Marked Disparities in COVID-19 Prevalence by Racial/Ethnic, Socioeconomic, Geographic, and Health Care Characteristics, United States, January – April, 2021

Data

We used pooled data from the seven consecutive rounds of the US Census Bureau's Household Pulse Survey (HPS) conducted from January 6 to April 26, 2021 (phase 3, weeks 22 through 27 and phase 3.1, week 28).^20,21 The HPS is a national sample household survey in which data on socioeconomic, demographic, self-assessed health, mental health, and health care characteristics during the COVID-19 pandemic are collected in near real time via email and internet. All information collected in the survey are based on self-reports. The HPS uses a systematic sample design and is representative of the civilian non-institutionalized population of the United States.

The HPS was developed as a rapid response survey in order to track the social and economic impacts of the COVID-19 pandemic on American households on a weekly or a bi-weekly basis in partnership with 11 federal statistical agencies: Bureau of Labor Statistics, National Center for Health Statistics, United States Department of Agriculture's Economic Research Service, National Center for Education Statistics, Centers for Disease Control and Prevention, Maternal and Child Health Bureau, National Institute for Occupational Safety and Health, Department of Defense, Department of Housing and Urban Development, Social Security Administration, and Bureau of Transportation Statistics.^20-23

The survey was implemented in four phases.²³ Phase 1 of the HPS was conducted every week starting April 23, 2020 and ended on July 21,2020. Data collection for Phase 2 of the HPS began on August 19, 2020 and ended on October 26, 2020. Data Collection for Phase 3 began on October 28, 2020 and ended on March 29, 2021.^21,23 Data Collection for Phase 3.1 began on April 14, 2021 and is scheduled to continue through the beginning of July, 2021.²¹ In Phase 1, individuals were asked about their experiences during the pandemic in terms of employment status, food security, housing, physical and mental health, health care access, and educational disruption. Phase 2 included additional questions on the application and receipt of benefits, spending patterns, and availability of financial resources, post-secondary education disruptions, capacity to telework, and travel practices. Phase 3 carries over many of the same questions, but starting with Week 22 additional questions on COVID-19 diagnosis and individuals' intention to receive a COVID-19 vaccine were added.²³

Measurement of COVID-19 Prevalence and Covariates

The prevalence of COVID-19 diagnosis was based on the question: “Has a doctor or other health care provider ever told you that you have COVID-19?”²⁰ The responses included “yes”, “no”, or “not sure”. Those responding “yes” were defined as having been diagnosed with COVID-19 and those with “no” response were considered as not having the disease diagnosis. The response “not sure” was excluded from the analysis.

The pooled sample size for the January 6 – April 26, 2021 HPS was 528,148 for adults aged 18 years and older. There were 6,945 (1.31%) observations with missing data or “not sure” response for the COVID-19 diagnosis question, excluding which resulted in an effective sample size of 521,203 for analysis.

Based on the social-determinants-of-health framework and previous research, the following covariates of COVID-19 prevalence were considered: age, gender, race/ethnicity, marital status, region of residence, educational attainment, household income, housing tenure, job/income loss during the pandemic, and health insurance status.^11,15 These variables are measured as shown in Table 1. The other/multiple race group consisted of AIANs, Native Hawaiians and Pacific Islanders, mixed-race individuals, and other races. Additionally, the COVID-19 prevalence was analyzed for the 15 largest Metropolitan Statistical Areas (MSAs) of the US that were available in the sample. The other metropolitan areas were not identified in the public-use file.

Statistical Analysis

Multivariate logistic regression was used to examine racial/ethnic, socioeconomic, and geographic inequalities in COVID-19 prevalence after controlling for selected socioeconomic and demographic characteristics. The Chi-square statistic was used to test the overall association between each covariate and COVID-19 prevalence, whereas the two-sample t test was used to test the difference in prevalence between any two groups or geographic areas. To account for the complex sample design of the HPS, SUDAAN software was used to conduct all statistical analyses, including the logistic modeling procedure RLOGIST.²⁴ Adjusted COVID-19 prevalence estimates were derived by the logistic model at the mean values of the covariates.

Disparities in Prevalence and Odds of COVID-19 by Race/Ethnicity and Socioeconomic Factors

During January-April 2021, 34.7 million or 14.2% of US adults reported being diagnosed with COVID-19 (data not shown). Those aged 18-24 and 45-54 were two times more likely to be diagnosed with COVID-19 than those aged ≥75 (Table 1). Disease prevalence ranged from 10.0% for Asians to 14.2% for Blacks/African Americans and 23.0% for Hispanics. Educational gradients in prevalence were marked, with adults with less than a high school education having twice the prevalence of COVID-19 than those with a graduate degree (19.0% vs. 9.4%). Adults experiencing household job/income losses during the pandemic had a 32% higher COVID-19 prevalence than those without job/income losses (16.5% vs. 12.5%). Those without health insurance had an 18% higher likelihood of being diagnosed with COVID-19 than those with insurance (15.4% vs. 13.1%).

Age, gender, race/ethnicity, education, household income, marital status, region of residence, health insurance, and household job/income loss during the pandemic were independent and significant predictors of COVID-19 diagnosis (Table 2). Those aged 18-24 and 45-54 had, respectively, 2.2 and 2.1 times higher adjusted odds of being diagnosed with COVID-19 than those aged ≥75 (Table 2). Controlling for covariates, non-Hispanic Whites, Blacks, and Hispanics had, respectively, 23%, 31%, and 134% higher adjusted odds of being diagnosed with COVID-19 than Asians. Controlling for covariates, those with less than a high school or high school education had, respectively, 66% and 48% higher odds of being diagnosed with COVID-19 than those with a graduate degree. Those experiencing household job/income losses during the pandemic had 18% higher odds of COVID-19 diagnosis than those without job/income losses, adjusting for other factors (Table 2).

Joint Effects of Age, Race/Ethnicity, and Education on COVID-19 Prevalence

Race-specific prevalence differed by age, with Asians reporting the lowest rate of COVID-19 prevalence and Hispanics, non-Hispanic Whites and Blacks reporting the highest prevalence for age groups 18-64 and ≥65 years. The prevalence of COVID-19 ranged 3-fold, from 7.3% for Asian aged ≥65 to 16.0% for Hispanics aged ≥65 and 23.9% for Hispanics aged 18-64 (Table 3). Educational gradients in prevalence were pronounced for both age groups; adults aged 18-64 with less than a high school education had 89% higher likelihood of being diagnosed and adults aged ≥65 had 166% higher likelihood of being diagnosed, compared to their counterparts with a graduate degree. Adults aged 18-64 years with less than a high school education had 3.5 times higher COVID-19 prevalence than those aged ≥65 years with a graduate degree (20.0% vs 5.8%) (Table 3).

Educational patterns in COVID-19 prevalence varied by race. Although lower educational attainment was generally associated with higher COVID-19 prevalence in all groups, educational gradients in prevalence were more consistent and pronounced for Asians, Hispanics and the other/multiple-race group (Table 3). For example, Hispanics with less than a high school education had a COVID-19 prevalence of 26.6%, 83% higher than the prevalence of 14.5% for Hispanics with a graduate degree. Asians with less than a high school education had a COVID-19 prevalence of 15.7%, 192% higher than the prevalence of 5.4% for Asians with a graduate degree. As for the full range of race-education disparities, the COVID-19 prevalence varied almost 5-fold, from a low of 5.4% for Asians with a graduate degree to a high of 26.6% for Hispanics with less than a high school education (Table 3).

Disparities in COVID-19 Prevalence among Large Metropolitan Areas

Among the 15 available largest metropolitan areas in the sample, prevalence of COVID-19 ranged from a low of 7.4% in Seattle-Tacoma and 8.3% in San Francisco-Oakland MSAs to a high of 18.3% in Phoenix-Mesa and 22.0% in Riverside-San Bernardino MSAs (Figure 1). For Hispanics living in Phoenix- Mesa, Los Angeles-Long Beach, and Riverside-San Bernardino MSAs, the COVID-19 prevalence was 28.0%, 28.6%, and 30.6% respectively (Table 4). For non-Hispanic Blacks in the Riverside-San Bernardino and Boston-Cambridge MSAs, the COVID-19 prevalence was 17.1% and 17.2% respectively (Table 4). For adults with less a high school education, the COVID-19 prevalence was 28.7% in Riverside- San Bernardino and 29.1% in Phoenix-Mesa MSAs. For adults with household income <$25,000, the COVID-19 prevalence was 24.7% in Riverside-San Bernardino.

Close

Figure 1:

Prevalence¹ and Adjusted² Odds Ratios for COVID-19 Diagnosis among Adults Aged ≥ 18 Years in 15 Largest Metropolitan Statistical Areas (MSAs), United States, January 6 – April 26, 2021

Notes: ¹Prevalence estimates for all MSAs except San Francisco-Oakland MSA were significantly higher than the prevalence for Seattle-Tacoma MSA at P<0.05. ²Adjusted by logistic regression for age, gender, race/ethnicity, marital status, education, household income, housing tenure, job/income loss, and insurance status. All adjusted odds ratios except for San Francisco-Oakland MSA were statistically significant at P<0.05. Source: Data derived from January 6-April 26, 2021 Household Pulse Survey

Export to PPT

Age, gender, race/ethnicity, education, household income, marital status, and job/income loss were independent and significant predictors of COVID-19 prevalence in metropolitan areas. After controlling for socioeconomic and demographic characteristics, compared with Seattle-Tacoma MSA, the odds of COVID-19 diagnosis were 2.4 times higher (OR=2.36; 95% CI= 1.98-2.82) for Riverside-San Bernardino, 2.3 times higher for Dallas-Fort Worth (OR=2.28; 95% CI=1.93-2.71), and 2.3 times higher (oR=2.33; 95% CI=1.97-2.75) for Phoenix -Mesa MSA. Both adjusted and unadjusted rates of COVID-19 prevalence were significantly higher for all metropolitan areas except for San Francisco- Oakland MSA, compared with the Seattle-Tacoma MSA (Figure 1).

Limitations

This study has limitations. HPS does not collect information on a number of risk factors that could help explain disparities in COVID-19 prevalence such as stress, social isolation, face mask use, social or physical distancing, lack of social support or connectedness, poor diet, physical inactivity, smoking, and alcohol consumption.³⁵ Second, ethnic detail in the public use file is limited as we are unable to identify AIANs and specific Asian/Pacific Islander and Hispanic subgroups who may be at greater risk of COVID-19 infection. Information is also lacking on immigrant groups who may be vulnerable to economic hardships and adverse health outcomes during the pandemic. Third, HPS is a cross-sectional survey; causality cannot be inferred, especially for the association of SES with COVID-19 diagnosis. However, educational-level disparities in COVID-19 prevalence are less likely to be affected by reverse causality as formal education is generally attained by age 25 by most individuals and is fairly stable over the life course.³⁶ Household income in HPS was measured as of 2019, and job and income losses in HPS were measured since the beginning of the pandemic in March 2020, and most likely precede the COVID-19 diagnosis in temporality. Fourth, the respondents in HSP are more likely to be women and non-Hispanic Whites and have higher education, compared with the American Community Survey.³⁷ This might have resulted in an underestimate of the magnitude of racial/ethnic and SES disparities in COVID-19 prevalence. However, we addressed disproportionate sampling of demographic characteristics by using survey weights, which rakes the demographics of the interviewed persons to education attainment/sex/age distributions and ethnicity/race/sex/age population distributions.²² Lastly, ethnic-minorities and adults with lower SES have significantly lower broadband internet and computer access in the US and are less likely to have participated in the internet-based HPS, which might have contributed to an underestimate of social inequalities in COVID-19 prevalence shown here.³⁸

Compliance with Ethical Standards