Patient outcomes in COVID-19 vary from asymptomatic infection to death. Using data from 38 hospitals in Michigan, the social vulnerability index, a composite measure of social disadvantage, was used to compare the outcomes of COVID-19 in patients living in low-vulnerability ZIP codes to those of patients living in high-vulnerability ZIP codes. Outcomes examined included acute organ dysfunction, organ failure, invasive mechanical ventilation, intensive care unit stay, mortality, and discharge disposition.
Visual Abstract. Neighborhood-Level Factors and COVID-19 Hospitalization.
Patient outcomes in COVID-19 vary from asymptomatic infection to death. Using data from 38 hospitals in Michigan, the social vulnerability index, a composite measure of social disadvantage, was used to compare the outcomes of COVID-19 in patients living in low-vulnerability ZIP codes to those of patients living in high-vulnerability ZIP codes. Outcomes examined included acute organ dysfunction, organ failure, invasive mechanical ventilation, intensive care unit stay, mortality, and discharge disposition.
Abstract
Background:
Although disparities in COVID-19 outcomes have been observed, factors contributing to these differences are not well understood.
Objective:
To determine whether COVID-19 hospitalization outcomes are related to neighborhood-level social vulnerability, independent of patient-level clinical factors.
Design:
Pooled cross-sectional study of prospectively collected data.
Setting:
38 Michigan hospitals.
Patients:
Adults older than 18 years hospitalized for COVID-19 in a participating site between March and December 2020.
Measurements:
COVID-19 outcomes included acute organ dysfunction, organ failure, invasive mechanical ventilation, intensive care unit stay, death, and discharge disposition. Social vulnerability was measured by the social vulnerability index (SVI), a composite measure of social disadvantage.
Results:
Compared with patients in low-vulnerability ZIP codes, those living in high-vulnerability ZIP codes were more frequently treated in the intensive care unit (29.0% vs. 24.5%); more frequently received mechanical ventilation (19.3% vs. 14.2%); and experienced higher rates of organ dysfunction (51.9% vs. 48.6%), organ failure (54.7% vs. 51.6%), and in-hospital death (19.4% vs. 16.7%). In mixed-effects regression analyses accounting for age, sex, and comorbid conditions, an increase in a patient's neighborhood SVI by 0.25 (1 quartile) was associated with greater likelihood of mechanical ventilation (increase of 2.1 percentage points), acute organ dysfunction (increase of 2.8 percentage points), and acute organ failure (increase of 2.8 percentage points) but was not associated with intensive care unit stay, mortality, or discharge disposition.
Limitation:
Observational data focused on hospitalizations in a single state.
Conclusion:
Hospitalized patients with COVID-19 from socially vulnerable neighborhoods presented with greater illness severity and required more intensive treatment, but once hospitalized they did not experience differences in hospital mortality or discharge disposition. Policies that target socially vulnerable neighborhoods and access to COVID-19 care may help ameliorate health disparities.
Primary Funding Source:
Blue Cross Blue Shield of Michigan (BCBSM) and Blue Care Network as part of the BCBSM Value Partnerships Program, the Michigan Public Health Institute, and the Michigan Department of Health & Human Services.
Disparities in COVID-19 incidence and outcomes related to patient characteristics (such as race or ethnicity) and geographic areas (such as neighborhoods) are well known (1–4). For example, in a recent systematic review, Black and Hispanic populations were found to experience disproportionate burdens of COVID-19 infection, hospitalization, and overall mortality (4). We previously found that U.S. counties with higher levels of social vulnerability or disadvantage—based on socioeconomic status, housing, and other factors—experienced greater COVID-19 incidence and mortality (3). Although we know that where a person lives affects their health, the interplays between individual- and neighborhood-level social, demographic, and health factors to COVID-19 outcomes are complex and understudied for hospital-based outcomes (5). Understanding the contributions of these domains to health outcomes is important for public health and health care policy.
Previous studies that have sought to understand disparities in COVID-19 health outcomes have been limited to cross-sectional or cohort studies of patients at a single health care system (6) or to ecological studies analyzing population-level as opposed to patient-level data (4). Cohort studies from multiple health care systems are uncommon, and those that are published have not been able to disentangle the contributions of a patient's individual clinical factors from neighborhood contextual effects on COVID-19 outcomes. In addition, less is known about what factors influence disparities in COVID-19 outcomes, whether related to greater exposure to COVID-19 infection, greater susceptibility to infection after exposure, or differential access to care (4).
The social vulnerability index (SVI), developed by the Centers for Disease Control and Prevention, provides an aggregate measure of neighborhood social factors known to affect public health crises, including disease outbreaks (7). Because it has been used frequently by public health authorities to investigate populations at higher risk for COVID-19, SVI represents an ideal tool with which to examine how social factors may or may not contribute to COVID-19 outcomes (7–9). Therefore, we used data from a multihospital cohort and ZIP code–linked SVI to quantify how individual- and neighborhood-level factors influenced outcomes after hospitalization for COVID-19. Our objective was to determine whether COVID-19 hospitalization outcomes were related to neighborhood-level social vulnerability or disadvantage, independent of patient-level clinical factors.
Methods
We performed a pooled cross-sectional study using data from patients hospitalized at 38 Michigan hospitals participating in a statewide collaborative quality improvement registry called MI-COVID19. Details regarding the MI-COVID19 registry (funded by the Blue Cross Blue Shield of Michigan/Blue Care Network of Michigan) have been previously published (10). This study was deemed “not regulated” by the University of Michigan Institutional Review Board (HUM00179611). In brief, trained abstractors collected data by reviewing patient medical records using a structured template. Patients were included in the study if they had either a positive COVID-19 test result during or up to 21 days before the hospital encounter; a negative COVID-19 test result during or up to 21 days before the hospital encounter with symptoms of cough, dyspnea, or fever or a discharge diagnosis of COVID-19 in the medical chart; or strong clinical suspicion of COVID-19 infection that was documented but could not be confirmed via testing because of logistic constraints. Patients were excluded if they were pregnant, were younger than 18 years, left against medical advice, entered comfort care or hospice within 3 hours of the hospital encounter, or had a length of stay greater than 120 days during the index encounter or if the patient discharge was within the 60-day follow-up window of a previously recorded or abstracted admission.
Sixty days after discharge, abstractors reviewed the medical records of patients to collect data on clinical events, including readmission (to the index hospital or any hospital viewable in the medical record) and postdischarge death. For this analysis, we excluded any patients who tested negative for COVID-19, who were discharged with an unconfirmed diagnosis of COVID-19, whose ZIP code was not within the state of Michigan, or who had a nonresidential ZIP code (for example, post office box). In addition, 144 patients from 12 participating hospitals with fewer than 25 patients with COVID-19 in the registry, classified as low-volume hospitals, were excluded from the main analyses. However, sensitivity analyses were performed including these patients, as noted in the following discussion. Figure 1 presents sample inclusion and exclusion criteria.
Figure. Analytic cohort construction: patients with COVID-19 from Michigan hospitals.
COVID-19 Hospitalization Outcomes
Our main COVID-19 outcomes included development of acute organ dysfunction, development of organ failure, use of invasive mechanical ventilation, intensive care unit stay, in-hospital death, and discharge disposition. Patients were classified as having acute organ dysfunction using the Centers for Disease Control and Prevention's Adult Sepsis Event definition as follows: acute renal dysfunction (creatinine level greater than 1.5 times baseline among patients without preexisting end-stage renal disease, where baseline is the lowest creatinine level during hospitalization); acute hematologic dysfunction (platelet count <100 × 109 cells/L, with ≥50% decrease compared with baseline); and acute liver dysfunction (total bilirubin >34.2 μmol/L [>2.0 mg/dL], with ≥50% increase compared with baseline). Patients were classified as having acute organ failure if they died during hospitalization or received at least 1 of the following therapies: heated high-flow nasal cannula, noninvasive ventilation (bilevel positive airway pressure or continuous positive airway pressure), invasive mechanical ventilation, dialysis or renal replacement therapy, or vasopressor support.
Neighborhood Social Disadvantage
Clinical data abstracted from patient charts (for example, patient characteristics, intensive care unit status, clinical characteristics) were merged with the SVI to understand how neighborhood factors influenced COVID-19 outcomes. Developed by the Centers for Diseases Control and Prevention, the SVI provides a composite measure of community susceptibility to adversities in the face of health shocks and includes 4 subindices: socioeconomic status, household composition and disability, racial or ethnic minority status and language, and housing type and transportation (7). See Appendix Table 1 for component measures for each subindex. The index is a percentile rank, ranging from 0 to 1, with higher values indicating greater social vulnerability or disadvantage. We transformed SVI reported at the census tract level into ZIP code level using a population-weighted average within each ZIP code. We hypothesized that patients from ZIP codes with higher SVI (that is, greater neighborhood disadvantage) would have poorer COVID-19 hospital outcomes. Thus, if neighborhood disadvantage effects on COVID-19 hospitalization outcomes are independent of individual patient clinical risk factors (for example, age, comorbid conditions), we would anticipate that SVI would be associated with poorer outcomes even after controlling for patient factors.
Appendix Table 1.
SVI Components and Data Sources
Covariates
Individual-level patient covariates included demographic characteristics (age, sex), baseline clinical characteristics (Charlson comorbidity index), clinical measurements on hospital admission (pulse oximetry, respiratory rate), and time period of hospital admission (March through May 2020, June through August 2020, and September through December 2020, corresponding to dates of COVID-19 surges in Michigan). Selection of clinical measurements was based on our team's previous work identifying risk factors for hospital mortality (11), with the exception of creatinine level, owing to greater than 10% missingness of this variable in our sample.
Statistical Analysis
Descriptive statistics were used to describe the patient cohort living in a ZIP code with an SVI rating in the highest quartile compared with all others. To determine whether COVID-19 hospitalization outcomes were related to neighborhood SVI, mixed-effects logistic regression models were fit for each of the outcomes using melogit in Stata (StataCorp). The composite SVI and its subindices were included as a continuous variable in separate models to avoid multicollinearity.
Our primary models controlled for time using a categorical variable corresponding to the COVID-19 surges in Michigan and clinical patient factors associated with COVID-19 outcomes in addition to a hospital-level random intercept to account for within-hospital correlation. To disentangle the individual effect of patient ZIP code SVI from the cluster-level effect of hospitals, hospital-level mean SVI exposures were included in all models. Postestimation predictive margins were used to estimate the absolute risk for each outcome (“baseline” percentage) for a patient living in a ZIP code with an overall or subindex SVI score of 0.5 and the change in risk associated with an increase in the index by 0.25 (percentage point change for an increase of 1 quartile in the SVI).
To ensure rigor, sensitivity analyses were conducted by repeating the analyses in a subsample excluding patients admitted through hospital transfer, and the full sample including the patients from a low-volume hospital and transferred patients. Additionally, we also repeated the analysis using a logistic regression model with cluster robust standard errors in the main analytic sample excluding patients from low-volume hospitals. We estimated E-values as the degree of association or confounding, on the relative risk (RR) scale, between an unobserved variable and the outcome and between that variable and SVI, that would have to be present to explain away the differences in outcomes associated with SVI. The estimated confounding RRs, from 1.8 to 2.5, suggest that an unmeasured confounder not already represented by observed covariates would need to be moderate or large to produce these significant associations. We could not identify such large confounders, and our results are robust to this source of bias. All analyses were performed in SAS, version 9.4 (SAS Institute), and Stata, v16 (StataCorp), with α set at 0.05.
Role of the Funding Source
The groups funding this research had no role in the design, conduct, or analysis of data for this manuscript. They also played no role in the authors' decision to submit the manuscript for publication.
Results
Data from 2678 patients with COVID-19 who were hospitalized between March and December 2020 were available. After exclusion criteria were applied, data from 2309 patients were included in the analysis. The distribution of the overall SVI index (median, 0.50; range, 0.04 to 0.96) and the 4 subindices, socioeconomic status (median, 0.48; range, 0.04 to 0.90), household composition and disability (median, 0.61; range, 0.07 to 0.99), minority status and language (median, 0.51; range, 0.08 to 0.92), and housing type and transportation (median, 0.55; range, 0.07 to 0.95) suggest significant variation in neighborhood social disadvantage. Similarly, the hospital mean SVI exposure for the overall SVI index (median, 0.48; range, 0.20 to 0.81) and the 4 subindices, socioeconomic status (median, 0.45; range, 0.28 to 0.74), household composition and disability (median, 0.63; range, 0.18 to 0.77), minority status and language (median, 0.51; range, 0.31 to 0.68), and housing type and transportation (median, 0.50; range, 0.19 to 0.79) also showed wide variability between hospitals in our sample. Appendix Table 2 shows the within-hospital variation among the participating hospitals along with the distribution of patients living in high- and low-vulnerability ZIP codes in the hospitals included in the analysis.
Appendix Table 2.
Variation of Patients’ SVI Within the Participating Hospitals
Patients living in high-vulnerability ZIP codes were younger, were more often Black or Hispanic, had more comorbid conditions, and more frequently had Medicaid insurance than patients in lower vulnerability ZIP codes (Table 1). These patients from high social vulnerability ZIP codes differed in pulse oximetry findings (5.4% in high-vulnerability vs. 3.5% in low-vulnerability ZIP codes had oxygen saturation ≤80%) and respiratory rate on admission (58.8% vs. 59.5%, respectively, had abnormal respiratory rates ≥20 breaths/min), compared with patients from other ZIP codes. Patients from high-vulnerability ZIP codes also more frequently were treated in the intensive care unit (29.0% vs. 24.5%), received mechanical ventilation (19.3% vs. 14.2%), and were discharged to home (62.1% vs. 60.1%). Compared with patients from low-vulnerability ZIP codes, those from high-vulnerability ZIP codes also had higher rates of acute organ dysfunction (51.9% vs. 48.6%), organ failure (54.7% vs. 51.6%), and in-hospital death (19.4% vs. 16.7%) in these unadjusted data.
Table 1.
Characteristics of Hospitalized Patients From High vs. Low SVI ZIP Codes
Association Between Neighborhood Social Disadvantage and COVID-19 Outcomes
In mixed-effects regression analyses adjusting for individual patient clinical characteristics, time period, and mean hospital SVI exposure, a patient's neighborhood SVI was associated with receipt of mechanical ventilation, development of acute organ dysfunction, and development of acute organ failure. For example, a patient living in a ZIP code with an SVI of 0.5 such as Ludington, Michigan (a small harbor town in Northern Michigan), was estimated to experience an absolute risk for mechanical ventilation of 14.7%, acute organ dysfunction of 48.8%, and acute organ failure of 52.1% (Table 2). In comparison, a patient living in a ZIP code in inner-city Detroit with an estimated increase in SVI by 0.25 (1 quartile above) had an increase in the risk for mechanical ventilation by 2.1 percentage points, acute organ dysfunction by 2.8 percentage points, and acute organ failure by 2.8 percentage points (Table 2).
Table 2.
Association of Patient-Level Neighborhood Social Disadvantage With COVID-19 Hospitalization Outcomes (n = 2309)*
Investigation of SVI subindices showed that patients living in a ZIP code with higher socioeconomic status subindex scores (that is, lower socioeconomic status) were at higher risk for requiring mechanical ventilation (change in risk with increase in SVI by 0.25 [Δrisk] = 2.3 percentage points) and developing acute organ failure (Δrisk = 2.3 percentage points) compared with those living in areas with lower socioeconomic status subindex scores. Likewise, patients living in ZIP codes with higher household and disability subindex scores also experienced greater risk for developing acute organ dysfunction (Δrisk = 3.3 percentage points) and acute organ failure (Δrisk = 3.3 percentage points); whereas patients living in ZIP codes with higher minority status and language subindex scores had greater risk for developing acute organ failure (Δrisk = 3.0 percentage points). The association of hospital SVI exposure on outcomes showed no significant association with COVID-19 outcomes across all models (Appendix Table 3).
Appendix Table 3.
Association of Hospital-Level Mean SVI Exposures With COVID-19 Hospitalization Outcomes in Analytic Sample (n = 2309)*
Sensitivity Analysis
In a sensitivity analysis, we excluded all patients who were transferred from another hospital (n = 130); in another sensitivity analysis, we included patients from low-volume hospitals that were not included in the main analyses. The association between a patient's SVI and COVID-19 outcomes was attenuated in these models compared with our main specification (Appendix Tables 4 and 5). The population transferred to another hospital was notably more severely ill than the baseline population (Appendix Table 4). Additional sensitivity analyses in the full sample, which included patients from low-volume hospitals and patients who were transferred from another hospital, did not show any major differences from the study findings (Appendix Tables 6 and 7). An alternative analytic approach of logistic regression models with cluster robust standard errors also did not show any significant variation from our main findings, demonstrating the robustness of our methods.
Appendix Table 5.
Association of Hospital-Level Mean SVI Exposures With COVID-19 Hospitalization Outcomes in Subsample Excluding Patients Transferred in From Another Hospital (n = 2179)*
Appendix Table 4.
Association of Patient-Level Neighborhood SVI With COVID-19 Hospitalization Outcomes in Subsample Excluding Patients Transferred in From Another Hospital (n = 2179)*
Appendix Table 6.
Association of Patient-Level Neighborhood SVI With COVID-19 Hospitalization Outcomes in Sample Including Patients From Hospitals Classified as Low Volume (n = 2453)*
Appendix Table 7.
Association of Hospital-Level Mean SVI Exposure With COVID-19 Hospitalization Outcomes in Sample Including Patients From Hospitals Classified as Low Volume (n = 2453)*
Discussion
In this multihospital study of patients hospitalized for COVID-19, we found that persons living in neighborhoods with greater social vulnerability were more likely to receive mechanical ventilation, experience acute organ dysfunction, and develop acute organ failure. These associations remained significant after adjustment for patient demographic and clinical characteristics, suggesting that much of the neighborhood social disadvantage effects we observed were independent of important individual-level factors related to patients' age and preexisting comorbid conditions. The association between patient ZIP code social vulnerability and COVID-19 hospitalization outcomes also remained significant after adjustment for hospital social vulnerability “case mix,” suggesting that patients' neighborhood social disadvantage influences outcomes more than variation across hospitals caring for patients from high- versus low-vulnerability areas. Taken together, these findings suggest that patients' neighborhood social disadvantage affects hospital outcomes, including the need for mechanical ventilation and severity of organ dysfunction.
Our findings shed important light on the various contributors to racial and ethnic disparities in outcomes after COVID-19 hospitalization. Whether these disparities are driven by greater exposure risk due to housing, transportation, or other factors; greater susceptibility to infection after exposure; patients' underlying medical conditions; or differential access to care such that some people delay seeking care and consequently present to the hospital sicker remains unclear (4). Although several prior studies (including those performed by our group) (1–4) have found that patient race or ethnicity and social vulnerability are associated with higher overall COVID-19 mortality, we found no significant association between neighborhood social vulnerability and in-hospital mortality in this analysis (3, 5, 12–14). This observation echoes the conclusion of a recent systematic review by Mackey and colleagues (4) who (despite disparities in overall mortality) also reported no association between race, ethnicity, and case-fatality rates among those confirmed to have COVID-19. Our findings instead suggest that patients from socially vulnerable neighborhoods may present to the hospital in a sicker state, leading to more intensive care in the hospital. However, we find that once patients were hospitalized, neighborhood social factors did not influence outcomes of mortality and discharge disposition.
Our study adds to a growing literature examining the impact of structural racism on COVID-19 outcomes (15–17). For example, a recent study from Minnesota found that persons belonging to racial or ethnic minority groups had higher COVID-19 mortality rates than White persons, related to living in less advantaged neighborhoods as well as to higher residual mortality even when living within the same level of neighborhood disadvantage (18). Thus, both the Minnesota study and our Michigan study point to the importance of neighborhood-level disadvantage in COVID-19 outcomes, but the Minnesota study also supports the notion that systemic and structural inequalities experienced by persons in racial and ethnic minority groups cannot be elucidated by neighborhood contextual factors alone. Rather, policymakers must consider both individual social risks, such as poor-quality and segregated housing and difficulty accessing care, and neighborhood social risks, such as poor transportation networks, when devising strategies to mitigate the impact of COVID-19 in specific populations. Attention to these “upstream,” prehospital aspects of health quality and health care delivery may offset “downstream” outcomes following hospitalization for COVID-19.
Our study has limitations, including a focus on hospitalizations in 1 state and the observational nature of the data. As well, potential missing documentation in chart abstraction and data reflecting trends related to changing COVID-19 variants remain a threat to inference. In addition, our study focuses on hospitalized patients and thus does not capture data from outpatient or postacute care sources, which may influence overall associations.
Despite these limitations, our study has important strengths, including a focus on type of care received during hospital admissions, not just rates of admission as examined in other studies (2–4, 19–28). Further, we add rigor by expanding from studies of single health care systems to a multihospital statewide cohort. By integrating data on individual patient clinical factors with neighborhood-level social disadvantage factors, we are able to understand not only aspects such as exposure to SARS-CoV-2 necessitating admission but also access to and experiences of health care once COVID-19 is suspected or diagnosed.
In conclusion, our findings demonstrate that hospitalized patients with COVID-19 from more socially vulnerable neighborhoods are more likely to present with greater illness severity and require more intensive treatment, but once hospitalized, they experience no differences in hospital mortality or discharge disposition. Policymakers should target more socially vulnerable neighborhoods to improve access to COVID-19 testing, treatment, and vaccination, as well as to identify and address social needs to ameliorate disparities in COVID-19 health outcomes.
Footnotes
This article was published at Annals.org on 22 February 2022.
References
- 1. Ogedegbe G, Ravenell J, Adhikari S, et al. Assessment of racial/ethnic disparities in hospitalization and mortality in patients with COVID-19 in New York City. JAMA Netw Open. 2020;3:e2026881. [PMID: ] doi: 10.1001/jamanetworkopen.2020.26881 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Price-Haywood EG, Burton J, Fort D, et al. Hospitalization and mortality among black patients and white patients with covid-19. N Engl J Med. 2020;382:2534-2543. [PMID: ] doi: 10.1056/NEJMsa2011686 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Karmakar M, Lantz PM, Tipirneni R. Association of social and demographic factors with COVID-19 incidence and death rates in the US. JAMA Netw Open. 2021;4:e2036462. [PMID: ] doi: 10.1001/jamanetworkopen.2020.36462 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Mackey K, Ayers CK, Kondo KK, et al. Racial and ethnic disparities in COVID-19-related infections, hospitalizations, and deaths: a systematic review. Ann Intern Med. 2021;174:362-373. [PMID: ] doi: 10.7326/M20-6306 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Robinson-Lane SG, Sutton NR, Chubb H, et al. Race, ethnicity, and 60-day outcomes after hospitalization with COVID-19. J Am Med Dir Assoc. 2021;22:2245-2250. [PMID: ] doi: 10.1016/j.jamda.2021.08.023 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Quan D, Luna Wong L, Shallal A, et al. Impact of race and socioeconomic status on outcomes in patients hospitalized with COVID-19. J Gen Intern Med. 2021;36:1302-1309. [PMID: ] doi: 10.1007/s11606-020-06527-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Flanagan BE, Gregory EW, Hallisey EJ, et al. A social vulnerability index for disaster management. Journal of Homeland Security and Emergency Management. 2011;8. doi:10.2202/1547-7355.1792
- 8. Srivastava T, Schmidt H, Sadecki E, Kornides M. Social vulnerability, disadvantage, and COVID-19 vaccine rationing: a review characterizing the construction of disadvantage indices deployed to promote equitable allocation of resources in the United States. SSRN. Preprint posted online 1 September 2021. doi:10.2139/ssrn.3882863
- 9. Schmidt H, Weintraub R, Williams MA, et al. Equitable allocation of COVID-19 vaccines in the United States. Nat Med. 2021;27:1298-1307. [PMID: ] doi: 10.1038/s41591-021-01379-6 [DOI] [PubMed] [Google Scholar]
- 10. Chopra V, Flanders SA, O’Malley M, et al. Sixty-day outcomes among patients hospitalized with COVID-19 [Letter]. Ann Intern Med. 2021;174:576-578. [PMID: ] doi: 10.7326/M20-5661 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Mann CZ, Abshire C, Yost M, et al. Derivation and external validation of a simple risk score to predict in-hospital mortality in patients hospitalized for COVID-19: A multicenter retrospective cohort study. Medicine (Baltimore). 2021;100:e27422. [PMID: ] doi: 10.1097/MD.0000000000027422 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Gross CP, Essien UR, Pasha S, et al. Racial and ethnic disparities in population-level COVID-19 mortality [Letter]. J Gen Intern Med. 2020;35:3097-3099. [PMID: ] doi: 10.1007/s11606-020-06081-w [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Khazanchi R, Beiter ER, Ganguli I. Methodological considerations for modeling social vulnerability and COVID-19 risk-response to Nayak et al [Letter]. J Gen Intern Med. 2021;36:1115-1116. [PMID: ] doi: 10.1007/s11606-021-06601-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Islam SJ, Nayak A, Hu Y, et al. Temporal trends in the association of social vulnerability and race/ethnicity with county-level COVID-19 incidence and outcomes in the USA: an ecological analysis. BMJ Open. 2021;11:e048086. [PMID: ] doi: 10.1136/bmjopen-2020-048086 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Egede LE, Walker RJ. Structural racism, social risk factors, and covid-19 - A dangerous convergence for Black Americans. N Engl J Med. 2020;383:e77. [PMID: ] doi: 10.1056/NEJMp2023616 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Tan SB, deSouza P, Raifman M. Structural racism and COVID-19 in the USA: a county-level empirical analysis. J Racial Ethn Health Disparities. 2021. [PMID: ] doi: 10.1007/s40615-020-00948-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Yearby R, Mohapatra S. Systemic racism, the government's pandemic response, and racial inequities in COVID-19. Emory Law Journal. 2021;70:1419-1473.
- 18. Wrigley-Field E, Garcia S, Leider JP, et al. COVID-19 mortality at the neighborhood level: racial and ethnic inequalities deepened in Minnesota in 2020. Health Aff (Millwood). 2021;40:1644-1653. [PMID: ] doi: 10.1377/hlthaff.2021.00365 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Adegunsoye A, Ventura IB, Liarski VM. Association of Black race with outcomes in COVID-19 disease: A retrospective cohort study [Letter]. Ann Am Thorac Soc. 2020;17:1336-1339. [PMID: ] doi: 10.1513/AnnalsATS.202006-583RL [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Azar KMJ, Shen Z, Romanelli RJ, et al. Disparities in outcomes among COVID-19 patients in a large health care system in California. Health Aff (Millwood). 2020;39:1253-1262. doi:10.1377/hlthaff.2020.00598 [DOI] [PubMed]
- 21. Ebinger JE, Achamallah N, Ji H, et al. Pre-existing traits associated with COVID-19 illness severity. PLoS One. 2020;15:e0236240. [PMID: ] doi: 10.1371/journal.pone.0236240 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Golestaneh L, Neugarten J, Fisher M, et al. The association of race and COVID-19 mortality. EClinicalMedicine. 2020;25:100455. [PMID: ] doi: 10.1016/j.eclinm.2020.100455 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Gu T, Mack JA, Salvatore M, et al. Characteristics associated with racial/ethnic disparities in COVID-19 outcomes in an academic health care system. JAMA Netw Open. 2020;3:e2025197. [PMID: ] doi: 10.1001/jamanetworkopen.2020.25197 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Killerby ME, Link-Gelles R, Haight SC, et al; CDC COVID-19 Response Clinical Team. Characteristics associated with hospitalization among patients with COVID-19 - metropolitan Atlanta, Georgia, March-April 2020. MMWR Morb Mortal Wkly Rep. 2020;69:790-794. [PMID: ] doi: 10.15585/mmwr.mm6925e1 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Lara OD, O’Cearbhaill RE, Smith MJ, et al. COVID-19 outcomes of patients with gynecologic cancer in New York City. Cancer. 2020;126:4294-4303. [PMID: ] doi: 10.1002/cncr.33084 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Petrilli CM, Jones SA, Yang J, et al. Factors associated with hospital admission and critical illness among 5279 people with coronavirus disease 2019 in New York City: prospective cohort study. BMJ. 2020;369:m1966. [PMID: ] doi: 10.1136/bmj.m1966 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Rentsch CT, Kidwai-Khan F, Tate JP, et al. COVID-19 testing, hospital admission, and intensive care among 2,026,227 United States veterans aged 54-75 years. medRxiv. Preprint posted online 14 April 2020. doi: 10.1101/2020.04.09.20059964 [DOI] [Google Scholar]
- 28. van Gerwen M, Alsen M, Little C, et al. Risk factors and outcomes of COVID-19 in New York City; a retrospective cohort study. J Med Virol. 2021;93:907-915. [PMID: ] doi: 10.1002/jmv.26337 [DOI] [PMC free article] [PubMed] [Google Scholar]