Abstract
Background
The purpose of this study is to examine the association between race/ethnicity and ischemic stroke severity in the United States.
Methods
We performed an analysis of adult hospital discharges in the National Inpatient Sample 2018-2021 with a primary discharge diagnosis of ischemic stroke. We stratified our cohort based on self-reported race/ethnicity and evaluated stroke severity using the NIH Stroke Scale (NIHSS). Age and sex-adjusted estimates of the NIHSS were derived from linear regression models.
Results
We included 231,396 stroke discharges with a mean NIHSS of 6.5±7.2. The cohort was 68.1% White, 17.4% Black, 8.2% Hispanic, and 6.3% other. The age and sex-adjusted NIHSS for White patients was 6.25 (95%CI 6.22-6.29), for Black patients was 7.12 (95%CI 7.05-7.19), for Hispanic patients was 6.86 (95%CI 6.76-6.97), and for patients of other races/ethnicities was 7.29 (95%CI 7.18-7.41). Further adjustment for the Charlson Comorbidity Index (CCI), socioeconomic factors, and poorly controlled hypertension or diabetes did not significantly alter these findings.
Conclusion
In a large, contemporary, and nationally representative sample of acute ischemic stroke (AIS) patients, we show an association between non-White race/ethnicity and higher stroke severity. These results are concerning for an underappreciated health disparity in AIS.
Keywords: acute ischemic stroke, race, ethnicity, stroke severity, nihss
Graphical Abstract
Introduction
Stroke is a leading cause of morbidity and mortality worldwide.1 The National Institutes of Health Stroke Scale (NIHSS) score is a validated tool to measure stroke severity, ranging from 0 (no symptoms) to a maximum of 42.2 Baseline NIHSS is the most important predictor of AIS outcome.3 Prior research indicates that racial/ethnic minorities often have higher baseline NIHSS scores compared to White populations, but the studies have not included nationally representative cohorts.4 In this study, we use data from the National Inpatient sample to investigate these racial/ethnic differences and whether differences in stroke severity between races/ethnicities persist after controlling for demographics, comorbidities, and socioeconomic variables.
Methods
Data used in this study are publicly available through the National Inpatient Sample; therefore, it was exempt from institutional review board (IRB) approval. No informed consent was required for this study. The analytic methods and study materials will be made available upon request to the corresponding author. We performed an analysis of non-elective adult (≥18 years) hospital admissions in the National Inpatient Sample (2018-2021) with a primary diagnosis of AIS (see Table S1) and a recorded NIHSS score (ICD-10-CM code R29.7x). CMS guidance indicates that the NIHSS recorded in R29.7 is the “initial [NIHSS] score documented”, indicating that for most patients, it would be an admission NIHSS. The primary outcome is NIHSS, which we treated as a continuous variable given its 43 levels (0-42). We fit linear regression models using self-reported race/ethnicity (White, Black, Hispanic, other) as the exposure. Model 1 was unadjusted; model 2 was adjusted for age and sex; and model 3 for age, sex, and CCI; model 4 for age, sex, CCI, median household income quartile by zip code, patient location on a rural-urban continuum, and expected primary payor; and model 5 for age, sex, CCI, complicated hypertension and diabetes. See Table S1 for more detailed definitions of study variables. We also included interaction terms in the linear regression models between race-ethnicity and the covariates. As a sensitivity analysis we defined the exposure as White, Black, Hispanic, Asian or Pacific Islander, Native American, or other, acknowledging that the resulting smaller sample size within the additional strata decreases the precision of the point estimates.5,6 We also evaluated for differences between those with and without missing NIHSS data (Table S2). This study was reported in accordance with the STROBE guidelines for observational studies.
Results
The cohort derivation is seen in Figure S1. There were 416,679 AIS, of which we excluded 10,580 (2.5%) for missing data on race/ethnicity and 174,705 (41.9%) for missing data on NIHSS. The resulting cohort of 231,396 patients included 48.5% women and had a mean age of 69.4±13.9 years. The mean NIHSS was 6.2±7.2 for White 6.8±7.0 for Black 6.6±7.2 for Hispanic, and 7.2±7.8 for other race/ethnicity patients (P<0.001). Between the race/ethnicity strata, there were statistically significant differences in all demographic and socioeconomic covariates (Table 1). Compared to White patients, the non-White patients were on average younger, with more medical comorbidities, less household income, higher rates of Medicaid insurance, and higher rates of living in a large metropolitan area.
Table 1.
Patient-level characteristics stratified by race/ethnicity.
| White (N=157,542) |
Black (N=40,366) |
Hispanic (N=18,960) |
Other (N=14,528) |
Total (N=231,396) |
P-value | |
|---|---|---|---|---|---|---|
| Demographics | ||||||
| Age, mean (SD) | 71.3 (13.4) | 64.1 (13.5) | 66.1 (14.7) | 68.3 (14.3) | 69.4 (13.9) | <0.001 |
| Male Sex, No. (%) | 81,055 (51.5%) | 20,211 (50.1%) | 10,235 (54.0%) | 7,727 (53.2%) | 119,228 (51.5%) | <0.001 |
| Medical Comorbidities | ||||||
| CCI, mean (SD) | 3.8 (2.2) | 4.2 (2.3) | 3.9 (2.2) | 3.9 (2.2) | 3.9 (2.2) | <0.001 |
| Complicated Hypertension | 42,513 (27.0%) | 14,584 (36.1%) | 5,102 (26.9%) | 3,795 (26.1%) | 65,994 (28.5%) | <0.001 |
| Complicated Diabetes | 31,715 (20.1%) | 12,175 (30.2%) | 5,973 (31.5%) | 3,877 (26.7%) | 53,740 (23.2%) | <0.001 |
| Socioeconomic Characteristics | ||||||
| Quartiles of Mean household income | <0.001 | |||||
| Highest Income | 33,790 (21.8%) | 3,996 (10.1%) | 2,677 (14.4%) | 4,870 (34.2%) | 45,333 (19.9%) | |
| High Medium | 40,832 (26.3%) | 6,505 (16.4%) | 4,311 (23.2%) | 3,651 (25.6%) | 55,299 (24.3%) | |
| Low Medium | 43,333 (27.9%) | 8,536 (21.5%) | 4,746 (25.6%) | 2,815 (19.8%) | 59,430 (26.1%) | |
| Lowest Income | 37,276 (24.0%) | 20,715 (52.1%) | 6,841 (36.8%) | 2,909 (20.4%) | 67,741 (29.7%) | |
| Expected Primary payer | <0.001 | |||||
| Medicare | 107,195 (68.1%) | 20,401 (50.7%) | 9,628 (50.8%) | 7,720 (53.2%) | 144,944 (62.7%) | |
| Medicaid | 9,795 (6.2%) | 6,980 (17.3%) | 3,342 (17.6%) | 2,242 (15.5%) | 22,359 (9.7%) | |
| Private Insurance | 31,058 (19.7%) | 8,695 (21.6%) | 3,572 (18.9%) | 3,351 (23.1%) | 46,676 (20.2%) | |
| Self-pay | 4,862 (3.1%) | 2,790 (6.9%) | 1,823 (9.6%) |
765 (5.3%) |
10,240 (4.4%) | |
| No Charge/Other | 4,413 (2.8%) | 1,409 (3.5%) | 571 (3.0%) | 430 (3.0%) | 6,823 (3.0%) | |
| Patient Location: Urban-Rural | <0.001 | |||||
| Large Metro (≥1M) | 32,664 (20.8%) | 17,866 (44.4%) | 9,756 (52.0%) | 6,826 (47.4%) | 67,112 (29.1%) | |
| Large Metro Suburb | 41,090 (26.2%) | 8,921 (22.2%) | 3,350 (17.9%) | 3,309 (23.0%) | 56,670 (24.6%) | |
| Metro (<1M) | 53,822 (34.3%) | 9,820 (24.4%) | 4,723 (25.2%) | 3,146 (21.8%) | 71,511 (31.0%) | |
| Rural/Micropolitan | 29,511 (18.8%) | 3,599 (9.0%) | 938 (5.0%) | 1,134 (7.9%) | 35,182 (15.3%) | |
| Outcome | ||||||
| NIHSS | 6.4 (7.2) | 6.8 (7.0) | 6.6 (7.2) | 7.2 (7.8) | 6.5 (7.2) | <0.001 |
SD= standard deviation, IQR= interquartile range, NIHSS= National Institute of Health Stroke Scale.
Continuous variables were analyzed using ANOVA test and categorical variables using chi-square test.
Income quartiles vary by year and can be seen at: https://hcup-us.ahrq.gov/db/vars/zipinc_qrtl/nisnote.jsp.
The age and sex-adjusted NIHSS for White patients was 6.25 (95%CI 6.22-6.29), for Black patients was 7.12 (95%CI 7.05-7.19), for Hispanic patients was 6.86 (95%CI 6.76-6.97), and for patients of other races/ethnicities was 7.29 (95%CI 7.18-7.41) (P<0.001 for all categories). Further adjustment for the CCI, socioeconomic variables, and poorly controlled hypertension and diabetes did not significantly alter these findings as can be seen in Figure 1. The sensitivity analysis with the addition of Asian/Pacific Islander and Native American as exposure categories continued to show higher NIHSS in non-White patients (Figure S2), with an NIHSS for Asian/Pacific Islanders of 7.23 (95%CI 7.07-7.40), for Native Americans of 6.896 (95%CI 6.51-7.38), and for the remaining other race/ethnicity patients of 7.42 (95% CI 7.24-7.59) (p<0.01 for all categories). We found minimal differences between groups with and without missing NIHSS data (Mean standardized difference was <0.1 in all groups, Table S2).
Figure 1.
Predicted NIH Stroke Scale score.
The interactions between race/ethnicity and the covariates were statistically significant (p<0.05) except for complicated hypertension (p=0.101) and diabetes (p=0.945). The age and sex-adjusted NIHSS within the strata resulting from the interactions are seen in Figure 2 and the NIHSS value with a 95% confidence interval and number of observations within the strata is seen in Table S3. Notable findings in these analyses include relatively higher NIHSS for older non-White patients, for Black female patients, for Black patients with less medical comorbidities, White patients with lower household income, Black patients with Medicare, and both Black and White patients living in rural or micropolitan areas.
Figure 2.
Interaction of predicted NIH Stroke Scale score with demographic and socioeconomic factors.
Discussion
We found a significant association between non-White race/ethnicity and higher AIS severity in a nationally representative sample. While prior studies have reported disparities in stroke incidence and mortality, our findings highlight disparities in stroke severity based on NIHSS. This association remains significant after adjusting for patient demographics, medical comorbidities, and socioeconomic variables. The mechanism for increased AIS severity among non-White patients is not well understood.
Vascular comorbidities, such as hypertension and diabetes, are more prevalent in non-White populations and are associated with stroke incidence.7 Our study adjusts for vascular comorbidities but cannot fully account for how well these comorbidities are managed. For example, while hypertension is a well-established risk factor for stroke in all racial/ethnic groups, significant disparities exist in its control.8 Similarly, diabetes is less controlled in Black and Hispanic populations.7 We found evidence of this in our cohort as seen in the higher rates of complicated hypertension and diabetes in non-White patients (Table 1), but adjusting for that does not mitigate their higher stroke severity. Black and Hispanic patients who are 65 years or older are at higher risk for developing disability compared to White populations.9 This finding was persistent but attenuated after controlling for health and socioeconomic differences.10
Unfortunately, in this analysis we cannot account for the possibility that higher pre-stroke disability in non-White patients may explain the observed higher stroke severity. Minority populations often face reduced access to healthcare facilities and are more frequently underinsured.7 Studies show that minorities present to the hospital with longer time elapsed from stroke onset and are less likely to use emergency medical services for transport to the hospital, which may cause an increase in baseline NIHSS as untreated strokes typically worsen in the first 12-24 hours after onset.11-13 Finally, disparities in stroke severity could also be influenced by cultural differences in stroke awareness, including recognizing its signs and symptoms, and the urgency of seeking medical attention. Immigrants may encounter language barriers, which can further exacerbate these disparities.7,11 To address these disparities, culturally tailored educational programs to promote stroke awareness should be considered for minority populations.11 Increasing healthcare availability in underserved areas is necessary for a more inclusive healthcare system. Future research should explore the underlying barriers contributing to reduced healthcare access and delayed care.7
Conclusion
Our study highlights significant disparities in stroke severity at hospital admission across different racial/ethnic groups. Acknowledging these disparities is essential to developing targeted strategies to reduce these inequities in our healthcare system.
Supplementary Material
Funding
Dr. de Havenon reports NIH/NINDS funding (K23NS105924, UG3NS130228, R01NS130189, R21NS138995). Dr. Sheth reports NIH/NINDS funding (U01NS106513, R01NS110721, R01NR018335, R01EB301114, R01MD016178, U24NS107215, U24NS107136, U24NS129500) and funding from the American Heart Association Bugher Award. Dr. Skolarus reports NIH funding (R01NS138072, R01MD019124, R01MD011516, R01AG059733, R01NS093870). Mr. Wong reports research funding from The Sumaira Foundation and The Siegel Rare Neuroimmune Association.
Non-standard Abbreviations and Acronyms
- NIHSS
National Institutes of Health Stroke Scale
- CCI
Charlson Comorbidity Index
- AIS
Acute Ischemic Stroke
Footnotes
Disclosures
Dr. de Havenon reports stock options in Cetrus and TitinKM; and grants from National Institutes of Health. He has received consultant fees from Novo Nordisk and royalty fees from UpToDate. Dr. Peng reports no conflicts of interest. Dr. Skolarus reports grants from American Heart Association. Dr. Sheth reports compensation from Philips, Sense and Zoll for data and safety monitoring services, and from Bexorg, Cerevasc, CSL Behring, Rhaeos, and Astrocyte for consultant services; grants from Hyperfine; and stock options in BrainQ. He also holds a patent for stroke wearables licensed to Alva Health. Mr. Wong reports travel support from University of Utah School of Medicine. Dr. Fayed reports no conflicts of interest.
References
- 1.Fan J, Li X, Yu X, Liu Z, Jiang Y, Fang Y, Zong M, Suo C, Man Q, Xiong L. Global Burden, Risk Factor Analysis, and Prediction Study of Ischemic Stroke, 1990–2030. Neurology. 2023;101(2):e137–e150. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Fonarow GC, Saver JL, Smith EE, Broderick JP, Kleindorfer DO, Sacco RL, Pan W, Olson DM, Hernandez AF, Peterson ED, et al. Relationship of National Institutes of Health Stroke Scale to 30-Day Mortality in Medicare Beneficiaries With Acute Ischemic Stroke. J Am Heart Assoc Cardiovasc Cerebrovasc Dis. 2012;1(1):42–50. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Rost NS, Bottle A, Lee J, Randall M, Middleton S, Shaw L, Thijs V, Rinkel GJE, Hemmen TM, Global Comparators Stroke GOAL collaborators. Stroke Severity Is a Crucial Predictor of Outcome: An International Prospective Validation Study. J Am Heart Assoc. 2016;5(1):e002433. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Jones EM, Okpala M, Zhang X, Parsha K, Keser Z, Kim CY, Wang A, Okpala N, Jagolino A, Savitz SI, et al. Racial disparities in post-stroke functional outcomes in young patients with ischemic stroke. Journal of Stroke and Cerebrovascular Diseases. 2020;29(8):104987. [DOI] [PubMed] [Google Scholar]
- 5.Gallacher KI, Jani BD, Hanlon P, Nicholl BI, Mair FS. Multimorbidity in Stroke. Stroke. 2019;50(7):1919–1926. [DOI] [PubMed] [Google Scholar]
- 6.Addo J, Ayerbe L, Mohan KM, Crichton S, Sheldenkar A, Chen R, Wolfe CDA, McKevitt C. Socioeconomic Status and Stroke. Stroke. 2012;43(4):1186–1191. [DOI] [PubMed] [Google Scholar]
- 7.Cruz-Flores S, Rabinstein A, Biller J, Elkind MSV, Griffith P, Gorelick PB, Howard G, Leira EC, Morgenstern LB, Ovbiagele B, Peterson E, Rosamond W, Trimble B, Valderrama AL. Racial-Ethnic Disparities in Stroke Care: The American Experience. Stroke. 2011;42(7):2091–2116. [DOI] [PubMed] [Google Scholar]
- 8.Abrahamowicz AA, Ebinger J, Whelton SP, Commodore-Mensah Y, Yang E. Racial and Ethnic Disparities in Hypertension: Barriers and Opportunities to Improve Blood Pressure Control. Curr Cardiol Rep. 2023;25(1):17–27. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Dunlop DD, Song J, Manheim LM, Daviglus ML, Chang RW. Racial/Ethnic Differences in the Development of Disability Among Older Adults. Am J Public Health. 2007;97(12):2209–2215. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Goyat R, Vyas A, Sambamoorthi U. Racial/Ethnic Disparities in Disability Prevalence. J Racial Ethn Health Disparities. 2016;3(4):635–645. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Willey JZ, Williams O, Boden-Albala B. Stroke literacy in Central Harlem: a high-risk stroke population. Neurology. 2009;73(23):1950–1956. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Otite FO, Saini V, Sur NB, Patel S, Sharma R, Akano EO, Anikpezie N, Albright K, Schmidt E, Hoffman H, et al. Ten-Year Trend in Age, Sex, and Racial Disparity in tPA (Alteplase) and Thrombectomy Use Following Stroke in the United States. Stroke. 2021;52(8):2562–2570. [DOI] [PubMed] [Google Scholar]
- 13.Springer MV, Labovitz DL, Hochheiser EC. Race-Ethnic Disparities in Hospital Arrival Time after Ischemic Stroke. Ethn Dis. 27(2):125–132. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Columbo JA, Daya N, Colantonio LD, Wang Z, Foti K, Hyacinth HI, Johansen MC, Gottesman R, Goodney PP, Howard VJ, et al. Derivation and Validation of ICD-10 Codes for Identifying Incident Stroke. JAMA Neurology. 2024;81:875–881. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.ICD-10 Code for National Institutes of Health Stroke Scale (NIHSS) score- R29.7- Codify by AAPC [Internet]. [cited 2023 Jan 19];Available from: https://www.aapc.com/codes/icd-10-codes/R29.7
- 16.Chang H-J, Chen P-C, Yang C-C, Su Y-C, Lee C-C. Comparison of Elixhauser and Charlson Methods for Predicting Oral Cancer Survival. Medicine. 2016;95:e2861. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Li B, Evans D, Faris P, Dean S, Quan H. Risk adjustment performance of Charlson and Elixhauser comorbidities in ICD-9 and ICD-10 administrative databases. BMC Health Serv Res. 2008;8:12. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Sharma N, Schwendimann R, Endrich O, Ausserhofer D, Simon M. Comparing Charlson and Elixhauser comorbidity indices with different weightings to predict in-hospital mortality: an analysis of national inpatient data. BMC Health Services Research. 2021;21:13. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.NIS Description of Data Elements [Internet]. [cited 2023 Jan 19];Available from: https://www.hcup-us.ahrq.gov/db/nation/nis/nisdde.jsp
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.