Racial Disparities in COVID-19 Outcomes Among Black and White Patients With Cancer

Key Points Question Among patients with cancer and COVID-19, do non-Hispanic Black patients have more severe COVID-19 at presentation and worse COVID-19–related outcomes compared with non-Hispanic White patients, after adjusting for demographic and clinical risk factors? Findings In this cohort study of 3506 patients, Black patients with cancer experienced significantly more severe COVID-19 outcomes compared with White patients with cancer, after adjustment for demographic and clinical risk factors. Meaning These findings suggest that, within the framework of structural racism in the US, having cancer and COVID-19 is associated with worse outcomes among Black patients compared with White patients.


Introduction
3][4][5][6] Although Black individuals represent 13% of the US population, they account for 20% of COVID-19 cases and 23% of COVID-19-related deaths. 63][14] At present, detailed data on racial disparities with respect to baseline prognostic factors, illness course, and outcomes among patients with cancer are limited. 12Before the COVID-19 pandemic, it was well described that Black patients with cancer have the highest death rates compared with all other racial and ethnic groups. 15Factors associated with racial disparities among patients with cancer are complex and likely constitute an interplay of socioeconomic status, preexisting comorbid conditions, access to care, and other social determinants of health (SDOH).
Racial and ethnic minority groups have long experienced cancer health disparities, with a disproportionately higher burden of exposure to factors known to be associated with increased cancer risk (eg, smoking, obesity, and unhealthy diet), decreased screening and cancer prevention (and, thus, delayed cancer detection), and fewer opportunities to receive advances in cancer treatment or standard of care. 67][18] Hence, we hypothesized that US Black patients with cancer and COVID-19 have worse baseline clinical characteristics, severity of COVID-19 presentation, complications, and outcomes compared with White patients after adjusting for demographic and clinical risk factors.

Study Population
The COVID-19 and Cancer Consortium (CCC19) registry (eAppendix 1 and eTable 1 in Supplement 1) captures detailed clinical characteristics, course of illness, and outcomes of COVID-19 among patients with cancer. 13,19,20Our target population included any patient with cancer and confirmed diagnosis of COVID-19 cared for at one of our participating institutions (eAppendix 2 in Supplement 1).For this registry-based, retrospective cohort study, we included all reports of laboratory-confirmed SARS-CoV-2 infection submitted to the CCC19 registry between March 17, 2020, and November 18, 2020, for US residents with current or past diagnosis of cancer and Black or White race.Race and ethnicity were derived from the patient's electronic health record (EHR), which was either self-reported or assigned by health care practitioner or triage personnel.Excluded cancers were precursor hematological malignant neoplasms (eg, monoclonal gammopathy of undetermined significance), in situ carcinoma (except bladder), and nonmelanomatous, noninvasive skin cancer 19 (eFigure 1 in Supplement 1).This study was approved by the Vanderbilt University institutional review board and participating sites.Informed consent was waived because the data were anonymous, and the study posed minimal risk to participants, in accordance with 45 CFR §46.Reporting of results follows the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline. 21

Study Framework
Patient-reported race and ethnicity were captured following the Centers for Disease Control and Prevention Race and Ethnicity codes. 22The primary outcome was a 5-level ordinal scale of COVID-19 severity based on a patient's most severe disease status: (1) none of the complications listed here, (2) hospital admission, (3) intensive care unit admission, (4), mechanical ventilation use, and ( 5) death from any cause. 23Index date for analysis was defined as diagnosis of COVID-19, and outcomes were assessed over the patients' total follow-up period.We performed an additional analysis with 30-day all-cause mortality as a secondary outcome.

JAMA Network Open | Oncology
The Institute of Medicine defines disparity as a difference in treatment provided to members of different racial (or ethnic) groups that is not due to access-related factors or needs or justified by the underlying health conditions or preferences of patients. 24Implementing the Institute of Medicine's definition of disparity, 24 only those factors not considered to be associated with disparity (ie, allowable covariates) were included in the analysis to account for the direction of the associations.

Statistical Analysis
A statistical analysis plan including model specification was predetermined before the analysis and was revised upon submission by the lead authors and the CCC19 Epidemiology, Biostatistics, and Informatics Cores (eAppendix 3 and eAppendix 4 in Supplement 1).We used standard descriptive statistics to summarize baseline covariates, rates of clinical complications (eg, cardiovascular and pulmonary complications or acute kidney injury), interventions after COVID-19 diagnosis (ie, supplemental oxygen, transfusion, remdesivir, hydroxychloroquine, and corticosteroids), and individual components of the ordinal severity outcome.We calculated unweighted and weighted absolute standardized mean differences for baseline covariates to evaluate the balance between Black and White patients; an absolute standardized mean difference less than 0.1 indicated balance.
Racial disparities in COVID-19 severity were estimated from minimally and fully adjusted multivariable ordinal logistic regression models and 30-day all-cause mortality from logistic and modified Poisson regression models. 27,28Treatment variables were binary indicators to account for patients receiving multiple therapies (eTable 2 in Supplement 1).The assumed functional form for continuous variables (age) was based on exploratory analyses (eFigure 2 in Supplement 1).Racestratified estimates for obesity, comorbidities, and cancer status were obtained from adjusted models with interaction terms between race and these factors, given their known association with cancer and COVID-19 health disparities. 6 addition, we estimated inverse probability of treatment weighted differences in COVID- We used the e-value to quantify sensitivity to unmeasured confounding. 29,30Additional details on evaluating proportional odds assumptions (eFigure 3 in Supplement 1), multiple imputation to impute missing and unknown data (10 iterations; missingness rates were <5%), and additional statistical methods are provided in the eAppendix 4 in Supplement 1. 31

Patient Characteristics
From The e-value for the unweighted mortality OR was 1.83, and the e-value for the 95% CI was 1.48.
Thus, an unobserved factor would need to be associated with both race and mortality with a risk ratio of at least 1.83 to fully attenuate the observed association; the risk ratio would need to be at least 1.48 for the null-hypothesized value (1.0) to be included in the 95% CI.Such an association is larger than most documented associations in the CCC19 cohort. 23The 95% CI from the weighted model for mortality included the null value.

Discussion
The COVID-19 pandemic has highlighted and exacerbated longstanding health and social inequities in the US.These are likely the same inequities that fuel the disparities seen across the cancer control continuum and that lead to higher mortality rates among Black patients. 32Given the paucity of data on one of the most vulnerable racial and ethnic groups, 15,33  Abbreviations: OR, odds ratio; RR, relative risk.
a ORs comparing COVID-19 severity between non-Hispanic Black vs non-Hispanic White patients were estimated from ordinal logistic regression models; ORs greater than 1 indicate higher COVID-19 severity.
b ORs comparing 30-day mortality between non-Hispanic Black vs non-Hispanic White patients were estimated from logistic regression models; ORs greater than 1 indicate higher odds of 30-day all-cause mortality.
c RRs comparing 30-day mortality between non-Hispanic Black vs non-Hispanic White patients were estimated from modified Poisson regression models; RRs greater than 1 indicate higher risk of 30-day all-cause mortality.
d Adjusted for age (linear and quadratic terms) and sex.
e Adjusted for age (linear and quadratic terms), sex, region of patient residence, smoking status, obesity, cardiovascular and pulmonary comorbidities, kidney disease, diabetes, type of malignant neoplasm, Eastern Cooperative Oncology Group performance status, cancer status, timing and modality of anticancer therapy, and month of COVID-19 diagnosis.
f Weighted by the reciprocal of the probability of receiving the treatment (ie, race) that was actually received, which was estimated from a propensity score model for race that included age, sex, region of patient residence, smoking status, obesity, cardiovascular and pulmonary comorbidities, kidney disease, diabetes, type of malignant neoplasm, Eastern Cooperative Oncology Group performance status, cancer status, timing and modality of anticancer therapy, and month of COVID-19 diagnosis.Inverse probability of treatment weighted analysis was conducted during manuscript revision.study validates and adds to their findings by examining a large cohort of Black patients with more granular data on the cancer status, severity of COVID-19 at presentation, course of illness, including systemic complications, and outcomes over longitudinal follow-up.Structural racism refers to the ways in which societies reinforce systems of health care, law enforcement, education, employment, benefits, media, and housing that perpetuate discriminatory distribution of resources and attitudes. 35,36The COVID-19 pandemic highlighted the health burden on racial and ethnic minority groups and the complexity of structural racism, which has led to disproportionately worse clinical outcomes among Black patients. 37,38Cancer health disparities have been well described elsewhere 6,18 ; for example, Black women have a disproportionately higher breast cancer-specific mortality compared with White women, and Black men have a higher incidence and mortality from prostate cancer compared with White men.Factors such as access to care and SDOH that contribute to racial disparities in patients with cancer are complex and potentially extend to disparities in COVID-19 outcomes as well.

Figure. Adjusted Associations of Key
How should these findings be interpreted?The inequity in access to quality health care that can lead to worse baseline clinical factors and ultimately severe complications among racial and ethnic minority groups is well known. 6Our findings are similar to the patterns of racial disparities observed among patients with cancer, which points to an overlap in the root causes of racial inequities between cancer and COVID-19.In addition, racial and ethnic minority groups are more likely to live in conditions that pose a challenge to social distancing and also are more likely to be essential workers; therefore, sheltering in place may not be a possibility for many of them, thus putting these populations at an increased risk for exposure to SARS-CoV-2. 6,39These and additional sociodemographic pressures such as underinsurance may lead to delays in seeking medical care, which would likely be associated with more severe COVID-19 upon presentation.
Unfortunately, if these same racial inequities in access to medical care hold for cancer screening, in the near future, we are likely to see worsening disparities in rates of advanced stage cancers at diagnosis. 6The COVID-19 pandemic has been especially challenging for the treatment of patients with cancer. 40In our study, the racial disparities in COVID-19 severity were sustained for patients receiving active cancer therapy (Table 3).In addition, there are a host of drug interactions between chemotherapeutic agents and COVID-19 therapies, leading to alterations in standard treatments.For patients receiving radiation, the pandemic has caused delays between radiotherapy sessions, possibly leading to a reduction in therapeutic efficacy. 41Finally, current and other studies 23,42,43

Table 1 .
Baseline Characteristics of Non-Hispanic Black and Non-Hispanic White Patients With Cancer and COVID-19 Diagnosis (continued) c Refers to patients reported to have obesity or to have a body mass index (calculated as weight in kilograms divided by height in meters squared) greater than 30.dPercentages could sum to greater than 100% because categories are not mutually exclusive.eModifiedKlabunde index is used.Klabunde is a modification of the Charlson Comorbidity Index.fRefers to within 3 months before COVID-19 diagnosis.gFivepercent of patients (n = 188) were receiving radiation treatment (52 Black patients [5%] and 136 White patients [6%]).JAMA Network Open | OncologyRacial Disparities in COVID-19 Outcomes Among Black and White Patients With Cancer JAMA Network Open.2022;5(3):e224304.doi:10.1001/jamanetworkopen.2022.4304(Reprinted) March 28, 2022 6/14 Downloaded From: https://jamanetwork.com/ on 08/30/2022

Table 2 .
12 this cohort study, we examined a range of clinically meaningful characteristics and outcomes in Black patients with cancer and COVID-19 in the US.We used a novel ordinal outcome of COVID-19 severity to capture the full spectrum of COVID-19 complications that may vary across different racial and ethnic groups and also explored the interactions between comorbidities and cancer to understand their synergistic effect on COVID-19 outcomes in different racial groups.34tes of Baseline Severity, Outcomes, Clinical Complications, and Interventions Received After COVID-19 DiagnosisCompared with White patients, Black patients had worse COVID-19 presentations and experienced significantly higher COVID-19 severity; this difference was consistent across different analysis methods.These findings are complementary to a recent EHR report that showed African American patients with cancer and COVID-19 had higher rates of hospitalization and death.12Our Abbreviation: ICU, intensive care unit.aRefers to number of patients with nonmissing data.b Mild denotes no hospitalization indicated, moderate denotes hospitalization indicated whether or not it occurred, and severe denotes ICU admission indicated, whether or not it occurred.c Included in primary ordinal COVID-19 severity outcome.d Refers to secondary outcome.e A full description of these complications is provided in eTable 4 in Supplement 1 and do not include "other."JAMA Network Open | Oncology Racial Disparities in COVID-19 Outcomes Among Black and White Patients With Cancer JAMA Network Open.2022;5(3):e224304.doi:10.1001/jamanetworkopen.2022.4304(Reprinted) March 28, 2022 7/14 Downloaded From: https://jamanetwork.com/ on 08/30/2022

Table 3 .
Unweighted and Weighted Analyses of Association of Racial Disparities With COVID-19 Severity (Primary Outcome) and 30-Day All-Cause Mortality (Secondary Outcome) Risk Factors With COVID-19 Severity and 30-Day All-Cause Mortality Stratified by Race and EthnicityData are shown for 3506 patients.Odds ratios (ORs) greater than 1 indicate higher COVID-19 severity or higher odds of 30-day all-cause mortality.ORs were adjusted for age (linear and quadratic terms), sex, region of patient residence, smoking status, obesity, cardiovascular and pulmonary comorbidities, kidney disease, diabetes, type of malignant neoplasm, Eastern Cooperative Oncology Group performance status, cancer status, timing and modality of anticancer therapy, and month of COVID-19 diagnosis.The contrast for cancer status is active and progressing cancer status vs remission or no evidence of disease; there was no evidence of effect modification for other categories (ie, active and responding, active and stable, unknown).

30/2022 limitation
current study has several strengths, including being, to our knowledge, the largest cohort to date of Black patients with cancer and COVID-19.Furthermore, we present data representative of Black patients across diverse age groups, cancer types, cancer status, geographical census regions, and academic and community centers, with longer-term follow-up beyond 30 days.Despite these strengths, there remain limitations associated with incomplete documentation of race and ethnicity, a commonly encountered problem in EHRs, and inherent lack of granularity in the Centers for Disease Control and Prevention classification schema chosen for the registry.We also did not examine the Racial Disparities in COVID-19 Outcomes Among Black and White Patients With Cancer of both the survey and underlying EHR data that is available to the respondents.Although a number of CCC19 participating sites are also National Cancer Institute Community Oncology Research Program sites, there is potentially incomplete capture of subpopulations of patients (eg, rural populations) who may not seek care at the mostly urban health care centers that compose the CCC19.In addition, CCC19 does not collect detailed data on SDOH, primarily because they are often not recorded within EHRs.46Because insurance (a surrogate of health care access) is a contributor of disparity, it was considered as a nonallowable covariate and, thus, was not included in the analysis.A sensitivity analysis including insurance showed a marginal decrease in the OR for COVID-19 severity outcome and a negligible change in OR or relative risk for mortality (eTable 6 in Supplement 1).Thus, much if not all the disparity in COVID-19 severity between Black and White patients with cancer could be explained by measurable and sometimes modifiable factors.ConclusionsOur study found that Black patients with cancer and COVID-19 have similar cancer status but worse preexisting comorbidities, severity of COVID-19 at presentation, and outcomes compared with White patients with cancer and COVID-19.Understanding and addressing the cumulative and synergistic association of racial inequities (eg, preexisting comorbidities, SDOH, and inadequate access to quality health care and cutting-edge research) on clinical outcomes is pivotal.This is a call for action to eradicate root causes of racial inequities, within the causal framework of structural racism, to reduce the disproportionate burden of diseases, such as COVID-19 and cancer, among Black patients and, possibly, other minority racial and ethnic groups.44.Flores LE, Frontera WR, Andrasik MP, et al.Assessment of the inclusion of racial/ethnic minority, female, and older individuals in vaccine clinical trials.JAMA Netw Open.2021;4(2):e2037640-e2037640.doi:10.1001/jamanetworkopen.2020.3764045.Painter EMUE, Ussery EN, Patel A, et al.Demographic characteristics of persons vaccinated during the first month of the COVID-19 vaccination program-United States, December 14, 2020-January 14, 2021.MMWR Morb Mortal Wkly Rep. 2021;70(5):174-177.doi:10.15585/mmwr.mm7005e146.Kim E, Rubinstein SM, Nead KT, Wojcieszynski AP, Gabriel PE, Warner JL.The evolving use of electronic health records (EHR) for research.Semin Radiat Oncol.2019;29(4):354-361.doi:10.1016/j.semradonc.2019.05.010CCC19 Data Collection and Quality Assurance eAppendix 2. Alphabetical List of Participants by Institution that Contributed at Least One Record to the Analysis eAppendix 3. Statistical Analysis Plan eAppendix 4. Statistical Methods eReferences eFigure 1. Flow Diagram eFigure 2. COVID-19 Severity by Age eFigure 3. Differences in Outcome Log Odds Between Univariable Logistic Regression Models for All Possible Cutpoints of the Ordinal COVID-19 Severity Outcome, Relative to the Ն1 Versus 0 Comparison eFigure 4. Distribution and Summary Statistics for Propensity Scores eFigure 5. Absolute Standardized Mean Differences for Demographic and Clinical Characteristics at COVID-19 Diagnosis Between Non-Hispanic Black and Non-Hispanic White Patients eTable 1. Metrics for Data Quality eTable 2. Patients on Multimodal Anticancer Therapy eTable 3. Type of Malignancy eTable 4. Laboratory Measurements Among Hospitalized Patients eTable 5. Rates of Cardiovascular, Pulmonary, and Gastrointestinal Complications eTable 6. Inverse Probability Treatment Weighting (IPTW) With Insurance Added association of COVID-19 with cancer among Hispanic Black patients.Gender was collected in the survey, including a third nonbinary option.For the purposes of this analysis, we mapped gender to sex: woman to female, man to male, and nonbinary to unknown or missing.We acknowledge that this may lead to a mismatch between biological sex and gender identity, which is an inherent JAMA Network Open | Oncology JAMA Network Open.2022;5(3):e224304.doi:10.1001/jamanetworkopen.2022.4304(Reprinted) March 28, 2022 9/14 Downloaded From: https://jamanetwork.com/ on 08/