key: cord-0998969-f4y3hu4v
authors: Fillmore, N. R.; La, J.; Szalat, R. E.; Tuck, D. P.; Nguyen, V.; Yildirim, C.; Do, N. V.; Brophy, M. T.; Munshi, N. C.
title: Prevalence and outcome of Covid-19 infection in cancer patients: a national VA study
date: 2020-08-24
journal: nan
DOI: 10.1101/2020.08.21.20177923
sha: c78200a44d5fb64d327720a62ec3285ad870da9c
doc_id: 998969
cord_uid: f4y3hu4v

Background: Emerging data suggest variability in susceptibility and outcome to Covid-19 infection. Identifying the risk-factors associated with infection and outcomes in cancer patients is necessary to develop healthcare recommendations. Methods: We analyzed electronic health records of the US National Veterans Administration healthcare system and assessed the prevalence of Covid-19 infection in cancer patients. We evaluated the proportion of cancer patients tested for Covid-19 and their confirmed positivity, with clinical characteristics, and outcome, and stratified by demographics, comorbidities, cancer treatment and cancer type. Results: Of 22914 cancer patients tested for Covid-19, 1794 (7.8%) were positive. The prevalence of Covid-19 was similar across all ages. Higher prevalence was observed in African-American (AA) (15%) compared to white (5.5%; P<.001), in Hispanic vs non-Hispanic population and in patients with hematologic malignancy compared to those with solid tumors (10.9% vs 7.7%; P<.001). Conversely, prevalence was lower in current smoker patients, patients with other co-morbidities and having recently received cancer therapy (<6 months). The Covid-19 attributable mortality was 10.9%. Highest mortality rates were observed in older patients, those with renal dysfunction, higher Charlson co-morbidity score and with certain cancer types. Recent (<6 months) or past treatment did not influence mortality. Importantly, AA patients had 3.5-fold higher Covid-19 attributable hospitalization, however had similar mortality rate as white patients. Conclusion: Pre-existence of cancer affects both susceptibility to Covid-19 infection and eventual outcome. The overall Covid-19 attributable mortality in cancer patients is affected by age, co-morbidity and specific cancer types, however, race or recent treatment including immunotherapy does not impact outcome.

The Covid-19 infection first reported in China, in December 2019 1 , has now spread worldwide affecting all demographics and regions. The emerging data suggests variability in susceptibility to the infection and ultimately outcome. A number of patient related factors, socio economic conditions, racial and ethnic differences and several comorbidities including obesity, diabetes and cardiovascular diseases have been associated with higher susceptibility and/or risk of mortality. [2] [3] [4] [5] The relatively higher transmission rate and associated greater risk of adverse outcome has highlighted the need to understand the epidemiologic characteristics of Covid-19 prevalence, and the risk factors associated with poor outcome and death, in order to establish the best possible public health policies. Cancer patients are considered to be at a higher risk of infections. This risk varies with functional status of the patient, the cancer type, and/or treatment modalities utilized. 6, 7 Thus, along with reducing exposure to the virus, other prophylactic as well as cancer-related risk factors may need to be addressed to decrease susceptibility to infection or to mitigate related complications in cancer patients. Small epidemiologic studies mainly from China and USA have also reported increased rates of death in cancer patients related to Covid-19. 8, 9 These observations have informed some changes and re-organization of cancer care worldwide 10 but larger studies are needed to understand the comprehensive cancer-related issues with Covid-19 infection. Here we investigated the prevalence and outcome of Covid-19 infection among cancer patients in a large cohort of patients from the nationwide Veterans Affairs (VA) healthcare system. Our report, besides suggesting that cancer patients are more vulnerable to Covid-19 infection, 11, 12 also highlights the prevalence of Covid-19 and outcome of the disease based on racial characteristics, comorbidities, type of cancer and related treatment in a cohort of 22914 cancer patients.

This analysis was conducted using data from the VA Corporate Data Warehouse (CDW), which centralizes EHR data for patients seen at VA facilities nationwide. The study population is defined as veterans with cancer who were tested for Covid-19 at the VA. Cancer patients were identified as patients with at least one occurrence of an International Classification of Diseases (ICD) code for cancer between January 1, 2010 and May 4, 2020. 26 Non-melanoma skin cancer and benign for use under a CC0 license. This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted August 24, 2020. . https://doi.org/10.1101/2020.08. 21 were considered to determine the type of therapy received. As regards to outcomes, hospitalization was defined as any inpatient visit after the patient's first Covid-19 test, and ICU admission was defined as a subset of hospitalizations where the specialty ward is either "Surgical ICU" or "Medical ICU". Respiratory support was determined based on the presence of a current procedural terminology (CPT) or ICD procedure code for intubation (CPT 94002, 94003, 94004, 94005, ICD-10 Z99.1, Z99.11, Z99.12) or mechanical ventilation (CPT 31500, ICD-10 J95.851) after the patient's first Covid-19 test. Missing data existed for race, ethnicity, and smoking status and was coded as a separate level in all analyses. difference of these two proportions. We used chi-squared tests to assess differences in proportion. Odds ratio (OR) as to Covid-19 positive status by cancer type was assessed using univariate and multivariate logistic regression, adjusting for race, ethnicity, and smoking status in the multivariate models.

We identified 22914 patients with history of cancer who were tested for Covid-19 infection on or before May 4, 2020, of whom 1794 patients (7.8%) reported positive ( Figure 1A) . The prevalence of Covid-19 among those tested was similar across all ages (<50 years to ≥80years; P=.158), although older patients with cancer were tested for Covid-19 more frequently (eTable 1 in Supplement 1). Importantly, there was a significant difference in prevalence of infection across race and ethnicity, with 14.95% of African Americans (AA) but only 5.49% of white patients testing positive for Covid-19 (P<.001), and 10.87% of Hispanic/Latino patients vs 7.71% of non-Hispanic/Latino patients testing positive (P<.001). We also observed a significantly lower prevalence in cancer patients who were current smokers compared to former smokers or those who never smoked (5.26% vs 9.49%; P<.001). Compared to the overall sample (7.8% positive), there was a small but statistically significant negative association between susceptibility to Covid-19 and concomitant congestive heart failure (6.68% positive; P<0.001), peripheral vascular disease (6.51%; P<0.001), chronic obstructive pulmonary disease (6.27%; P<.001), and moderate/severe liver disease (5.15%; P=.023). Higher body mass index (BMI) was associated with increased prevalence of infection (P<.001).During the period of observation, Covid-19 was much more frequent in the North Atlantic (14.77%) than other regions (6.25% Continental, 7.85% Midwest, 2.37% Pacific, 3.21% Southeast; P<.001). Differences in susceptibility across race persisted within each region (eTable 6).

We further investigated cancer types and prevalence of Covid-19 ( Figure 1B) This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. 

We next evaluated the impact of Covid-19 infection on outcome in cancer patients and calculated the difference between frequency in positive vs negative as the Covid-19 attributable contribution.

Overall, compared to Covid-19 negative cancer patients, Covid-19 positive cancer patients have higher frequency of hospitalizations (31.5% versus 43.8% respectively; an excess of 12.3%

Covid-19 attributable), ICU admissions (7.8% vs 19.7% respectively; 11.9% Covid-19 attributable), respiratory support (1.3% vs 7.9% respectively; 6.6% Covid-19 attributable). Covid-19 attributable death was 10.9% with 14% death in Covid-19 positive compared to 3.1% in Covid-19 negative patients ( Figure 2 ).

To further assess the impact of patient and disease-related features on Covid-19 attributable outcomes, we evaluated each of the demographic and comorbidity strata described above ( Figure 3A and Figure 4A ). In general outcomes occur more frequently among Covid-19 positive cancer patients, however the difference attributable to Covid-19 infection varies widely across strata. Covid-19 attributable mortality is strongly associated with age, ranging from 0.23% among for use under a CC0 license.

This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted August 24, 2020. . patients <50 years old to 20.51% among patients ≥80 years old (P<.001). Presence of other comorbidities is also associated with increased Covid-19 attributable death, ranging from 3.07% among patients with Charlson score 0, to 14.96% among patients with Charlson score ≥5 (P<.001). Covid-19 attributable ICU admissions were more common in obese patients (14.34%) than in patients with normal BMI (5.87%, P<.001), as is need for respiratory support (8.94% vs 4.78%, P=.014); however mortality was not higher (10.45% in obese, 13.47% in normal; P=.21).

Interestingly, Covid-19 attributable mortality was lower in current smokers (6.45%) compared to former/never smokers (11.99%; P=.002). Similarly, we do not observe a significant difference in Covid-19 attributable outcomes for Hispanic/Latino compared to Non-Hispanic/Latino (eTable 4).

Outcomes were variable based on type of cancer ( Figure 3B and Figure 4B ). Covid-19 attributable deaths were observed in most but not all cancer types, with the highest attributable mortality in acute leukemia (25.64%), male genital (20.79%), and thyroid cancer (19.88%). Covid-19 attributable mortality is similar in patients recently treated for cancer (≤6 months ago, 14.17%) compared to those treated >6 months ago (13.06%) or never treated with systemic therapy (10.09%; P=.250). There is a suggestive relationship between type of treatment and outcome attributable to Covid-19 infection. In patients receiving ICI within the last 6 months, Covid-19 attributable mortality (7.10%) is less than half that observed in patients receiving chemotherapy (14.02%), hormone therapy (16.21%), or targeted therapy (14.13%), though significance is not reached because of fewer observed deaths.

We investigated a large cohort of cancer patients evaluated for Covid-19 at the VA Healthcare System from across the United States in order to identify risk factors for Covid-19 susceptibility for use under a CC0 license.

This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. Perhaps the most striking observation in our data is the significant racial and ethnic disparity in prevalence of Covid-19 infection. African-American (15%) and Hispanic (10.9%) cancer patients had significantly higher rates of Covid-19 infection in comparison to white cancer patients (5.5%), and also had higher rates of hospitalization. The biological and/or social basis for this observation, also recently reported in the general population, 17-19 remains unclear. Overall, the observation of this disparity across all the regions of the country (eTable 6) rules out the possible role of regional conditions in explaining this difference. Covid-19 was more frequent among patients with prostate cancer which is also more frequent in African-American patients. 20 Although after adjusting for race, ethnicity, and smoking status, this difference is no longer observed. The possible explanations requiring further investigation include socio-economic factors or genetic predisposition. Importantly, although AA cancer patients were more frequently admitted to the hospital, with equal access to care in the VA healthcare system, similar mortality rate was observed between AA and white cancer patients. Similar trend was also observed in Hispanic/Latino population. These results suggest distinct factors modulating Covid-19 susceptibility and the related-outcome.

The significantly reduced prevalence of Covid-19 infection in cancer patients with current smoking history, compared to those who have quit smoking or never smoked, is intriguing. This observation is also confirmed by observed reduced frequency of Covid-19 infection among patients with lung (4.31%, P<0.001), SCCHN (5.52%, P<0.001), and urothelial (6.46, P=0.013) cancers, all malignancies associated with smoking. These differences might also be related to race or ethnicity (eTables 3 and 4), or to socio-economics factors. An earlier preliminary report has also suggested association of active smoking with significantly lower rates of Covid-19 infections. 21 for use under a CC0 license.

This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted August 24, 2020. . This negative association requires further investigation. The role of nicotine, local epithelial cell changes or inflammatory environment may play a role. 22, 23 We did not observe significant impact of cardiopulmonary and other comorbidities as well as age of patients on rate of infection ( Figure   1 ). The lack of significant impact of the well-described influence of comorbidities on Covid-19 in our patients may suggest that pre-existence of cancer may outweigh the other comorbidities in regard to susceptibility to Covid-19. Moreover, it is possible that sick patients and patients under active cancer treatment may have been more cautious and thus less exposed to Covid-19, and this may affect the vulnerability. While we are unable to directly quantify severity of cancer illness in our dataset, we included two proxies for severity of disease that are available: (1) whether or not each patient was recently treated for cancer (≤6 months ago), treated >6 months ago, or never treated with systemic therapy, which serves to some extent as a proxy for severity; and (2) the Charlson score which reflects severity of overall health condition prior to Covid-19 diagnosis. Our analysis does reveal significantly lower rate of infection in patients treated ≤6 months ago, though no significant difference in Covid-19 attributable mortality or other outcomes, and higher rates of infection in patients with higher Charlson score was observed.

Outcome of cancer patients infected by Covid-19 was characterized by higher rates of mortality.

Although the frequency of Covid-19 positivity was lower in patients receiving cancer related therapy within 6 months of the Covid-19 infection, the overall mortality was not significantly different in comparison to no treatment or treatment prior to 6 months groups. These mortality rates are higher than the mortality rates reported in the global population 8,24 and confirm higher vulnerability of cancer patients to Covid-19 infection. Therapies including conventional chemotherapy, targeted therapies with small molecules or monoclonal antibodies were associated with higher mortality rates in this study. Immune checkpoint inhibitor treatment was associated with a lower rate of infection however, only a small number of patients are in this group suggesting both caution in interpreting this data and also a need for further focused investigation in patients receiving ICI therapy and with Covid-19 positivity. The lower mortality rate in patients receiving this treatment confirms the recent report about the impact of this therapy on Covid-19 outcome. 25 Although frequency of Covid-19 infection was not significantly different, higher rates of mortality were observed in elderly patients (19.5% in patients > 70 y.o.), patients with low Charlson score, underweight patients and patients with comorbidities, mainly cardio-vascular and renal disease (Figure 3) . While hematological malignancies were associated with higher rates of infection, similar Covid-19 attributable mortality was observed in comparison to solid malignancies. Fatality rate related to each cancer type was variable with breast, male genital, for use under a CC0 license. This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted August 24, 2020. . https://doi.org/10.1101/2020.08.21.20177923 doi: medRxiv preprint acute leukemia and thyroid cancer being associated with highest rate of Covid-19 related mortality ( Figure 3) .

Our study has several limitations. First, the veteran population is primarily male and hence the study represents various trends and associations whose interpretations are restricted to male.

Second, patients tested outside the VA system would not be included in this analysis. However, this number is likely to be very small as most patients who get cancer care in the VA do return for their healthcare needs to the VA. Third, it is possible that analyses are confounded by indication for testing. However, we believe that our comparison of those who tested positive to those who tested negative remains relevant, especially because Covid-19 testing criteria at VA hospitals was informed by centralized guidance from the VA Central Office, increasing consistency of testing criteria nationally. 27

In conclusion, the presence of cancer changes the susceptibility to Covid-19 infection and affects overall outcome. The overall disease behavior is modulated by patient-related as well as cancerrelated factors which needs to be considered in development of Covid-19 preventative strategies as well as modulation of cancer therapies to optimize the patient care. Importantly, having equal access to care is an important component to improving overall outcome.

Plan: Incident-specific Annex to the VHA High Consequence Infection (HCI) Base Plan. Version 16 March 23, 2020 https://www.va.gov/opa/docs/VHA_COVID_19_03232020_vF_1.pdf. This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted August 24, 2020. .

Percent of Covid-19 positive patients among cancer patients tested for Covid-19, stratified by demographics, comorbidities, cancer type, and cancer therapy are represented. The dashed line indicates the overall percent positive (7.8%). The two rows at top show the number of Covid-19 positive patients, and the total number of cancer patients tested. In addition to the percent positive in each group, a 95% confidence interval and P value are shown. (* = P < .05; ** = P < .01; *** = P < .001).

for use under a CC0 license. This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted August 24, 2020. .

Percent of patients experiencing hospitalization, ICU visits, respiratory support, and death in Covid-19 positive (blue) and negative (red) cancer patients are represented. The Covid-19 attributable risk of experiencing each outcome, i.e., the difference of the percent experiencing each outcome in Covid-19 positive compared to negative patients is also shown (darker red), along with 95% confidence intervals.

for use under a CC0 license. This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted August 24, 2020. .

Covid-19 attributable mortality defined as the difference in the percent mortality in Covid-19 positive compared to negative patients, is shown, along with 95% confidence intervals. The dashed line shows the Covid-19 attributable mortality in the overall cohort (14.4%), and the dotted line marks 0, the point where there is no Covid-19 attributable mortality. The four rows at top show the number of Covid-19 positive patients who died, the total number of Covid-19 positive patients, the number of Covid-19 negative patients who died, and the total number of Covid-19 negative patients.

for use under a CC0 license. This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted August 24, 2020. . This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted August 24, 2020. . https://doi.org/10.1101/2020.08.21.20177923 doi: medRxiv preprint

The geographical region of the VA hospital each patient was tested in was extracted from structured data, with regions defined using the following mapping from the Corporate Data Warehouse:

• for use under a CC0 license.

This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted August 24, 2020. . This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted August 24, 2020. This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted August 24, 2020. . This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted August 24, 2020. This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted August 24, 2020. This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted August 24, 2020. . This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted August 24, 2020. This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted August 24, 2020. . https://doi.org/10.1101/2020.08.21.20177923 doi: medRxiv preprint

A Novel Coronavirus from Patients with Pneumonia in China

Is diabetes mellitus associated with mortality and severity of COVID-19? A meta-analysis

Clinical Characteristics of Covid-19 in

Presenting Characteristics, Comorbidities, and Outcomes Among 5700 Patients Hospitalized With COVID-19 in the

Clinical course and risk factors for mortality of adult inpatients with COVID-19 in Wuhan, China: a retrospective cohort study

Antimicrobial prophylaxis and outpatient management of fever and neutropenia in adults treated for malignancy: American Society of Clinical Oncology clinical practice guideline

Incidence of Infection-related mortality in cancer patients: Trend and survival analysis

Clinical Characteristics of Coronavirus Disease 2019 in China

Cancer patients in SARS-CoV-2 infection: a nationwide analysis in China

Case Fatality Rate of Cancer Patients with COVID-19

COVID-19 More Frequent, Severe in Cancer Patients. Cancer Discov

comorbidity: An R package for computing comorbidity scores

Validating the use of veterans affairs tobacco health factors for assessing change in smoking status: accuracy, availability, and approach

Patients In A Large Health Care System In California. Health Aff (Millwood)

Racial Health Disparities and Covid-19 -Caution and Context

Hospitalization and Mortality among Black Patients and White Patients with Covid-19

Prostate cancer in men of African origin

Current Smoking is Not Associated with COVID-19

COVID-19 and Nicotine as a Mediator of ACE-2

Is Nicotine the Hidden Link?

World Health Organization. Coronavirus disease (COVID-2019) situation reports

blockade on severity of COVID-19 in patients with lung cancers. Cancer Discov

Cancer incidence among patients of the U.S. Solid and Heme

Demographics of Cancer Patients Tested for Covid-19 across Regions Region Overall, n Black, n (% of Overall) White, n (% of Overall)

This work was supported by the VA Office of Research and Development, Cooperative Studies