- Email: [email protected]
Race and Income Disparity in Ischemic Stroke Care: Nationwide Inpatient Sample Database, 2002 to 2008
Race and Income Disparity in Ischemic Stroke Care: Nationwide Inpatient Sample Database, 2002 to 2008 Matthew M. Kimball, MD,* Dan Neal, MS,† Michael F. Waters, MD, PhD,‡x and Brian L. Hoh, MD*
Background: Health care disparities exist between demographic groups with stroke. We examined whether patients of particular ethnicity or income levels experienced reduced access to or delays in receiving stroke care. Methods: We studied all admissions for ischemic stroke in the Nationwide Inpatient Sample (NIS) database between 2002 and 2008. We used statistical models to determine whether median income or race were associated with intravenous (IV) thrombolysis treatment, inhospital mortality, discharge disposition, hospital charges, and LOS in high- or low-volume hospitals. Results: There were a total of 477,474 patients with ischemic stroke: 10,781 (2.3%) received IV thrombolysis, and 380,400 (79.7%) were treated in high-volume hospitals. Race (P , .0001) and median income (P , .001) were significant predictors of receiving IV thrombolysis, and minorities and low-income patients were less likely to receive IV thrombolysis. Median income was a predictor of access to high-volume hospitals (P , .0001), with wealthier patients more likely to be treated in high-volume hospitals, which had lower mortality rates (P 5 .0002). Patients in high-volume hospitals were 1.84 times more likely to receive IV thrombolysis (P , .0001). Conclusions: African Americans, Hispanics, and low median income patients are less likely to receive IV thrombolysis for ischemic stroke. Low median income patients are less likely to be treated at high-volume hospitals. High-volume hospitals have lower mortality rates and a higher likelihood of treating patients with IV thrombolysis. There is evidence for an influence of socioeconomic status and racial disparity in the treatment of ischemic stroke. Key Words: Hospital charges—ischemic stroke—length of hospitalization—socioeconomic— thrombolytic. Ó 2014 by National Stroke Association
Individuals with lower median income or who are of certain ethnicities may have limited access to care for acute stroke.1,2-5 Intravenous (IV) thrombolysis is a treatment
From the *Departments of Neurosurgery; †Biostatistics; ‡Neurology; and xNeuroscience, University of Florida, Gainesville, Florida. Received February 19, 2012; revision received May 7, 2012; accepted June 4, 2012. Drs. Waters and Hoh were supported by the National Institutes of Health. Address correspondence to Matthew M. Kimball, MD, Department of Neurosurgery, University of Florida, Box 100265, Gainesville, FL 32610-0261. E-mail: [email protected]. 1052-3057/$ - see front matter Ó 2014 by National Stroke Association http://dx.doi.org/10.1016/j.jstrokecerebrovasdis.2012.06.004
option for acute ischemic stroke that is readily available, and the current guidelines recommend treatment within up to 4.5 hours after the onset of stroke symptoms.6-8 Given the narrow therapeutic window for thrombolytic therapy, timely transport to an appropriate medical facility is critical. Factors that contribute to disparity in stroke treatment may include a lack of awareness of stroke symptoms and the necessity for emergent care, social limitations, such as language barriers, inequality of income, and lack of insurance.2 Certain minorities or low-income groups may arrive in a delayed fashion, decreasing the likelihood that they will arrive within the time window for IV thrombolysis, or they may lack access to high-volume hospitals that achieve better outcomes in stroke.3-5
Journal of Stroke and Cerebrovascular Diseases, Vol. 23, No. 1 (January), 2014: pp 17-24
17
M.M. KIMBALL ET AL.
18
Materials and Methods We collected data regarding hospitalizations for ischemic stroke between 2002 and 2008 using the Nationwide Inpatient Sample (NIS) database obtained from the Agency for Healthcare Quality and Research’s Healthcare Cost and Utilization Project (Rockville, MD). These data included all patients with International Classification of Diseases, 9th revision (ICD-9) codes for ischemic stroke (e.g., 433.01, 433.11, 433.21, 433.31, 433.81, 433.91, 434.01, 434.11, 434.91, and 437.1). We studied the effect of race and income on: (1) treatment in a hospital with a high stroke case volume and (2) treatment with IV thrombolysis. We adjusted each analysis for age, gender, and medical comorbidity score. We also analyzed whether treatment in a hospital with a high stroke case volume has an effect on discharge disposition, hospital charges, or length of stay (LOS). The NIS database is the largest all-payer hospital inpatient database in the United States, which is an administrative claim–based database and contains data approximating a 20% stratified sample of US hospitals. For each sampled hospital, all inpatient admissions for the year are contained in the NIS, permitting annual case volumes for hospitals to be calculated for each year from 2002 to 2008. Hospital annual case volumes for stroke admissions were determined by ranking the total stroke admissions across all sampled hospitals. Each hospital’s volume of stroke admissions was defined as ‘‘high volume’’ if its stroke case volume met or exceeded the 67th percentile (148 stroke cases per year) and ‘‘low volume’’ if below this threshold. Treatment with IV thrombolysis was determined by ICD-9 code 99.10. There is some inherent limitation of using this code for thrombolysis outcome, because coding error(s) exist in the NIS database. We adjusted each analysis for age, gender, and medical comorbidity score. Medical comorbidity score has been described and validated previously.9 The NIS classifies median income into quartiles: low, low to middle, middle to high, and high. This is determined by the estimated median household income of residents in the patient’s ZIP code. Median ZIP code income has been shown to have the highest correlation with individual income, and median income has been seen as a sensible single aggregate measure to use to estimate socioeconomic status.10 These quartiles and values are updated annually in the NIS database. Race was classified as white, African American, Hispanic, Asian/Pacific Islander, Native American, or other. Secondary outcomes were: (1) inflation-adjusted hospital charges, (2) LOS, and (3) discharge disposition. We defined discharge disposition as ‘‘favorable’’ if the patient was sent home, transferred to a short-term facility, or left the hospital against medical advice. We defined discharge disposition as ‘‘poor’’ if the patient was sent to a long-term facility or to hospice, and discharge disposition as ‘‘deceased’’ if the patient died in the hospital, at home, or in another
medical facility. The primary payer was classified as Medicaid, Medicare, private insurance, self-pay, no charge, or other. Hospital bed size was classified as small, medium, or large. Hospital location was classified as Northeast, Midwest, South, and West.
Statistical Methods For the primary analysis, there were 2 outcome variables of interest: IV thrombolysis (yes or no) and stroke case volume of the hospital (high or low). There were 2 predictor variables of interest: race and median income. Race and median income were studied independently to avoid confounding effects. We used general estimating equations (GEEs; SAS PROC GENMOD function [version 9.1; SAS, Inc., Cary, NC), performing 4 analyses on the dataset that contained only hospitalizations in which the patient had suffered a stroke. We assumed a binomial distribution for the outcome variable and used a logit link function. Along with the primary predictor variable, we included gender, age, and comorbidity index as covariates. To account for the clustering of observations by hospital, we treated hospital as a repeated factor, and we assumed an exchangeable working correlation matrix. We retained the default convergence criteria set by the SAS system when fitting the model. The secondary analyses included 4 outcomes: (1) inhospital mortality; (2) total hospital charges; (3) LOS; and (4) discharge disposition. There were 2 predictors of interest: hospital stroke case volume and IV thrombolysis. For all outcome measures, our independent variables were the predictor of interest and gender, age, comorbidity score, year of admission, median income, primary payer, hospital size, and region. For the in-hospital mortality outcome, discharge disposition, and LOS, we used GEE, assuming a binary distribution and logit link function. For total charges, we first adjusted all charges for inflation by adding 3% per year for each year before 2008. For adjusted charges and LOS, we took the natural log of adjusted charges (or of LOS) as our outcome variable to meet the assumptions of a general linear model. We then used GEE, assuming a normal distribution and identity link function. For the discharge disposition outcome, we used GEE, assuming a multinomial distribution and a cumulative logit link function.
Results Between 2002 and 2008, there were 477,474 patients in the NIS database with ICD-9 codes for ischemic stroke (433.01, 433.11, 433.21, 433.31, 433.81, 433.91, 434.01, 434.11, 434.91, and 437.1). Patient demographics and hospital characteristics by race and median income by ZIP code are shown in Tables 1 and 2, respectively. Of 477,474 patients, 10,781 (2.3%) received IV thrombolysis and 380,400 (79.7%) were admitted to hospitals with high stroke case volumes.
Age, y (mean 6 SD) Gender Male Female Median income in the patient’s ZIP code Low Low to middle Middle to high High Primary payer Medicare Medicaid Private insurance Self-pay No charge Other Hospital bed size Small Medium Large Hospital location Rural Urban Hospital region Northeast Midwest South West Annual hospital case volume, n 5 1833 hospitals (mean 6 SD)* High-volume hospital (.147 cases per year) No Yes Thrombolytics used No Yes Comorbidity score, mean 6 SD
Overall (N 5 349,691)
White (n 5 249,110)
Black (n 5 64,659)
Hispanic (n 5 21,089)
Asian (n 5 5187)
Native American (n 5 1339)
Other (n 5 8307)
71.3 6 14.8
73.5 6 14.0
65.0 6 15.1
66.8 6 15.8
69.2 6 15.4
66.4 6 15.4
68.6 6 15.9
158,503 (45.3) 191,172 (54.7)
112,136 (45.0) 136,963 (55.0)
28,529 (44.1) 36,126 (55.9)
10,570 (50.1) 10,518 (49.9)
2585 (49.8) 2602 (50.2)
638 (47.7) 701 (52.3)
4045 (48.7) 4262 (51.3)
92,611 (30.7) 7137 (25.5) 68,327 (22.6) 64,021 (21.2)
50,644 (23.6) 58,196 (27.1) 53,438 (24.9) 52,806 (24.6)
30,045 (53.8) 11,887 (21.3) 8353 (15.0) 5553 (9.9)
8828 (48.0) 3947 (21.4) 3537 (19.2) 2098 (11.4)
651 (14.5) 1090 (24.3) 1003 (22.4) 1741 (38.8)
640 (53.1) 261 (21.6) 164 (13.6) 141 (11.7)
1803 (25.5) 1756 (24.8) 1832 (25.9) 1682 (23.8)
241,644 (69.2) 21,704 (6.2) 63,447 (18.2) 14,405 (4.1) 2056 (0.6) 5988 (1.7)
185,249 (74.4) 8415 (3.4) 43,952 (17.7) 6803 (2.7) 824 (0.3) 3620 (1.5)
36,528 (56.6) 8488 (13.2) 12,754 (19.8) 4567 (7.1) 702 (1.1) 1465 (2.3)
11,522 (54.7) 3018 (14.3) 3618 (17.2) 1930 (9.2) 411 (2.0) 577 (2.7)
2893 (55.8) 657 (12.7) 1147 (22.1) 350 (6.8) 30 (0.6) 108 (2.1)
800 (59.8) 159 (11.9) 223 (16.7) 90 (6.7) 17 (1.3) 48 (3.6)
4672 (56.3) 967 (11.7) 1753 (21.1) 665 (8.0) 72 (0.9) 170 (2.1)
36,929 (10.6) 88,450 (25.3) 224,169 (64.1)
27,984 (11.2) 63,219 (25.4) 157,801 (63.4)
5560 (8.6) 15,393 (23.8) 43,706 (67.6)
1469 (7.0) 5486 (26.0) 14,134 (67.0)
963 (18.6) 1662 (32.2) 2544 (49.2)
146 (10.9) 447 (33.4) 746 (55.7)
807 (9.7) 2243 (27.1) 5238 (63.2)
25,702 (11.0) 208,106 (89.0)
21,140 (12.6) 146,282 (87.4)
3065 (7.2) 39,768 (92.8)
761 (5.3) 13,604 (94.7)
229 (6.9) 3082 (93.1)
219 (33.6) 432 (66.4)
288 (5.5) 4938 (94.5)
102,462 (29.3) 48,908 (14.0) 180,558 (51.6) 17,763 (5.1) 150 6 192
77,140 (31.0) 38,878 (15.6) 119,994 (48.2) 13,098 (5.3) 379 6 326
14,464 (22.4) 7717 (11.9) 41,571 (64.3) 907 (1.4) 461 6 359
6032 (28.6) 351 (1.7) 13,227 (62.7) 1479 (7.0) 449 6 434
1672 (32.2) 171 (3.3) 1630 (31.4) 1714 (33.0) 363 6 312
301 (22.5) 135 (10.1) 734 (54.8) 169 (12.6) 323 6 262
2852 (34.3) 1656 (19.9) 3402 (41.0) 396 (4.8) 404 6 363
66,393 (19.0) 283,298 (81.0)
51,545 (20.7) 197,565 (79.3)
7902 (12.2) 56,757 (87.8)
3871 (18.4) 17,218 (81.6)
1070 (20.6) 4117 (79.4)
334 (24.9) 1005 (75.1)
1671 (20.1) 6636 (79.9)
341,596 (97.7) 8095 (2.3) 2.5 6 1.5
242,993 (97.5) 6117 (2.5) 2.5 6 1.5
63,484 (98.2) 1175 (1.8) 2.7 6 1.5
20,637 (97.9) 452 (2.1) 2.5 6 1.5
5056 (97.5) 131 (2.5) 2.2 6 1.4
1318 (98.4) 21 (1.6) 2.5 6 1.5
8108 (97.6) 199 (2.4) 2.4 6 1.5 19
Abbreviation: SD, standard deviation. All values are n (%) unless otherwise indicated. *Overall average is per hospital. Averages for individual races are per hospitalization.
THROMBOLYTIC USE: RACIAL AND INCOME DISPARITY
Table 1. Patient and hospital characteristics by race
20
Table 2. Patient and hospital characteristics by median income in the patient’s ZIP code Income classification Low (n 5 124,626)
Low to middle (n 5 108,330)
Middle to high (n 5 95,363)
High (n 5 83,441)
71.3 6 14.9
69.6 6 15.0
71.2 6 14.8
72.0 6 14.7
73.3 6 14.6
186,253 (45.2) 225,478 (54.8)
55,815 (44.8) 68,800 (55.2)
49,213 (45.4) 59,106 (54.6)
43,155 (45.3) 52,203 (54.7)
38,070 (45.6) 45,369 (54.4)
282,833 (68.8) 25,279 (6.2) 77,016 (18.7) 16,722 (4.1) 1996 (0.5) 7361 (1.8)
83,428 (67.1) 11,959 (9.6) 18,741 (15.1) 6714 (5.4) 898 (0.7) 2657 (2.1)
74,665 (69.0) 6585 (6.1) 19,627 (18.1) 4717 (4.4) 531 (0.5) 2062 (1.9)
66,105 (69.4) 4315 (4.5) 19,637 (20.6) 3234 (3.4) 367 (0.4) 1579 (1.7)
58,635 (70.3) 2420 (2.9) 19,011 (22.8) 2057 (2.5) 200 (0.2) 1063 (1.3)
43,190 (10.5) 101,667 (24.7) 266,425 (64.8)
10,259 (8.3) 30,091 (24.2) 84,031 (67.6)
11,487 (10.6) 25,989 (24.0) 70,707 (65.4)
11,645 (12.2) 22,867 (24.0) 60,770 (63.8)
9799 (11.7) 22,720 (27.2) 50,917 (61.0)
33,539 (12.9) 225,725 (87.1)
18,450 (23.9) 56,682 (76.1)
11,122 (16.6) 56,048 (83.4)
3118 (5.1) 58,123 (94.9)
849 (1.6) 52,872 (98.4)
89,651 (21.8) 97,615 (23.7) 193,593 (47.0) 30,901 (7.5) 150 6 192
19,167 (15.4) 23,400 (18.8) 75,479 (60.6) 6580 (5.3) 362 6 322
18,913 (17.5) 28,097 (25.9) 52,575 (48.5) 8745 (8.1) 354 6 325
20,698 (21.7) 26,864 (28.2) 38,997 (40.9) 8804 (9.2) 399 6 343
30,873 (37.0) 19,254 (23.1) 26,542 (31.8) 6772 (8.1) 422 6 310
81,712 (19.8) 330,048 (80.2)
30,158 (24.2) 94,468 (75.8)
25,079 (23.2) 83,251 (76.9)
16,621 (17.4) 78,742 (82.6)
9854 (11.8) 73,587 (88.2)
401,923 (97.6) 9837 (2.4) 2.5 6 1.5
122,175 (98.0) 2451 (2.0) 2.5 6 1.5
105,847 (97.7) 2483 (2.3) 2.5 6 1.5
92,874 (97.4) 2489 (2.6) 2.5 6 1.5
81,027 (97.1) 2414 (2.9) 2.4 6 1.5
Abbreviation: SD, standard deviation. All values are n (%) unless otherwise indicated. *Overall average is per hospital. Averages for individual income levels are per hospitalization.
M.M. KIMBALL ET AL.
Age, y (mean 6 SD) Gender Male Female Primary payer Medicare Medicaid Private insurance Self-pay No charge Other Hospital bed size Small Medium Large Hospital location Rural Urban Hospital region Northeast Midwest South West Annual hospital case volume, n 5 1833 hospitals (mean 6 SD)* High-volume hospital (.147 cases per year) No Yes Thrombolytics used No Yes Comorbidity score, mean 6 SD
Overall (N 5 411,760)
THROMBOLYTIC USE: RACIAL AND INCOME DISPARITY
Effect of Race and Income on IV Thrombolysis Race was a highly significant predictor for IV thrombolysis (P , .0001). Relative to whites, the odds ratio (OR) of IV thrombolysis for African Americans was 0.58 (95% confidence interval [CI] 0.52-0.66; P , .001), for Native Americans 0.59 (95% CI 0.39-0.91; P 5 .0166), and for Hispanics 0.76 (95% CI 0.64-0.91; P 5 .0020). The odds for Asians and ‘‘other’’ were not significantly different than whites. The odds of IV thrombolysis by race are seen in Table 3. Median income was a significant predictor of IV thrombolysis (P , .001). Compared to patients from the poorest ZIP codes, the odds of IV thrombolysis for patients from the wealthiest ZIP codes was 1.55 (95% CI 1.371.76; P , .0001), for middle to high 1.48 (95% CI 1.331.64; P , .0001), and for low to middle 1.26 (95% CI 1.14-1.39; P , .0001). The odds of IV thrombolysis by median income are seen in Table 4.
21
Table 4. Odds of intravenous thrombolysis use by median income when compared to the poorest incomes Median income quartile
OR
95% CI
P value
High Middle to high Low to middle
1.55 1.48 1.26
1.37-1.76 1.33-1.64 1.14-1.39
,.0001 ,.0001 ,.0001
Abbreviations: CI, confidence interval; OR, odds ratio.
Relationship between Race and Thrombolysis Use and ICD-9 Codes for Stroke
Relative to whites, Native Americans were less likely to be treated at a high stroke case volume hospital (OR 0.73; 95% CI 0.65-0.83; P ,.0001), while African Americans (OR 1.8; 95% CI 1.76-1.85; P , .0001) and Hispanics (OR 1.12; 95% CI 1.08-1.17; P , .0001) were more likely to be treated at a high stroke case volume hospital. The odds of treatment at a high-volume hospital for Asians and those in race category ‘‘other’’ were not significantly different than whites. The odds of being treated in a high stroke case volume hospital by race can be seen in Table 5. Median income was a predictor of treatment at a high stroke case volume hospital (P , .0001). Relative to patients from the poorest ZIP codes, the OR of being treated in a high-volume hospital for those in the wealthiest ZIP codes was 2.54 (95% CI 2.48-2.60; P , .0001), for middle to high income ZIP codes 1.57 (95% CI 1.53-1.60; P , .0001), and for low to middle income ZIP codes 1.08 (95% CI 1.061.10; P , .0001) The odds of being treated in a high stroke case volume hospital by median income can be seen in Table 6.
ICD-9 codes were divided into 5 groups: group 1 (433.01, 433.11, and 433.21); group 2 (433.31, 433.81, and 433.91); group 3 (434.01); group 4 (434.11); and group 5 (434.91 and 437.1). The proportion of patients that received thrombolysis varied significantly across the 5 diagnosis groups, with group 5 less likely to get tissue plasminogen activator (tPA; P , .0001). The odd ratios of receiving thrombolysis among the groups when compared to group 5 was 1.9 (95% CI 1.7-2.1; P , .0001) for group 1, 1.5 (95% CI 1.2-2.0; P 5 .0007) for group 2, 4.2 (95% CI 3.6-4.8; P , .0001) for group 3, and 2.5 (95% CI 2.3-2.7; P , .0001) for group 4. The distribution of race also varied significantly across the groups, with group 5 having a higher percentage of African Americans (P , .0001). Nevertheless, even when controlling for diagnosis group, race remains a significant predictor of thrombolytic use, with ORs only slightly less extreme than those reported in the primary analysis. This result is largely caused by the fact that 80% of the original hospitalizations fall into group 5, and therefore dramatic differences in thrombolysis and race between the relatively small groups 1 to 4 and group 5 would be required to substantially affect the conclusions of the primary analysis; the differences in thrombolysis and race among the groups, while statistically significant, are far from dramatic. The vast majority of stroke patients (80%) have 1 of 2 ICD-9 diagnoses (434.91 or 437.1), so any grouping by stroke type is unlikely to change the basic conclusions of the analysis.
Table 3. Odds of intravenous thrombolysis use by race when compared to white patients
Table 5. Odds of treatment at high stroke case volume hospitals by race when compared to white patients
Effect of Race and Income on Access to High Stroke Case Volume Hospitals
Race
OR
95% CI
P value
Race
OR
95% CI
P value
African American Native American Hispanic Asian Other
0.58 0.59 0.76 0.96 0.91
0.52-0.66 0.39-0.91 0.64-0.91 0.74-1.23 0.75-1.11
,.001 .017 .002 .725 .343
Native American African American Hispanic Other Asian
0.73 1.80 1.12 1.01 1.00
0.65-0.83 1.76-1.85 1.08-1.17 0.71-1.43 0.54-1.84
,.0001 ,.0001 ,.0001 .963 .987
Abbreviations: CI, confidence interval; OR, odds ratio.
Abbreviations: CI, confidence interval; OR, odds ratio.
M.M. KIMBALL ET AL.
22
Table 6. Odds of treatment at high stroke case volume hospitals by median income when compared to the poorest incomes Median income quartile
OR
95% CI
P value
High Middle to high Low to middle
2.54 1.57 1.08
2.48-2.60 1.53-1.60 1.06-1.10
,.0001 ,.0001 ,.0001
Abbreviations: CI, confidence interval; OR, odds ratio.
Effect of Hospital Stroke Case Volume on Mortality, Outcome, Hospital Charges, and LOS Hospital stroke case volume was a significant predictor of mortality and receiving IV thrombolysis. Patients treated in low-volume hospitals were more likely to die in the hospital (OR 1.10; 95% CI 1.05-1.16; P 5 .0002). Patients treated in high-volume hospitals had 1.84 times the odds of IV thrombolysis than those in low-volume hospitals (95% CI 1.55-2.18; P , .0001). Hospital volume was a predictor of total charges (P , .0001) and LOS (P , .0001). Average inflationadjusted charges for patients treated in high-volume hospitals were estimated to be $21,163 (95% CI $20,131$22,026) compared to $16,155 (95% CI $15,522-$16,815) in low-volume hospitals. The average LOS for patients in high-volume hospitals was estimated to be 4.7 days (95% CI 4.6-4.9]) compared to 4.3 days (95% CI 4.2-4.4) for low-volume hospitals. These estimated averages for charges and LOS are converted from a log scale, which was the scale of analysis for both outcomes.
Discussion Stroke represents the fourth leading cause of death and the leading cause of adult disability in the United States.11 Treatments for acute ischemic stroke include IV thrombolysis, intra-arterial thrombolysis, angioplasty, and thrombectomy. Previous studies have shown that certain factors were associated with higher rates of thrombolysis. In a previous study using the NIS database between 1999 and 2004, Schumacher et al12 reported that thrombolysis rates are highest at academic medical centers and hospitals that have higher stroke treatment volume, and also found that thrombolysis rates were higher in white patients.12 Our study has shown that African American, Hispanic, Native American, and low-income patients have decreased access to IV thrombolysis, and those who live in poorer neighborhoods have decreased access to high stroke case volume hospitals. Part of the disparity may be that a disproportionate number of certain minorities and the poor live in rural areas, and these areas are often remote from large medical centers. This is likely the case for Native Americans, whose largest populations are
located in the remote Four Corners region of the southwest United States and rural Oklahoma. Approximately 25% of the US population lives in rural areas, where they tend to be older, have a higher prevalence of stroke risk factors, and lack access to preventive care.9,13,14 Remote location does not appear to be the only factor. Large populations of minorities and low-income patients live in urban centers, yet our study found disparities in stroke treatment among race and income level across all hospital locations and types. Racial and ethnic minorities in the United States now represent 28% of the population. African American men have the lowest average life expectancy (61.7 years), and the most significant factor leading to this disparity is cardiac disease and stroke.15 Other studies have shown that race and socioeconomic disparities exist for access to preventive health care and availability of acute stroke treatment.1-3,11,13-16,17,18,19,20-25 In Washington, DC, African Americans were less likely to arrive within 3 hours of stroke symptoms and to meet inclusion criteria for IV thrombolysis.20 In patients meeting criteria, African Americans were half as likely to receive IV thrombolysis as whites.20 Other studies have shown that minorities who present with stroke symptoms have longer wait times in emergency department waiting rooms than whites,3,23 and also have a wariness toward the medical system,2 which may translate to lower rates of consenting to IV thrombolysis. One study21 evaluated stroke patients admitted to academic medical centers to see if ethnicity was an influence on treatment. Of those meeting criteria for IV thrombolysis, African Americans were one-fifth as likely to receive that treatment when compared to whites (P 5 .001).21 In the present study, African Americans and Hispanics were more likely to be evaluated in high-volume hospitals where IV thrombolysis was more likely, but less likely to receive IV thrombolysis than whites. This may be because minorities are presenting to high-volume hospitals in a delayed fashion. In our study, patients from high-income ZIP codes are more likely to receive IV thrombolysis; however, patients with lower income have a higher prevalence of stroke risk factors and stroke incidence.2,4,17-18,24,26 In the United States, African Americans and Hispanics on average, have lower income levels than whites, and there may be an association between race and income.24 We attempted to clarify this by studying race and income independently from one another. High stroke volume hospitals are consistently associated with lower stroke mortality.4 High-volume hospitals are a source of many subspecialists, including stroke neurologists, neurointerventionalists, and neurosurgeons. When treated and evaluated by a neurologist, acute stroke patients are 3.7 times more likely to receive IV thrombolysis.2 This was consistent among African American and white patients; however, only 10.6% of African American patients were treated by a neurologist
THROMBOLYTIC USE: RACIAL AND INCOME DISPARITY
compared with 20.3% of white patients. Hospitals that have been declared designated primary stroke centers have been shown to have lower mortality and morbidity rates and a greater use of thrombolytic therapy.4,10,27,28,29 Since the implementation of designated stroke centers, many states have improved the use of streamlining patients with stroke symptoms to be directly transferred to primary stroke centers. More recently, California has used a county-wide ‘‘spoke-and-hub’’ stroke system that incorporates several comprehensive stroke center recommendations through many smaller linked hospitals to increase thrombolysis rates and the use of endovascular treatments.30 In South Carolina, clinicians have been using telemedicine to increase the thrombolysis rates in remote hospitals through contact with a primary stroke center.31 In the present study, IV thrombolysis was more likely if treated at a high-volume hospital, and high-volume hospitals were associated with better outcomes. A NIS study conducted between 1999 and 2002 reached similar conclusions; 70% of patients who received thrombolysis were treated at high-volume centers.32 Low-income patients were less likely to be treated in a high-volume hospital, suggesting that a socioeconomic disparity in health care exists for ischemic stroke patients. There is a possibility that some patients arrived at small medical centers first and received IV thrombolysis in the emergency department there, and were later transferred to larger stroke centers for continued care. The NIS unit of analysis is the hospitalization, not the patient. Anything that happens to the patient during a specific hospitalization is attributed to that hospital. Anything that happens to the patient before or after the hospitalization is unavailable in the data recorded for that hospitalization, and the NIS offers no way to track a patient from one hospitalization to another. If a patient visited a low-volume hospital and received recombinant tPA there, the recombinant tPA use would be attributed to that hospital, regardless of whether the patient was transferred. If the patient was transferred to another hospital, the previous recombinant tPA would not show up in the record for the new hospitalization. Our data suggest that a disparity exists for minorities and lower income patients with ischemic stroke. Lowincome patients lack access to high-volume hospitals, which are more likely to treat with IV thrombolysis, and are associated with improved outcomes. African Americans and Hispanics are more likely to be seen at high-volume hospitals but are less likely to be treated with IV thrombolysis. Delayed arrival to appropriate medical facilities may be an important factor for minority and lower income patients, making them ineligible for thrombolysis. With the increasing number of primary stroke centers and the availability of telemedicine for smaller rural facilities, there will likely continue to be an increase in the
23
number of patients arriving within the 4.5-hour time window for IV thrombolysis. However, researchers also need to focus on educating the public about the signs and symptoms of stroke and the acuity in which they need to be treated. Health care providers must become aware of the disparity that exists in these populations, and efforts should be focused on narrowing the gap.
Limitations There are limitations to the study. This is a retrospective study, and there is therefore the potential for selection bias. A prospective randomized trial cannot be performed, because a patient cannot be randomized to race or income level. There may be risks of coding error(s). In the NIS, a potential exists for variability in coding. The NIS database does not contain data on neurologic conditions of patients, and the analysis could not be adjusted for stroke severity. Stroke severity as recorded by the National Institutes of Health Stroke Score (NIHSS) or admission or discharge disability scale as recorded by modified Rankin Scale (mRS) are not available in the NIS database—therefore, we could not determine whether these were factors that had an effect on our findings. There is also a lack of information regarding inpatient care for these patients, such as other medical or surgical problems during their hospitalization. Exclusion criteria for IV thrombolysis cannot be determined from the NIS database, including time of stroke symptom onset. It would be important to know if the limitation of receiving IV thrombolysis was arriving outside of the time window, medical contraindications, or refusal of consent for treatment. We only used IV thrombolysis as an outcome for treatment. The use of code 99.10 for IV thrombolysis also has a limitation, because it could have been miscoded and may be underreported. Other endovascular procedures were not included in this study because of inconsistent coding of the procedures.
References 1. Zweifler RM, McClure LA, Howard VJ, et al. Racial and geographic differences in prevalence, awareness, treatment and control of dyslipidemia: the reasons for geographic and racial differences in stroke (REGARDS) study. Neuroepidemiology 2011;37:39-44. 2. Cruz-Flores S, Rabinstein A, Biller J, et al. Racial-ethnic disparities in stroke care: The American experience: A statement for healthcare professionals from the American Heart Association/American Stroke Association. Stroke 2011;42:2091-2116. 3. Lacy CR, Suh DC, Bueno M, et al. Delay in presentation and evaluation for acute stroke: Stroke Time Registry for Outcomes Knowledge and Epidemiology (S.T.R.O.K.E.). Stroke 2001;32:63-69. 4. Saposnik G, Jeerakathil T, Selchen D, et al. Socioeconomic status, hospital volume, and stroke fatality in Canada. Stroke Outcome Research Canada Working Group. Stroke 2008;39:3360-3366.
24 5. Saposnik G, Baibergenova A, O’Donnell M, et al. Hospital volume and stroke outcome: Does it matter? Neurology 2007;69:1142-1151. 6. Marler JR, Tilley BC, Lu M, et al. National Institute of Neurological Disorders and Stroke: Early stroke treatment associated with better outcome: The NINDS rt-PA stroke study. Neurology 2000;55:1649-1655. 7. Del Zoppo GJ, Saver JL, Jauch EC, et al. Expansion of the time window for treatment of acute ischemic stroke with intravenous tissue plasminogen activator: A science advisory from the American Heart Association/American Stroke Association. Stroke 2009;40:2945-2948. 8. Wahlgren N, Ahmed N, D avalos A, et al. Thrombolysis with alteplase 3-4.5 h after acute ischaemic stroke (SITSISTR): An observational study. Lancet 2008;372:1303-1309. 9. Elixhauser A, Steiner C, Harris DR, et al. Comorbidity measures for use with administrative data. Med Care 1998;36:8-27. 10. Geronimus AT, Bound J. Use of census-based aggregate variables to proxy for socioeconomic group: evidence from national samples. Am J Epidemiol 1998;148:475-486. 11. Lloyd-Jones D, Adams RJ, Brown TM, et al. Executive summary: Heart disease and stroke statistics—2010 update: A report from the American Heart Association. American Heart Association Statistics Committee and Stroke Statistics Subcommittee. Circulation 2010;121:948-954. 12. Schumacher HC, Bateman BT, Boden-Albala B, et al. Use of thrombolysis in acute ischemic stroke: Analysis of the Nationwide Inpatient Sample 1999 to 2004. Ann Emerg Med 2007;50:99-107. 13. Pearson TA, Lewis C. Rural epidemiology: Insights from a rural population laboratory. Am J Epidemiol 1998; 148:949-957. 14. Eberhardt MS, Pamuk ER. The importance of place of residence: Examining health in rural and nonrural areas. Am J Public Health 2004;94:1682-1686. 15. Murray CJ, Kulkarni SC, Michaud C, et al. Eight Americas: Investigating mortality disparities across races, counties, and race-counties in the United States. PLoS Med 2006; 3:e260. 16. Levine DA, Neidecker MV, Kiefe CI, et al. Racial/ethnic disparities in access to physician care and medications among US stroke survivors. Neurology 2011;76:53-61. 17. Liao Y, Greenlund KJ, Croft JB, et al. Factors explaining excess stroke prevalence in the US Stroke Belt. Stroke 2009;40:3336-3341. 18. Cesaroni G, Agabiti N, Forastiere F, et al. Socioeconomic differences in stroke incidence and prognosis under a universal healthcare system. Stroke 2009;40:2812-2819.
M.M. KIMBALL ET AL. 19. Kapral MK, Wang H, Mamdani M, et al. Effect of socioeconomic status on treatment and mortality after stroke. Stroke 2002;33:268-273. 20. Hsia AW, Edwards DF, Morgenstern LB, et al. Racial disparities in tissue plasminogen activator treatment rate for stroke: A population-based study. Stroke 2011; 42:2217-2221. 21. Johnston SC, Fung LH, Gillum LA, et al. Utilization of intravenous tissue-type plasminogen activator for ischemic stroke at academic medical centers: The influence of ethnicity. Stroke 2001;32:1061-1068. 22. Leira EC, Hess DC, Torner JC, et al. Rural-urban differences in acute stroke management practices: A modifiable disparity. Arch Neurol 2008;65:887-891. 23. Karve SJ, Balkrishnan R, Mohammad YM, et al. Racial/ ethnic disparities in emergency department waiting time for stroke patients in the United States. J Stroke Cerebrovasc Dis 2011;20:30-40. 24. Howard G, Anderson RT, Russell G, et al. Race, socioeconomic status, and cause-specific mortality. Ann Epidemiol 2000;10:214-223. 25. Trimble B, Morgenstern LB. Stroke in minorities. Neurol Clin 2008;26:1177-1190. 26. Reiner-Deitemyer V, Teuschl Y, Matz K, et al. Austrian Stroke Unit Registry Collaborators. Helicopter transport of stroke patients and its influence on thrombolysis rates: data from the Austrian Stroke Unit Registry. Stroke 2011; 42:1295-1300. 27. Gropen T, Magdon-Ismail Z, Day D, et al. Regional implementation of the stroke systems of care model: Recommendations of the northeast cerebrovascular consortium. Stroke 2009;40:1793-1802. 28. Xian Y, Holloway RG, Chan PS, et al. Association between stroke center hospitalization for acute ischemic stroke and mortality. JAMA 2011;305:373-380. 29. Prabhakaran S, McNulty M, O’Neill K, et al. Intravenous thrombolysis for stroke increases over time at primary stroke centers. Stroke 2012;43:875-877. 30. Cramer SC, Stradling D, Brown DM, et al. Organization of a United States county system for comprehensive acute stroke care. Stroke 2012;43:1089-1093. 31. Lazaridis C, Desantis SM, Jauch EC, et al. Telestroke in South Carolina. J Stroke Cerebrovasc Dis 2011 Dec 22 [Epub ahead of print]. 32. Bateman BT, Schumacher HC, Boden-Albala B, et al. Factors associated with in-hospital mortality after administration of thrombolysis in acute ischemic stroke patients: An analysis of the nationwide inpatient sample 1999 to 2002. Stroke 2006;37:440-446.
Background: Health care disparities exist between demographic groups with stroke. We examined whether patients of particular ethnicity or income levels experienced reduced access to or delays in receiving stroke care. Methods: We studied all admissions for ischemic stroke in the Nationwide Inpatient Sample (NIS) database between 2002 and 2008. We used statistical models to determine whether median income or race were associated with intravenous (IV) thrombolysis treatment, inhospital mortality, discharge disposition, hospital charges, and LOS in high- or low-volume hospitals. Results: There were a total of 477,474 patients with ischemic stroke: 10,781 (2.3%) received IV thrombolysis, and 380,400 (79.7%) were treated in high-volume hospitals. Race (P , .0001) and median income (P , .001) were significant predictors of receiving IV thrombolysis, and minorities and low-income patients were less likely to receive IV thrombolysis. Median income was a predictor of access to high-volume hospitals (P , .0001), with wealthier patients more likely to be treated in high-volume hospitals, which had lower mortality rates (P 5 .0002). Patients in high-volume hospitals were 1.84 times more likely to receive IV thrombolysis (P , .0001). Conclusions: African Americans, Hispanics, and low median income patients are less likely to receive IV thrombolysis for ischemic stroke. Low median income patients are less likely to be treated at high-volume hospitals. High-volume hospitals have lower mortality rates and a higher likelihood of treating patients with IV thrombolysis. There is evidence for an influence of socioeconomic status and racial disparity in the treatment of ischemic stroke. Key Words: Hospital charges—ischemic stroke—length of hospitalization—socioeconomic— thrombolytic. Ó 2014 by National Stroke Association
Individuals with lower median income or who are of certain ethnicities may have limited access to care for acute stroke.1,2-5 Intravenous (IV) thrombolysis is a treatment
From the *Departments of Neurosurgery; †Biostatistics; ‡Neurology; and xNeuroscience, University of Florida, Gainesville, Florida. Received February 19, 2012; revision received May 7, 2012; accepted June 4, 2012. Drs. Waters and Hoh were supported by the National Institutes of Health. Address correspondence to Matthew M. Kimball, MD, Department of Neurosurgery, University of Florida, Box 100265, Gainesville, FL 32610-0261. E-mail: [email protected]. 1052-3057/$ - see front matter Ó 2014 by National Stroke Association http://dx.doi.org/10.1016/j.jstrokecerebrovasdis.2012.06.004
option for acute ischemic stroke that is readily available, and the current guidelines recommend treatment within up to 4.5 hours after the onset of stroke symptoms.6-8 Given the narrow therapeutic window for thrombolytic therapy, timely transport to an appropriate medical facility is critical. Factors that contribute to disparity in stroke treatment may include a lack of awareness of stroke symptoms and the necessity for emergent care, social limitations, such as language barriers, inequality of income, and lack of insurance.2 Certain minorities or low-income groups may arrive in a delayed fashion, decreasing the likelihood that they will arrive within the time window for IV thrombolysis, or they may lack access to high-volume hospitals that achieve better outcomes in stroke.3-5
Journal of Stroke and Cerebrovascular Diseases, Vol. 23, No. 1 (January), 2014: pp 17-24
17
M.M. KIMBALL ET AL.
18
Materials and Methods We collected data regarding hospitalizations for ischemic stroke between 2002 and 2008 using the Nationwide Inpatient Sample (NIS) database obtained from the Agency for Healthcare Quality and Research’s Healthcare Cost and Utilization Project (Rockville, MD). These data included all patients with International Classification of Diseases, 9th revision (ICD-9) codes for ischemic stroke (e.g., 433.01, 433.11, 433.21, 433.31, 433.81, 433.91, 434.01, 434.11, 434.91, and 437.1). We studied the effect of race and income on: (1) treatment in a hospital with a high stroke case volume and (2) treatment with IV thrombolysis. We adjusted each analysis for age, gender, and medical comorbidity score. We also analyzed whether treatment in a hospital with a high stroke case volume has an effect on discharge disposition, hospital charges, or length of stay (LOS). The NIS database is the largest all-payer hospital inpatient database in the United States, which is an administrative claim–based database and contains data approximating a 20% stratified sample of US hospitals. For each sampled hospital, all inpatient admissions for the year are contained in the NIS, permitting annual case volumes for hospitals to be calculated for each year from 2002 to 2008. Hospital annual case volumes for stroke admissions were determined by ranking the total stroke admissions across all sampled hospitals. Each hospital’s volume of stroke admissions was defined as ‘‘high volume’’ if its stroke case volume met or exceeded the 67th percentile (148 stroke cases per year) and ‘‘low volume’’ if below this threshold. Treatment with IV thrombolysis was determined by ICD-9 code 99.10. There is some inherent limitation of using this code for thrombolysis outcome, because coding error(s) exist in the NIS database. We adjusted each analysis for age, gender, and medical comorbidity score. Medical comorbidity score has been described and validated previously.9 The NIS classifies median income into quartiles: low, low to middle, middle to high, and high. This is determined by the estimated median household income of residents in the patient’s ZIP code. Median ZIP code income has been shown to have the highest correlation with individual income, and median income has been seen as a sensible single aggregate measure to use to estimate socioeconomic status.10 These quartiles and values are updated annually in the NIS database. Race was classified as white, African American, Hispanic, Asian/Pacific Islander, Native American, or other. Secondary outcomes were: (1) inflation-adjusted hospital charges, (2) LOS, and (3) discharge disposition. We defined discharge disposition as ‘‘favorable’’ if the patient was sent home, transferred to a short-term facility, or left the hospital against medical advice. We defined discharge disposition as ‘‘poor’’ if the patient was sent to a long-term facility or to hospice, and discharge disposition as ‘‘deceased’’ if the patient died in the hospital, at home, or in another
medical facility. The primary payer was classified as Medicaid, Medicare, private insurance, self-pay, no charge, or other. Hospital bed size was classified as small, medium, or large. Hospital location was classified as Northeast, Midwest, South, and West.
Statistical Methods For the primary analysis, there were 2 outcome variables of interest: IV thrombolysis (yes or no) and stroke case volume of the hospital (high or low). There were 2 predictor variables of interest: race and median income. Race and median income were studied independently to avoid confounding effects. We used general estimating equations (GEEs; SAS PROC GENMOD function [version 9.1; SAS, Inc., Cary, NC), performing 4 analyses on the dataset that contained only hospitalizations in which the patient had suffered a stroke. We assumed a binomial distribution for the outcome variable and used a logit link function. Along with the primary predictor variable, we included gender, age, and comorbidity index as covariates. To account for the clustering of observations by hospital, we treated hospital as a repeated factor, and we assumed an exchangeable working correlation matrix. We retained the default convergence criteria set by the SAS system when fitting the model. The secondary analyses included 4 outcomes: (1) inhospital mortality; (2) total hospital charges; (3) LOS; and (4) discharge disposition. There were 2 predictors of interest: hospital stroke case volume and IV thrombolysis. For all outcome measures, our independent variables were the predictor of interest and gender, age, comorbidity score, year of admission, median income, primary payer, hospital size, and region. For the in-hospital mortality outcome, discharge disposition, and LOS, we used GEE, assuming a binary distribution and logit link function. For total charges, we first adjusted all charges for inflation by adding 3% per year for each year before 2008. For adjusted charges and LOS, we took the natural log of adjusted charges (or of LOS) as our outcome variable to meet the assumptions of a general linear model. We then used GEE, assuming a normal distribution and identity link function. For the discharge disposition outcome, we used GEE, assuming a multinomial distribution and a cumulative logit link function.
Results Between 2002 and 2008, there were 477,474 patients in the NIS database with ICD-9 codes for ischemic stroke (433.01, 433.11, 433.21, 433.31, 433.81, 433.91, 434.01, 434.11, 434.91, and 437.1). Patient demographics and hospital characteristics by race and median income by ZIP code are shown in Tables 1 and 2, respectively. Of 477,474 patients, 10,781 (2.3%) received IV thrombolysis and 380,400 (79.7%) were admitted to hospitals with high stroke case volumes.
Age, y (mean 6 SD) Gender Male Female Median income in the patient’s ZIP code Low Low to middle Middle to high High Primary payer Medicare Medicaid Private insurance Self-pay No charge Other Hospital bed size Small Medium Large Hospital location Rural Urban Hospital region Northeast Midwest South West Annual hospital case volume, n 5 1833 hospitals (mean 6 SD)* High-volume hospital (.147 cases per year) No Yes Thrombolytics used No Yes Comorbidity score, mean 6 SD
Overall (N 5 349,691)
White (n 5 249,110)
Black (n 5 64,659)
Hispanic (n 5 21,089)
Asian (n 5 5187)
Native American (n 5 1339)
Other (n 5 8307)
71.3 6 14.8
73.5 6 14.0
65.0 6 15.1
66.8 6 15.8
69.2 6 15.4
66.4 6 15.4
68.6 6 15.9
158,503 (45.3) 191,172 (54.7)
112,136 (45.0) 136,963 (55.0)
28,529 (44.1) 36,126 (55.9)
10,570 (50.1) 10,518 (49.9)
2585 (49.8) 2602 (50.2)
638 (47.7) 701 (52.3)
4045 (48.7) 4262 (51.3)
92,611 (30.7) 7137 (25.5) 68,327 (22.6) 64,021 (21.2)
50,644 (23.6) 58,196 (27.1) 53,438 (24.9) 52,806 (24.6)
30,045 (53.8) 11,887 (21.3) 8353 (15.0) 5553 (9.9)
8828 (48.0) 3947 (21.4) 3537 (19.2) 2098 (11.4)
651 (14.5) 1090 (24.3) 1003 (22.4) 1741 (38.8)
640 (53.1) 261 (21.6) 164 (13.6) 141 (11.7)
1803 (25.5) 1756 (24.8) 1832 (25.9) 1682 (23.8)
241,644 (69.2) 21,704 (6.2) 63,447 (18.2) 14,405 (4.1) 2056 (0.6) 5988 (1.7)
185,249 (74.4) 8415 (3.4) 43,952 (17.7) 6803 (2.7) 824 (0.3) 3620 (1.5)
36,528 (56.6) 8488 (13.2) 12,754 (19.8) 4567 (7.1) 702 (1.1) 1465 (2.3)
11,522 (54.7) 3018 (14.3) 3618 (17.2) 1930 (9.2) 411 (2.0) 577 (2.7)
2893 (55.8) 657 (12.7) 1147 (22.1) 350 (6.8) 30 (0.6) 108 (2.1)
800 (59.8) 159 (11.9) 223 (16.7) 90 (6.7) 17 (1.3) 48 (3.6)
4672 (56.3) 967 (11.7) 1753 (21.1) 665 (8.0) 72 (0.9) 170 (2.1)
36,929 (10.6) 88,450 (25.3) 224,169 (64.1)
27,984 (11.2) 63,219 (25.4) 157,801 (63.4)
5560 (8.6) 15,393 (23.8) 43,706 (67.6)
1469 (7.0) 5486 (26.0) 14,134 (67.0)
963 (18.6) 1662 (32.2) 2544 (49.2)
146 (10.9) 447 (33.4) 746 (55.7)
807 (9.7) 2243 (27.1) 5238 (63.2)
25,702 (11.0) 208,106 (89.0)
21,140 (12.6) 146,282 (87.4)
3065 (7.2) 39,768 (92.8)
761 (5.3) 13,604 (94.7)
229 (6.9) 3082 (93.1)
219 (33.6) 432 (66.4)
288 (5.5) 4938 (94.5)
102,462 (29.3) 48,908 (14.0) 180,558 (51.6) 17,763 (5.1) 150 6 192
77,140 (31.0) 38,878 (15.6) 119,994 (48.2) 13,098 (5.3) 379 6 326
14,464 (22.4) 7717 (11.9) 41,571 (64.3) 907 (1.4) 461 6 359
6032 (28.6) 351 (1.7) 13,227 (62.7) 1479 (7.0) 449 6 434
1672 (32.2) 171 (3.3) 1630 (31.4) 1714 (33.0) 363 6 312
301 (22.5) 135 (10.1) 734 (54.8) 169 (12.6) 323 6 262
2852 (34.3) 1656 (19.9) 3402 (41.0) 396 (4.8) 404 6 363
66,393 (19.0) 283,298 (81.0)
51,545 (20.7) 197,565 (79.3)
7902 (12.2) 56,757 (87.8)
3871 (18.4) 17,218 (81.6)
1070 (20.6) 4117 (79.4)
334 (24.9) 1005 (75.1)
1671 (20.1) 6636 (79.9)
341,596 (97.7) 8095 (2.3) 2.5 6 1.5
242,993 (97.5) 6117 (2.5) 2.5 6 1.5
63,484 (98.2) 1175 (1.8) 2.7 6 1.5
20,637 (97.9) 452 (2.1) 2.5 6 1.5
5056 (97.5) 131 (2.5) 2.2 6 1.4
1318 (98.4) 21 (1.6) 2.5 6 1.5
8108 (97.6) 199 (2.4) 2.4 6 1.5 19
Abbreviation: SD, standard deviation. All values are n (%) unless otherwise indicated. *Overall average is per hospital. Averages for individual races are per hospitalization.
THROMBOLYTIC USE: RACIAL AND INCOME DISPARITY
Table 1. Patient and hospital characteristics by race
20
Table 2. Patient and hospital characteristics by median income in the patient’s ZIP code Income classification Low (n 5 124,626)
Low to middle (n 5 108,330)
Middle to high (n 5 95,363)
High (n 5 83,441)
71.3 6 14.9
69.6 6 15.0
71.2 6 14.8
72.0 6 14.7
73.3 6 14.6
186,253 (45.2) 225,478 (54.8)
55,815 (44.8) 68,800 (55.2)
49,213 (45.4) 59,106 (54.6)
43,155 (45.3) 52,203 (54.7)
38,070 (45.6) 45,369 (54.4)
282,833 (68.8) 25,279 (6.2) 77,016 (18.7) 16,722 (4.1) 1996 (0.5) 7361 (1.8)
83,428 (67.1) 11,959 (9.6) 18,741 (15.1) 6714 (5.4) 898 (0.7) 2657 (2.1)
74,665 (69.0) 6585 (6.1) 19,627 (18.1) 4717 (4.4) 531 (0.5) 2062 (1.9)
66,105 (69.4) 4315 (4.5) 19,637 (20.6) 3234 (3.4) 367 (0.4) 1579 (1.7)
58,635 (70.3) 2420 (2.9) 19,011 (22.8) 2057 (2.5) 200 (0.2) 1063 (1.3)
43,190 (10.5) 101,667 (24.7) 266,425 (64.8)
10,259 (8.3) 30,091 (24.2) 84,031 (67.6)
11,487 (10.6) 25,989 (24.0) 70,707 (65.4)
11,645 (12.2) 22,867 (24.0) 60,770 (63.8)
9799 (11.7) 22,720 (27.2) 50,917 (61.0)
33,539 (12.9) 225,725 (87.1)
18,450 (23.9) 56,682 (76.1)
11,122 (16.6) 56,048 (83.4)
3118 (5.1) 58,123 (94.9)
849 (1.6) 52,872 (98.4)
89,651 (21.8) 97,615 (23.7) 193,593 (47.0) 30,901 (7.5) 150 6 192
19,167 (15.4) 23,400 (18.8) 75,479 (60.6) 6580 (5.3) 362 6 322
18,913 (17.5) 28,097 (25.9) 52,575 (48.5) 8745 (8.1) 354 6 325
20,698 (21.7) 26,864 (28.2) 38,997 (40.9) 8804 (9.2) 399 6 343
30,873 (37.0) 19,254 (23.1) 26,542 (31.8) 6772 (8.1) 422 6 310
81,712 (19.8) 330,048 (80.2)
30,158 (24.2) 94,468 (75.8)
25,079 (23.2) 83,251 (76.9)
16,621 (17.4) 78,742 (82.6)
9854 (11.8) 73,587 (88.2)
401,923 (97.6) 9837 (2.4) 2.5 6 1.5
122,175 (98.0) 2451 (2.0) 2.5 6 1.5
105,847 (97.7) 2483 (2.3) 2.5 6 1.5
92,874 (97.4) 2489 (2.6) 2.5 6 1.5
81,027 (97.1) 2414 (2.9) 2.4 6 1.5
Abbreviation: SD, standard deviation. All values are n (%) unless otherwise indicated. *Overall average is per hospital. Averages for individual income levels are per hospitalization.
M.M. KIMBALL ET AL.
Age, y (mean 6 SD) Gender Male Female Primary payer Medicare Medicaid Private insurance Self-pay No charge Other Hospital bed size Small Medium Large Hospital location Rural Urban Hospital region Northeast Midwest South West Annual hospital case volume, n 5 1833 hospitals (mean 6 SD)* High-volume hospital (.147 cases per year) No Yes Thrombolytics used No Yes Comorbidity score, mean 6 SD
Overall (N 5 411,760)
THROMBOLYTIC USE: RACIAL AND INCOME DISPARITY
Effect of Race and Income on IV Thrombolysis Race was a highly significant predictor for IV thrombolysis (P , .0001). Relative to whites, the odds ratio (OR) of IV thrombolysis for African Americans was 0.58 (95% confidence interval [CI] 0.52-0.66; P , .001), for Native Americans 0.59 (95% CI 0.39-0.91; P 5 .0166), and for Hispanics 0.76 (95% CI 0.64-0.91; P 5 .0020). The odds for Asians and ‘‘other’’ were not significantly different than whites. The odds of IV thrombolysis by race are seen in Table 3. Median income was a significant predictor of IV thrombolysis (P , .001). Compared to patients from the poorest ZIP codes, the odds of IV thrombolysis for patients from the wealthiest ZIP codes was 1.55 (95% CI 1.371.76; P , .0001), for middle to high 1.48 (95% CI 1.331.64; P , .0001), and for low to middle 1.26 (95% CI 1.14-1.39; P , .0001). The odds of IV thrombolysis by median income are seen in Table 4.
21
Table 4. Odds of intravenous thrombolysis use by median income when compared to the poorest incomes Median income quartile
OR
95% CI
P value
High Middle to high Low to middle
1.55 1.48 1.26
1.37-1.76 1.33-1.64 1.14-1.39
,.0001 ,.0001 ,.0001
Abbreviations: CI, confidence interval; OR, odds ratio.
Relationship between Race and Thrombolysis Use and ICD-9 Codes for Stroke
Relative to whites, Native Americans were less likely to be treated at a high stroke case volume hospital (OR 0.73; 95% CI 0.65-0.83; P ,.0001), while African Americans (OR 1.8; 95% CI 1.76-1.85; P , .0001) and Hispanics (OR 1.12; 95% CI 1.08-1.17; P , .0001) were more likely to be treated at a high stroke case volume hospital. The odds of treatment at a high-volume hospital for Asians and those in race category ‘‘other’’ were not significantly different than whites. The odds of being treated in a high stroke case volume hospital by race can be seen in Table 5. Median income was a predictor of treatment at a high stroke case volume hospital (P , .0001). Relative to patients from the poorest ZIP codes, the OR of being treated in a high-volume hospital for those in the wealthiest ZIP codes was 2.54 (95% CI 2.48-2.60; P , .0001), for middle to high income ZIP codes 1.57 (95% CI 1.53-1.60; P , .0001), and for low to middle income ZIP codes 1.08 (95% CI 1.061.10; P , .0001) The odds of being treated in a high stroke case volume hospital by median income can be seen in Table 6.
ICD-9 codes were divided into 5 groups: group 1 (433.01, 433.11, and 433.21); group 2 (433.31, 433.81, and 433.91); group 3 (434.01); group 4 (434.11); and group 5 (434.91 and 437.1). The proportion of patients that received thrombolysis varied significantly across the 5 diagnosis groups, with group 5 less likely to get tissue plasminogen activator (tPA; P , .0001). The odd ratios of receiving thrombolysis among the groups when compared to group 5 was 1.9 (95% CI 1.7-2.1; P , .0001) for group 1, 1.5 (95% CI 1.2-2.0; P 5 .0007) for group 2, 4.2 (95% CI 3.6-4.8; P , .0001) for group 3, and 2.5 (95% CI 2.3-2.7; P , .0001) for group 4. The distribution of race also varied significantly across the groups, with group 5 having a higher percentage of African Americans (P , .0001). Nevertheless, even when controlling for diagnosis group, race remains a significant predictor of thrombolytic use, with ORs only slightly less extreme than those reported in the primary analysis. This result is largely caused by the fact that 80% of the original hospitalizations fall into group 5, and therefore dramatic differences in thrombolysis and race between the relatively small groups 1 to 4 and group 5 would be required to substantially affect the conclusions of the primary analysis; the differences in thrombolysis and race among the groups, while statistically significant, are far from dramatic. The vast majority of stroke patients (80%) have 1 of 2 ICD-9 diagnoses (434.91 or 437.1), so any grouping by stroke type is unlikely to change the basic conclusions of the analysis.
Table 3. Odds of intravenous thrombolysis use by race when compared to white patients
Table 5. Odds of treatment at high stroke case volume hospitals by race when compared to white patients
Effect of Race and Income on Access to High Stroke Case Volume Hospitals
Race
OR
95% CI
P value
Race
OR
95% CI
P value
African American Native American Hispanic Asian Other
0.58 0.59 0.76 0.96 0.91
0.52-0.66 0.39-0.91 0.64-0.91 0.74-1.23 0.75-1.11
,.001 .017 .002 .725 .343
Native American African American Hispanic Other Asian
0.73 1.80 1.12 1.01 1.00
0.65-0.83 1.76-1.85 1.08-1.17 0.71-1.43 0.54-1.84
,.0001 ,.0001 ,.0001 .963 .987
Abbreviations: CI, confidence interval; OR, odds ratio.
Abbreviations: CI, confidence interval; OR, odds ratio.
M.M. KIMBALL ET AL.
22
Table 6. Odds of treatment at high stroke case volume hospitals by median income when compared to the poorest incomes Median income quartile
OR
95% CI
P value
High Middle to high Low to middle
2.54 1.57 1.08
2.48-2.60 1.53-1.60 1.06-1.10
,.0001 ,.0001 ,.0001
Abbreviations: CI, confidence interval; OR, odds ratio.
Effect of Hospital Stroke Case Volume on Mortality, Outcome, Hospital Charges, and LOS Hospital stroke case volume was a significant predictor of mortality and receiving IV thrombolysis. Patients treated in low-volume hospitals were more likely to die in the hospital (OR 1.10; 95% CI 1.05-1.16; P 5 .0002). Patients treated in high-volume hospitals had 1.84 times the odds of IV thrombolysis than those in low-volume hospitals (95% CI 1.55-2.18; P , .0001). Hospital volume was a predictor of total charges (P , .0001) and LOS (P , .0001). Average inflationadjusted charges for patients treated in high-volume hospitals were estimated to be $21,163 (95% CI $20,131$22,026) compared to $16,155 (95% CI $15,522-$16,815) in low-volume hospitals. The average LOS for patients in high-volume hospitals was estimated to be 4.7 days (95% CI 4.6-4.9]) compared to 4.3 days (95% CI 4.2-4.4) for low-volume hospitals. These estimated averages for charges and LOS are converted from a log scale, which was the scale of analysis for both outcomes.
Discussion Stroke represents the fourth leading cause of death and the leading cause of adult disability in the United States.11 Treatments for acute ischemic stroke include IV thrombolysis, intra-arterial thrombolysis, angioplasty, and thrombectomy. Previous studies have shown that certain factors were associated with higher rates of thrombolysis. In a previous study using the NIS database between 1999 and 2004, Schumacher et al12 reported that thrombolysis rates are highest at academic medical centers and hospitals that have higher stroke treatment volume, and also found that thrombolysis rates were higher in white patients.12 Our study has shown that African American, Hispanic, Native American, and low-income patients have decreased access to IV thrombolysis, and those who live in poorer neighborhoods have decreased access to high stroke case volume hospitals. Part of the disparity may be that a disproportionate number of certain minorities and the poor live in rural areas, and these areas are often remote from large medical centers. This is likely the case for Native Americans, whose largest populations are
located in the remote Four Corners region of the southwest United States and rural Oklahoma. Approximately 25% of the US population lives in rural areas, where they tend to be older, have a higher prevalence of stroke risk factors, and lack access to preventive care.9,13,14 Remote location does not appear to be the only factor. Large populations of minorities and low-income patients live in urban centers, yet our study found disparities in stroke treatment among race and income level across all hospital locations and types. Racial and ethnic minorities in the United States now represent 28% of the population. African American men have the lowest average life expectancy (61.7 years), and the most significant factor leading to this disparity is cardiac disease and stroke.15 Other studies have shown that race and socioeconomic disparities exist for access to preventive health care and availability of acute stroke treatment.1-3,11,13-16,17,18,19,20-25 In Washington, DC, African Americans were less likely to arrive within 3 hours of stroke symptoms and to meet inclusion criteria for IV thrombolysis.20 In patients meeting criteria, African Americans were half as likely to receive IV thrombolysis as whites.20 Other studies have shown that minorities who present with stroke symptoms have longer wait times in emergency department waiting rooms than whites,3,23 and also have a wariness toward the medical system,2 which may translate to lower rates of consenting to IV thrombolysis. One study21 evaluated stroke patients admitted to academic medical centers to see if ethnicity was an influence on treatment. Of those meeting criteria for IV thrombolysis, African Americans were one-fifth as likely to receive that treatment when compared to whites (P 5 .001).21 In the present study, African Americans and Hispanics were more likely to be evaluated in high-volume hospitals where IV thrombolysis was more likely, but less likely to receive IV thrombolysis than whites. This may be because minorities are presenting to high-volume hospitals in a delayed fashion. In our study, patients from high-income ZIP codes are more likely to receive IV thrombolysis; however, patients with lower income have a higher prevalence of stroke risk factors and stroke incidence.2,4,17-18,24,26 In the United States, African Americans and Hispanics on average, have lower income levels than whites, and there may be an association between race and income.24 We attempted to clarify this by studying race and income independently from one another. High stroke volume hospitals are consistently associated with lower stroke mortality.4 High-volume hospitals are a source of many subspecialists, including stroke neurologists, neurointerventionalists, and neurosurgeons. When treated and evaluated by a neurologist, acute stroke patients are 3.7 times more likely to receive IV thrombolysis.2 This was consistent among African American and white patients; however, only 10.6% of African American patients were treated by a neurologist
THROMBOLYTIC USE: RACIAL AND INCOME DISPARITY
compared with 20.3% of white patients. Hospitals that have been declared designated primary stroke centers have been shown to have lower mortality and morbidity rates and a greater use of thrombolytic therapy.4,10,27,28,29 Since the implementation of designated stroke centers, many states have improved the use of streamlining patients with stroke symptoms to be directly transferred to primary stroke centers. More recently, California has used a county-wide ‘‘spoke-and-hub’’ stroke system that incorporates several comprehensive stroke center recommendations through many smaller linked hospitals to increase thrombolysis rates and the use of endovascular treatments.30 In South Carolina, clinicians have been using telemedicine to increase the thrombolysis rates in remote hospitals through contact with a primary stroke center.31 In the present study, IV thrombolysis was more likely if treated at a high-volume hospital, and high-volume hospitals were associated with better outcomes. A NIS study conducted between 1999 and 2002 reached similar conclusions; 70% of patients who received thrombolysis were treated at high-volume centers.32 Low-income patients were less likely to be treated in a high-volume hospital, suggesting that a socioeconomic disparity in health care exists for ischemic stroke patients. There is a possibility that some patients arrived at small medical centers first and received IV thrombolysis in the emergency department there, and were later transferred to larger stroke centers for continued care. The NIS unit of analysis is the hospitalization, not the patient. Anything that happens to the patient during a specific hospitalization is attributed to that hospital. Anything that happens to the patient before or after the hospitalization is unavailable in the data recorded for that hospitalization, and the NIS offers no way to track a patient from one hospitalization to another. If a patient visited a low-volume hospital and received recombinant tPA there, the recombinant tPA use would be attributed to that hospital, regardless of whether the patient was transferred. If the patient was transferred to another hospital, the previous recombinant tPA would not show up in the record for the new hospitalization. Our data suggest that a disparity exists for minorities and lower income patients with ischemic stroke. Lowincome patients lack access to high-volume hospitals, which are more likely to treat with IV thrombolysis, and are associated with improved outcomes. African Americans and Hispanics are more likely to be seen at high-volume hospitals but are less likely to be treated with IV thrombolysis. Delayed arrival to appropriate medical facilities may be an important factor for minority and lower income patients, making them ineligible for thrombolysis. With the increasing number of primary stroke centers and the availability of telemedicine for smaller rural facilities, there will likely continue to be an increase in the
23
number of patients arriving within the 4.5-hour time window for IV thrombolysis. However, researchers also need to focus on educating the public about the signs and symptoms of stroke and the acuity in which they need to be treated. Health care providers must become aware of the disparity that exists in these populations, and efforts should be focused on narrowing the gap.
Limitations There are limitations to the study. This is a retrospective study, and there is therefore the potential for selection bias. A prospective randomized trial cannot be performed, because a patient cannot be randomized to race or income level. There may be risks of coding error(s). In the NIS, a potential exists for variability in coding. The NIS database does not contain data on neurologic conditions of patients, and the analysis could not be adjusted for stroke severity. Stroke severity as recorded by the National Institutes of Health Stroke Score (NIHSS) or admission or discharge disability scale as recorded by modified Rankin Scale (mRS) are not available in the NIS database—therefore, we could not determine whether these were factors that had an effect on our findings. There is also a lack of information regarding inpatient care for these patients, such as other medical or surgical problems during their hospitalization. Exclusion criteria for IV thrombolysis cannot be determined from the NIS database, including time of stroke symptom onset. It would be important to know if the limitation of receiving IV thrombolysis was arriving outside of the time window, medical contraindications, or refusal of consent for treatment. We only used IV thrombolysis as an outcome for treatment. The use of code 99.10 for IV thrombolysis also has a limitation, because it could have been miscoded and may be underreported. Other endovascular procedures were not included in this study because of inconsistent coding of the procedures.
References 1. Zweifler RM, McClure LA, Howard VJ, et al. Racial and geographic differences in prevalence, awareness, treatment and control of dyslipidemia: the reasons for geographic and racial differences in stroke (REGARDS) study. Neuroepidemiology 2011;37:39-44. 2. Cruz-Flores S, Rabinstein A, Biller J, et al. Racial-ethnic disparities in stroke care: The American experience: A statement for healthcare professionals from the American Heart Association/American Stroke Association. Stroke 2011;42:2091-2116. 3. Lacy CR, Suh DC, Bueno M, et al. Delay in presentation and evaluation for acute stroke: Stroke Time Registry for Outcomes Knowledge and Epidemiology (S.T.R.O.K.E.). Stroke 2001;32:63-69. 4. Saposnik G, Jeerakathil T, Selchen D, et al. Socioeconomic status, hospital volume, and stroke fatality in Canada. Stroke Outcome Research Canada Working Group. Stroke 2008;39:3360-3366.
24 5. Saposnik G, Baibergenova A, O’Donnell M, et al. Hospital volume and stroke outcome: Does it matter? Neurology 2007;69:1142-1151. 6. Marler JR, Tilley BC, Lu M, et al. National Institute of Neurological Disorders and Stroke: Early stroke treatment associated with better outcome: The NINDS rt-PA stroke study. Neurology 2000;55:1649-1655. 7. Del Zoppo GJ, Saver JL, Jauch EC, et al. Expansion of the time window for treatment of acute ischemic stroke with intravenous tissue plasminogen activator: A science advisory from the American Heart Association/American Stroke Association. Stroke 2009;40:2945-2948. 8. Wahlgren N, Ahmed N, D avalos A, et al. Thrombolysis with alteplase 3-4.5 h after acute ischaemic stroke (SITSISTR): An observational study. Lancet 2008;372:1303-1309. 9. Elixhauser A, Steiner C, Harris DR, et al. Comorbidity measures for use with administrative data. Med Care 1998;36:8-27. 10. Geronimus AT, Bound J. Use of census-based aggregate variables to proxy for socioeconomic group: evidence from national samples. Am J Epidemiol 1998;148:475-486. 11. Lloyd-Jones D, Adams RJ, Brown TM, et al. Executive summary: Heart disease and stroke statistics—2010 update: A report from the American Heart Association. American Heart Association Statistics Committee and Stroke Statistics Subcommittee. Circulation 2010;121:948-954. 12. Schumacher HC, Bateman BT, Boden-Albala B, et al. Use of thrombolysis in acute ischemic stroke: Analysis of the Nationwide Inpatient Sample 1999 to 2004. Ann Emerg Med 2007;50:99-107. 13. Pearson TA, Lewis C. Rural epidemiology: Insights from a rural population laboratory. Am J Epidemiol 1998; 148:949-957. 14. Eberhardt MS, Pamuk ER. The importance of place of residence: Examining health in rural and nonrural areas. Am J Public Health 2004;94:1682-1686. 15. Murray CJ, Kulkarni SC, Michaud C, et al. Eight Americas: Investigating mortality disparities across races, counties, and race-counties in the United States. PLoS Med 2006; 3:e260. 16. Levine DA, Neidecker MV, Kiefe CI, et al. Racial/ethnic disparities in access to physician care and medications among US stroke survivors. Neurology 2011;76:53-61. 17. Liao Y, Greenlund KJ, Croft JB, et al. Factors explaining excess stroke prevalence in the US Stroke Belt. Stroke 2009;40:3336-3341. 18. Cesaroni G, Agabiti N, Forastiere F, et al. Socioeconomic differences in stroke incidence and prognosis under a universal healthcare system. Stroke 2009;40:2812-2819.
M.M. KIMBALL ET AL. 19. Kapral MK, Wang H, Mamdani M, et al. Effect of socioeconomic status on treatment and mortality after stroke. Stroke 2002;33:268-273. 20. Hsia AW, Edwards DF, Morgenstern LB, et al. Racial disparities in tissue plasminogen activator treatment rate for stroke: A population-based study. Stroke 2011; 42:2217-2221. 21. Johnston SC, Fung LH, Gillum LA, et al. Utilization of intravenous tissue-type plasminogen activator for ischemic stroke at academic medical centers: The influence of ethnicity. Stroke 2001;32:1061-1068. 22. Leira EC, Hess DC, Torner JC, et al. Rural-urban differences in acute stroke management practices: A modifiable disparity. Arch Neurol 2008;65:887-891. 23. Karve SJ, Balkrishnan R, Mohammad YM, et al. Racial/ ethnic disparities in emergency department waiting time for stroke patients in the United States. J Stroke Cerebrovasc Dis 2011;20:30-40. 24. Howard G, Anderson RT, Russell G, et al. Race, socioeconomic status, and cause-specific mortality. Ann Epidemiol 2000;10:214-223. 25. Trimble B, Morgenstern LB. Stroke in minorities. Neurol Clin 2008;26:1177-1190. 26. Reiner-Deitemyer V, Teuschl Y, Matz K, et al. Austrian Stroke Unit Registry Collaborators. Helicopter transport of stroke patients and its influence on thrombolysis rates: data from the Austrian Stroke Unit Registry. Stroke 2011; 42:1295-1300. 27. Gropen T, Magdon-Ismail Z, Day D, et al. Regional implementation of the stroke systems of care model: Recommendations of the northeast cerebrovascular consortium. Stroke 2009;40:1793-1802. 28. Xian Y, Holloway RG, Chan PS, et al. Association between stroke center hospitalization for acute ischemic stroke and mortality. JAMA 2011;305:373-380. 29. Prabhakaran S, McNulty M, O’Neill K, et al. Intravenous thrombolysis for stroke increases over time at primary stroke centers. Stroke 2012;43:875-877. 30. Cramer SC, Stradling D, Brown DM, et al. Organization of a United States county system for comprehensive acute stroke care. Stroke 2012;43:1089-1093. 31. Lazaridis C, Desantis SM, Jauch EC, et al. Telestroke in South Carolina. J Stroke Cerebrovasc Dis 2011 Dec 22 [Epub ahead of print]. 32. Bateman BT, Schumacher HC, Boden-Albala B, et al. Factors associated with in-hospital mortality after administration of thrombolysis in acute ischemic stroke patients: An analysis of the nationwide inpatient sample 1999 to 2002. Stroke 2006;37:440-446.