusions drawn from this study are nationally applicable. State-level analyses – which are useful to guide state-level changes to hospital resource and asset distribution, for example – require different and more specific sources of data. Lastly, while we were able to adjust for insurance, hospital rural/urban location, and zip code income quartile, we were unable to adjust for a variety of other relevant socioeconomic factors such as patient language spoken, educational level, and health practices.
Trauma in the United States is a disease that differs depending on your race. White and Black trauma patients differ by age, mechanism, socioeconomics, and the hospitals that treat them. Our study results suggest that for Black and White patients, the best mortality outcomes occur in hospitals that treat those patients most frequently. This suggests possible racial bias at the institution level, although these factors are complex and likely multifactorial. It is imperative to continue to examine the interplay between race and socioeconomic and demographic factors, such as insurance status, geographical location, and injury patterns, to isolate and challenge the specific etiologies and outcomes related to this racial disparity.29 Efforts must be made to promote equitable treatment for any patient who is injured.
Supplementary Material
Acknowledgements:
This publication was made possible by the Clinical and Translational Science Collaborative of Cleveland, KL2TR002547 from the National Center for Advancing Translational Sciences (NCATS) component of the National Institutes of Health and NIH roadmap for Medical Research. Its contents are solely the responsibility of the authors and do not necessarily represent the official views of the NIH.
Footnotes
Presentations:
This manuscript was presented at the 80th Annual Meeting of the American Association for the Surgery of Trauma (AAST) in Atlanta, Georgia, September 29 – October 2, 2021.
Conflicts of Interest:
Dr. Towe is a consultant for Medtronic, Zimmer Biomet, Atricure, and Astra Zeneca.
Dr. Ho’s spouse is a consultant for Medtronic, Zimmer Biomet, Atricure, and Astra Zeneca.
Dr. Kishawi, Dr. Tseng, Ms. Adomshick, and Dr. Ho have no conflicts of interest or financial ties to disclose.
References
- 1.Heron M Deaths: Leading Causes for 2018. Natl Vital Stat Rep. 2021;70(4):1–115. http://www.ncbi.nlm.nih.gov/pubmed/34029179. [PubMed] [Google Scholar]
- 2.Haider AH, Chang DC, Efron DT, Haut ER, Crandall M, Cornwell EE. Race and insurance status as risk factors for trauma mortality. Arch Surg. 2008;143(10):945–949. [DOI] [PubMed] [Google Scholar]
- 3.Moreira CC, Farber A, Rybin D, Doros G, Kalish J, Eberhardt RT, Siracuse JJ, Hamburg NM. Racial Differences in Treatment Approaches and Mortality Following Arterial Trauma. Vasc Endovascular Surg. 2015;49(7):180–187. [DOI] [PubMed] [Google Scholar]
- 4.Haider AH, Weygandt PL, Bentley JM, Monn MF, Rehman KA, Zarzaur BL, Crandall ML, Cornwell EE, Cooper LA. Disparities in trauma care and outcomes in the United States: A systematic review and meta-analysis. J Trauma Acute Care Surg. 2013;74(5):1195–1205. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Lucas FL, Stukel TA, Morris AM, Siewers AE, Birkmeyer JD. Race and surgical mortality in the United States. Ann Surg. 2006;243(2):281–286. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Morris AM, Billingsley KG, Baxter NN, Baldwin LM, Stamos MJ. Racial Disparities in Rectal Cancer Treatment: A Population-Based Analysis. Arch Surg. 2004;139(2):151–156. [DOI] [PubMed] [Google Scholar]
- 7.Birkmeyer JD, Siewers AE, Finlayson EV, Stukel TA, Lucas FL, Batista I, Welch HG, Wennberg DE. Hospital Volume and Surgical Mortality in the United States. N Engl J Med. 2002;346(15):1128–1137. [DOI] [PubMed] [Google Scholar]
- 8.Birkmeyer JD, Stukel TA, Siewers AE, Goodney PP, Wennberg DE, Lucas FL. Surgeon Volume and Operative Mortality in the United States. N Engl J Med. 2003;349(22):2117–2127. [DOI] [PubMed] [Google Scholar]
- 9.Skinner J, Chandra A, Staiger D, Lee J, McClellan M. Mortality after acute myocardial infarction in hospitals that disproportionately treat black patients. Circulation. 2005;112(17):2634–2641. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Howell EA, Egorova N, Balbierz A, Zeitlin J, Hebert PL. Black-white differences in severe maternal morbidity and site of care. Am J Obstet Gynecol. 2016;214(1):122.e1–122.e7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Jha AK, Orav EJ, Li Z, Epstein AM. Concentration and quality of hospitals that care for elderly black patients. Arch Intern Med. 2007;167(11):1177–1182. [DOI] [PubMed] [Google Scholar]
- 12.Osler T, Glance LG, Li W, Buzas JS, Wetzel ML, Hosmer DW. Trauma care does not discriminate: The association of race and health insurance with mortality following traumatic injury. J Trauma Acute Care Surg. 2015;78(5):1026–1033. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.National Inpatient Sample. Healthcare Cost and Utilization Project (HCUP). http://www.hcup-us.ahrq.gov/nisoverview.jsp. Agency Healthc Res Qual. 2017. https://www.hcup-us.ahrq.gov/nisoverview.jsp. [Google Scholar]
- 14.Clark DE, Black AW, Skavdahl DH, Hallagan LD. Open-access programs for injury categorization using ICD-9 or ICD-10. Inj Epidemiol. 2018;5(1):11. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Elixhauser A, Steiner C, Harris DR, Coffey RM. Comorbidity Measures for Use with Administrative Data. Med Care. 1998;36(1):8–27. [DOI] [PubMed] [Google Scholar]
- 16.Ly DP, Lopez L, Isaac T, Jha AK. How do black-serving hospitals perform on patient safety indicators?: Implications for national public reporting and pay-for-performance. Med Care. 2010;48(12):1133–1137. [DOI] [PubMed] [Google Scholar]
- 17.Shavers VL, Fagan P, Jones D, Klein WM, Boyington J, Moten C, Rorie E. The state of research on racial/ethnic discrimination in the receipt of health care. Am J Public Health. 2012;102(5):953–966. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Falcone RA, Martin C, Brown RL, Garcia VF. Despite overall low pediatric head injury mortality, disparities exist between races. J Pediatr Surg. 2008;43(10):1858–1864. [DOI] [PubMed] [Google Scholar]
- 19.Chen H, Scarborough J, Zens T, Brummeyer B, Agarwal S, Haines KL. Race and Insurance Status as Predictors of Bicycle Trauma Outcome in Adults. J Surg Res. 2020;245:198–204. [DOI] [PubMed] [Google Scholar]
- 20.Singer MB, Liou DZ, Clond MA, Bukur M, Mirocha J, Margulies DR, Salim A, Ley EJ. Insurance-and race-related disparities decrease in elderly trauma patients. J Trauma Acute Care Surg. 2013;74(1):312–316. [DOI] [PubMed] [Google Scholar]
- 21.Hicks CW, Hashmi ZG, Velopulos C, Efron DT, Schneider EB, Haut ER, Cornwell EE 3rd, Haider AH. Association between race and age in survival after trauma. JAMA Surg. 2014;149(7):642–647. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Haider AH, Ong'uti S, Efron DT, Oyetunji TA, Crandall ML, Scott VK, Haut ER, Schneider EB, Powe NR, Cooper LA, et al. Association between hospitals caring for a disproportionately high percentage of minority trauma patients and increased mortality: A nationwide analysis of 434 hospitals. Arch Surg. 2012;147(1):63–70. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Fu CY, Bajani F, Tatebe L, Butler C, Starr F, Dennis A, Kaminsky M, Messer T, Schlanser V, Kramer K, et al. Right hospital, right patients: Penetrating injury patients treated at high-volume penetrating trauma centers have lower mortality. J Trauma Acute Care Surg. 2019;86(6):961–966. [DOI] [PubMed] [Google Scholar]
- 24.Tang A, Chehab M, Ditillo M, Asmar S, Khurrum M, Douglas M, Bible L, Kulvatunyou N, Joseph B. Regionalization of trauma care by operative experience: Does the volume of emergent laparotomy matter? J Trauma Acute Care Surg. 2021;90(1):11–20. [DOI] [PubMed] [Google Scholar]
- 25.Lipsky AM, Karsteadt LL, Gausche-Hill M, Hartmans S, Bongard FS, Cryer HG, Ekhardt PB, Loffredo AJ, Farmer PD, Whitney SC, et al. A comparison of rural versus urban trauma care. J Emergencies, Trauma Shock. 2014;7(1):41–46. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Alsan M, Garrick O, Graziani G. Does Diversity Matter for Health? Experimental Evidence from Oakland. Vol 109. Cambridge, MA; 2019. doi: 10.1257/aer.20181446 [DOI] [Google Scholar]
- 27.Martin KD, Roter DL, Beach MC, Carson KA, Cooper LA. Physician communication behaviors and trust among black and white patients with hypertension. Med Care. 2013;51(2):151–157. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Hofmann LJ, Lee S, WaddelL B, Davis KG. Effect of race on colon cancer treatment and outcomes in the department of defense healthcare system. Dis Colon Rectum. 2010;53(1):9–15. [DOI] [PubMed] [Google Scholar]
- 29.Kahn KL, Pearson ML, Harrison ER, Desmond KA, Rogers WH, Rubenstein LV, Brook RH, Keeler EB. Health Care for Black and Poor Hospitalized Medicare Patients. JAMA J Am Med Assoc. 1994;271(15):1169–1174. [PubMed] [Google Scholar]
- 30.Newgard CD, Fu R, Bulger E, Hedges JR, Mann NC, Wright DA, Lehrfeld DP, Shields C, Hoskins G, Warden C, et al. Evaluation of Rural vs Urban trauma patients served by 9-1-1 emergency medical services. JAMA Surg. 2017;152(1):11–18. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Centers for Disease Control and Prevention. National Violent Death Reporting System (NVDRS). https://www.cdc.gov/violenceprevention/datasources/nvdrs/index.html. [Google Scholar]
- 32.Centers for Disease Control and Prevention. National Death Index. https://www.cdc.gov/nchs/ndi/index.htm. [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.