Rapid real‐world data analysis of patients with cancer, with and without COVID‐19, across distinct health systems

Abstract Background The understanding of the impact of COVID‐19 in patients with cancer is evolving, with need for rapid analysis. Aims This study aims to compare the clinical and demographic characteristics of patients with cancer (with and without COVID‐19) and characterize the clinical outcomes of patients with COVID‐19 and cancer. Methods and Results Real‐world data (RWD) from two health systems were used to identify 146 702 adults diagnosed with cancer between 2015 and 2020; 1267 COVID‐19 cases were identified between February 1 and July 30, 2020. Demographic, clinical, and socioeconomic characteristics were extracted. Incidence of all‐cause mortality, hospitalizations, and invasive respiratory support was assessed between February 1 and August 14, 2020. Among patients with cancer, patients with COVID‐19 were more likely to be Non‐Hispanic black (NHB), have active cancer, have comorbidities, and/or live in zip codes with median household income <$30 000. Patients with COVID‐19 living in lower‐income areas and NHB patients were at greatest risk for hospitalization from pneumonia, fluid and electrolyte disorders, cough, respiratory failure, and acute renal failure and were more likely to receive hydroxychloroquine. All‐cause mortality, hospital admission, and invasive respiratory support were more frequent among patients with cancer and COVID‐19. Male sex, increasing age, living in zip codes with median household income <$30 000, history of pulmonary circulation disorders, and recent treatment with immune checkpoint inhibitors or chemotherapy were associated with greater odds of all‐cause mortality in multivariable logistic regression models. Conclusion RWD can be rapidly leveraged to understand urgent healthcare challenges. Patients with cancer are more vulnerable to COVID‐19 effects, especially in the setting of active cancer and comorbidities, with additional risk observed in NHB patients and those living in zip codes with median household income <$30 000.

Since the first cases of a pneumonia of unknown etiology in Wuhan, China, were reported in late 2019, more than 9 million confirmed cases of the novel coronavirus identified as severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) and over 231 000 related deaths have occurred in the United States. 1 Questions remain about the clinical epidemiology of coronavirus disease , including the characteristics of infected populations and the factors that influence susceptibility and disease severity or mortality risk, including the need for intensive care unit (ICU) care or mechanical ventilation.
In the United States, various reports suggest that COVID-19 incidence and related outcomes vary by race, ethnicity, and socioeconomic status. [2][3][4][5][6] Higher incidence of COVID-19 infection has been reported among black and Latino Americans 2,3,6 and in counties with either more diverse populations or a higher proportion of adults with less education than a high school diploma. 6 Higher mortality has also been reported in counties in which black Americans comprise a larger proportion of the population or counties with larger percentages of residents living below the poverty level, on Medicaid, or living with a disability. 6,7 The constellation of factors associated with increased rates and/or severity of COVID-19 infection in under-represented minority populations reinforce the role that social determinants of health play in health outcomes.
Patients with a history of cancer and those undergoing active treatment for malignancies who contract COVID-19 are also susceptible to poor outcomes. Selected studies conducted in New York and Northern California have estimated that cancer was present in a minority (5%-6%) of hospitalized COVID-19 patients [8][9][10] ; yet research indicates that cancer in patients diagnosed with COVID-19 is associated with increased risk of severe events (ICU admission, mechanical ventilation, death), with greater risk among patients with metastatic disease, recent treatment (past month) with chemotherapy, immune checkpoint inhibitor therapy (90 days), or surgery. [10][11][12][13][14][15][16][17] Additionally, quicker progression to severe events was found among patients with cancer than among those without cancer. 14,18,19 In the United States, data characterizing COVID-19 in patients with cancer are still quite limited and generally obtained within single health systems or from voluntary surveillance registries or surveys. 10,12,13,[20][21][22][23][24][25] The impact of race, health status, and socioeconomic factors with COVID-19-related incidence or outcomes in patients with cancer is not well described. To expand the available evidence, the current study utilizes the ability for rapid COVID-19 case identification through access to integrated, detailed, longitudinal clinical data from two large Midwestern health systems to examine the differential risk for infection and severe outcomes among adults with a history of cancer and those undergoing active cancer treatment.

| Outcome measures
Patients were followed from February 1, 2020, (the index date) to August 14, 2020. The primary outcome of this study was all-cause mortality. Dates of death were obtained from the integrated health system data directly if available or from linkage to hospital tumor registries, digitized obituaries, the Social Security Death Index (SDI), and chart abstraction conducted by Certified Tumor Registrars (CTRs).
Hospital tumor registries were accessed through MetriQ and CNExT.

| Covariates
For patients that were hospitalized, inpatient treatment with the following medications was assessed: vasopressors; azithromycin or other antibiotics; hydroxychloroquine, remdesivir, or other antiviral drugs; famotidine; and tocilizumab. We evaluated the following acute COVID-19 complications: respiratory distress or failure; sepsis; renal failure; kidney injury; liver injury; arrhythmia; conduction disorders; cardiac arrest; cardiomyopathy, myocarditis, pericarditis; coagulopathy; chronic pulmonary disease; and cytokine release syndrome.

| Statistical analysis
Demographic, clinical, and socioeconomic characteristics of patients with active cancer or history of cancer and COVID-19 were compared to the characteristics of patients without recorded COVID-19. We used multivariable logistic regression to estimate odds ratios (OR) and evaluate the association between clinical factors and allcause mortality. Models were adjusted for sex, age, race/ethnicity, median household income, CCI, recent surgery, hypertension, coagulopathy, pulmonary circulation disorders, obesity, cancer type, cancer status, and cancer treatment type. All analyses were performed in R programming language, version 3.3.2 (R Foundation for Statistical Computing, Vienna, Austria). A P-value of <.05 was considered statistically significant.

| RESULTS
The study population consisted of 147 969 patients with cancer (Table 1). Compared to patients without COVID-19 (n = 146 702), a greater proportion of patients with COVID-19 (n = 1267) were NHB, lived in a zip code with median annual household income of < $30 000, had a CCI of 2 or greater, had diabetes, or had active cancer.

| Patients with cancer and COVID-19
Among patients with cancer and COVID-19, patients with active cancer were more likely than patients with a history of cancer to be male (48% vs 35%) or have a CCI of 1 or greater (79% vs 51%) ( Table 2). Patients living in areas with median household income below $30 000 were more likely to be diagnosed with COVID-19 by ICD codes alone (62% vs 49%). Among patients with COVID-19, the incidence of death, hospital admission, and use of invasive respiratory support was greatest among patients with active cancer, or those residing in zip codes with median household income below $30 000; NHB patients were more likely to receive invasive respiratory support than the other racial/ethnic groups ( Table 2). Females were less likely to have active cancer or comorbidities or to use antihypertensive medications. Females were also younger than males, and at lower risk of mortality, hospital admission, and of requiring invasive respiratory support (Table S1).
Admission diagnoses, treatments, and clinical outcomes for patients with COVID-19 who were hospitalized are presented in Table S2. Within this cohort, the distributions of the 20 most common hospital admission diagnoses were similar for patients with active cancer and history of cancer. Breathing abnormalities were more common for males than females. Patients in zip codes with median household income below $30 000 were twice as likely as patients in zip codes with median household income above $30 000 to be hospitalized for pneumonia, fluid balance disorders, cough, respiratory failure, or acute renal failure. Additionally, NHB patients were more likely than other groups to be admitted for breathing abnormalities, pneumonia, fluid balance disorders, cough, acute renal failure or chronic kidney disease, and fever. NHB patients were more likely than other groups to be treated with hydroxychloroquine alone or in combination with azithromycin. In terms of acute COVID-19 complications, patients in zip codes with median household income below $30 000 and males were more likely to experience respiratory distress or failure, as well as kidney and liver injury (Figure 1). Patients with a history of cancer and NHB patients were more likely to be released to self-care, while patients in zip codes with median household income below $30 000 were more likely to expire while hospitalized (Table S3).
In adjusted multivariable analyses, we found that male sex, increasing age, living in a zip code with median household income below $30 000, history of pulmonary circulatory disorder, and recent treatment with immune checkpoint inhibitors or chemotherapy were all significantly associated with greater odds of all-cause mortality (Table 3). While NHB and Hispanic/Latino race/ethnicity categories were not significantly associated with mortality compared to NHW, we found that a category that combined Asian/Native Hawaiian/ Pacific Islander, other or unknown race/ethnicities, was associated with a lower all-cause mortality.

| DISCUSSION
Cancer treatment increases the potential for adverse health outcomes, including susceptibility to infections. As the COVID-19 T A B L E 1 Demographic, clinical, and socioeconomic characteristics of the study population of patients with cancer, with and without COVID-19   higher burden of comorbidities and frailty than the general population.
Immunosuppressive cancer therapy may impair patients' ability to mount an effective antiviral immune response or mitigate deleterious multisystem effects. Consistent with this hypothesis, COVID-19-infected patients who have active cancer or recent cancer treatment have been found to have an increased risk of mortality, 24,29,30 and in our study, recent chemotherapy and immune checkpoint inhibitor therapy were both independently associated with mortality. While some studies have reported an association between immune checkpoint inhibitor therapy and COVID-19 severity or mortality, 11,12 others have not found evidence of such an association. 31 Understanding the contribution of potential social and biological determinants of COVID-19 severity requires study beyond the scope of this retrospective analysis. While we found that race/ethnicity were not significantly associated with higher all-cause mortality in multivariable models, our findings highlight that race and socioeconomic factors identify a cancer population that is vulnerable to greater infection rates and morbidity. Addressing and supporting these vulnerabilities will be critical to minimize mortality risk from COVID-19 in cancer patients. writing-review and editing.

CONFLICT OF INTEREST
The authors have no conflicts of interest to disclose.
T A B L E 3 Odds ratios (OR) and 95% confidence intervals for allcause mortality among patients with cancer and COVID-19

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available from the corresponding author upon reasonable request.

ETHICS STATEMENT
This study was performed through a research collaboration agreement

SUPPORTING INFORMATION
Additional supporting information may be found online in the Supporting Information section at the end of this article.