Secondary Logo

Journal Logo

RESEARCH

The Association of Race, Socioeconomic Status, and Insurance on Trauma Mortality

Mikhail, Judy N. PhD, MSN, MBA, RN; Nemeth, Lynne S. PhD, RN; Mueller, Martina PhD; Pope, Charlene PhD, MPH, RN; NeSmith, Elizabeth G. PhD, APRN-BC; Wilson, Kenneth L. MD; McCann, Michael DO; Fakhry, Samir M. MD

Author Information
doi: 10.1097/JTN.0000000000000246
  • Free

Abstract

Trauma is the leading cause of death in the United States for age groups from 1 to 44 years (Centers for Disease Control and Prevention, 2012). However, injury is not distributed equally across society and unduly affects minorities, poor, and young populations. A recent meta-analysis of studies assessing the effects of race, insurance, and socioeconomic status (SES) on trauma revealed consistent evidence of disparities including that Black patients were 19% more likely to die after trauma than White patients, uninsured patients were twice as likely to die than insured patients, and that median income as a measure of SES is predictive of higher mortality rate (Haider et al., 2013). Because race is highly correlated with both SES and insurance status, it is crucial to determine whether racial disparities are independent of SES and insurance disparities. Haider et al. (2013) found mixed results in that some studies comparing Black and White trauma patients, while adjusting for insurance status, revealed that Black race predicts mortality independent of SES and other studies found that insurance has a stronger predictive value on mortality than race. Previous trauma disparities research has overwhelmingly focused on race, insurance, and SES in isolation of each other and often without proper risk adjustment methodology that is essential to control for case mix to allow examination of a single risk factor while controlling for all others (Haider et al., 2014; Shine, 2012). It is hypothesized that patients experiencing multiple forms of disadvantage (minority race, uninsured, low SES) experience greater trauma mortality than when each disadvantage is studied separately. The purpose of this study was to determine the association of race, SES, and insurance on trauma mortality in a single trauma center with a racially and socioeconomically diverse population.

METHODS

This is a retrospective, cross-sectional analysis of trauma patients from Hurley Medical Center, a 465-bed, American College of Surgeons-verified Level I trauma center in Flint, MI, between 2000 and 2009. Included patients were aged 18–64 years with moderate to severe injuries, defined by an injury severity score (ISS) of 9 or greater, caused by blunt or penetrating trauma. This age range was chosen to focus on the population with the maximum burden of injury and avoid confounding issues associated with extremes of age. Burn patients, patients pronounced dead on arrival, and transferred patients were excluded on the basis of known differences in mortality rate among these populations. The study was approved by the Hurley Medical Center institutional review board.

Demographic data were assessed by patient age, sex, race, SES, and insurance. Patient race was grouped by self-reported Black and White race, with all other race/ethnicities excluded because of limited numbers. Insurance was collected at patient discharge and grouped as Private (commercial), Uninsured (private pay/not billed), Medicare, Medicaid, No-Fault (State mandated autoinsurance coverage, regardless of fault), and Other (Tricare (military), Managed Care, and Workman's Compensation). Using techniques previously described (Diez Roux et al., 2001; Birkmeyer, Gus, Baser, Morris, & Birkmeyer, 2008), SES was assessed using a summary measure of data on income, education, and occupation for each zip code from the 2000 U.S. Census that was linked to the patient's zip code of residence at the time of the injury obtained from the hospital record. In brief, the summary measure consists of six census variables, which include three measures of wealth/income, two measures of education, and one measure of occupation/employment, for which a z-score for each zip code was established and then summed for each of the six variables, and sorted into quintiles grouped as low, middle, and high. For analysis purposes, the middle and high SES groups were combined for comparison with the low SES group and the uninsured were grouped as an insurance type. The primary outcome measure was inpatient mortality. Missing data ranged from 1.7% for systolic blood pressure and to 11.2% for insurance status. Missing data were addressed by multiple imputation, a validated statistical technique that replaces missing data with substituted values (Oyetunji et al., 2011).

Statistical significance was set at p = .05. Logistic regression modeling was performed, initially with simple regression analysis modeling each independent variable (SES, insurance, race) on mortality. A second level of models included SES, race, and insurance status in pairs, with race and insurance included in the first model, race and SES in a second model, and insurance and SES in a third model. In a subsequent step, the same models were used with an additional interaction term including (a) race-by-insurance; (b) race-by-SES; and (c) insurance-by-SES. The previously developed models were then adjusted for a set of covariates known to predict trauma mortality including age, sex, Glasgow motor score (GCS-M), hypotension, ISS, head Abbreviated Injury Score (AIS), and mechanism of injury (Haider et al., 2012) as described by Cohen and colleagues (Cohen, Cohen, West, & Aiken, 2002). The final model included the three independent variables as well as the set of clinical covariates. Because of a lack of valid comorbidity data in the existing trauma data set, a sensitivity analysis focusing on the 18- to 40-year-old age group was performed to investigate the effect of confounding by comorbidities (younger patients tend to have fewer comorbidities). Results are presented as odds ratios (ORs), and p values, with 95% confidence intervals. Statistical Package for the Social Sciences for Windows, Version 22 (IBM SPSS, New York), was used for all statistical analysis.

RESULTS

A total of 4,007 adult trauma patients aged 18–64 years were admitted during the 10-year study period from 2000 to 2009. The population was predominantly male (75%), with a mean age of 37.8 years (median, 37.5 years). White and Black patients comprised 64% and 36%, respectively, of the sample. Most of the patients (78%) had blunt trauma, whereas penetrating injuries accounted for the remaining 22%. The median ISS was 13.0, mean GCS-M was 5.29, and 9.9% of the population was hypotensive on admission. Socioeconomic status was divided into low SES and non-low SES, which was 44.2% and 55.8%, respectively. Insured patients accounted for 68.6% of the population. Insurance by type was 31.4% Uninsured, 25.4% Private, 16.9% Medicaid, 14.3% No-Fault, 5.9% Medicare, and 6.0% Other. Overall in-hospital mortality rate was 10%. As expected, patients with a higher ISS, penetrating injury, lower motor GCS, and higher head AIS were more likely to die (Table 1). Regression diagnostics revealed no collinearity among covariates. Interaction effects were tested between race and insurance, race and SES, and insurance and SES; both alone and with the clinical covariates and no interaction effects were noted.

TABLE 1 - Population Characteristics by Mortality
Overall (N = 4,007) Discharged (n = 3,599) (89.8%) Died (n = 404) (10.1%) p
Age, mean (SD), year 37.86 (12.8) 38.07 (12.8) 36.14 (12.7) .004
Sex .000
 Male (%) 3,019 (75.3) 2,681 (88.9) 335 (11.1)
 Female (%) 988 (24.7) 918 (93.0) 69 (7.0)
Race .000
 White (%) 2,581 (64.4) 2,358 (65.5) 220 (54.5)
 Black (%) 1,426 (35.6) 1,241 (34.5) 184 (45.5)
Mechanism of injury .000
 Blunt (%) 3,127 (78.0) 2,893 (80.4) 231 (57.2)
 Penetrating (%) 880 (22.0) 706 (19.6) 173 (42.8)
Physiologic/anatomic severity
 Injury Severity Score, mean (SD) 17.39 15.52 (8.5) 33.98 (16.9) .000
 Injury Severity Score .000
 9–15 Moderate injury 2,222 (55.5) 2,197 (61.0) 25 (6.2)
 6–24 Moderately severe 954 (23.8) 894 (24.8) 59 (14.6)
 ≥25 Severe 828 (20.7) 508 (14.2) 320 (79.2)
 Systolic blood pressure, mean (SD) mmHg 128.42 (40.6) 135.54 (28.3) 61.03 (68.5) .000
 Hypotensive on arrival (%) 397 (9.9) 171 (4.8) 226 (59.9) .000
 Glasgow Motor Score, mean (SD) 5.29 (1.65) 5.67 (1.1) 1.83 (1.7) .000
 Glasgow Motor Score .000
 High Function (6) 3,200 (81.0) 3,153 (88.6) 47 (12.0)
 Moderate Function (2–5) 282 (7.1) 246 (6.9) 36 (9.2)
 Low Function (1) 471 (11.9) 161 (4.5) 310 (78.9)
 Severe Head Injury (AIS ≥ 3) (%) 2.96 (1.17) 1,042 (53.7) 226 (94.2) .000
Insurance status .000
 Uninsured (%) 1,260 (31.4) 1,039 (28.9) 220 (54.5)
 Private (%) 1,019 (25.4) 956 (26.6) 62 (15.3)
 Medicaid (%) 679 (16.9) 641 (17.8) 36 (8.9)
 No-Fault (%) 572 (14.3) 547 (15.2) 25 (6.2)
 Medicare (%) 235 (5.9) 211 (5.9) 24 (5.9)
 Other (Tricare, MC, WC) 242 (6.0) 205 (5.7) 37 (9.2)
Socioeconomic status .028
 Low (%) 1,770 (44.2) 1,569 (43.6) 199 (49.4)
 Non-low (%) 2,233 (55.7) 2,027 (56.4) 204 (50.6)
Note. AIS = abbreviated injury score; BCBS = Blue Cross Blue Shield; MC = Managed Care; SD = standard deviation; WC = Workman's Compensation.

Bivariate analysis revealed 10 characteristics significantly associated with mortality, including race, insurance, SES, age, gender, ISS, GCS-M, head AIS, mechanism of injury, and hypotension (Table 1). Black patients were 1.5 times more likely to die than White patients (p = .000), low SES patients were 1.2 times more likely to die than patients with non-low SES (p = .02), and the uninsured had 3.2 greater odds of dying than privately insured patients (p = .000) (Table 2). When race was combined with insurance status, their relationships with mortality held; while SES was no longer statistically significantly associated with mortality when included in models with either variable. The final model, including race, SES, insurance and clinical covariates, found that insurance type remained statistically significant, with Medicare (OR = 3.63; p = .006) and Other Insurance (OR = 3.02; p = .007) predictive of mortality compared with Private patients (Table 3). Neither race nor SES contributed to mortality in the final model although both were significantly associated with mortality when considered individually.

TABLE 2 - Individual Regression Analyses of Race, SES, and Insurance by Mortality
OR 95% CI p
Race (reference: White)
 Black 1.59 [1.29, 1.95] .000
SES (reference: non-Low)
 Low 1.26 [1.02, 1.54] .020
Insurance (reference: private)
 None 3.26 [2.43, 4.38] .000
 Medicare 1.75 [1.07, 2.87] .020
 No-Fault 0.70 [0.43, 1.13] .150
 Medicaid 0.86 [0.56, 1.32] .500
 Other 2.78 [1.80, 4.29] .000
Note. CI = confidence interval; OR = odds ratio.

TABLE 3 - Final Logistic Regression Model for Mortality
OR 95% CI p
Race (reference: White)
 Black 1.07 [0.61, 1.89] .790
SES (reference: non-Low)
 Low 1.16 [0.69, 1.92] .560
Insurance (reference: private)
 None 1.75 [0.96, 3.19] .060
 Medicare 3.63 [1.44, 9.10] .000
 No-Fault 0.49 [0.22, 1.09] .080
 Medicaid 1.56 [0.72, 3.41] .250
 Other 3.02 [1.36, 6.73] .000
 Age 1.02 [1.00, 1.04] .010
 Injury Severity Score 1.06 [1.03, 1.08] .000
 Glasgow Motor Score 0.56 [0.50, 0.62] .000
 Head Abbreviated
   Injury Score
1.61 [1.28, 2.01] .000
 Mechanism of injury 5.74 [2.78, 11.84] .000
 Hypotension 4.80 [2.78, 8.29] .000
Gender (reference: Female)
 Male 1.29 [0.76, 2.18] .330
Note. CI = confidence interval; OR = odds ratio.

While Blacks made up only 35.5% of the total trauma center population, more Blacks compared were Whites were low SES (73% vs. 28.3%; p = .000), Uninsured (42.1% vs. 25.5%, p = .000), had penetrating injury (46.2% vs. 8.6%; p = .000), severe head injury (63.6% vs. 56.1%, p = .000), were hypotensive (14.3% vs. 7.7%; p = .001), and suffered in-hospital mortality (12.9% vs. 8.5%; p = .000). Patients with more Low SES compared with non-low SES patients were hypotensive (12.4% vs. 8.2%; p = .000) and had penetrating injury (33.6% vs. 12.8%; p = .000). Hospital mortality by insurance type ranged from 4.4% for No-Fault to 17.5% for the Uninsured group, with Other insurance ranking second highest for mortality at 15.3%. The two insurance categories with the highest percentage of Whites were Private (79.2%) and No-Fault (78.3%), while the two insurance categories with the highest percentage of Blacks were Medicaid (49.6%) and Uninsured (47.7%). Mortality was inversely associated with age (OR = 1.01 [1/0.988]; P = .004) and corresponded with higher percentages of penetrating injury (31.8%, 32.2%), lower GCS-M (14.1%; 13.7%), and hypotension (10.3%; 12.5%) seen among the younger age groups of 18–34 years and 25–34 years, respectively.

Further examination of different model variations of the Uninsured group revealed a trending toward significance when compared with Private insured (Table 4). For example, when SES and Gender were removed from the full model, the Uninsured group became statistically significant for its contribution to mortality at (OR = 1.81; P = .048); this relationship was even stronger when race was removed in addition to SES and gender (OR = 1.83; p = .043). A sensitivity analysis comparing patients aged 18–40 versus 41–64 (Table 5) revealed that the influence of insurance status remains even when analyzing younger patients who likely have minimal comorbidity. However, Uninsured and Other insurance were statistically significantly associated with mortality only in the younger age group, while perhaps not surprisingly, Medicare remained statistically significant only for the older age group. The clinical variables remained statistically significant in both age groups. Imputation of incomplete data resulted in similar results in all analyses.

TABLE 4 - Model Variations of the Uninsured Group
1 2 3 4 5 6 7 8
Insurance Insurance Insurance Insurance Insurance Insurance Insurance Insurance
Race Race Race Race
SES SES SES SES
Gender Gender Gender Gender
Age Age Age Age Age Age Age Age
ISS ISS ISS ISS ISS ISS ISS ISS
GCS-M GCS-M GCS-M GCS-M GCS-M GCS-M GCS-M GCS-M
HeadAIS HeadAIS HeadAIS HeadAIS HeadAIS HeadAIS HeadAIS HeadAIS
Mechanism Mechanism Mechanism Mechanism Mechanism Mechanism Mechanism Mechanism
Hypotension Hypotension Hypotension Hypotension Hypotension Hypotension Hypotension Hypotension
OR 1.75 1.19 1.77 1.80 1.80 1.79 1.81 1.83
CI 0.96; 3.19 0.75, 1.89 0.97, 3.21 0.99, 3.27 0.99, 3.27 0.99, 3.25 1.00, 3.28 1.01, 3.31
p .066 .451 .059 .053 .053 .052 .048 .043
−2 LL 586.54 589.74 590.64 590.63 590.70 590.90 591.56 591.81
Note. AIS = Abbreviated Injury Score; CI = confidence interval; GCS-M = Glasgow Motor Score; ISS = Injury Severity Score; −2LL = Log Likelihood; SES = socioeconomic status.

TABLE 5 - Sensitivity Analysis of Age Groups Comparisons (18–40 Years) and (41–64 Years)
Age 18–40 Years Age 41–64 Years Age 18–64 Years
OR 95% CI p OR 95% CI p OR 95% CI p
Insurance (REF: private)
 Uninsured 2.36 [0.98, 5.71] .05 1.25 [0.53, 2.95] .60 1.75 [0.96, 3.19] .06
 Medicare 1.14 [0.18, 6.97] .88 4.49 [1.54, 13.0] .006 3.63 [1.44, 9.10] .00
 No-Fault 0.65 [0.21, 1.99] .45 0.366 [0.10, 1.23] .10 0.49 [0.22, 1.09] .08
 Medicaid 2.30 [0.66, 8.01] .18 1.13 [0.40, 3.18] .81 1.56 [0.72, 3.41] .25
 Other 5.96 [1.8, 19.78] .003 1.39 [0.43, 4.44] .57 3.02 [1.36, 6.73] .00
SES (REF: Non-Low)
 Low 1.16 [0.54, 2.47] .70 1.02 [0.50, 2.09] .94 1.16 [0.69, 1.92] .56
Race (REF: White)
 Black 1.14 [0.47, 2.75] .76 0.84 [0.38, 1.85] .67 1.07 [0.61, 1.89] .79
Gender (REF: Female)
 Male 1.09 [0.49, 2.46] .81 1.32 [0.65, 2.68] .43 1.29 [0.76, 2.18] .330
Injury Severity Score 1.05 [1.02, 1.08] .000 1.06 [1.03, 1.10] .000 1.06 [1.03, 1.08] .000
Head AIS Score 1.52 [1.10, 2.11] .010 1.65 [1.19, 2.30] .003 1.61 [1.28, 2.01] .000
Glasgow Motor Score 0.51 [0.43, 0.61] .000 0.62 [0.52, 0.71] .000 0.56 [0.50, 0.62] .000
Mechanism of injury 4.59 [1.74, 12.1] .002 8.58 [2.34, 31.39] .001 5.74 [2.78, 11.84] .000
Hypotension 7.00 [3.07, 15.95] .000 4.47 [2.01, 9.94] .000 4.80 [2.78, 8.29] .000
Note. AIS = Abbreviated Injury Score; CI = confidence interval; OR = odds ratio; REF = Reference; SES = socioeconomic status.

DISCUSSION

This analysis of adult trauma patients demonstrates that when all three variables are modeled in combination with clinical variables, insurance type, not race, or SES, is associated with different mortality outcomes and varies with age. Compared with Private insurance, only Medicare and Other Insurance contributed to mortality while the Uninsured, No-Fault, and Medicaid insurances demonstrated no difference. This study's results differ from previous research that grouped insurances into two or three broad categories and consistently found that lack of insurance predicts trauma mortality (Bell & Zarzaur, 2013; Crandall et al., 2011; Dozier et al., 2010; Falor et al., 2014; Haider et al., 2008; Maybury et al., 2010; Salim et al., 2010; Schoenfeld, Belmont, See, Bader, & Bono, 2013; Short et al., 2013; Tepas, Pracht, Orban, & Flint, 2011). In contrast, we found that our results more closely resemble the only other study to use expanded insurance types. Weygandt et al. (2012), using the National Trauma Data Bank, examined the effect of 10 insurance types on adult (ages 18–64 years) blunt trauma. Their results partially mirrored our own in that Medicare and Other Government Insurance (Tricare, Supplemental Security Income, and Federal Employees) predicted mortality while Medicaid was protective. Our study confirms their previous observation that how insurances are grouped yield varying associations with trauma mortality. Therefore, grouping insurances may not provide the granularity necessary to study trauma outcomes by payer status. Medicare predicted higher mortality in our study, which was expected because of eligibility requirements (suffering a disabling injury or illness), given that our sample was restricted to ages 18–64 years. Medicaid was found to be protective in our study and is likely related to the timing of insurance data collection, which was performed on patient discharge, possibly creating survival bias. Most trauma centers aggressively enroll uninsured patients during hospitalization with Medicaid conversion rates ranging from 2.5% to 9% following injury (Zaloshnja, Miller, Coben, & Steiner, 2012).

Further analysis of the Other insurance group revealed that it was predictive of mortality in the younger (18–40 years) but not older (41–64 years) age group (Table 5). Additional subgroup analysis (Table 6) revealed that while the Tricare group was small (n = 24), it was also deadly. It was significantly younger (62.5%), male (79.2%), and White (91.7%), with a higher percentage of severe head injury as noted by GCS motor score of 1 (50.0%), Head AIS ≥ 3 (89.5%), and a mortality rate of 66.6% (P = .000). Suicide e-codes were reported in only two of the 24 cases (8.3%). Therefore, the lethality of the Tricare group leads to the possibility that former military is at higher risk for subsequent lethal injury as civilians. Preliminary evidence suggests that veterans are at increased risk for engaging in high risk and/or violent behavior following deployment (Killgore et al., 2008; Macmanus et al., 2013). To our knowledge, this has yet to be studied and is a research gap.

TABLE 6 - Subgroup Analysis of Other Insurance
Other Insurance (N = 242) (6.0% of Total)
Variable Managed Care (N = 121) (3%) Workman's Compensation (N = 97) (2.4%) Tricare (Military) (N = 24) (0.6%) p
Discharged (%) 104 (86.0) 93 (95.9) 8 (33.3) .000
Died (%) 17 (14.0) 4 (4.1) 16 (66.6) .000
Head AIS ≥3 (%) 46 (59.0) 27 (69.2) 17 (89.5) .000
GCS-M = 1 (%) 23 (19.3) 7 (7.3) 12 (50.0) .000
Hypotension (%) 18 (15.3) 4 (4.2) 1 (4.2) .000
Penetrating (%) 21 (17.4) 6 (6.2) 1 (4.2) .000
Age (Years) .000
 18–35 57 (48.3) 51 (53.1) 15 (62.5)
 36–64 61 (51.7) 45 (46.9) 9 (37.5)
SES .000
 Low 56 (46.3) 24 (24.7) 5 (20.8)
 Non Low 65 (53.7) 73 (75.3) 19 (79.2)
Race .000
 White 76 (62.8) 83 (85.6) 22 (91.7)
 Black 45 (37.2) 14 (14.4) 2 (8.3)
Gender .000
 Female 46 (38.0) 13 (13.4) 5 (20.8)
 Male 75 (62.0) 84 (86.6) 19 (79.2)
Note. AIS = Abbreviated Injury Score; GCS-M = Glasgow Motor Score; SES = socioeconomic status.

To our knowledge, this is the first trauma study to use multiple area-level measures of SES with multiple insurance types. We used a composite of six census area-level SES measures, including three measures of wealth/income, two measures of education, and one measure of occupation/employment. While multiple dimensions of SES, including both individual and area-based measures, are preferred (Cubbin & Smith, 2002), area-based measures are commonly accepted as valid measures of SES (Bell, Arrington, & Adams, 2015; Signorello et al., 2014). When analyzed together, our study found that race and SES were not significant contributors to mortality while insurance type was. Our results differ from previous research that found that SES is related to racial disparities in injury risk (Braver, 2003; Cubbin, LeClere, & Smith, 2000; Ladha, Young, Ng, Efron, & Haider, 2011) or that controlling for SES lessens but does not abolish racial disparities in injury (Fabio, Sauber-Schatz, Barbour, & Li, 2009; Steenland, Halperin, Hu, & Walker, 2003). Varying results are not unexpected as race-specific SES differences are known to be influenced by the underlying study population race and SES homogeneity (Aldridge, Canavan, Cherlin, & Bradley, 2015; Signorello et al., 2014). Perhaps not surprisingly, our study finding that both race and SES were nonsignificant predictors of mortality when combined with clinical variables was performed in a setting where race-specific SES differences were minimal (Braveman et al., 2011; Signorello et al., 2007). Past trauma disparities research has suffered from inadequate risk adjustment, grouped race/ethnicity and/or insurances, and the lack of logistic regression and interaction analysis between race, insurance, SES, and injury, all of which can contribute to the perpetuation of racial stereotypes (Adler, Bush, & Pantell, 2012; Cheng, Goodman, & Committee on Pediatric, 2015).

Few studies have been performed on the effect of race, SES, and insurance in trauma using multivariable regression analysis with the covariates necessary for mortality risk adjustment. Five studies were identified for comparison to our study, and all used the U.S. Census median household income by zip code as the area-level SES measure. Of the five studies, the first two support our study findings. Hazlitt, Hill, Gunter, and Guillamondegui (2013), using a single center registry with good risk adjustment and three insurance groups, reported that in traumatic hollow viscus injuries, mortality was not associated with race, SES, or insurance. Similarly, Taghavi et al. (2012), using a single center registry with fair risk adjustment (no shock or head injury severity scoring), and two insurance groups, reported that in penetrating trauma, mortality was not associated with race, SES, or insurance. Rangel, Burd, Falcone, and Multicenter Child Abuse Disparity Group (2010), in a multicenter study of infant child abuse using adequate risk adjustment, with two insurance groups, found that SES and insurance, not race, predicted mortality. The remaining two large studies by Arthur, Hedges, Newgard, Diggs, and Mullins (2008) and Ali et al. (2013) used the Nationwide Inpatient Sample (NIS) administrative data set, which is devoid of the physiologic injury severity indices necessary for robust risk adjustment, and found mixed results. Arthur et al. (2008) examined adult trauma patients, using six insurance groups, and found that while controlling for SES and insurance, race predicted mortality with Blacks and Asians more likely to die than Whites, whereas Ali et al. (2013) examined adult trauma patients also with six insurance groups and found that after controlling for race and insurance, SES predicted mortality. Comparisons are made difficult by the heterogeneity within and across trauma studies, including populations, ages, mechanism of injury, insurance grouping, race/ethnicity grouping, the variety, and validity of SES constructs, such as the use of race/ethnicity and/or insurance as measures of SES (Kaplan & Bennett, 2003).

Another important finding of our study is the identification that the association of insurance type with mortality varies with age (Table 5). In the full model, Medicare and Other were the only two insurances that predicted mortality. However, when stratified into age groups of 18–40 and 41–64 years, the insurance influence varied by age. Uninsured and Other insurance groups drove mortality in the younger age group, while Medicare drove mortality in the older age group. These findings somewhat complement emerging research that demonstrates that racial disparities in trauma mortality appear to be age dependent, where insurance status is an included covariate (Hicks et al., 2014,2015; Singer et al., 2013), whereas our inclusion of insurance type as a covariate using multiple insurance categories may explain why we found that the association of insurance type with mortality varies by age, not race. Future studies should examine disparities across multiple age groups and insurance types.

This study has some important limitations. The single-center retrospective nature of the study limits generalizability. The cross-sectional study design, while helpful in identifying associations for further study, precludes conclusions regarding underlying disparities etiologies. There is likely trauma center variation in insurance capture due to both the timing of data collection and the coding of insurance because there is no established national standard for grouping insurances. The use of zip code measures of SES suffers more from within-unit variation than Census tract measures and increases the possibility of aggregation bias (Bilheimer & Klein, 2010). Finally, we were unable to adjust for comorbidities, which are known confounders of trauma outcomes (Patel, Malinoski, Nguyen, & Hoyt, 2011) and frequently suffer from inconsistent data capture within trauma registeries (Arabian et al., 2015; Dente et al., 2016).

CONCLUSION

Understanding how race, SES, and insurance interact to create disparities in trauma is critical to erasing them. By using detailed insurance-type information and a composite of multiple measures of SES, this analysis demonstrates that when analyzed in combination, specific insurance types, not race, or SES, are associated with trauma mortality and that this relationship varies with age. It is increasingly apparent that one size does not fit all in the study of trauma disparities. Disparities likely vary across trauma populations and by the sensitivity of the measures used to study them. Grouping insurances and/or race/ethnicities is discouraged as it may mask underlying disparities. For example, Tricare patients may represent an emerging vulnerable population at risk for trauma recidivism. The importance of our study finding that insurances' association with trauma mortality held over that of multiple census measures of SES, including income, education, and occupation, is informative. Insurance appears to capture unmeasured confounders of preinjury health status in trauma patients that perhaps traditional SES measures cannot. Future trauma disparities work should expand measures of SES beyond median income to consider a range of measures to capture life-course adversity and structural disadvantage including percentage of population below the poverty line (Krieger, Waterman, Chen, Soobader, & Subramanian, 2003; Subramanian, Chen, Rehkopf, Waterman, & Krieger, 2006) area unemployment rate, percentage of lone parent families, percentage of non–high school graduates (Bell et al., 2015), the Gini coefficient (income inequality), and the Dissimilarity Index (racial residential segregation) (Nuru-Jeter & LaVeist, 2011). Future research should also consider other potential confounders including time to death, patient comorbidities, frailty, both admission and discharge insurances, as well as provider bias and hospital-level organizational factors and quality.

KEY POINTS

  • This trauma disparities study showed that when race, multiple census measures of SES, and insurance types are analyzed together, specific insurances, not race, or SES, are associated with trauma mortality.
  • The association of insurance type with trauma mortality varies with age, with the uninsured and Tricare insurance driving mortality in the younger age group and Medicare driving mortality in the older age group.
  • Military veterans as civilian trauma patients may represent an emerging vulnerable population at risk for trauma recidivism and death, which requires further study.

REFERENCES

Adler N., Bush N. R., Pantell M. S. (2012). Rigor, vigor, and the study of health disparities. Proceedings of the National Academy of Sciences of the United States of America, 109(Suppl. 2), 17154–17159. doi:10.1073/pnas.1121399109
Aldridge M., Canavan M., Cherlin E., Bradley E. (2015). Has hospice use changed? 2000–2010 utilization patterns. Medical Care, 53(1), 95–101.
Ali M. T., Hui X., Hashmi Z. G., Dhiman N., Scott V. K., Efron D. T., Haider A. H. (2013). Socioeconomic disparity in inpatient mortality after traumatic injury in adults. Surgery, 154(3), 461–467. doi:10.1016/j.surg.2013.05.036
Arabian S. S., Marcus M., Captain K., Pomphrey M., Breeze J., Wolfe J., Rabinovici R. (2015). Variability in interhospital trauma data coding and scoring: A challenge to the accuracy of aggregated trauma registries. The Journal of Trauma and Acute Care Surgery, 79(3), 359–363. doi:10.1097/TA.0000000000000788
Arthur M., Hedges J. R., Newgard C. D., Diggs B. S., Mullins R. J. (2008). Racial disparities in mortality among adults hospitalized after injury. Medical Care, 46(2), 192–199.
Bell N., Arrington A., Adams S. A. (2015). Census-based socioeconomic indicators for monitoring injury causes in the USA: a review. Injury Prevention, 21(4), 278–284. doi:10.1136/injuryprev-2014-041444
Bell T. M., Zarzaur B. L. (2013). Insurance status is a predictor of failure to rescue in trauma patients at both safety net and non-safety net hospitals. The Journal of Trauma and Acute Care Surgery, 75(4), 728–733.
Bilheimer L. T., Klein R. J. (2010). Data and measurement issues in the analysis of health disparities. Health Services Research, 45(5, Pt 2), 1489–1507. doi:10.1111/j.1475-6773.2010.01143.x
Birkmeyer N. J., Gus N., Baser O., Morris A. M., Birkmeyer J. D. (2008). Socioeconomic status and surgical mortality in the elderly. Medical Care, 46, 893–899.
Braveman P. A., Kumanyika S., Fielding J., LaVeist T., Borrell L. N., Manderscheid R., Troutman A. (2011). Health disparities and health equity: The issue is justice. American Journal of Public Health, 101(S1), S149–S155. doi:10.2105/ajph.2010.300062
Braver E. R. (2003). Race, Hispanic origin, and socioeconomic status in relation to motor vehicle occupant death rates and risk factors among adults. Accident Analysis and Prevention, 35(3), 295–309. doi:10.1016/s0001-4575(01)00106-3
Centers for Disease Control and Prevention. (2012). Youth violence facts at a glance. Retrieved from http://www.cdc.gov/ViolencePrevention/pdf/YV-DataSheet-a.pdf
Cheng T. L., Goodman E., & the Committee on Pediatric Research. (2015). Race, ethnicity, and socioeconomic status in research on child health. Pediatrics, 135(1), e225–e237. doi:10.1542/peds.2014-3109
Cohen J., Cohen P., West S. G., Aiken L. S. (2002). Applied multiple regression/correlation analysis for the behavioral sciences (3rd ed.). New York, NY: Routledge Taylor & Francis Group.
Crandall M., Sharp D., Brasel K., Carnethon M., Haider A., Esposito T. (2011). Lower extremity vascular injuries: Increased mortality for minorities and the uninsured? Surgery, 150(4), 656–664. doi:10.1016/j.surg.2011.07.052
Cubbin C., LeClere F. B., Smith G. S. (2000). Socioeconomic status and the occurrence of fatal and nonfatal injury in the United States. American Journal of Public Health, 90(1), 70–77.
Cubbin C., Smith G. S. (2002). Socioeconomic inequalities in injury: Critical issues in design and analysis. Annual Review of Public Health, 23, 349–375. doi:10.1146/annurev.publhealth.23.100901.140548
Dente C. J., Ashley D. W., Dunne J. R., Henderson V., Ferdinand C., Renz B., ... GRIT Study Group. (2016). Heterogeneity in trauma registry data quality: Implications for regional and national performance improvement in trauma. Journal of the American College of Surgeons, 222(3), 288–295. doi:10.1016/j.jamcollsurg.2015.11.035
Diez Roux A. V., Merkin S. S., Arnett D., Chambless L., Massing M., Nieto F. J., Watson R. L. (2001). Neighborhood of residence and incidence of coronary heart disease. New England Journal of Medicine, 345(2), 99–106. doi:10.1056/NEJM200107123450205
Dozier K. C., Miranda M. A. Jr.,, Kwan R. O., Cureton E. L., Sadjadi J., Victorino G. P. (2010). Insurance coverage is associated with mortality after gunshot trauma. Journal of the American College of Surgeons, 210(3), 280–285. doi:10.1016/j.jamcollsurg.2009.12.002
Fabio A., Sauber-Schatz E. K., Barbour K. E., Li W. (2009). The association between county-level injury rates and racial segregation revisited: A multilevel analysis. American Journal of Public Health, 99(4), 748–753. doi:10.2105/ajph.2008.139576
Falor A., Kim D., Bricker S., Neville A., Bongard F., Putnam B., Plurad D. (2014). Insurance status predicts survival for trauma patients undergoing urgent intervention. Journal of Surgical Research, 188(1), 238–242
Haider A. H., Chang D. C., Efron D. T., Haut E. R., Crandall M., Cornwell E. (2008). Race and insurance status as risk factors for trauma mortality. Archives of Surgery, 143(10), 945–949. doi:10.1001/archsurg.143.10.945
Haider A. H., Hashmi Z. G., Zafar S. N., Castillo R., Haut E. R., Schneider E. B., Efron D. T. (2014). Developing best practices to study trauma outcomes in large databases: An evidence-based approach to determine the best mortality risk adjustment model. The Journal of Trauma and Acute Care Surgery, 76(4), 1061–1069.
Haider A. H., Saleem T., Leow J. J., Villegas C. V., Kisat M., Schneider E. B., Efron D. T. (2012). Influence of the National Trauma Data Bank on the study of trauma outcomes: Is it time to set research best practices to further enhance its impact? Journal of the American College of Surgeons, 214(5), 756–768.
Haider A., Weygandt P. L., Bentley J. M., Monn M. F., Rehman K. A., Zarzaur B. L., Cooper L. (2013). Disparities in trauma care and outcomes in the United States: A systematic review and meta-analysis. Journal of Trauma and Acute Care Surgery, 74(5), 1195–1205. doi:10.1097/TA.0b013e31828c331d
Hazlitt M., Hill J. B., Gunter O. L., Guillamondegui O. D. (2013). Disparities in trauma: The impact of socioeconomic factors on outcomes following traumatic hollow viscus injury. Journal of Surgical Research, 191(1), 6–11. doi:10.1016/j.jss.2013.05.052
Hicks C. W., Hashmi Z. G., Hui X., Velopulos C., Efron D. T., Schneider E. B., Haider A. H. (2015). Explaining the paradoxical age-based racial disparities in survival after trauma: The role of the treating facility. Annals of Surgery, 262(1), 179–183. doi:10.1097/SLA.0000000000000809
Hicks C. W., Hashmi Z. G., Velopulos C., Efron D. T., Schneider E. B., Haut E. R., Haider A. H. (2014). Association between race and age in survival after trauma. JAMA Surgery, 149(7), 642–647. doi: 10.1001/jamasurg.2014.166
Kaplan J. B., Bennett T. (2003). Use of race and ethnicity in biomedical publication. JAMA, 289(20), 2709–2716.
Killgore W. D., Cotting D. I., Thomas J. L., Cox A. L., McGurk D., Vo A. H., Hoge C. W. (2008). Post-combat invincibility: Violent combat experiences are associated with increased risk-taking propensity following deployment. Journal of Psychiatric Research, 42(13), 1112–1121. doi:10.1016/j.jpsychires.2008.01.001
Krieger N., Waterman P. D., Chen J. T., Soobader M. J., Subramanian S. V. (2003). Monitoring socioeconomic inequalities in sexually transmitted infections, tuberculosis, and violence: Geocoding and choice of area-based socioeconomic measures—the public health disparities geocoding project (US). Public Health Reports, 118(3), 240–260.
Ladha K. S., Young J. H., Ng D. K., Efron D. T., Haider A. H. (2011). Factors affecting the likelihood of presentation to the emergency department of trauma patients after discharge. Annals of Emergency Medicine, 58(5), 431–437. doi:10.1016/j.annemergmed.2011.04.021
Macmanus D., Dean K., Jones M., Rona R. J., Greenberg N., Hull L., Fear N. T. (2013). Violent offending by UK military personnel deployed to Iraq and Afghanistan: A data linkage cohort study. Lancet, 381(9870), 907–917. doi:10.1016/s0140-6736(13)60354-2
Maybury R. S., Bolorunduro O. B., Villegas C., Haut E. R., Stevens K., Cornwell E., Haider A. (2010). Pedestrians struck by motor vehicles further worsen race- and insurance-based disparities in trauma outcomes: The case for inner-city pedestrian injury prevention programs. Surgery, 148(2), 202–208.
Nuru-Jeter A. M., LaVeist T. A. (2011). Racial segregation, income inequality, and mortality in US metropolitan areas. Journal of Urban Health, 88(2), 270–282. doi:10.1007/s11524-010-9524-7
Oyetunji T. A., Crompton J. G., Ehanire I. D., Stevens K. A., Efron D. T., Haut E. R., Haider A. H. (2011). Multiple imputation in trauma disparity research. Journal of Surgical Research, 165(1), e37–e41. doi:10.1016/j.jss.2010.09.025
Patel M. S., Malinoski D. J., Nguyen X. M., Hoyt D. B. (2011). The impact of select chronic diseases on outcomes after trauma: A study from the National Trauma Data Bank. Journal of the American College of Surgeons, 212(1), 96–104. doi:10.1016/j.jamcollsurg.2010.09.028
Rangel E. L., Burd R. S., Falcone R. A. Jr.,, Multicenter Child Abuse Disparity Group. (2010). Socioeconomic disparities in infant mortality after nonaccidental trauma: A multicenter study. Journal of Trauma, 69(1), 20–25. doi:10.1097/TA.0b013e3181bbd7c3
Salim A., Ottochian M., DuBose J., Inaba K., Teixeira P., Chan L. S., Margulies D. R. (2010). Does insurance status matter at a public, level I trauma center? Journal of Trauma, 68(1), 211–216. doi:10.1097/TA.0b013e3181a0e659
Schoenfeld A. J., Belmont P. J. Jr.,, See A. A., Bader J. O., Bono C. M. (2013). Patient demographics, insurance status, race, and ethnicity as predictors of morbidity and mortality after spine trauma: A study using the National Trauma Data Bank. Spine Journal, 13(12), 1766–1773. doi:10.1016/j.spinee.2013.03.024
Shine D. (2012). Risk-adjusted mortality: Problems and possibilities. Computational and Mathematical Methods in Medicine, 1–5. Retrieved from doi:10.1155/2012/829465
Short S. S., Liou D. Z., Singer M. B., Bloom M. B., Margulies D. R., Bukur M., Ley E. J. (2013). Insurance type, not race, predicts mortality after pediatric trauma. Journal of Surgical Research, 184(1), 383–387. doi:10.1016/j.jss.2013.03.042
Signorello L. B., Cohen S. S., Williams D. R., Munro H. M., Hargreaves M. K., Blot W. J. (2014). Socioeconomic status, race, and mortality: A prospective cohort study. American Journal of Public Health, 104(12), e98–e107. doi:10.2105/AJPH.2014.302156
Signorello L. B., Schlundt D. G., Cohen S. S., Steinwandel M. D., Buchowski M. S., McLaughlin J. K., Blot W. J. (2007). Comparing diabetes prevalence between African Americans and Whites of similar socioeconomic status. American Journal of Public Health, 97(12), 2260–2267. doi:10.2105/ajph.2006.094482
Singer M. B., Liou D. Z., Clond M. A., Bukur M., Mirocha J., Margulies D. R., Ley E. J. (2013). Insurance- and race-related disparities decrease in elderly trauma patients. The Journal of Trauma and Acute Care Surgery, 74(1), 312–316.
Steenland K., Halperin W., Hu S., Walker J. T. (2003). Deaths due to injuries among employed adults: The effects of socioeconomic class. Epidemiology, 14(1), 74–79. doi:10.1097/01.ede.0000037973.98536.da
Subramanian S. V., Chen J. T., Rehkopf D. H., Waterman P. D., Krieger N. (2006). Comparing individual- and area-based socioeconomic measures for the surveillance of health disparities: A multilevel analysis of Massachusetts births, 1989–1991. American Journal of Epidemiology, 164(9), 823–834. doi:10.1093/aje/kwj313
Taghavi S., Jayarajan S. N., Duran J. M., Gaughan J. P., Pathak A., Santora T. A., Goldberg A. J. (2012). Does payer status matter in predicting penetrating trauma outcomes? Surgery, 152(2), 227–231. doi:10.1016/j.surg.2012.05.039
Tepas J. J. III,, Pracht E. E., Orban B. L., Flint L. M. (2011). Insurance status, not race, is a determinant of outcomes from vehicular injury. Journal of the American College of Surgeons, 212(4), 722–727; discussion 727–729. doi:10.1016/j.jamcollsurg.2010.12.016
Weygandt P. L., Losonczy L. I., Schneider E. B., Kisat M. T., Licatino L. K., Cornwell E. E. III, Haider A. H. (2012). Disparities in mortality after blunt injury: Does insurance type matter? Journal of Surgical Research, 177(2), 288–294. doi:10.1016/j.jss.2012.07.003
Zaloshnja E., Miller T. R., Coben J., Steiner C. (2012). How often do catastrophic injury victims become Medicaid recipients? Medical Care, 50(6), 513–519.
Keywords:

Health care disparities; Insurance type; Race; Socioeconomic status; Trauma mortality

© 2016 Society of Trauma Nurses