Racial and Ethnic Disparities in Covid-19 Illness Severity Accounting for Concomitant Comorbidities: A Cohort Study

Background Multiple reports have highlighted important racial and ethnic differences in the degree to which Americans may be vulnerable to severe forms of Covid-19 illness. Whether or not racial or ethnic disparities are related to variations in the underlying burden of comorbidities or other predisposing factors remains unclear. Methods We identied patients diagnosed with Covid-19, based on a positive PCR for SARS-CoV-2, from the electronic health record of a large multi-hospital system located in Southern California. We developed an illness severity score, based on the level of care each patient required (not admitted to the hospital; required hospital admission but never required intensive care; required intensive level care but never intubation; and, required intubation during hospitalization) and assessed for associations with clinical and demographic factors for each patient using ordinal logistic regression. Results A total of 571 patients with Covid-19 were identied a majority of whom were male (56%), with a mean age of 55±21 years. There were 81 (14%) patient who identied as African American, and 101 (18%) as Hispanic. A total of 202 (36%) patients required hospitalization without need for intensive care, 43 (8%) required intensive care without intubation, and 64 (11%) required intubation while also receiving intensive care. Of the total sample, African American race (OR 2.33, 95% CI 1.44-3.78, P=0.001) and Hispanic ethnicity (OR 1.97, 95% CI 1.14-3.12, P=0.004) were associated with greater illness severity. Conclusions Racial and ethnic disparities in the severity of Covid-19 illness persist, even when controlling for baseline comorbidities. It remains unclear if these differences are related to variations in physiologic response to SARS-CoV-2, differential timing of presentation or disparities in care. Values are presented as mean (SD) for continuous variables and n (%) for categorical variables. P values are for groupwise comparisons, using One-way ANOVA for normal distributed data, Kruskal-Wallis Rank Sum Test for non-normal distributed data, and Chi-square Test for categorical data by default. Outcome of Covid-19 illness severity score in hospitalized patients was dened as an ordinal variable wherein: 0 = No need for admission, 1 = required admission but never ICU level care, 2 = required ICU level care but never intubate, 3 = required intubation. Odds ratios (OR) are from one multivariable adjusted ordinal logistic regression models adjusted for age, sex, obesity, hypertension, diabetes mellitus, Elixhauser comorbidity score, prior myocardial infarction or heart failure, and prior COPD or asthma.


Abstract
Background Multiple reports have highlighted important racial and ethnic differences in the degree to which Americans may be vulnerable to severe forms of Covid-19 illness. Whether or not racial or ethnic disparities are related to variations in the underlying burden of comorbidities or other predisposing factors remains unclear.
Methods We identi ed patients diagnosed with Covid-19, based on a positive PCR for SARS-CoV-2, from the electronic health record of a large multi-hospital system located in Southern California. We developed an illness severity score, based on the level of care each patient required (not admitted to the hospital; required hospital admission but never required intensive care; required intensive level care but never intubation; and, required intubation during hospitalization) and assessed for associations with clinical and demographic factors for each patient using ordinal logistic regression. Results A total of 571 patients with Covid-19 were identi ed a majority of whom were male (56%), with a mean age of 55±21 years. There were 81 (14%) patient who identi ed as African American, and 101 (18%) as Hispanic. A total of 202 (36%) patients required hospitalization without need for intensive care, 43 (8%) required intensive care without intubation, and 64 (11%) required intubation while also receiving intensive care. Of the total sample, African American race (OR 2.33, 95% CI 1.44-3.78, P=0.001) and Hispanic ethnicity (OR 1.97, 95% CI 1.14-3.12, P=0.004) were associated with greater illness severity.
Conclusions Racial and ethnic disparities in the severity of Covid-19 illness persist, even when controlling for baseline comorbidities. It remains unclear if these differences are related to variations in physiologic response to SARS-CoV-2, differential timing of presentation or disparities in care.

Methods
The enlarging population of Covid-19 patients cared for by our U.S. multi-center health system allowed us to examine the severity of Covid-19 illness by race and ethnicity, controlling for known and presumed risk factors. We curated demographic and clinical data for all Covid-19 con rmed patients from the combined electronic health record of the multi-site Cedar-Sinai Health System, in Los Angeles, California, which includes a catchment area of 1.8 million individuals, 80% of whom identify as a racial or ethnic minority. Laboratory testing for SARS-CoV-2 was performed using reverse transcriptase polymerase chain reaction of extracted RNA from nasopharyngeal swabs. To capture variation in comorbid status, in a way that is not represented by distinct medical history variables alone, we calculated the Elixhauser Comorbidity Index with van Walraven weighting for all patients based on all available data.(4) Our primary outcome was an illness severity score, with higher values assigned to more intensive levels of clinical care based on the following stepwise categories: 0 = clinically deemed to not require admission; 1 = required hospital admission but never required intensive care; 2 = required intensive level care but never intubation; and, 3 = required intubation during hospitalization. The Cedars-Sinai institutional review board approved all protocols. Per convention, we considered a two-sided P < 0.05 as statistically signi cant for association analyses, and P < 0.10 as statistically signi cant for interaction analyses. (5) Patients in this study were included in a prior, preliminary analysis of our cohort. (6) The results of the current study are derived from a now much larger sample, including individuals not previously studied.

Results
Of the total 571 patients with con rmed Covid-19 studied, the mean age was 55 ± 21 years; the study sample included 318 (56%) males, 81 (14%) African Americans, and 101 (18%) Hispanics with the clinical characteristics shown in Table 1. Of the total sample, 202 (36%) required hospitalization without need for intensive care, 43 (8%) required intensive care without intubation, and 64 (11%) required intubation while also receiving intensive care. In ordinal logistic regression models, adjusting for all covariates shown in Fig. 1, both African American race (OR 2.33, 95% CI 1.44-3.78, P = 0.001) and Hispanic ethnicity (OR 1.97, 95% CI 1.14-3.12, P = 0.004) were associated with greater illness severity. While there were no signi cant interactions by race, we observed a paradoxical interaction with ethnicity such that diabetes was associated with greater illness severity in non-Hispanics compared to Hispanics. Values are presented as mean (SD) for continuous variables and n (%) for categorical variables.
P values are for groupwise comparisons, using One-way ANOVA for normal distributed data, Kruskal-Wallis Rank Sum Test for non-normal distributed data, and Chi-square Test for categorical data by default. Outcome of Covid-19 illness severity score in hospitalized patients was de ned as an ordinal variable wherein: 0 = No need for admission, 1 = required admission but never ICU level care, 2 = required ICU level care but never intubate, 3 = required intubation. Odds ratios (OR) are from one multivariable adjusted ordinal logistic regression models adjusted for age, sex, obesity, hypertension, diabetes mellitus, Elixhauser comorbidity score, prior myocardial infarction or heart failure, and prior COPD or asthma.

Discussion
In summary, we found that among patients cared for Covid-19 in a U.S. metropolitan healthcare system, African American race and Hispanic ethnicity were both associated with greater illness severity irrespective of age, sex, and overall burden of comorbid conditions. Our ndings are consistent with recent reports of marked susceptibility to Covid-19 among African Americans, and we now con rm that this trend exists even after accounting for concomitant risk factors. (7)  Availability of data and materials: The data that support the ndings of this study are available from Cedars-Sinai Medical Center, upon reasonable request. The data are not publicly available due to the contents including information that could compromise research participant privacy/consent.

Figure 1
Characteristics Associated with Overall Covid-19 Illness Severity by Race and Ethnicity. *P for interaction values were calculated from likelihood ratio test between models with and without the interaction term. For each variable in the list, race and ethnicity interaction terms are implemented in multivariable adjusted models, with other covariates representative of the entire cohort. Outcome of Covid-19 illness severity score in hospitalized patients was de ned as an ordinal variable wherein: 0 = No need for admission, 1 = required admission but never ICU level care, 2 = required ICU level care but never intubate, 3 = required intubation.