SARS-CoV-2 Outcomes: The Role of Patient Characteristics and Chronic Disease Comorbidities

Introduction

Evidence has shown that people with chronic conditions such as diabetes, hypertension, or heart disease have increased severity of illness and mortality when infected with SARS-CoV-2. There is growing evidence that racial minorities are also at higher risk of severe illness when infected with the virus. This could be linked to the higher prevalence of chronic diseases such as diabetes. Data from the Centers for Disease Control and Prevention found that diabetes prevalence (diagnosed plus undiagnosed) was nearly 18% among Hispanics and 16.8% among non-Hispanic Blacks compared to 10% among non-Hispanic Whites. ¹

A systematic review found that Black and Hispanic populations have higher rates of SARS-CoV-2 infection, hospitalization, and mortality than non-Hispanic White populations, but not higher case-fatality rates. Asian populations have similar outcomes to non-Hispanic White groups. Of 3686 COVID-positive patients, adjusted hospitalization rates were higher for male (odds ratio [OR] 1.61), Black (OR 1.47), Hispanic (OR 1.42), and Asian (OR 1.47) groups compared with non-Hispanic White patients. ² There were no racial differences in inpatient mortality or total mortality. There also was a recent systematic overview of the association between race and ethnicity and use of health care services among COVID-19 patients. ²

Although several studies have examined the impact of race and hospital admission among COVID patients, most have been conducted at a local (city, health system) level. One exception is the Boston University and COVID tracking project that examined racial differences in mortality and hospitalizations by race. ³ These data show higher rates of mortality and hospitalizations among African Americans, Hispanics, American Indians, and other racial minorities compared to Whites. For example, mortality rates among Blacks per 100,000 were 178 compared to 154 among Hispanics and 124 among Whites. One of the underlying factors that could account for these differences is the higher rates of chronic disease prevalence among racial minorities. These comorbidities may result in an increase in the severity of COVID-19 and with it rates of hospitalization.

This paper extends previous research in 2 directions. First, this study examines a national database of COVID-19 patients starting in May 2019. These claims data include information on hospitalizations and intensive care unit (ICU) use, among other health care services. Second, this study examines the impact of comorbid chronic conditions on rate of hospitalization, ICU use, and length of hospital stay.

What is not known is whether the increased risk of SARS-CoV-2 morbidity or mortality relates to the presence of the underlying disease itself or whether that underlying disease is under control or not. This study examines how the number of chronic health conditions among COVID-19 patients affects the likelihood of being hospitalized, admitted to the ICU, and length of hospital stay. It also examines how patient characteristics such as race, income, health insurance and age are associated with hospitalizations, ICU use, and length of stay among COVID-19 patients.

Methods

A population of SARS-CoV-2 patients with and without relevant comorbidities (eg, diabetes, hypertension) was identified. Logistic regressions are used to estimate the probability that number of comorbidities (ie, diabetes, hypertension, hyperlipidemia, kidney disease, heart disease, cancer, chronic obstructive pulmonary disease) and patient characteristics (eg, race, age, gender, source of health insurance, medication adherence [medication possession ratio]) have on the probability of hospital use and ICU use. A generalized linear model with log link is used for hospital length of stay and marginal effects are reported.

The analysis is completed using a specially collected set of data on COVID-19 patients. The Symphony data set contains claims data but also includes information on patient demographics starting in May 2019 to the present. Symphony has one of the largest existing repositories of patient-level integrated data, including retail pharmacy data, medical claims, and remittance data. It has vast coverage of the market, including: 92% of retail pharmacy claims, 71% of mail order pharmacy claims, 65% of specialty pharmacy activity, more than 280 million patients, 1.8 million prescribers, and more than 16,000 health plans. The data cover May 2019 to the present. The data contain 1,006,692 counts of COVID-19 patients identified using International Classification of Diseases, Tenth Revision (ICD-10) code U07.1. Symphony received data from clearinghouses and, by the nature of its agreement with suppliers, was not able to disclose patient names.

This study focuses on several highly prevalent chronic conditions among COVID-19 patients – including diabetes, hypertension, hyperlipidemia, chronic kidney disease, and chronic obstructive pulmonary disease – to compare the likelihood of hospitalization, admission to the ICU, and length of stay compared to patients with no comorbid conditions.

Symphony data are used to examine the following outcomes as dependent variables in the regression modeling. COVID-19 is defined by focusing on ICD-10 code U07.1 in the analysis. Utilization outcomes include: admission to the hospital, admission to the ICU, and total length of stay in hospital. Admission to the ICU is identified by critical care Current Procedural Terminology codes 99291 or 99292.

This study also examines interactions between race and the number of comorbid conditions for the 3 utilization outcome variables. The dependent variables are both continuous as well as dichotomous. Dichotomous dependent variables are most appropriately estimated using a logistic or probit regression. For the main analysis without interactions, this study relies on logistic regressions. However, introducing interactions into these models complicates the interpretation of marginal effects. As a result, a linear probability model is used for the models with dichotomous dependent variables. These models have some statistical issues (predictions may be more or less than 0 and 1, and heteroscedasticity). However, they do provide a reasonable estimate of how the race/ethnicity variables and number of comorbid conditions are associated with the utilization measures.

Results

Table 1 presents ORs and marginal effects on the probability of a hospitalization for all identified COVID-19 patients. Female COVID-19 patients were 25% less likely to be hospitalized compared to males. Similar to earlier research, this study finds that minorities have a higher likelihood of hospital admission compared to White adults and children. The odds of a hospital admission were 35% higher among Black patients compared to White COVID-19 patients. Similarly, the odds of a hospital admission were higher among Hispanic patients and among Asian Americans compared to White patients. Relative to commercially insured patients, both Medicare and Medicaid COVID-19 patients were more likely to be hospitalized (Table 1).

Hospital admissions also varied by income. Compared to adults in lower income households (less than $30,000) COVID-19 patients earning more than $100,000 were a third less likely to have a hospital admission. Households earning $75,000 to $99,000 and those earning $50,000 to $74,000 were less likely to be admitted to the hospital compared to the lower income households (Table 1).

The likelihood of hospitalization also varied by age. COVID-19 patients aged 75 and older were more likely to be hospitalized compared to patients aged 50 to 64 years, as were patients aged 65 to 74 years. Patients younger than age 18 have nearly 80% percent lower odds of admission relative to COVID-19 patients aged 50 to 64 years; patients aged 18 to 34 years and 35 to 49 years also had lower odds than patients aged 50 to 64 years (Table 1).

COVID-19 patients with chronic comorbidities also had higher odds of a hospital admission compared to patients with no comorbidities. COVID-19 patients with 1 comorbid condition were more than 2 times more likely to be admitted to the hospital compared to patients with no comorbid conditions. COVID-19 patients with 2, 3, 4, or 5 chronic conditions were more likely to be hospitalized compared to patients with no comorbidities (Table 1).

A very similar pattern emerged for the characteristics of patients admitted to the ICU. Female COVID-19 patients were nearly 40% less likely to be admitted to the ICU compared to males. There also were substantial differences in ICU admission by race. Black, Asian, and Hispanic COVID-19 patients were more likely to admitted to the ICU compared to White COVID-19 patients (Table 1).

However, a different pattern of hospitalization emerged for Medicare patients as they were 20% less likely to be admitted to the ICU compared to privately insured patients. However, this is largely offset by the higher odds of a ICU admission among patients aged 65 and older. The most significant difference was among cash/uninsured patients, who were 23% more likely to be admitted to the ICU compared to privately insured patients. This could reflect differences in underlying comorbid risk factors and health status among the uninsured compared to privately insured COVID-19 patients. Patients with other forms of government insurance were less likely to be admitted to the ICU.

ICU admission also was associated with income. Compared to COVID-19 patients earning less than $30,000, those earning more than $30,000 were less likely to be admitted to the ICU. Adults earning more than $100,000 were more than 20% percent less likely to be admitted compared to patients earning less than $30,000. Those earning $30,000 to $49,000, $50,000 to $74,000 and $75,000 to $99,000 also were less likely to be admitted compared to those with less than $30,000 in income (Table 1).

The likelihood of an ICU admission also increased with age. Those younger than age 18, those aged 18 to 34 years, and those aged 35 to 49 years had lower odds of ICU admissions compared to the group aged 50 to 64 years. Finally, patients aged 65 to 74 years and those older than age 75 were more likely to be admitted to the ICU compared to those aged 50 to 64 (Table 1).

COVID-19 patients with more comorbidities also were more likely to be admitted to the ICU. Patients with 1 comorbid condition were more than 2 times more likely to be admitted to the ICU compared to patients with no comorbid conditions. Moreover, patients with 5 comorbid conditions were more than 7 times more likely to be admitted to the ICU compared to patients with no comorbidities. Each additional comorbidity increased the odds of ICU admission (Table 1).

Finally, length of hospital stay displayed a slightly different pattern. There were no significant differences in length of stay between females and males. Although racial minorities and lower income patients were more likely to be admitted to the hospital and an ICU unit, their length of hospital stay was shorter than White COVID-19 patients (Table 1).

Privately insured patients had the longest length of stay except for Medicare patients. Length of stay among Medicaid patients and stays of uninsured (cash) patients were shorter than for privately insured patients (Table 1).

There also were significant differences in length of stay by age. Relative to hospitalized COVID-19 patients age 50 to 64, length of stay among those younger than age 18 and adults aged 18 to 34 years were shorter. The oldest age cohort, adults aged 75 and older and seniors aged 65 to 74 years had longer lengths of stay than the group aged 50 to 64 years (Table 1).

Finally, number of comorbid chronic conditions was associated with longer lengths of stay. A patient with 3 to 5 comorbid conditions stayed almost 2 days longer compared to patients with no comorbid conditions. Patients with 1 or 2 comorbid conditions stayed longer than patients with no comorbid conditions (Table 1).

Because coding was not available to identify patient mortality in the hospital, it is not clear whether these differences reflect true differences in length of stay or variations in mortality by these characteristics.

This study next examined whether the observed utilization differences by race were confounded by variations in the number of comorbid conditions. To do this a linear probability model was estimated that interacted each of the race categories with the dummy variables for the number of comorbid conditions. Controls were the same as described in the Methods section. To do this White adult COVID-19 patients are compared with each of the racial groups by number of comorbid conditions. The “test” command in Stata (StataCorp LLC, College Station, TX) is used to determine whether the relative changes across racial groups are significantly different. ⁴ Wald tests in Stata are 2-sided tests of equality.

Table 2 shows the results of the interaction between race and number of comorbid conditions. The probability of a hospital admission for White COVID-19 patients was higher regardless of the number of comorbid conditions. White COVID-19 patients were approximately 2 percentage points less likely to be admitted compared to Black and Hispanic patients. The most dramatic difference was between White and Black patients with 3 or more comorbid conditions. Here White patients were 5-6 percentage points less likely to be admitted compared to Black COVID-19 patients.

A similar pattern exists for ICU admission. White patients with no comorbid conditions were less likely to be admitted compared to Black and Hispanic patients. This difference between White, Black, and Hispanic patients increased by a greater percentage for those with 3 and 4 comorbid conditions (Table 2). The only exception was for White patients with 5 or more comorbid conditions, who were nearly 2 percentage points more likely to be admitted to the ICU that Asian patients.

Finally, patterns of length of stay were similar to those presented in Table 1. Results show that longer length of stay among White COVID-19 patients was not related to differences in the number of comorbid conditions by race. Among White patients, length of stay compared to Black patients was a half day longer to nearly a full day (.89 days) longer. Similar patterns were observed for White patients compared to Asian and Hispanic patients.

Discussion

This study found significant differences in the use of hospital services among minority patients with COVID-19. The odds of a hospital admission were 35% higher for Black patients, 21% higher for Hispanics, and 30% higher for Asian Americans compared to White COVID-19 patients. The longer lengths of stay among White COVID-19 patients persisted when controlling for the number of comorbid conditions. Number of comorbid conditions also was associated with substantial differences in the odds of hospital admissions, ICU use, and longer lengths of hospital stay.

COVID-19 patients with 1 or more comorbid conditions had a greater likelihood of a hospital admission, were more likely to be admitted to the ICU, and had longer lengths of stay. The results continue to highlight the importance of the public health strategy of vaccinating older adults and even younger adults with comorbid chronic health care conditions.

To the authors' knowledge, this is the first national study examining the role of key patient demographics and comorbid chronic conditions on the use of health care services among COVID-19 patients. It highlights the need to target resources selectively to patients at risk for the most severe reaction to COVID-19. It also highlights the need for equitable distribution of resources given that the odds of more severe reactions were highest among low-income minority populations. The results provide ongoing lessons for the prevention and treatment of COVID-19 as well as planning lessons for future pandemics.