The Effect of Insurance Status, Race, and Gender on ED Disposition of Persons With Traumatic Brain Injury ANBESAW WOLDE SELASSIE, DRPH,* EMILY ELISABETH PICKELSIMER, DA,* LEROY FRAZIER, JR., MSPH,† AND PAMELA LYNN FERGUSON, PHD*,† The objective of this study was to assess the effect of insurance status and demographic characteristics on ED disposition among patients with traumatic brain injury (TBI). Statewide hospital discharge and ED datasets in South Carolina, 1996-2001, were analyzed by primary or secondary diagnosis of TBI in a multivariable logistic regression model. Of 70,671 unduplicated patients with TBI evaluated in the ED, 76% were treated and released; 26% had no insurance. The strongest predictors of hospital admission were TBI severity and preexisting health conditions. However, the uninsured and black females were less likely to be hospitalized after adjusting for demographic, clinical, and hospital characteristics (odds ratio [OR], 0.52; 95% confidence interval [CI], 0.48-0.55 and OR, 0.79; CI, 0.72-0.87, respectively). Although this study does not infer causality, insurance status, race, and gender were significant predictors of hospital admission. These results suggest that inpatient resources are not equitably used. (Am J Emerg Med 2004;22:465-473. © 2004 Elsevier Inc. All rights reserved.)
Uninsured patients are more likely to make ambulatory visits and overuse health resources than patients with insurance,1,2 use EDs for their ambulatory health care,3 and to accumulate higher medical costs.4 The Emergency Medical Treatment and Active Labor Act (EMTALA), included in the 1985 Consolidated Omnibus Budget Reconciliation Act (COBRA), requires Medicare-participating hospitals with EDs to screen and stabilize patients and states that screening cannot be delayed while hospital personnel verify the patients’ insurance.5 Despite access to ED care, however, numerous studies report that insurance status,6-17 race,18-28
From the *Department of Biometry and Epidemiology, Medical University of South Carolina, Charleston, South Carolina; and the †Division of Injury and Disability Prevention, South Carolina Department of Health and Environmental Control, Columbia, South Carolina. Manuscript received April 1, 2003, accepted July 26, 2003. This study was supported in part by grant nos. U17/CCU414981 and U17/CCU411892 from the Division of Injury and Disability Outcome Programs; the National Center for Injury Prevention and Control; and the Centers for Disease Control and Prevention (CDC). Its contents are solely the responsibility of the authors and do not necessarily represent the official views of CDC. Partial data from this article were presented at the Third World Congress on Brain Injury, Quebec City, Quebec, Canada, June 1999. Address correspondence to Anbesaw Wolde Selassie, DrPH, Medical University of South Carolina, Department of Biometry and Epidemiology, 135 Cannon St., PO Box 250835, Charleston, SC 29425. Email:
[email protected]. Key Words: Brain injury, emergency department, insurance status, race and gender. © 2004 Elsevier Inc. All rights reserved. 0735-6757/04/2206-0006$30.00/0 doi:10.1016/j.ajem.2004.07.024
age,22,24,25 and gender22,26-29 are associated with greater risk of unequal access to appropriate medical care. The proportion of the U.S. population without health insurance has remained nearly constant for the past 5 years. The Census Bureau estimates that 41.2 million people in the United States, or 14.6% of the population, had no health insurance coverage for any part of 2001; 14.2% were uninsured during 2000; 15.5% were uninsured during 1999; 16.3% had no health insurance in 1998; 16.1% were uninsured during 1997; and 15.6% were uninsured during 1996. South Carolina’s uninsured rates during those years were 12.2%, 12%, 15.6%, 13.8%, 16.8%, and 17.1%, respectively.30-34 Young adult males, the highest risk group for injury, were less likely to have health insurance coverage.31 In addition, recent welfare reform has moved people from the welfare rolls to “nonstandard” jobs, thus creating a class of working people who often have no health coverage.28,29 Hospital-based EDs in the United States treat and release an estimated one million people with traumatic brain injury (TBI) each year.35 Approximately 80% of these individuals are not hospitalized.36 The discharge rates for TBI-related hospitalizations declined nationally approximately 48% between 1980 and 1995.37 During the past 10 years, South Carolina’s discharge rate for people with TBI declined 47%, whereas the mortality rate increased 5%.38 Most studies to date have documented inequities in the receipt of specific procedures or services among different patient subgroups; relatively few have examined determinants of hospital admission among TBI patients.39,40 The purpose of this study was to examine the effect of insurance, race, and gender on the likelihood of hospitalization among patients who presented to any nonfederal, hospital-based ED in South Carolina for treatment of a TBI between 1996 and 2001. METHODS Data Sources We used the January 1, 1996, through December 31, 2001, State of South Carolina Hospital Discharge and Emergency Department Visit Data Sets maintained by the State’s Budget and Control Board. South Carolina law mandates that all federal and nonfederal hospitals, including self-standing EDs, report discharge and ED visit uniform billing abstracted data to the state. The data files contain patient identifiers (that is, medical record number, personal unique ID, first and last name, address and zip code), demographics (date of birth, race, gender, place of residence), dates of admission/ED visit, 10 International Clas465
466
AMERICAN JOURNAL OF EMERGENCY MEDICINE ■ Volume 22, Number 6 ■ October 2004
sification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM)41 diagnostic codes (one primary and nine secondary codes), external causes of injury codes, length of stay, discharge disposition, source of admission, principal payer, hospital identification number, and trauma center level. The South Carolina Injury Surveillance System verifies the accuracy and completeness rate of these data as part of a routine evaluation protocol. The abstracted data are 99% accurate and complete.42 To acquire additional information, the TBI Surveillance System annually reviews the medical records of 1,000 randomly selected TBI discharges and 3,500 ED visits. Because personal identifiers are included in the dataset, we were able to avoid inclusion of repeat visits for the same injury event. Study Population This study was reviewed and exempted by the Institutional Review Board. All patients who sustained a TBI that resulted in an ED visit to any nonfederal, hospital-based ED located in South Carolina between January 1, 1996, through December 31, 2001, were eligible for the study. We defined a case of TBI as any discharge or ED visit with a primary or secondary diagnosis of skull fracture and/or intracranial injury in accordance with the Centers for Disease Control and Prevention case definition for TBI.43 For case identification purposes, the ICD-9-CM nature of injury codes (Ncodes) included were 800.0-801.9, 803.0-804.9, 850.0854.1, and 959.01. We excluded TBIs coded as complications (N-code 958) and late effects (N-code 907), patients who died on arrival or during resuscitation in the ED, and patients with TBI who were directly admitted to the hospital without being seen in the ED. In addition, patients who left the ED against medical advice were excluded. For individuals with more than one isolated event of TBI during the study period, we retained the index hospitalizations for those admitted and the highest severity for those treated and released from the ED. Overall, 945 patients (1.34%) had more than one TBI-related ED visit or hospitalization. There was significant variation in the timing of these later ED visits and hospitalizations: 24% occurred within 1 month, 15% occurred between 1 and 6 months, 18% occurred between 7 and 12 months, 28% occurred between 1 and 2 years, and 15% occurred more than 2 years after the initial ED visit or hospitalization. Statistical Analysis The dependent variable, type of care, was defined as “hospitalized” for patients who were admitted to the hospital and “ED-managed” for those treated and released from the ED. We dichotomized South Carolina’s counties into “urban” and “rural” according to the federal metropolitan statistical area designation. Patients from other states were grouped as “out-of-state.” Injury severity was determined by translating ICD-9-CM codes into Abbreviated Injury Scale (AIS)44 scores using the ICDMAP-90 software.45 For patients with more than one diagnosis of TBI, we selected the TBI with the highest AIS score. Severity of TBI was classified on the basis of the AIS score as “severe” with a score of 4 to 6, “moderate” with a score of 3, and “mild” with a score of 2. TBI patients with other trauma were
categorized on the number of body regions involved as, “three or more body regions”, “two body regions”, and “one body region.” Patients with no other trauma except TBI were classified as “TBI only.” Because only 2.6% of the study population is other races, we combined them with blacks and created four categories of race and gender: black female, black male, white female, and white male. To compare the diagnosis types of TBI by the type of care, we used the head region of the Barell Injury Diagnosis Matrix (see Table 1 for the diagnosis listing).46 ICD-9-CM diagnostic codes were also used to measure the number of serious preexisting conditions each patient had. Elixhauser and colleagues developed a comorbidity scale for use with administrative data that identifies 30 conditions that are significant predictors of in-hospital mortality and resources.47 All 10 ICD-9-CM diagnosis fields were searched for any of these 30 conditions, and the total number of conditions present was used to reflect each patient’s underlying preexisting health. The relationship between type of care and different patient, injury, and hospital characteristics was examined for all TBI patients who were hospitalized or treated and released from the ED using a chi-squared test statistic.48 Type of care was also modeled as a function of age, race and gender, insurance status, TBI severity, number of body regions injured, number of preexisting conditions, place of residence, year of discharge, and trauma center level using unconditional multivariable logistic regression techniques49 with the statistical package, PC SAS.50 The categorization of the independent variables was based on their relationship with type of care (see Table 2 for coding of variables). Multicollinearity among the independent variables was evaluated by assessing the deviations of the regression coefficients and their standard errors in the fitted univariate and multivariate models.51 The independent variables were entered simultaneously into the model. The unadjusted (from the univariate models) as well as the adjusted odds ratios (from the final multivariate model that includes all independent variables) and 95% confidence intervals are reported. RESULTS During the 6-year study period, a total of 70,671 persons were evaluated in the ED for TBI. Seventy-six percent were treated and released. Approximately one-fourth (26%) of the study population was uninsured. The uninsured and Medicaid-insured had the lowest rates of hospitalization, respectively, 15% and 18%; privately insured 23%, and Medicare/other- government-insured 44%. The uninsured were predominantly male (65%), residing in urban areas (64%), and between the ages of 15 and 44 (73%). Among the uninsured, the elderly (age ⱖ65) accounted for only 1% and children (age ⱕ14) accounted for 15%. Although blacks constituted 33% of the total TBI patients, they represented 41% of the uninsured and 53% of the Medicaid-insured. Similarly, although children constituted 27% of the total TBI patients, they represented 66% of the Medicaid-insured. Approximately 43% of patients with TBI under Medicare/other government insurance had preexisting conditions (Table 2). Table 3 compares TBI severity and type of care between
SELASSIE ET AL ■ BRAIN INJURY AND INSURANCE STATUS IN THE ED
TABLE 1.
467
Type of Traumatic Brain Injury by Type of Care (N ⫽ 70,671) Total No. of Patients
Type of Traumatic Brain Injury A. Type I 1. Fracture of vault of the skull with intracranial injury and/or loss of consciousness ⱖ1 hr 2. Fracture of base of the skull with intracranial injury and/or loss of consciousness ⱖ1 hr 3. Dural or arachnoid hemorrhage without open intracranial wound 4. Unqualified skull fracture with intracranial injury and/or loss of consciousness ⱖ1 hr 5. Cerebral laceration and/or contusion 6. Other and unspecified intracranial hemorrhage without open wound 7. Multiple fracture of the skull with intracranial injury and/or loss of consciousness ⱖ1 hr 8. Intracranial injury of other and unspecified nature 9. Concussion with loss of consciousness ⱖ1 hr B. Type II 10. Fracture of vault of the skull without intracranial injury or loss of consciousness ⬍1 hr 11. Fracture of base of the skull without intracranial injury or loss of consciousness ⬍1 hr 12. Unqualified skull fracture without intracranial injury or loss of consciousness ⬍1 hr 13. Multiple fracture of the skull without intracranial injury or loss of consciousness ⬍1 hr 14. Concussion with loss of consciousness ⬍1 hr C. Type III 15. Fracture of vault of the skull without intracranial injury and loss of consciousness 16. Fracture of base of the skull without intracranial injury and loss of consciousness 17. Unqualified skull fracture without intracranial injury and loss of consciousness 18. Multiple fracture of the skull without intracranial injury and loss of consciousness D. Unspecified head 19. Unspecified head injuries*
Median AIS for the Head Region
21,116
Hospitalized (Percent) 48.2
636
4
93
1,301
4
91
3,112
4
90
360 2,198
4 4
89 86
940
4
82
38 12,213 318 19,476
4 2 3
76 21 14 21.1
276
3
71
1,121
3
63
250
2
28
19 17,810 1,520
2 2
21 18 51.1
323
2
61
988
3
53
196
2
29
13 28,559 28,559
2
23 6 6
approximately 2
*The minimum AIS of 2 was assigned for all head injuries coded with 959.01.
the uninsured (persons without any health insurance) and insured (persons with any health insurance). The uninsured had significantly lower rates of admission for both moderate and severe injuries than persons with insurance (P ⬍.01). Conversely, the insured had significantly higher rates of admission for mild injuries than the uninsured (P ⬍.01). Table 1 presents the frequency distribution of the types of TBI, the median AIS score for the head region noted in our study, and the proportion hospitalized. The most serious TBIs are classified under type I. When the large number of unspecified intracranial injuries (ICD-9 code 854.0) was excluded, the median AIS score for TBIs classified as type I was 4 (severe), and 9 of 10 patients were hospitalized. Among the TBIs classified under type II, concussions with less than 1 hour loss of consciousness (“mild” concussion) constituted 91% of the injuries; when these “mild” concussions were excluded, the median AIS score was 3 and 58% were hospitalized. TBIs classified under type III are skull fractures without evidence of intracranial injury and loss of consciousness. They are generally less severe than TBIs under types I and II. Head injuries coded as unspecified
(ICD-9-CM code 959.01) accounted for 40% of the total study population. Overall, unspecified head injuries and unspecified intracranial injuries without open wound (ICD-9-CM code 854.0) accounted for 58% of the study population. Table 4 presents the unadjusted and adjusted odds of hospital admission after TBI. The observed associations in both the unadjusted and adjusted measures are in the same direction and the degree of confounding noted for all variables, except the number of body regions involved and residence, is either toward the null or very minimal. As a result, the following discussions pertain only to the findings of the multivariable logistic analysis. After adjusting for demographic, clinical, and hospital characteristics, the uninsured were less likely to be hospitalized than those with private insurance. Overall, uninsured patients with TBI were 48% less likely to be hospitalized than those with private insurance (odds ratio [OR], 0.52; 95% confidence interval [CI], 0.48-0.55). Furthermore, this insurance-related difference among the various insurance categories remained across all levels of injury severity (see Figure 1). Conversely, TBI patients covered by Medicare/
AMERICAN JOURNAL OF EMERGENCY MEDICINE ■ Volume 22, Number 6 ■ October 2004
468
TABLE 2.
Patient Characteristics by Insurance Type (N ⫽ 70,671) Insurance Status
Characteristics Type of care Hospitalized ED treated and released Severity Severe (AIS 4-6) Moderate (AIS 3) Mild (AIS 2) No. of preexisting health conditions Three or more Two One None Body regions injured other than traumatic brain injury ⱖ3 body regions 2 body regions 1 body region Traumatic brain injury only Residence Urban Out-of-state Rural Trauma level designation Level I Level II Level III Undesignated Year of discharge 2001 2000 1999 1998 1997 1996 Race and gender Black female Black male White male White female Age group 65 and over 45-64 25-44 15-24 0-14 Mean (standard deviation)
Percent Uninsured (n ⫽ 18,046)
Percent Medicaid (n ⫽ 10,696)
Percent Medicare and Other Government (n ⫽ 11,395)
Percent Private (n ⫽ 30,534)
15.0 85.0
18.0 82.0
43.9 56.1
22.9 77.1
6.7 5.8 87.5
7.9 3.4 88.7
24.1 7.5 68.4
9.0 5.4 85.6
0.3 1.3 6.7 91.7
0.7 2.0 6.6 90.7
8.0 12.4 22.3 57.4
0.6 2.1 7.9 89.4
4.2 10.4 28.8 56.6
2.7 5.8 17.7 73.8
4.3 9.6 26.1 59.9
5.4 10.7 25.7 58.2
63.5 4.2 32.3
56.5 0.8 42.7
59.0 3.7 37.3
57.4 5.4 37.1
23.7 10.3 40.1 25.9
24.4 10.0 33.7 32.0
27.0 10.9 35.3 26.8
24.1 11.0 36.3 28.7
18.9 20.4 14.0 18.0 14.7 20.0
23.6 21.5 15.3 16.8 11.6 11.2
22.1 21.1 14.0 15.6 13.2 14.0
19.8 20.0 14.7 17.3 14.4 13.8
13.5 27.5 37.4 21.6
22.0 30.6 26.4 21.0
10.0 13.7 33.7 42.6
9.3 15.7 46.2 28.8
1.1 11.5 43.1 29.8 14.5 28.8 (⫾14.4)
0.5 5.8 11.3 16.2 66.1 13.6 (⫾14.9)
62.1 13.0 12.8 6.9 5.2 62.3 (⫾24.7)
2.5 15.1 27.6 26.5 28.3 26.2 (⫾18.0)
other government insurance programs were more likely to be hospitalized (OR, 1.41; 95% CI, 1.29-1.54). The severity of TBI and number of preexisting conditions a person had were the strongest predictors of hospitalization. Furthermore, the patterns of hospitalization showed biologic gradient, that is, with increased severity of injury and increased number of preexisting conditions, odds of hospital admission strongly increased. Similar biologic gradient was also noted with number of body regions injured, yielding over a twofold increase in the probability of hos-
pitalization for one additional body region involved. Patients who presented to hospitals that were designated as level I trauma centers were almost six times more likely to be hospitalized compared with hospitals in which the trauma center level was undesignated. Overall, 28% of the TBI patients were evaluated in hospitals without any trauma center level designation. Rates of hospitalization also varied significantly by race and gender. After adjusting for all covariates listed in Table 4, black females were 21% less likely to be hospitalized
SELASSIE ET AL ■ BRAIN INJURY AND INSURANCE STATUS IN THE ED
TABLE 3.
469
TBI Severity and Type of Care Between Health Insured and Uninsured Persons Uninsured Hospitalized
All Insured ED T&R
Hospitalized
ED T&R
Total
Traumatic Brain Injury Severity
No.
Percent
No.
Percent
No.
Percent
No.
Percent
No.
Percent
Severe (AIS 4-6) Moderate (AIS ⫽ 3) Mild (AIS ⫽ 2) Total
995 479 1,239 2,713
82 46 8 15
222 664 14,547 15,333
18 64 92 86
6,720 2,032 6,160 13,912
90 71 14 26
617 849 37,247 38,713
10 29 86 74
7,554 3,924 59,193 70,671
11 6 84 100
compared with white females (OR, 0.79; 95% CI, 0.720.87). Regardless of race, males had higher odds of hospital admission than females. Furthermore, TBI patients residing in urban areas or out-of-state had significantly lower odds of hospital admission than those residing in rural areas (urban OR, 0.83; 95% CI, 0.74-0.95; out-of-state OR, 0.54; 95% CI, 0.51-0.75). Table 4 further shows the decrease in the rates of hospital admission as the calendar years progressed. The number of TBI patients, as a proportion of the total TBI patients treated and released from the ED, was the lowest in 1996 and the highest in 2001, with a significant trend effect (chi-square test of trend P ⬍.01). The odds of hospital admission were significantly lower by 24% in 1997, 25% in 1998, 35% in 2000, and 43% in 2001 than in 1996. A lower effect of 5% decrease was noted in 1999 but was not statistically significant. DISCUSSION This study examined the delivery of inpatient care among patients with TBI. It evaluated the influence of insurance status, race, and gender on access to inpatient care. Generally, the types of decision errors encountered as a result of differences in the delivery of inpatient care could be one of under admission or over admission. Although both are plausible explanations for the findings noted in this study, errors pertaining to underadmission deserve special attention as a result of the long-term consequences for TBI patients who are suboptimally observed and treated during the acute stages of the condition. Although the differences noted in this study could be justified on the grounds of clinical characteristics and patient-specific conditions, the variability noted by demographic and/or insurance status suggest that healthcare services are not equitably distributed. After controlling for demographic, clinical, and hospital characteristics, patients who were uninsured were consistently less likely to be admitted, regardless of the severity of injury. The results extended previous reports on rates of hospitalization for patients with TBI by comparing the rate of ED visits with the rate of hospital admissions.39,40 Svenson and Spurlock found that among patients with a less severe head injury, uninsured patients were significantly less likely to be admitted compared with those with private insurance.40 In contrast, this study provides information on the whole spectrum of injury severity and types of TBI in a statewide population over an extended period of time and measures the relative effect of insurance status on the rate of hospital admission after controlling for other factors. Important strengths of the study include the use of data from a
well-defined, population-based data system that are routinely validated as part of an ongoing surveillance program and the inclusion of a standard, widely used measure of injury severity, the AIS, which allowed us to adjust for this key influence on the likelihood of hospital admission. The disparities noted in hospital admission rates as a function of insurance status raises major concerns for two reasons. First, a disproportionate burden of injury falls on the economically disadvantaged and the uninsured.52-56 For instance, the proportion of South Carolina residents who were uninsured during the study period ranged from 12% to 17%.30-34 In contrast, 26% of all study patients during the same period were uninsured. It is therefore likely that inpatient resources that are skewed toward the insured might keep the uninsured medically disadvantaged. Second, although there is a disagreement over the admission of persons with mild to moderate TBI, most physicians would agree that severe head trauma requires hospitalization. In our study, approximately 11% of patients who sustained moderately severe to severe injury were not admitted. The uninsured were disproportionately represented (18%) among persons with moderately severe to severe injuries who were treated and released from the ED. Although it is possible to conclude that insurance-related differences in admission rates among those with a mild TBI could represent overadmissions among the insured, the data we examined strongly suggest the underadmission rate of the uninsured as a cogent explanation. Our findings suggest that among those who sustain severe TBI, uninsured patients were less likely than patients with insurance to receive inpatient care. TBI patients covered by Medicare and other government insurance plans were significantly more likely to be hospitalized at all levels of severity than those with private insurance, Medicaid, or the uninsured. Although 43% of the elderly had significant preexisting health problems, the higher hospitalization rate noted in this study is independent of comorbid patterns. This higher level of admission of Medicare patients in contrast to other insurance categories could be attributed to the physicians’ recognition of the vulnerability of the elderly for serious complications and the protracted healing time needed for good recovery.54 This study found race and gender to be factors independently accounting for hospital admission decisions. However, the effect was not significant when TBI was severe. Other researchers noted significantly higher provision of diagnostic tests and therapeutic procedures for males and whites compared with females and blacks.19,20,57-59 The interaction effect of race and gender had been reported previously.60-62 A study by Schulman and colleagues re-
AMERICAN JOURNAL OF EMERGENCY MEDICINE ■ Volume 22, Number 6 ■ October 2004
470
TABLE 4.
Adjusted and Unadjusted Odds of Hospital Admission Among Patients With Traumatic Brain Injury (N ⫽ 70,671) Type of Care Characteristics
Insurance status Uninsured Medicaid Medicare and other government Private Severity Severe (AIS 4-6) Moderate (AIS 3) Mild (AIS 2) Number of preexisting conditions Three or more Two One None Body regions injured other than traumatic brain injury ⱖ3 body regions 2 body regions 1 body regions Traumatic brain injury only Residence Urban Out-of-state Rural Trauma level designation Level I Level II Level III Undesignated Year of discharge* 2001 2000 1999 1998 1997 1996 Race and gender Black female Black male White male White female Age group 65 and over 45-64 25-44 15-24 0-14
Odds Ratio (95% Confidence Interval)
Hospitalized (n ⫽ 16,625)
ED T&R (n ⫽ 54,046)
Unadjusted
Adjusted
16.3 11.6 30.1 42.0
28.4 18.2 11.8 43.6
0.60 (0.57-0.63) 0.74 (0.70-0.78) 2.64 (2.52-2.76) Reference
0.52 (0.48-0.55) 1.01 (0.93-1.10) 1.41 (1.29-1.54) Reference
40.4 15.1 44.5
1.5 2.6 95.9
56.02 (51.94-60.44) 12.44 (11.60-13.34) Reference
38.09 (34.92-41.54) 7.90 (7.27-8.59) Reference
6.6 10.5 22.3 60.6
0.2 1.4 5.9 92.5
42.52 (35.36-51.13) 11.56 (10.59-12.63) 5.79 (5.50-6.10) Reference
36.33 (29.18-45.24) 9.40 (8.32-10.63) 4.38 (4.06-4.72) Reference
12.4 17.1 28.6 41.9
2.1 7.4 24.3 66.2
9.52 (8.81-10.28) 3.63 (3.44-3.84) 1.86 (1.78-1.94) Reference
10.98 (9.91, 12.16) 4.21 (3.91, 4.55) 1.95 (1.84, 2.07) Reference
58.3 4.8 36.9
59.4 3.9 36.7
1.24 (1.14-1.35) 0.98 (0.94-1.01) Reference
0.83 (0.74, 0.95) 0.54 (0.51, 0.57) Reference
50.1 13.7 21.5 14.7
16.6 9.7 41.4 32.3
6.65 (6.32-7.01) 3.12 (2.92-3.33) 1.14 (1.08-1.21) Reference
5.75 (5.34, 6.20) 3.31 (3.02, 3.62) 1.33 (1.23, 1.43) Reference
17.8 17.7 16.5 16.0 14.4 17.6
21.4 21.3 13.8 17.6 13.7 12.2
0.58 (0.55-0.62) 0.57 (0.54-0.61) 0.83 (0.78-0.88) 0.63 (0.60-0.67) 0.73 (0.69-0.78) Reference
0.57 (0.52, 0.62) 0.65 (0.60, 0.71) 0.95 (0.87, 1.04) 0.75 (0.69, 0.83) 0.76 (0.70, 0.84) Reference
8.6 21.7 43.0 26.7
13.6 20.4 37.7 18.3
0.68 (0.63-0.72) 1.14 (1.08-1.19) 1.22 (1.17-1.27) Reference
0.79 (0.72, 0.87) 1.15 (1.07, 1.24) 1.18 (1.11, 1.26) Reference
21.5 18.5 28.1 19.8 14.1
8.4 11.2 26.3 23.4 30.7
5.54 (5.21-5.90) 3.19 (2.99-3.39) 2.31 (2.19-2.44) 1.83 (1.73-1.94) Reference
1.00 (0.88, 1.13) 1.08 (0.98, 1.19) 1.25 (1.16, 1.36) 1.16 (1.07, 1.26) Reference
*Chi-squared test of trend P ⬍ .01.
ported that black females were significantly less likely to be recommended for cardiac catheterization compared with white males, whereas no significant differences were found for black males or white females.61 The finding reported in this study for decreased likelihood of hospital admission after TBI among black females is consonant with findings on gender and minority disparities by the researchers cited previously. The results of this study also indicate that out-of-state and urban residents were significantly less likely to be admitted than patients who resided in rural counties. Out-of-state
patients could opt to be transferred to their home state rather than be admitted to a hospital in another state. In addition, the portability of some insurance policies across state jurisdictions could be limited, necessitating transfer to an instate hospital after medical stabilization. Although it is not clear why urban residents were less likely to be admitted compared with rural residents, one possible explanation for this is that the inpatient bed capacity in relation to patient demand could be limited in urban as compared with rural areas. Our findings further suggest reduced admission rates with advancing calendar years. This might be attributed to
SELASSIE ET AL ■ BRAIN INJURY AND INSURANCE STATUS IN THE ED
FIGURE 1.
471
Adjusted odds of hospital admission by insurance status and traumatic brain injury severity.
increasingly strict hospital admission policies to curtail healthcare costs. Hospitals have staged strong cost containment efforts to offset the financial fallout from reduced reimbursement for trauma patients63 amid rising operational costs. Therefore, hospitals that operate under competitive market pressure could institute policies to restrict the admission of uninsured patients.7,64 Despite the underadmission rates associated with insurance, race and gender, year of event, and place of residence, the strongest predictors of hospital admission were the severity of the head trauma, the number of preexisting conditions, and the number of body regions involved in addition to the head. The more severe the TBI, the higher the probability of being admitted because severely injured patients need more supervision and care than those with less severe injuries. Similarly, the simultaneous diagnosis of one or more preexisting conditions with the TBI significantly increased the likelihood of hospital admission, suggesting the uncertainties of the outcome of trauma among persons in poor health. Other researchers have also noted the increased likelihood of death after hospital admission and poor functional outcomes among trauma patients with preexisting conditions.66 Finally, our study found that TBI patients who presented to designated trauma centers (levels I–III) were significantly more likely to be admitted compared with patients who were evaluated in hospitals without a trauma level designation regardless of injury severity. This might
be the result of the more specialized services offered in designated trauma centers than in undesignated hospitals. Recent studies also suggest that designated trauma centers offer more specialized services, tend to be larger, and are more likely to be public hospitals and teaching institutions.67 LIMITATIONS Our analysis has several limitations. First, this study mainly depended on administrative data that are readily available, inexpensive to acquire, and encompass large populations. However, their use to evaluate quality of care is very inadequate because they fail to supply critical information on many aspects of care such as the interpersonal quality of care, the technical quality of care, or the appropriateness of the care received. Hence, we do not know if reduced hospitalization rates among uninsured patients with TBI had any effect on either short-term or long-term outcomes. The possibility remains that uninsured patients who were not hospitalized might have fared well after being treated and released from the ED. Furthermore, our analysis lacks the ability to adequately discriminate whether the reduced hospitalization rate among the uninsured is the result of the relative overadmission of the insured. There are no trend data to determine the pattern of admission practices as a function of insurance status and time.
472
AMERICAN JOURNAL OF EMERGENCY MEDICINE ■ Volume 22, Number 6 ■ October 2004
Second, our findings might have been influenced by other variables with significant bearing on ED disposition but not included in our model. As mentioned, our variables were from an administrative dataset and thus were not originally collected for the purposes of this study. As such, neither the variables nor the study itself was designed to address causality. Furthermore, the observed relationship between race– gender and hospital admission might be the result of other related, but unavailable, variables. Examples are numerous and could include patient preferences for outpatient care, self-initiated out-of-state referral, timing of ED visit, secondary payer status, the experience and skill of the provider, family support, availability of sick leave, family income, level of education, distance from home to the hospital, availability of a bed in the hospital, and patient perception of injury severity, among others. Finally, our analysis depended on severity measures translated from ICD-9-CM codes assigned at the time of discharge for billing purposes. The assignment of these codes could have been influenced by coding practices that prefer codes with higher severity as a way to maximize reimbursement among patients who have insurance. With increased use of prospective payment systems, coding practices appear to gravitate toward using codes that supply higher severity scores.68-70 As a result, the accuracy of our estimate of the level of injury severity is contingent on the correctness of the ICD-9-CM diagnostic coding and on accurate classification of these injuries by ICDMAP-90 software. CONCLUSIONS The results of this study suggest the importance of insurance status and demographic factors as potential determinants for hospital admission among patients with TBI. Although the study relied on administrative data, its limitations should affect all patient subgroups equally and not bias one subgroup more than another. Although injury severity and preexisting health status remain the strongest determinants of hospital admission after TBI, it is unsettling that black females and persons without insurance were significantly less likely to be admitted compared with white females and the privately insured with similar characteristics. However, this relationship does not suggest or imply causal relationship between the gender–race of the patient and the admission bias observed. Although the findings did not exclude the possibility that the observed differences could have resulted from overadmission of some patient groups, one should expect comparable rates of hospital admission for persons with severe TBI across all insurance and demographic categories. Although the findings of this study did not establish if persons treated as inpatient had better functional outcomes than persons treated and released from the ED, these preliminary findings warrant prospective comparison of patients with TBI as a function of the type of care received. ACKNOWLEDGMENT The authors recognize Georgette Demian and Lynnore Liggins fronm the South Carolina Department of Health and Environmental Control for their coordination of the abstraction process and for validating the accuracy of the medical record review report. Dr.
Linda Veldheer, Director of the Head and Spinal Cord Injury Division of the South Carolina Department of Disabilities and Special Needs, assisted with data oversight for the project.
REFERENCES 1. Weissman JS, Gatsonis C, Epstein AM: Rates of avoidable hospitalization by insurance status in Massachusetts and Maryland. JAMA 1992;268:2388-2394 2. Pappas G, Hadden WC, Kozak LJ, et al: Potentially avoidable hospitalizations: inequalities in rates between US socioeconomic groups. Am J Public Health 1997;87:811-816 3. Baker DW, Stevens CD, Brook RH: Regular source of ambulatory care and medical care utilization by patients presenting to a public hospital emergency department. JAMA 1994;271:1909-1912 4. Epstein AM, Stern RS, Weissman JS: Do the poor cost more? A multihospital study of patients’ socioeconomic status and use of hospital resources. N Engl J Med 1990;322:1122-1128 5. 42 USC Sec. 1395dd 6. Rowland D, Feder J, Keenan PS: Uninsured in America: the causes and consequences. In: Altman SH, Reinhardt UE, Shields AE, eds. The Future US Healthcare System: Who Will Care for the Poor and Uninsured? Chicago: Health Administration Press; 1998: 25-44 7. Weissman JS, Epstein AM: Falling Through the Safety Net: Insurance Status and Access to Health Care. Baltimore: Johns Hopkins University Press; 1994 8. Burstin HR, Lipsitz SR, Brennan TA: Socioeconomic status and risk for substandard medical care. JAMA 1992;268:2383-2387 9. Hadley J, Steinberg EP, Feder J: Comparison of uninsured and privately insured hospital patients: condition on admission, resource use, and outcome. JAMA 1991;265:374-379 10. Baker DW, Shapiro MF, Schur CL: Health insurance and access to care for symptomatic conditions. Arch Intern Med 2000; 160:1269-1274 11. Canto JG, Rogers WJ, French WJ, et al: Payer status and the utilization of hospital resources in acute myocardial infarction; a report from the National Registry of Myocardial Infarction 2. Arch Intern Med 2000;160:817-823 12. Faulkner LA, Schauffler HH: The effect of health insurance coverage on the appropriate use of recommended clinical preventive services. Am J Prev Med 1997;13:453-458 13. Ayanian JZ, Weissman JS, Schneider EC, et al: Unmet health needs of uninsured adults in the United States. JAMA 2000;284: 2061-2069 14. Himmelstein DU, Woolhandler S: Care denied: US residents who are unable to obtain needed medical services. Am J Public Health 1995;85:341-344 15. Holl JL, Szilagyi PG, Rodewalk LE, et al: Profile of uninsured children in the United States. Arch Pediatr Adolesc Med 1995;149: 398-406 16. Schoen C, Lyons B, Rowland D, et al: Insurance matters for low-income adults: results from a five-state survey. Health Aff (Millwood) 1997;16:163-171 17. Sox CM, Burstin HR, Edwards RA, et al: Hospital admissions through the emergency department: does insurance status matter? Am J Med 1998;105:506-512 18. Blendon RJ, Aiken LH, Freeman HE, et al: Access to medical care for black and white Americans: a matter of continuing concern. JAMA 1989;261:278-281 19. Kressin NR, Petersen LA: Racial differences in the use of invasive cardiovascular procedures: review of the literature and prescription for future research. Ann Intern Med 2001;135:352-366 20. Lee AJ, Gelbach S, Hosmer DW, et al: Medicare treatment differences for blacks and whites. Med Care 1997;35:1173-1189 21. Pope HJ, Aufderheide TP, Ruthazer R, et al: Missed diagnoses of acute cardiac ischemia in the emergency department. N Engl J Med 2000;342:1163-1170 22. Kjellstrand CM: Age, sex and race inequality in renal transplantation. Arch Intern Med 1988;148:1305-1309 23. Wenneker MB, Epstein AM: Racial inequalities in the use of procedures for patients with ischemic heart disease in Massachusetts. JAMA 1989;261:253-257 24. Bennett CL, Greenfield S, Aronow H, et al: Patterns of care related to age of men with prostate cancer. Cancer 1991;67:26332641
SELASSIE ET AL ■ BRAIN INJURY AND INSURANCE STATUS IN THE ED
25. Greenfield S, Blanco DM, Elashoff RM, et al: Patterns of care related to age of breast cancer patients. JAMA 1987;257:2766-2770 26. Ayanian JZ, Epstein AM: Difference in the use of procedures between women and men hospitalized for coronary artery disease. N Engl J Med 1991;325:221-225 27. Steingart RM, Packer M, Hamm P, et al: Sex differences in the management of coronary artery disease. N Engl J Med 1991; 325:226-230 28. Kuttner R: The American health care system: health insurance coverage. N Engl J Med 1999;340:163-168 29. Mishel L, Bernstein J, Schmitt J: The state of working America, 1998-1999. Washington, DC: Economic Policy Institute; 1998: 243 30. Mills RJ: Health Insurance Coverage: 2001. Census Bureau, US Department of Commerce, P60-220, Sept 2002 31. Mills RJ: Health Insurance Coverage: 2000. Census Bureau, US Department of Commerce, P60-215, Sept 2001 32. US Census Bureau. Current Population Survey, March 1999, 2000, and 2001 33. Campbell JA: Health Insurance Coverage: 1998. Census Bureau, US Department of Commerce, P60-208, Oct 1998 34. US Census Bureau. Current Population Survey, March 1996, 1997, and 1998 35. US Department of Health and Human Services, Centers for Disease Control and Prevention, Division of Acute Care, Rehabilitation Research, and Disability Prevention: Traumatic Brain Injury in the United States: A Report to Congress. Atlanta: Centers for Disease Control and Prevention; May 5, 1999 36. Guerrero JL, Thurman J, Sniezek JE: Emergency department visits associated with traumatic brain injury: United States, 19951996. Brain Inj 2000;14:181-186 37. Thurman D, Guerrero J: Trends in hospitalization associated with traumatic brain injury. JAMA 1999;282:954-957 38. South Carolina Traumatic Head and Spinal Cord Injury Information System, Department of Disabilities and Special Needs, Annual Report, Columbia, SC; 1999 39. McCarthy ML, Serpi T, Kufera JA, et al: Factors influencing admission among children with a traumatic brain injury. Acad Emerg Med 2002;9:684-693 40. Svenson JE, Spurlock CW: Insurance status and admission to hospital for head injuries: are we part of a two-tiered medical system? Am J Emerg Med 2001;19:19-24 41. International Classification of Diseases, 9th Revision, Clinical Modification, 6th ed. Washington, DC: US Department of Health and Human Services; 1996 42. Division of Injury and Disability Prevention: Traumatic Brain Injury Surveillance System Data Evaluation Report. Columbia, SC: South Carolina Department of Health and Environmental Control; 1999 43. Thurman DJ, Sniezek JE, Johnson D, et al: Guidelines for Surveillance of Central Nervous System Injury. Atlanta: Centers for Disease Control and Prevention; 1995 44. Association for the Advancement of Automotive Medicine: The Abbreviated Injury Scale, 1990 Revision. Des Plaines, IL: Association for the Advancement of Automotive Medicine; 1990 45. Center for Injury Research and Policy of The Johns Hopkins University School of Public Health: ICDMAP-90 Software. Baltimore: The Johns Hopkins University and Tri-Analytics Inc; 1997 46. Barell V, Aharonson-Daniel L, Fingerhut L, et al: An introduction to the Barell body region by nature of injury diagnosis matrix. Inj Prev 2002;8:91-96 47. Elixhauser A, Steiner C, Harris DR, et al: Comorbidity measures for use with administrative data. Med Care 1998;36:8-27 48. Fleiss JL: Statistical Methods for Rates and Proportions, 2nd ed. New York: John Wiley & Sons; 1998:58-61 49. Breslow NE, Day NE: Statistical Methods in Cancer Re-
473
search, vol 1. The analyses of case-control studies. Lyon, France: International Agency for Research on Cancer (IARC scientific publications no. 32); 1980:192-246 50. Statistical Analytical Software, version 8.1. Cary, NC: SAS Institute Inc; 2000 51. Darlington GA: Collinearity. In: Armitage P, Colton T, eds. Encyclopedia of Biostatistics. Chichester, West Sussex, UK: John Wiley & Sons; 1998:788-789 52. Cubbin C, LeClere FB, Smith GS: Socioeconomic status and the occurrence of fatal and nonfatal injury in the United States. Am J Public Health 2000;90:70-77 53. Cubbin C, LeClere FB, Smith GS: Socioeconomic status and injury mortality: individual and neighborhood determinants. J Epidemiol Community Health 2000;54:517-524 54. Baker SP, O’Neill B, Ginsburg MJ, et al: The Injury Fact Book, 2nd ed. New York: Oxford University Press; 1992 55. The LEAP Study Group, MacKenzie EJ, Bosse MJ, et al: Characterization of patients with high-energy lower extremity trauma. J Orthop Trauma 2000;14:455-466 56. Marcin JP, Schembri MS, He J, et al: A population-based analysis of socioeconomic status and insurance status and their relationship with pediatric trauma hospitalization and mortality rates. Am J Public Health 2003;93:461-466 57. Mayberry RM, Mili F, Ofili E: Racial and ethnic differences in access to medical care. Med Care Res Rev 2000;7(suppl 1):108-145 58. Chandra NC, Ziegelstein RC, Rogers WJ, et al: Observations of the treatment of women in the United States with myocardial infarction. Arch Intern Med 1998;158:981-988 59. Harris DR, Andrews R, Elixhauser A: Racial and gender differences in use of procedures for black and white hospitalized adults. Ethn Dis 1997;7:91-105 60. Schneider EC, Leape LL, Weissman JS, et al: Racial differences in cardiac revascularization rates: does ’overuse’ explain higher rates among white patients? Ann Intern Med 2001;135:328337 61. Schulman KA, Berlin JA, Harless W, et al: The effect of race and sex on physicians’ recommendations for cardiac catheterization. N Engl J Med 1999;340:618-626 62. Giles WH, Anda R, Casper ML, et al: Race and sex differences in rates of invasive cardiac procedures in US hospitals, data from the national hospital discharge survey. Arch Intern Med 1995; 155:318-324 63. Schwab CW, Young G, Civil I, et al: DRG reimbursement for trauma: the decline of the trauma center (the use of ISS groupings as an early predictor of total hospital cost). J Trauma 1988;28:939945 64. Wrenn K: Sounding board: no insurance, no admission. N Engl J Med 1985;312:373-374 65. Morris JA, MacKenzie EJ, Edelstein SL: The effect of preexisting conditions on mortality in trauma patients. JAMA 1990;263: 1942-1946 66. McCarthy ML, MacKenzie EJ, Bosse MJ, et al: Functional status following orthopedic trauma in young women. J Trauma 1995;39:828-837 67. MacKenzie EJ, Hoyt DB, Sacra JC, et al: National inventory of hospital trauma centers. JAMA 2003;289:1515-1522 68. Hsia DC, Ahern CA, Ritchie BP, et al: Medicare reimbursement accuracy under the prospective payment system, 1985 to 1988. JAMA 1992;268:896-899 69. Goldfarb MG, Coffey RM: Change in Medicare case-mix index in the 1980s and the effect of the prospective payment system. Health Serv Res 1992;27:385-415 70. Assaf AR, Lapane KL, McKenney JL, et al: Possible influence of the prospective payment system on the assignment of discharge diagnoses for coronary heart disease. N Engl J Med 1993;329:931935