Understanding contributors to racial and ethnic inequities in COVID-19 incidence and mortality rates.

BACKGROUND: Racial inequities in Coronavirus 2019 (COVID-19) have been reported over the course of the pandemic, with Black, Hispanic/Latinx, and Native American individuals suffering higher case rates and more fatalities than their White counterparts.
METHODS: We used a unique statewide dataset of confirmed COVID-19 cases across Missouri, linked with historical statewide hospital data. We examined differences by race and ethnicity in raw population-based case and mortality rates. We used patient-level regression analyses to calculate the odds of mortality based on race and ethnicity, controlling for comorbidities and other risk factors.
RESULTS: As of September 10, 2020 there were 73,635 confirmed COVID-19 cases in the State of Missouri. Among the 64,526 case records (87.7% of all cases) that merged with prior demographic and health care utilization data, 12,946 (20.1%) were Non-Hispanic (NH) Black, 44,550 (69.0%) were NH White, 3,822 (5.9%) were NH Other/Unknown race, and 3,208 (5.0%) were Hispanic. Raw cumulative case rates for NH Black individuals were 1,713 per 100,000 population, compared with 2,095 for NH Other/Unknown, 903 for NH White, and 1,218 for Hispanic. Cumulative COVID-19-related death rates for NH Black individuals were 58.3 per 100,000 population, compared with 38.9 for NH Other/Unknown, 19.4 for NH White, and 14.8 for Hispanic. In a model that included insurance source, history of a social determinant billing code in the patient's claims, census block travel change, population density, Area Deprivation Index, and clinical comorbidities, NH Black race (OR 1.75, 1.51-2.04, p<0.001) and NH Other/Unknown race (OR 1.83, 1.36-2.46, p<0.001) remained strongly associated with mortality.
CONCLUSIONS: In Missouri, COVID-19 case rates and mortality rates were markedly higher among NH Black and NH Other/Unknown race than among NH White residents, even after accounting for social and clinical risk, population density, and travel patterns during COVID-19.

Introduction

Racial inequities in case rates and clinical outcomes from the Coronavirus 2019 (COVID-19) have been reported over the course of the pandemic, with Black, Hispanic/Latinx, and Native American individuals suffering markedly higher case rates and more fatalities than their White counterparts [1]. There are a number of potential contributors to racial inequities in COVID-19 [2]. These include risk factors for exposure to the disease such as essential worker status, use of public transportation, and population density [3-5], as well as risk factors for poor outcomes among people with established disease, including comorbidities and social determinants of health such as poverty, education, access to care, and insurance coverage that are known to be associated with adverse health outcomes more broadly [6-11].

Understanding diverse contributors to these inequities and their mechanisms is a crucial step in designing clinical and policy initiatives that might alleviate them. Specifically, if the majority of the excess risk associated with minority race or ethnicity is a result of differential exposure to the disease, and therefore differential case rates, interventions might include better protection for essential workers in terms of personal protective equipment or paid sick leave when quarantining, or better access to testing among certain groups. On the other hand, if the majority of the excess risk is a result of differential clinical outcomes among those who contract the disease, better attention to access to novel therapies, access to high-quality health care, or a broader focus on improving health among at-risk groups would be more promising strategies.

Prior studies examining racial and ethnic inequities in COVID-19 case rates and clinical outcomes have often been limited to inpatients, or to either case rates or clinical outcomes but not both. Using a unique statewide dataset of confirmed COVID-19 cases across Missouri, linked with historical statewide hospital data, we examined differences by race and ethnicity in raw population-based case and mortality rates, and then used patient-level regression analyses to calculate the odds of mortality based on race and ethnicity. We also aimed to understand which risk factors, specifically social and clinical risk, population density, and travel patterns during COVID-19, were most strongly associated with racial and ethnic differences in COVID-19 incidence and outcomes.

Methods

Data

Case and fatality records were obtained from the Missouri Department of Health and Senior Services (DHSS) from March 1 to September 20, 2020. Cases were linked to administrative hospital claims data from the Hospital Industry Data Institute, a statewide repository of all inpatient and emergency department (ED) encounters in the state. We used a master patient identifier from Novetta Entity Analytics (Novetta LLC, McLean, VA) based on patient name, address, date of birth and gender to match individuals with confirmed COVID-19 to one or more prior hospital encounters during the previous two-year period. We successfully matched 64,526 out of 73,635 cases, for a match rate of 87.7% (see for flow diagram). This linkage allowed us to determine demographics (age, race, ethnicity), payer information, and home ZIP code among any person with any prior health care encounter in an inpatient or outpatient hospital setting, regardless of whether he or she were hospitalized for COVID-19. From these clinical data, we also extracted information on comorbidities as defined using the Chronic Conditions Warehouse methodology from the U.S. Centers for Medicare and Medicaid Services, and whether or not the patient had ever had a billing code indicative of a social determinant of health (SDOH) challenge using International Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM) z-codes Z55-Z65, which include codes for housing instability, lack of social support, history of abuse, and other important issues [12, 13].

We used patient address to geocode confirmed COVID-19 cases to the census block group-level and ZIP Code Tabulation Area-level to merge the case and hospital encounter data with 2015 Area Deprivation Index data from the University of Wisconsin Neighborhood Atlas [14], and 2018 population and population density estimates from the U.S. Census American Community Survey [15]. We also linked the data with 2019–2020 mobility data from the SafeGraph COVID-19 Data Consortium [16], Social Distancing Metrics, which links cellular phone locations to points of interest and allowed us to determine whether, in each census block group, the median time spent at home went up or down during the first four months of the pandemic in Missouri (March-June, 2020) compared to the same time period during 2019, a proxy for whether people were able to restrict travel.

Predictors and outcomes

Our primary predictor was race/ethnicity as obtained from individuals’ hospital data, defined as Non-Hispanic Black or African-American (henceforth NH Black), Non-Hispanic White (henceforth NH White), Hispanic, and Non-Hispanic Other or Unknown race (henceforth NH Other/Unknown) which were the categories available from the harmonized hospital dataset, and reflect hospital registration data. The Other or Unknown group includes six individual race categories (American Indian/Alaska Native, Asian, Pacific Islander, Other, Multiple Races, and Unknown or Refused), in addition to individuals with multiple races reported in the hospital claims data where no dominant mode was detected (e.g. an individual with two claims, one coded White and one coded Asian, ). We considered census blocks in the highest two quintiles of Area Deprivation Index to be highly deprived, and those in the highest quintile of population density to be highly dense. Our primary outcomes were case rate per 100,000 population and case fatality rates using deceased vital status codes included in the DHSS data, which include COVID-19-related expirations occurring in any setting.

Statistical analysis

We first calculated patient characteristics overall and by race/ethnicity group. We then plotted cumulative case and mortality rates by race/ethnicity over the study period. We then estimated multivariable associations of demographic, community, SDOH and clinical covariates on the log-odds of experiencing COVID-19 related mortality using logistic and hierarchical generalized linear methods. Three models were fit under each specification to evaluate stepwise differences in fixed and random effects for race and ethnicity associated with each model: demographic only; demographic and SDOH; and demographic, SDOH and clinical factors. Our primary analysis included race and ethnicity as fixed effects in our logistic models to quantify the log-odds of case fatality associated with NH Black, NH Other/Unknown and Hispanic individuals compared to White cases.

In secondary analyses, in order to estimate the extent to which observed disparities in COVID-19 related mortality persisted after accounting for individual risk factors, we constructed a ratio that captures the excess risk of belonging to a non-White racial group (beyond the risks that relate to covarying individual characteristics). To do this, we estimated mixed models in which race and ethnicity are modeled as random effects, to capture the individual-level risks within each racial group (the rate predicted by the model, which allows for explanatory variables that often covary by race). We also calculated the expected rate holding individual risk constant (i.e. the expected value determined by evaluating the linear combination of the individual’s other demographic and comorbidity data and the coefficient vector). When aggregated over all members of a racial group, this yields the expected rate based only on non-racial explanatory variables. The ratio of predicted-to-expected case fatality rates for each race/ethnicity category provides an estimate of what may be termed the “otherwise unexplained” racial disparities in COVID-19-related mortality.

This study was approved by the Office of Human Research Protection at the Washington University School of Medicine. The requirement for informed consent and HIPAA notification were waived due to the de-identified nature of the data. SAS version 9.4, Cary North Carolina was used for all analyses. For each primary outcome, a p value less than 0.05/3 = 0.017 was considered statistically significant to account for multiple comparisons.

Results

Case characteristics

As of September 10, 2020 there were 73,600 confirmed COVID-19 cases in the State of Missouri. Among the 64,526 case records (87.7% of all cases) that merged with prior demographic and health care utilization data, 12,946 (20.1%) were NH Black, 44,550 (69.0%) were NH White, 3,822 (5.9%) were NH Other/Unknown, and 3,208 (5.0%) were Hispanic (). NH White patients were the oldest on average (47.2 years, compared to 44.4 for NH Black, 38.2 for NH Other/Unknown, and 37.2 for Hispanic patients), and roughly 40% of each group was male. NH White patients were also the least likely to be insured by Medicaid (8.0%, compared to 23.2% for NH Black patients, 19.0% for NH Other/Unknown patients, and 28.3% for Hispanic patients) or to be uninsured (12.7%, compared to 29.4% for NH Black patients, 24.1% for NH Other/Unknown patients, and 33.1% for Hispanic patients). NH White and NH Other/Unknown patients were less likely to have SDOH Z-codes in their billing records, indicating a code for a relevant social risk factor, at 3.8% and 4.0% respectively, compared to 9.5% for NH Black patients and 5.6% for Hispanic patients. The 9,074 cases that did not merge with prior hospital records were younger, more often male, and particularly likely to be of other or unknown race (reflecting data collected with the test itself rather than via our demographic matching).

Regarding comorbidities, NH White patients were the most likely to have evidence of a psychiatric disorder, but less likely to have most other comorbidities including asthma, chronic kidney disease, heart failure, hypertension, obesity, tobacco use, or uncontrolled diabetes than NH Black patients. Patterns for other-race patients and Hispanic patients were variable.

For neighborhood characteristics, NH Black and Hispanic patients were much more likely to live in areas with high levels of deprivation as defined by the Area Deprivation Index, and to live in high-population-density areas. White patients were the most likely to live in census block groups where the median time spent at home, as measured by cell phone activity, was higher in March-June 2020 than March-June 2019, suggesting higher levels of change in travel patterns associated with local stay-at-home orders.

Population-based case and mortality rates

Examining case rates over time, there were early and persistent inequities across racial and ethnic groups, particularly for the NH Black and NH Other/Unknown groups ().

Cumulative COVID-19 related cases per 100,000 by race and ethnicity, Missouri residents, Mar. 1, 2020 –Sep. 10, 2020.

Source: Authors’ analysis of COVID-19 case data from the Missouri Department of Health and Senior Services and 2018 population estimates from the U.S. Census American Community Survey.

Patterns for population-based mortality were similar (). Cumulatively, the NH Black COVID-19-related death rate was 58.3 per 100,000 population in the state, compared with 38.9 for NH Other/Unknown, 19.4 for NH White, and 14.8 for Hispanic. These inequities appeared within the first month of the pandemic and grew in magnitude until mid-May 2020, with the gap between Black and White persisting thereafter.

Cumulative COVID-19 related deaths per 100,000 by race and ethnicity, Missouri residents, Mar. 1, 2020 –Sep. 10, 2020.

Source: Authors’ analysis of COVID-19 case data from the Missouri Department of Health and Senior Services and 2018 population estimates from the U.S. Census American Community Survey.

Multivariate regression models for mortality

There were 1,437 deaths in the sample, for a raw mortality rate of 2.2%. In logistic regression models that included demographics alone, age (odds ratio [OR] 1.09, 95% confidence interval [CI] 1.09–1.10 per one year increase in age) and male sex (OR 2.00 (1.78–2.24)) were associated with higher odds of mortality (). NH Black race was associated with a 2.82 times higher odds of dying (2.48–3.21), NH Other/Unknown race with 1.78 times higher odds (1.34–2.36), and Hispanic ethnicity with 1.54 times higher odds (1.09–2.16) of dying as compared to NH White individuals. After adding risk factors including insurance source, history of an SDOH Z-code in the patient’s claims, census block travel change, population density, and Area Deprivation Index to the model, NH Black race (OR 2.07, 1.79–2.39) and NH Other/Unknown race (OR 1.58, 1.18–2.10) remained associated with higher odds of mortality. Finally, after adding clinical comorbidities to the model, NH Black race (OR 1.75, 1.51–2.04) and NH Other/Unknown race (OR 1.83, 1.36–2.46) remained strongly associated with higher odds of mortality.

The ratio analyses based on random effects models revealed similar results. Observed case fatality rates for NH Black cases were 1.6 times the rate of NH White cases (). After accounting for age and sex, risk-adjusted case fatality rates for NH Black cases increased to 2.4 times the rate for NH White cases. Case fatality disparities were moderated but persisted after accounting for demographic and social factors (NH Black case fatality 1.8 times NH White case fatality) and more so when clinical characteristics were added to the model (NH Black case fatality 1.5 times NH White case fatality). Analyses of the intraclass coefficient from each model suggested that race explained roughly 5.4% or the variability in mortality in a model that included demographics alone, but that figure dropped to 2.9% when SDOH were added to the model, and 2.4% when clinical comorbidities were added ().

Observed and risk-adjusted case fatality rates.

Abbreviations: SDOH, social determinants of health. Brackets depict ratio of case fatality rate for the group of interest versus White patients, using the random effects model described in the methods.

Discussion

We found that major inequities in case rates and mortality for confirmed COVID-19 cases across Missouri emerged early and persisted well into the pandemic. Confirmed case rates for NH Black Missourians were markedly higher than for their NH White counterparts, as were case and mortality rates among the relatively small number of NH Other/Unknown race patients in our dataset. Additionally, analysis of patient-level data suggested that NH Black and NH Other/Unknown race were strongly associated with a higher risk of mortality, even after accounting for age and sex, social risk, neighborhood characteristics, mobility as assessed by cell phone data, and medical comorbidities.

The reasons for the markedly different case diagnosis rate by race and ethnicity are likely many. Black and Hispanic individuals, as well as those of other racial groups, are more likely to be essential workers, and to be in high-contact essential employment positions, like health care, education, and public transportation, as well as to use public transportation for job access [3, 17]. NH Black and Hispanic individuals also live in more population-dense ZIP codes and denser housing conditions, on average, than NH White individuals in Missouri, which can also impact one’s likelihood of coming into contact with someone with known or unknown COVID-19 infection. Another possible mediator of exposure that we were not able to measure in our study is the prevalence of intergenerational households [18]. The nature and number of intergenerational contacts varies by culture, which has implications for exposure to COVID-19 [19]. It is possible that the high case rates seen in the NH Black and NH Other/Unknown race populations in Missouri were related to exposure through one or more of these means, though the lower case rates among Hispanic individuals counters this hypothesis somewhat.

In patient-level regression analyses, mortality was much higher among Black and Other race individuals than White ones, with odds ratios near two for mortality among confirmed cases even after accounting for a host of potential confounders and mediators, including social risk factors like insurance status and Area Deprivation Index as well as clinical comorbidities. It is likely that the residual differences we see by race even after including explanatory variables reflect not biological differences in likelihood of poor outcomes based on skin color per se, but differences in likelihood of poor outcomes because of the many downstream impacts of exposure to individual and structural racism, though we cannot test each of these directly. For example, there is convincing evidence that the experience of racism and discrimination leads to inflammation and elevated levels of circulating biomarkers associated with poor health outcomes, which may hinder people’s ability to fight COVID-19 [20-23]. Racism and discrimination are also associated with poor health in general; claims-based comorbidity assessment cannot determine differences in risk within diagnoses, for example failing to differentiate between severe longstanding diabetes and more mild or recent diabetes, risk differentials that might inaccurately capture clinical severity particularly in minority groups.

It is also possible that worse outcomes for NH Black and NH Other/Unknown race individuals represent interpersonal racism leading to poorer quality care within the same hospitals, for example, clinicians electing to treat Black or other race patients differently from White patients. While treatment data were unavailable in our dataset, future research is needed to examine the types of treatments that Black and Other race patients compared to White patients have received over the course of the pandemic, and to ensure there has been equity in access to clinical trials and other treatments such as convalescent plasma.

The choice of hospital may matter as well: one recent study suggested that much of the difference in mortality between Black and White patients might be explained by the hospitals in which each group receives care [24]. Prior studies have shown that Black and Hispanic individuals tend to receive care in lower-quality hospitals, which could reflect structural racism in differential resourcing of hospitals that traditionally serve these groups, or other barriers to accessing high-quality care such as transportation [25]. Transportation and access issues may be particularly acute in rural parts of Missouri and other states; rural areas of the state are disproportionately White, suggesting that disparities might be even greater in urban areas where access to referral centers is higher.

Another possibility for the patterns we report is that cases may be differentially confirmed by testing based on race or ethnicity. While we had a statewide sample of cases regardless of hospitalization status, it is likely that many cases of COVID-19 go un-diagnosed. Prior studies have shown that access to testing differs by race, which could mean that more cases go unconfirmed in communities of color. If diagnosis rates differ by race, this could create selection bias, whereby the Black, Hispanic, or Other/Unknown race patients in our sample are comprised of a sicker, more symptomatic subset of total COVID-19 cases. Worse outcomes among confirmed cases in these groups would in that case represent selection rather than biology. However, our population-based analyses, which demonstrate markedly higher fatality rates per 100,000 population, show similar if not more striking disparities, and these numbers should not suffer from selection bias to the same degree.

Our study should be interpreted in light of other extant findings in the literature. A study of over 11,000 patients from the Premier network found no significant differences in outcomes for hospitalized patients by race after adjusting for comorbidities, neighborhood characteristics, and insurance status [26]. Studies from the Veterans Health Administration reported a higher incidence of testing and hospitalization among Black veterans but no difference in mortality by race after adjusting for clinical characteristics [27, 28]. Studies from New York City and Wisconsin found no difference in mortality among Black patients after accounting for comorbidities and presenting severity of illness [29]. We suspect that the differences in these findings compared to ours reflect the use of clinical data to adjust risk and our use of a statewide rather than hospital-based sample, though we cannot be sure.

Our study has limitations. We used hospital registration data to obtain race and ethnicity data, which have a number of limitations. Because we relied on claims data, we lack detailed clinical information such as blood pressure or oxygen saturation at admission with which to risk adjust. Area Deprivation Index and American Community Survey data are only updated periodically, and shifts in demographics may have occurred between 2015 and the present. Some individuals lack cell phones, which could bias ZIP code-level cell phone data. Our data reflect the experience of a single state; while it is a diverse state with urban and rural areas, a sizeable minority population, and a range of hospital types, patterns we noted here may not apply to other states with different demographics, a different time course of COVID-19, or different patterns or mitigators of structural racism compared to Missouri. Our NH Other/Unknown and Hispanic groups were small, and likely reflect a wide diversity of nationalities, backgrounds, and socioeconomic circumstances. Our findings regarding Hispanic individuals run counter to findings in other parts of the country with larger Hispanic populations, and therefore should be interpreted with caution. Data limitations and sample size precluded further breakdown of these groups into subgroups by country of origin or immigration status, but such detailed analysis is an important direction for future work. We were unable to match roughly 12 percent of cases to prior health care utilization in hospital inpatient or outpatient departments, and this group undoubtedly represents a healthier subset of cases, with lower raw mortality rates. If members of racial or ethnic minority groups were less likely to have health care interactions prior to their COVID-19 diagnosis due to limitations in access, whether due to financial or insurance-related concerns, this could bias our sample, as well as limit our full ascertainment of comorbidities. Finally, death data may lag in reporting.

Our findings suggest that, to reduce or eliminate such inequities in future waves of COVID-19 or in future pandemics, attention needs to be paid to reducing exposure among communities of color, perhaps by strategies like providing PPE, sick leave, and testing to essential workers, and also to improving outcomes among Black and other minority patients once exposed. Black and other minority patients diagnosed with COVID-19 should be considered to be at higher risk than their White counterparts, and interventions or enhanced monitoring directed at this population may have the potential to improve outcomes and reduce inequities. More broadly, interventions that help reduce racism and discrimination, improve access to high quality hospitals, or improve the quality of care delivered to minority patients, should be prioritized.

In conclusion, COVID-19 case rates and mortality rates were markedly higher among NH Black and NH Other/Unknown race than among NH White residents, even after accounting for social and clinical risk, population density, and travel patterns during COVID-19.

Flow diagram of patient inclusion and exclusion.

Abbreviations: MO = Missouri; MO DHSS = Missouri Department of Health and Social Services.

(TIF)

Click here for additional data file.

Further breakdown of “Other” race category.

Numbers represent hospital claims rather than unique individuals, since patients could have one race reported on one claim and a different one reported on another claim. If a patient had an even number of claims with two different races (for example, two with White and two with Asian), he or she was left in the “other” category because we could not determine which was the appropriate category.

(DOCX)

Click here for additional data file.

Impact of additional adjustment on ICC associated with race variable.

Abbreviations: ICC, intraclass coefficient; SDOH, social determinants of health; SE, standard error.

(DOCX)

Click here for additional data file.

Transfer Alert

This paper was transferred from another journal. As a result, its full editorial history (including decision letters, peer reviews and author responses) may not be present.

2 Jul 2021

PONE-D-21-03586

Understanding Contributors to Racial and Ethnic Inequities in COVID-19 Incidence and Mortality Rates

PLOS ONE

Dear Dr. Joynt Maddox,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

Please address the following journal, additional editor, and reviewer comments.

Please submit your revised manuscript by Aug 16 2021 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see:  http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols . Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at  https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols .

We look forward to receiving your revised manuscript.

Kind regards,

Fernando A. Wilson, PhD

Academic Editor

PLOS ONE

Additional Editor Comments:

- There is inconsistent capitalization of ‘white’ within text and Fig. 3 caption.

- I suggest adding data labels to the Fig. 3 columns to be consistent with Fig. 1.

- Confirm that the PLOS ONE guideline on Acknowledgements has been followed (link below):

https://journals.plos.org/plosone/s/submission-guidelines#loc-acknowledgments

- Please verify references follow PLOS ONE formatting requirements. For example, all journal names must be abbreviated using the NCBI database.  Please refer to the following link:

https://journals.plos.org/plosone/s/submission-guidelines#loc-references

Journal Requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

2. Thank you for stating the following in the Competing Interests section:

"I have read the journal's policy and the authors of this manuscript have the following competing interests: Dr. Joynt Maddox receives research support from the National Heart, Lung, and Blood Institute (R01HL143421) and National Institute on Aging (R01AG060935, R01AG063759, and R21AG065526), and previously did contract work for the US Department of Health and Human Services. The other authors report no disclosures."

Please confirm that this does not alter your adherence to all PLOS ONE policies on sharing data and materials, by including the following statement: "This does not alter our adherence to  PLOS ONE policies on sharing data and materials.” (as detailed online in our guide for authors http://journals.plos.org/plosone/s/competing-interests).  If there are restrictions on sharing of data and/or materials, please state these. Please note that we cannot proceed with consideration of your article until this information has been declared.

Please include your updated Competing Interests statement in your cover letter; we will change the online submission form on your behalf.

Please know it is PLOS ONE policy for corresponding authors to declare, on behalf of all authors, all potential competing interests for the purposes of transparency. PLOS defines a competing interest as anything that interferes with, or could reasonably be perceived as interfering with, the full and objective presentation, peer review, editorial decision-making, or publication of research or non-research articles submitted to one of the journals. Competing interests can be financial or non-financial, professional, or personal. Competing interests can arise in relationship to an organization or another person. Please follow this link to our website for more details on competing interests: http://journals.plos.org/plosone/s/competing-interests

3. We note that you have indicated that data from this study are available upon request. PLOS only allows data to be available upon request if there are legal or ethical restrictions on sharing data publicly. For information on unacceptable data access restrictions, please see http://journals.plos.org/plosone/s/data-availability#loc-unacceptable-data-access-restrictions.

In your revised cover letter, please address the following prompts:

a) If there are ethical or legal restrictions on sharing a de-identified data set, please explain them in detail (e.g., data contain potentially identifying or sensitive patient information) and who has imposed them (e.g., an ethics committee). Please also provide contact information for a data access committee, ethics committee, or other institutional body to which data requests may be sent.

b) If there are no restrictions, please upload the minimal anonymized data set necessary to replicate your study findings as either Supporting Information files or to a stable, public repository and provide us with the relevant URLs, DOIs, or accession numbers. Please see http://www.bmj.com/content/340/bmj.c181.long for guidelines on how to de-identify and prepare clinical data for publication. For a list of acceptable repositories, please see http://journals.plos.org/plosone/s/data-availability#loc-recommended-repositories.

We will update your Data Availability statement on your behalf to reflect the information you provide.

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: Partly

Reviewer #3: Partly

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Yes

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: No

Reviewer #2: Yes

Reviewer #3: No

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: GENERAL COMMENTS

The authors present a well written study of racial/ethnic inequities in COVID-19 incidence and mortality. The use of state-wide data is a major strength of this manuscript and the discussion of observed disparities is generally thoughtful. I provide a few specific comments below.

MAJOR COMMENTS

1. The analysis of case distribution by race/ethnicity (fig 1) does not take advantage of individual-level information on race/ethnicity and instead only presents COVID-19 case rates by ZIP-code level %Black/African American. I believe you have the data necessary to examine case rates by individual-level race/ethnicity and should present this in addition to (or in lieu of) the current analysis. This would more directly answer the first part of the question set out in the Introduction: whether the case rates are different by race/ethnicity or if among those who are diagnosed with the disease, the mortality rate differs by race/ethnicity, or both.

2. The Discussion is generally well-written and thoughtful, but primary focuses on the results for the NH Black/African American group and does not provide a discussion for the NH Other and Hispanic groups, for which elevated risks were also observed.

a. The “NH Other” group had ORs higher than the NH Black group in the fully adjusted model. The authors should provide more discussion here. Is there any other information of who comprises this “Other” group (e.g., Asian, Pacific Islanders, American Indians)? What hypotheses do they have for the observed results for this group?

b. The mortality rate among Hispanic cases was elevated but was attenuated in further adjusted models. How do these results compare with the literature? Hispanics are large diverse group, with many different countries of ancestry. Are there additional data on the Hispanic individuals in these data or in Missouri in general that could inform why different results were (or were not) observed for these analysis (e.g., Hispanic origin, immigration status, etc.)?

OTHER COMMENTS

1. Line 133: What is meant by “race/ethnicity risk factors”? Do you mean race/ethnicity as categorical variables, or factors that vary with race/ethnicity? Please specify.

2. Lines 155-158/Fig 1: Add confidence intervals for effect estimates to the figure and include in the text for effect estimates stated (i.e., 870.4 and 1,877.9).

3. Figure1: Results for mortality per 100,000 population (grey bars in Fig 1) are not mentioned in the text (neither in the Results section or in the Statistical Analysis description). Include this in the text in these two sections.

4. Lines 233-234/Fig 3: If possible with this type of analysis, add confidence intervals for effect estimates to the figure and include in the text for effect estimates stated (i.e., 1.6, 2.4, etc.).

5. Lines 332-333: The strategies listed all relate to proximal intervention strategies (i.e., closer to the individual-level—using PPE, taking sick leave, getting tested). Earlier in the Discussion you provided very clear statements about how racism and discrimination and social determinants of health may have affected COVID-19 outcomes. As such, you could include in this sentence some potential interventions targeting these more distal factors related to health.

Reviewer #2: The authors report their analysis of covid19 infection rates and outcomes using a state-wide dataset. The following are some questions and comments:

1. Figure 1 – this is an excellent figure. However, graphically, it does not clearly show that not only are there increased case number and deaths in zip codes with higher percentage of Blacks, but also the percentage of deaths among patients diagnosed with covid increases too.

2. Figure 2 – it is interesting that there is such a discrepancy between Hispanic population and the other minorities. What do the authors think accounts for this? The data presented suggest that Hispanic and Black patients had same likelihood of being insured by Medicaid, being uninsured, and having social risk factors. Based on table 1, it would seem like the differences in outcomes was attributable to prevalence of co-morbidities? In the US, both Blacks and Hispanics have higher rates of obesity and diabetes, but in the current cohort, Hispanics appear to have lower rates of obesity and uncontrolled diabetes. Does this reflect a bias in the data capture and perhaps an issue with access to care?

3. Line 206-211 – White patients had lower levels of change in travel patterns – could this also suggest less cautious behavior, or reflect differences in “essential worker” status?

4. Did the authors consider the effect of rural location or distance from a major health care center? Those factors may also impact access to care.

5. Although the authors report a 87% match rate, there were still 9000 patients who were not matched. Based on the methods, these unmatched patients did not have a hospitalization in the prior 2 years, which suggests that they were very healthy (did not require hospitalization) or did not have access to necessary medical care.

Reviewer #3: Line Comment

25 COVID-19

32 Unclear what 99.1% represents - % of total COVID cases +/- address info?

Throughout abstract Use COVID-19, not just COVID

37 COVID-19

40 Non-Hispanic other race

Throughout Manuscript Use COVID-19, not just COVID

83 Do you ever use MPID again? If not, do not need to define abbreviation

98-102 References for sources

105 (March-June, 2020)

110 Place Non-Hispanic Other at the end of the list, since it is a catch-all category

129 Inconsistent race terminology throughout paper.

154 Line glosses over important details – how many total cases, how many with complete address information? Methods cites 73,600 cases. Need a flow diagram, and more transparent text here and in methods to understand analytic denominator of study.

155 Provide total number of deaths, and overall case fatality rate.

Fig 1. Unclear what n=xx represents. Clarify legend.

Fig 2. X axis labels crowded and difficult to read

Table 1. Age needs units, e.g., year, which implies continuity

Add number of deaths row

Race terminology does not match rest of the paper, e.g., Other/Unknown – harmonize.

Why were quintiles used in table/model? No mention in Methods why/how you collapsed categories.

Do you mean to say 32.6% of people spent less time at home in 2020 vs 2019, or that 67.4% spent more time at home during COVID-19.

179 “Matched Patients” is a confusing term in the title. Remove term, flow chart would clarify your analytic sample.

199 Limit P value to < 0.001

201, 206 Starting consecutive paragraphs with “In terms of…” is awkward – change both.

Figure 3 Narrow brackets showing bars that you are comparing – appears that you are comparing blue and orange bars, but does not line up exactly

Table 2 Age needs units label, e.g., years. Why was age modeled as a continuous linear variable? Many other models find it is not linear, and use categories to define age risk, e.g., King PLOS One 2020, or splines, e.g., Mathur Lancet 2021.

251 “… white or Hispanic...”?

252 You are now being vague about a precise terminology for race category used throughout your paper.

276-291 After stating that the reasons for this are unclear, the authors charge ahead into a discussion about racism as the etiology. They never mention the term disparities a description, instead jumping to racism as a causal explanation. The underlying theme seems to be worse outcomes for Blacks = racism.

313 Add Rentsch 2021 PLOS Medicine paper for another take on VA race issue.

315 Do you mean “laboratory data” when you state “clinical data”? Rentsch paper did not use laboratory data, but rather pre-existing diagnoses, the same as your analysis.

327 Line implies that structural racism is casual in this paper, which the authors do not prove, and admit there are other explanations. Suggest disparities – a descriptor, rather than racism, a hypothesized cause.

What about patients dropped who could not be matched? 2018 census data, with population shifts in the interim? 2015 Area Depravation Index data, with changes over the last 5 years? Those without cell phones? Lack of outpatient diagnoses to determine comorbidities. How reliable is your death data?

Throughout the document, authors are inconsistent about the use COVID vs COVID-19. Change all COVID to COVID-19.

Reference 22 has odd format, prune.

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: No

Reviewer #3: No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

16 Sep 2021

We very much appreciate the thoughtful and thorough reviews from your reviewers and editors. In response, we have removed the ZIP-level analyses and now present only the linked patient-level analyses in the manuscript to increase clarity. We have also updated Figure 1 to reflect these changes. Lastly, we have added additional text to the discussion and limitations regarding the Non-Hispanic Other Race category and Hispanic individuals. We think these changes have meaningfully strengthened the paper and hope that you agree.

Submitted filename: Racial Inequities in COVID PLOS ONE R1 Response 8-14-21.docx

Click here for additional data file.

8 Nov 2021

Understanding Contributors to Racial and Ethnic Inequities in COVID-19 Incidence and Mortality Rates

PONE-D-21-03586R1

Dear Dr. Joynt Maddox,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Fernando A. Wilson, PhD

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

Reviewers' comments:

19 Nov 2021

PONE-D-21-03586R1

Understanding Contributors to Racial and Ethnic Inequities in COVID-19 Incidence and Mortality Rates

Dear Dr. Joynt Maddox:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

If we can help with anything else, please email us at plosone@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. Fernando A. Wilson

Academic Editor

PLOS ONE

Table 1

Patient characteristics by race and ethnicity.

	Non-Hispanic			Hispanic	Total	Unmerged with prior hospital records
	Black or African American	White	Other / Unknown Races
COVID-19 Cases (n)	12,946	44,550	3,822	3,208	64,526	9,074
Number of deaths	425	911	62	39	1437	717
Case-Fatality Rate	3.3%*	2.0%*	1.6%^	1.2%*	2.2%	0.79%*
DEMOGRAPHICS
Age (mean, in years)	44.4*	47.2*	38.2*	37.2*	45.6	37.0*
Male	37.2%*	39.1%	41.2%^	41.9%^	39.0%	52.0%*
Female	62.8%*	60.9%	58.8%^	58.1%^	61.0%	48.0%*
NH Black	100.0%	0.0%	0.0%	0.0%	20.1%	11.1%*
NH White	0.0%	100.0%	0.0%	0.0%	69.0%	45.0%*
NH Other/Unknown	0.0%	0.0%	100.0%	0.0%	5.9%	38.6%*
Hispanic	0.0%	0.0%	0.0%	100.0%	5.0%	5.3%
INDIVIDUAL SOCIAL DETERMINANTS
Medicaid	23.2%*	8.0%*	19.0%*	28.3%*	12.7%	na
Uninsured	29.4%*	12.7%*	24.1%*	33.1%*	17.7%	na
SDOH Z-Code	9.5%*	3.8%*	4.0%^	5.6%	5.0%	na
INDIVIDUAL COMORBIDITIES
Psychiatric Disorder	20.9%^	23.2%*	12.3%*	18.0%*	21.9%	na
Alzheimer’s/Dementia	5.9%	5.9%*	2.5%*	3.1%*	5.6%	na
Asthma	14.4%*	6.9%*	5.4%*	7.1%^	8.3%	na
Chronic Kidney Disease	20.6%*	13.0%*	8.7%*	11.8%*	14.2%	na
COPD	8.1%	8.2%*	3.3%*	5.3%*	7.8%	na
Heart Failure	8.5%*	6.0%^	2.6%*	3.5%*	6.2%	na
Hypertension	41.6%*	31.0%*	18.3%*	21.6%*	31.9%	na
Lung Cancer	0.4%	0.5%^	0.1%^	0.3%	0.4%	na
Obesity	22.3%*	12.6%*	8.5%*	12.0%*	14.3%	na
Tobacco Use	16.2%*	11.1%*	8.7%*	9.2%*	11.9%	na
Uncontrolled Diabetes	9.2%*	4.6%*	3.7%*	6.1%	5.6%	na
AREA-LEVEL DATA
Census Block Group Area Deprivation Index q4-5	69.4%*	44.2%*	56.5%*	70.7%*	51.3%	47.0%*
Census Block Group More Time Spent at Home (Mar-Jun 2020 vs. 2019)	67.4%*	75.9%*	71.9%^	65.7%*	73.5%	75.8%*
Urban (highest quintile of ZIP Code Population Density)	90.8%*	56.7%*	70.3%*	67.8%^	64.9%	67.0%*

Source: Authors’ analysis of COVID-19 case data from the Missouri Department of Health and Senior Services merged with 2018–2020 hospital claims data from the Hospital Industry Data Institute, 2015 Area Deprivation Index data from the University of Wisconsin Neighborhood Atlas, 2019–2020 mobility data from Safe Graph, and 2018 population estimates from the U.S. Census American Community Survey. Abbreviations: SDOH, Social Determinant of Health; q, Quintile; COPD, Chronic Obstructive Pulmonary Disease.

*P<0.001

^P<0.05.

Table 2

Logistic regression models for mortality.

	Demographic Model	Demographic and Neighborhood Model	Demographic, Neighborhood, and Comorbidity Model
Effect	OR (95% CI)	OR (95% CI)	OR (95% CI)
Age (continuous)	1.09 (1.09–1.10)	1.09 (1.09–1.10)	1.07 (1.07–1.08)
Male	2.00 (1.78–2.24)	2.00 (1.78–2.24)	1.93 (1.71–2.17)
Non-Hispanic Black	2.82 (2.48–3.21)	2.07 (1.79–2.39)	1.75 (1.51–2.04)
Non-Hispanic Other/Unknown	1.78 (1.34–2.36)	1.58 (1.18–2.10)	1.83 (1.36–2.46)
Hispanic	1.54 (1.09–2.16)	1.32 (0.93–1.86)	1.25 (0.88–1.78)
Medicaid		1.29 (1.01–1.64)	1.06 (0.83–1.35)
Uninsured		0.90 (0.72–1.13)	0.98 (0.78–1.23)
SDOH Z-Code		2.02 (1.69–2.41)	1.01 (0.84–1.22)
Census Block Group Area Deprivation Index q4-5		1.07 (0.95–1.22)	1.05 (0.92–1.19)
Census Block Group Less Median Time Spent at Home (Mar-Jun 2020 vs. 2019)		0.86 (0.74–0.99)	0.85 (0.73–0.99)
ZIP Code Population Density q5		2.14 (1.83–2.49)	1.79 (1.53–2.10)
Psychiatric Disorder			1.28 (1.13–1.46)
Alzheimer’s/Dementia			2.23 (1.95–2.55)
Asthma			0.88 (0.71–1.10)
Chronic Kidney Disease			2.75 (2.38–3.19)
COPD			1.23 (1.06–1.42)
Heart Failure			1.27 (1.11–1.46)
Hypertension			1.10 (0.92–1.31)
Lung Cancer			2.39 (1.64–3.47)
Obesity			1.30 (1.12–1.50)
Tobacco Use			1.19 (1.00–1.42)
Uncontrolled Diabetes			1.27 (1.09–1.49)
Observations	64,526	64,526	64,526
COVID-19-Related Deaths (dep. variable)	1,437	1,437	1,437
C-Statistic	0.912	0.918	0.936

Source: Authors’ analysis of COVID-19 case data from the Missouri Department of Health and Senior Services merged with 2018–2020 hospital claims data from the Hospital Industry Data Institute, 2015 Area Deprivation Index data from the University of Wisconsin Neighborhood Atlas, 2019–2020 mobility data from Safe Graph, and 2018 population estimates from the U.S. Census American Community Survey. Abbreviations: SDOH, Social Determinant of Health; q, Quintile; COPD, Chronic Obstructive Pulmonary Disease.