Racial and Ethnic Disparities in Coexisting Cancer and COVID-19 Mortality

INTRODUCTION

Patients with cancer during the COVID-19 pandemic faced impaired access to care and increased risk of severe infection, which may have increased risk of death from co-existing cancer and COVID-19.^1,2 There are well-documented racial disparities in pre-pandemic cancer outcomes and in risk of COVID-19 infection, but data are more limited assessing outcomes disparities for patients with cancer during the pandemic, particularly by underlying cause of death.^2–5

METHODS

Multiple causes of death data were obtained from the National Center for Health Statistics (NCHS) from 2020 to 2021 (with 2021 data considered provisional). NCHS data include all recorded deaths among the entire US population, where causes of death and race and ethnicity are based on ICD-10 codes and death certificate information. Mortality data were obtained for individuals whose underlying causes of death included cancer or both cancer and COVID-19. There were three primary outcomes: (1) non-COVID-19-related cancer mortality, where the underlying cause of death was cancer excluding events where COVID-19 was a secondary cause of death; (2) COVID-19-associated cancer mortality, where the underlying cause of death was cancer with COVID-19 as a secondary cause of death; and (3) cancer-associated COVID-19 mortality, where the underlying cause of death was COVID-19 with cancer as a secondary cause of death. Mortality event data, consisting of the number of events within a defined population and the number of individuals in the given population, were obtained for every combination of age group (at the time of death), sex, metropolitan residence status, year, and race and ethnicity. Missing mortality data (suppressed when number of deaths is < 10) was imputed using multiple imputation.⁶ Binomial logistic regression was utilized to estimate temporal trends and racial and ethnic disparities in mortality rates adjusted for age, sex, and metropolitan residence status. Inclusion of other covariates was not feasible due to excessive data suppression with further stratification. Due to the nature of the publicly available aggregate data, the research was exempt from Institutional Review Board oversight.

RESULTS

There were 1,198,840 non-COVID-19-related cancer deaths, 8632 COVID-19-associated cancer deaths, and 29,828 cancer-associated COVID-19 deaths in 2020–2021. Increases in non-COVID-19-related cancer mortality from 2020 to 2021 were observed among Asian (OR 1.05, 95% CI 1.03–1.07) and Hispanic (OR 1.03, 1.02–1.05) populations (Fig. 1A; Table 1). There was higher non-COVID-19-related cancer mortality in the Black relative to non-Hispanic White population (OR 1.08, 1.07–1.09). Cancer-associated COVID-19 mortality was higher in Black (OR 1.44, 1.39–1.49) and Hispanic (OR 1.17, 1.13–1.22) populations when compared to the non-Hispanic White population (Fig. 1B). There were also higher rates of COVID-19-associated cancer mortality in the Black (OR 1.40, 1.32–1.49) population (Fig. 1C). The Asian and more than one race populations had lower cancer and COVID-19 mortality than the non-Hispanic White population.

Figure 1.

Non-COVID-19-related cancer mortality (A), cancer-associated COVID-19 mortality (B), and COVID-19-associated cancer mortality (C) by race and ethnicity. Legend: AIAN American Indian or Alaskan Native, NHOPI Native Hawaiian or Other Pacific Islander. Point estimates are the mortality rates provided by the National Center for Health Statistics, with bars representing 95% confidence intervals. Note that some confidence intervals are poorly visualized due to estimate precision. Age-adjusted rates are given where available (not available for the non-COVID-19-related cancer deaths, which were derived, and for NHOPI COVID-19-associated cancer deaths, which were not calculated by the NCHS given “unreliable” estimates owing to the number of deaths between 10 and 20). The numbers of deaths for each racial and ethnic group (by mortality event type) are as follows: American Indian or Alaska Native (non-COVID-19-related cancer deaths = 6577; COVID-19-associated cancer deaths = 65; cancer-associated COVID-19 deaths = 269), Asian (37,634; 180; 793), Black or African American (141,218; 1243; 4184), Native Hawaiian or Other Pacific Islander (1703; 17; 35), White (1,005,745; 7088; 24,419), more than one race (5963; 39; 128), Hispanic or Latino (90,704; 796; 3328), not Hispanic or Latino (1,138,164; 8123; 26,421).

Table 1.

Cancer and COVID-19 Mortality and Racial and Ethnic Disparities

	Non-COVID-19-related cancer mortality*				Cancer-associated COVID-19 mortality†,‡		COVID-19-associated cancer mortality†,§
	Change from 2020 to 2021		Disparity during pandemic		Disparity during pandemic		Disparity during pandemic
	OR||,¶	p-value	OR||,#	p-value	OR||,#	p-value	OR||,#	p-value
Non-Hispanic White**	1.00 (0.99–1.00)	0.174	Reference		Reference		Reference
American Indian or Alaska Native	1.04 (0.99–1.09)	0.165	0.64 (0.62–0.65)	<.001	0.99 (0.88–1.12)	0.87	0.83 (0.65–1.06)	0.135
Asian	1.05 (1.03–1.07)	<.001	0.59 (0.58–0.59)	<.001	0.56 (0.52–0.60)	<.001	0.42 (0.36–0.49)	<.001
Black	0.99 (0.98–1.00)	0.016	1.08 (1.07–1.09)	<.001	1.44 (1.39–1.49)	<.001	1.40 (1.32–1.49)	<.001
Native Hawaiian or Other Pacific Islander	1.05 (0.96–1.16)	0.318	0.92 (0.87–0.96)	<.001	0.76 (0.54–1.06)	0.103	1.24 (0.77–2.00)	0.368
More than one race	1.03 (0.98–1.09)	0.211	0.42 (0.41–0.43)	<.001	0.38 (0.32–0.45)	<.001	0.39 (0.28–0.53)	<.001
Hispanic or Latino	1.03 (1.02–1.05)	<.001	0.66 (0.65–0.66)	<.001	1.17 (1.13–1.22)	<.001	0.93 (0.86–1.00)	0.044

^*Underlying cause of death is any malignant neoplasm, minus any deaths where COVID-19 was listed as a secondary cause of death

^†Due to data suppression when the number of events for a given population (in our case, age-sex-race-residence-year units) is < 10, we utilized multiple imputation (30 imputations, obtained using the R package Amelia, which allowed us to incorporate a prior and restrict imputed estimates to < 10). The imputations were scaled such that the total number of events for a given race/ethnicity matched the number of events based on aggregate counts, as the distribution of missing values differs by race and ethnicity largely as a function of differing population sizes (e.g., without scaling, if the imputation procedure favors values closer to 10, races and ethnicities with more suppressed values could have an incorrectly high number of imputed events, biasing estimates of disparities)

^‡Underlying cause of death is COVID-19, with any malignant neoplasm as a secondary cause of death

^§Underlying cause of death is any malignant neoplasm, with COVID-19 as a secondary cause of death

^||OR = odds ratio. An odds ratio greater than 1.0 demonstrates a higher mortality rate (i.e., more deaths) from the given cause(s)

^¶Regression model included the following covariates: race, ethnicity, year (dummy variables for 2021), interaction terms between year and race and ethnicity, age, sex, and metropolitan residence. The estimates given are the changes in mortality rates from 2020 to 2021, which were derived using linear combinations of the coefficients for race, ethnicity, time, and time by race and ethnicity interaction terms

^#Regression models included the following covariates: race, ethnicity, year (dummy variable for 2021), age, sex, and metropolitan residence. The estimates given are the regression coefficients corresponding to a given race or ethnicity, with non-Hispanic/White serving as the reference category

^**While most subgroups refer to either race or ethnicity, the reference category in our regression analyses was non-Hispanic White due to the nature of the data, which was stratified separately by both race and ethnicity, and the nature of the conditional regression coefficient estimates

DISCUSSION

Non-COVID-19-related cancer mortality increased during the COVID-19 pandemic for Asian and Hispanic populations. Rising cancer mortality is concerning given recent celebrated downtrends.⁵ Although exact causality of this rise is outside the scope of this study, a likely contributing factor is impaired access to care, such as avoiding medical appointments due to concerns of infection risk or difficulty completing appointments in a healthcare system with shifting priorities, ultimately leading to delayed diagnosis or inadequate treatment.¹ This decreased access may have been differentially worse in communities with unique burdens during the pandemic. There were disparities in co-existing cancer and COVID-19 mortality, most notably for Black and Hispanic populations relative to the White population. These findings likely reflect the disproportionate impact of COVID-19 on minorities, such as higher rates of infection and worse hospitalization outcomes, possibly arising due to vaccine hesitancy and adverse social determinants of health including living and working conditions that are not conducive to social distancing.^3,4

This is the first study to our knowledge to utilize nationwide data to explore racial and ethnic disparities in COVID-19 and cancer mortality during the COVID-19 pandemic. Limitations of this retrospective study include its use of provisional data and the possibility of confounding factors outside the limited covariates included, such as area-level COVID-19 disease burden, vaccination levels, and socioeconomic factors. Additionally, the ecological nature of the study precludes individual-level inferences. Further efforts to minimize the effects of the COVID-19 pandemic on racial and ethnic minorities are warranted.

Funding

Dr. F. Chino is funded in part through the NIH/NCI Support Grant P30 CA008748 for work outside the present manuscript.

Data Availability:

Data are publicly available from the Centers for Disease Control and Prevention WONDER database: https://wonder.cdc.gov/mcd.html.

Declarations:

Conflict of Interest:

The authors declare that they do not have a conflict of interest.