COVID-19 Exposure, Susceptibility, and Treatment Among Racial and Ethnic Minorities in Pennsylvania, USA: Evidence from a Cross-Sectional, Non-Probability Web Panel Survey

Abstract

This study adopted a survey instrument based on a consistent conceptual framework to investigate racial and ethnic disparities in COVID-19 exposure, susceptibility, and treatment among Pennsylvania residents. A cross-sectional design was implemented through primary data collection using a non-probability web panel. Quota sampling ensured statewide representativeness by region and combined age/gender categories, yielding a final sample of 1043 residents across 64 counties who completed the survey between February and April 2021. Propensity score matching was utilized to estimate average effects of race/ethnicity on COVID-19-related outcomes. Results indicated that racial/ethnic minority respondents were more likely than non-Hispanic White respondents to live in apartments or group quarters. Non-Hispanic Blacks and Asians were disproportionately urban residents, while non-Hispanic Blacks faced increased COVID-19 exposure risks due to reliance on public transportation. Additionally, Non-Hispanic Others experienced higher exposure risks from insufficient sick leave and caregiving responsibilities. The study found limited evidence for disparities in susceptibility and treatment. These findings highlight how disparities in COVID-19 exposure conditions likely contribute significantly to the differing COVID-19 outcomes observed between racial/ethnic minorities and non-Hispanic Whites in Pennsylvania.

1. Introduction

Coronavirus disease 2019 (COVID-19) has become one of the worst disease outbreaks to have ravaged humanity. As of August 2025, there were more than 778.4 million confirmed cases and 7.1 million deaths of COVID-19 all around the world, according to the World Health Organization [1]. The United States has seen more than 103 million COVID-19 cases and 1.2 million deaths due to COVID-19 [1]. Pennsylvania has been one of the hardest-hit states in terms of COVID-19 confirmed cases and deaths. In Pennsylvania, by 10 March 2023—the last day the Johns Hopkins Coronavirus Resource Center reported state totals—there were more than 3.5 million COVID-19 confirmed cases and 50,398 deaths in Pennsylvania [2].

While most victims of COVID-19 are older adults and have underlying medical conditions, racial and ethnic minorities also appear to be bearing the disproportionate burden of COVID-19. Age-adjusted cumulative COVID-19 infection rates were approximately 10% higher among non-Hispanic Black (NHB) and 50% higher among Hispanic compared with the non-Hispanic White (NHW), based on updated CDC data through May 2023 [3]. Disparities in hospitalization outcomes were even more pronounced: hospitalization rates were approximately twice as high for NHB and 1.8 times higher for Hispanic relative to NHW [3]. These statistics illustrate racial and ethnic disparities in COVID-19 exposure and severity.

Although the completeness of COVID-19 data reporting by race and ethnicity has improved considerably since early in the pandemic, disparities may still be underestimated due to ongoing gaps in reporting. As of April 2021, only 22 states reported COVID-19 hospitalization data broken down by race and ethnicity. Only nine states provided similar data for COVID-19 treatments [4]. By November 2021, however, reporting improved significantly, with all 50 states and D.C. providing racial and ethnic breakdowns for COVID-19 cases and deaths, and more states began to share hospitalization data as well [4]. Nevertheless, during COVID-19, the reporting had limitations. Especially regarding data on outpatient treatments (antiviral therapies or interventions), which are still rarely reported publicly at the state level. Consequently, current assessments of racial and ethnic disparities in COVID-19 outcomes, particularly treatment access, likely continue to underestimate the true extent of inequities [5].

Many existing studies on the racial and ethnic disparities in COVID-19 outcomes analyzed factors associated with the disparities on an ad hoc basis without a consistent conceptual framework [6]. To address these limitations, this study developed a new survey instrument based on a consistent conceptual framework to investigate the disparities in COVID-19 exposure, susceptibility, and treatment between racial/ethnic minorities and non-Hispanic Whites in Pennsylvania. This study examined whether racial and ethnic minority residents in Pennsylvania experienced higher levels of COVID-19 exposure, susceptibility, and barriers to treatment access compared to non-Hispanic White residents during the early state of the COVID-19 pandemic.

2. Methods

2.1. Data

This study conducted a cross-sectional analysis using primary data collection obtained through a non-probability web panel managed by Marketing Systems Group of Horsham, Pennsylvania. The online survey panel was established using a double opt-in recruitment methodology. This process involves an initial expression of interest by prospective participants, who voluntarily submit their email addresses. Subsequently, a confirmation email is dispatched to inform them about the panel’s objectives and to verify their intent to participate. Only after the recipient affirms their consent is their enrollment in the panel finalized. The study utilized quotas to establish a statewide sample of Pennsylvania residents that was representative by region and, separately, by age/gender combined categories using the U.S. Census Bureau’s 1 July 2019 State Population Estimates. Region quotas were developed by totaling Pennsylvania’s population by county and then determining what proportion of the state’s residents lived in the counties represented by each region. The survey guarded against duplicates through IP filtering, using cookies to prevent multiple submissions, and manually reviewing responses for similarities in demographics and open-ended responses. Respondents were required to answer all questions to be included in the final dataset; however, they were informed that they had the right to end the survey at any time. Altogether, 1043 adults living in 64 counties of Pennsylvania completed a self-administered web survey between 23 February and 20 April 2021. The survey’s final participation rate was 9.0% This was calculated using the American Association of Public Opinion Research’s Response Rate 3 (RR3) formula. RR3 is obtained by dividing the number of completed interviews by the sum of the numbers of completed interviews, partially completed interviews, refusals, and non-contacts. The participation rate is then adjusted by estimating the proportion of cases of unknown eligibility based on the known proportion of eligible cases of all cases for which eligibility was determined.

2.2. Conceptual Framework

Potential sources of COVID-19 racial and ethnic disparities were systemically analyzed using a consistent conceptual framework adapted from Blumenshine et al. (2008) [7].

Table 1. Measures in exposure, susceptibility, and treatment to COVID-19.

Disparities	Measures
Exposure	Residential setting (single-family home/townhouse/apartment/other); imposed social distancing (yes/no); barriers to social distancing (working remotely was not allowed/no payment if not able to go to work/sick leave is not available/job is not secure if stays home from work/public transportation)
Susceptibility	Heart disease; hypertension; cancer; diabetes; asthma; lung disease; immunosuppression
Treatment	Had COVID-19 symptoms (difficulty breathing/shortness of breath/loss of smell or taste/diarrhea/sore throat/tiredness or exhaustion/muscle or joint ache/cough); received COVID-19 testing (yes/no); reasons why could not receive COVID-19 testing (cost of testing/cost of care/location and mode of test facility/messaging on the need for a gateway provider)

Source: Adapted from health disparities in pandemic influenza [7].

shows the measures in this study. The first factor is racial and ethnic disparities in exposure to communicable disease. Racial and ethnic minorities are more likely to be exposed to communicable diseases than the non-Hispanic White population due to living in urban areas [8] and working as essential workers [6], having unhealthy working environments [9], and lacking paid sick leave [10].

The second factor is racial and ethnic disparities in susceptibility to communicable disease. Racial and ethnic minorities have more chronic health conditions that may lead to severe outcomes from communicable diseases, including heart disease [11], hypertension [12], obesity [13], and diabetes [14], than the non-Hispanic White population.

The third factor is racial and ethnic disparities in access to timely and effective care. Racial and ethnic minorities are more likely to have limited access to care due to inadequate insurance coverage [15], language barriers [16], and a lack of a usual source of care [17].

2.3. Measures

The structural factors associated with COVID-19 racial and ethnic disparities were identified by using a consistent conceptual framework adapted from Blumenshine et al. (2008) [7]. Firstly, the instrument included the following structural measures of living conditions to examine COVID-19 exposure: (1) Living in an urban area; and (2) living in an apartment or group quarters. In addition, the instrument measured the following barriers to social distancing among those who were not able to always practice social distancing during the pandemic: (1) Required to work closely with coworkers outside of the home; (2) not paid if staying home from work; (3) no paid sick leave; (4) could lose job or business if not able to work; (5) reliance on public transportation; and (6) need to take care of friends and family.

Secondly, the instrument asked whether respondents had been diagnosed with the following conditions to predict the likelihood of severe illness due to COVID-19: (1) heart disease; (2) hypertension; (3) cancer; (4) diabetes; (5) asthma; (6) suppressed immune system; and (7) chronic lung disease.

Thirdly, the instrument measured whether respondents had any exposure to a known coronavirus case or had experienced specific symptoms, such as difficulty breathing, shortness of breath, change in sense of smell or taste, diarrhea, sore throat, tiredness or exhaustion, muscle or joint pain, or cough. For those who had COVID-19 exposure or symptoms but were not able to obtain a test, the instrument measured the following barriers to treatment: (1) Cost of testing; (2) unable to get to or use a testing facility (access to testing facility, e.g., transportation); (3) unavailable when they tried to get tested (availability of testing); (4) do not know where to get tested (knowledge of testing facility); (5) do not want to miss work if they receive a positive result (work requirements); (6) it takes too long to receive results (test-result waiting); (7) told by someone (e.g., doctor, employer) that they do not need a test (need of testing). The instrument also collected other demographic and socioeconomic factors, such as age, gender, education, household income, employment status, marital status, and poverty status. The race and ethnicity of respondents were categorized as five mutually exclusive groups, including Hispanic, non-Hispanic Black (NHB), non-Hispanic Asian (NHA), non-Hispanic Other (NHO), and non-Hispanic White (NHW) population. The NHO group includes American Indian and Alaska Native, non-Hispanic Native Hawaiians and Other Pacific Islanders, and multi-race respondents. The survey results carry a margin of error of plus or minus 3.0 percentage points, based on a standard 95% confidence level commonly used in statistical analysis.

Table A1. Effects of race/ethnicity on COVID-19 exposure, susceptibility, and treatment for non-Hispanic Blacks ^{a, b}.

	ATET	SE	Z	p-Value	Lower 95% CI	Upper 95% CI
Disparities in COVID-19 Exposure
Living in an urban area	0.203	0.083	2.440	0.015	0.040	0.367
Living in an apartment	0.157	0.125	1.260	0.209	−0.088	0.403
Practiced social distancing	−0.103	0.103	−1.000	0.315	−0.305	0.098
Not practiced—do not want	−0.295	0.053	−5.530	0.000	−0.399	−0.190
Not practiced—work required	−0.075	0.155	−0.480	0.630	−0.378	0.229
Not practiced—not paid	0.014	0.133	0.100	0.917	−0.247	0.274
Not practiced—no sick leave	−0.093	0.176	−0.520	0.600	−0.439	0.253
Not practiced—lose job	−0.003	0.136	−0.020	0.985	−0.268	0.263
Not practiced—public transportation	0.266	0.120	2.210	0.027	0.030	0.502
Not practiced—family care	0.026	0.156	0.170	0.868	−0.281	0.333
Disparities in COVID-19 susceptibility
Heart disease	−0.011	0.012	−0.930	0.350	−0.034	0.012
High blood pressure	0.066	0.094	0.710	0.480	−0.118	0.250
Diabetes	0.099	0.038	2.640	0.008	0.026	0.173
Asthma	0.003	0.062	0.050	0.962	−0.119	0.125
Cancer	−0.034	0.019	−1.770	0.077	−0.072	0.004
Suppressed immune system	−0.029	0.019	−1.550	0.122	−0.066	0.008
Chronic lung disease	−0.015	0.012	−1.260	0.209	−0.039	0.009
Disparities in COVID-19 treatment
No barriers to testing	−0.069	0.114	−0.600	0.547	−0.293	0.155
Do not want testing	−0.025	0.051	−0.500	0.615	−0.125	0.074
Cost of testing	−0.049	0.021	−2.300	0.022	−0.091	−0.007
Unable to get to a testing facility	0.054	0.015	3.620	0.000	0.025	0.084
Testing was unavailable	0.179	0.035	5.150	0.000	0.111	0.247
Do not know where to get tested	−0.108	0.041	−2.620	0.009	−0.189	−0.027
Do not want to miss work	−0.026	0.019	−1.400	0.163	−0.063	0.011
Too long to receive results	−0.034	0.048	−0.710	0.480	−0.129	0.060
Told that I do not need a test	−0.042	0.055	−0.760	0.447	−0.149	0.066
Something else	−0.053	0.023	−2.260	0.024	−0.098	−0.007

^a Each racial/ethnic minority group was matched with non-Hispanic Whites to achieve a better balance in confounders, including age, sex, education, marital status, employment status, household income, and poverty status. The 1:3 propensity score matching was performed by using the nearest-neighbor method. The “teffects psmatch” command was used for propensity score matching in STATA SE version 16. ^b Average treatment effect for the treated (ATET), standard errors (SE), Z-statistic, p-value and 95% confidence intervals (CI) for each race/ethnicity variable were reported in this table.

,

Table A2. Average effects of race/ethnicity on COVID-19 exposure, susceptibility, and treatment for Hispanics ^{a, b}.

	ATET	SE	Z	p-Value	Lower 95% CI	Upper 95% CI
Disparities in COVID-19 Exposure
Living in an urban area	−0.070	0.132	−0.530	0.596	−0.328	0.188
Living in an apartment	0.207	0.118	1.750	0.080	−0.025	0.439
Practiced social distancing	−0.096	0.112	−0.860	0.392	−0.315	0.124
Not practiced—do not want	−0.077	0.190	−0.410	0.684	−0.449	0.295
Not practiced—work required	0.295	0.123	2.400	0.017	0.054	0.537
Not practiced—not paid	−0.138	0.161	−0.860	0.391	−0.453	0.177
Not practiced—no sick leave	0.114	0.224	0.510	0.610	−0.325	0.553
Not practiced—lose job	0.019	0.257	0.070	0.941	−0.486	0.524
Not practiced—public transportation	0.143	0.189	0.760	0.449	−0.227	0.513
Not practiced—family care	0.060	0.208	0.290	0.775	−0.348	0.467
Disparities in COVID-19 susceptibility
Heart disease	−0.018	0.012	−1.540	0.123	−0.041	0.005
High blood pressure	−0.008	0.110	−0.080	0.940	−0.224	0.207
Diabetes	0.015	0.075	0.200	0.844	−0.133	0.162
Asthma	−0.095	0.073	−1.290	0.196	−0.239	0.049
Cancer	−0.013	0.013	−0.970	0.332	−0.039	0.013
Suppressed immune system	−0.034	0.026	−1.310	0.189	−0.085	0.017
Chronic lung disease	−0.032	0.023	−1.380	0.168	−0.077	0.013
Disparities in COVID-19 treatment
No barriers to testing	0.386	0.107	3.600	0.000	0.176	0.596
Do not want testing	−0.103	0.037	−2.800	0.005	−0.175	−0.031
Cost of testing	−0.107	0.050	−2.140	0.032	−0.206	−0.009
Unable to get to a testing facility	−0.086	0.051	−1.680	0.093	−0.187	0.014
Testing was unavailable	−0.126	0.048	−2.630	0.009	−0.220	−0.032
Do not know where to get tested	−0.030	0.026	−1.150	0.252	−0.081	0.021
Do not want to miss work	−0.062	0.032	−1.960	0.050	−0.124	0.000
Too long to receive results	−0.076	0.041	−1.840	0.066	−0.157	0.005
Told that I do not need a test	−0.130	0.047	−2.780	0.005	−0.221	−0.038
Something else	−0.074	0.109	−0.680	0.495	−0.288	0.139

^a Each racial/ethnic minority group was matched with non-Hispanic Whites to achieve a better balance in confounders, including age, sex, education, marital status, employment status, household income, and poverty status. The 1:3 propensity score matching was performed by using the nearest-neighbor method. The “teffects psmatch” command was used for propensity score matching in STATA SE version 16. ^b Average treatment effect for the treated (ATET), standard errors (SE), Z-statistic, p-value and 95% confidence intervals (CI) for each race/ethnicity variable were reported in this table.

,

Table A3. Average effects of race/ethnicity on COVID-19 exposure, susceptibility, and treatment for non-Hispanic Asians ^{a, b}.

	ATET	SE	Z	p-Value	Lower 95% CI	Upper 95% CI
Disparities in COVID-19 Exposure
Living in an urban area	0.217	0.043	5.050	0.000	0.133	0.301
Living in an apartment	0.240	0.117	2.050	0.040	0.011	0.469
Practiced social distancing	0.139	0.029	4.800	0.000	0.082	0.196
Not practiced—do not want	−0.338	0.054	−6.270	0.000	−0.443	−0.232
Not practiced—work required	0.007	0.149	0.040	0.964	−0.286	0.299
Not practiced—not paid	−0.323	0.176	−1.830	0.067	−0.669	0.023
Not practiced—no sick leave	−0.125	0.216	−0.580	0.563	−0.549	0.299
Not practiced—lose job	−0.274	0.209	−1.310	0.189	−0.684	0.135
Not practiced—public transportation	0.042	0.164	0.250	0.799	−0.280	0.363
Not practiced—family care	−0.075	0.183	−0.410	0.682	−0.434	0.284
Disparities in COVID- 19 susceptibility
Heart disease	−0.009	0.009	0.960	0.337	−0.027	0.009
High blood pressure	−0.122	0.056	−2.190	0.029	−0.231	−0.013
Diabetes	−0.063	0.093	−0.680	0.499	−0.246	0.120
Asthma	−0.086	0.080	−1.080	0.282	−0.244	0.071
Cancer	−0.034	0.017	−1.970	0.049	−0.067	0.000
Suppressed immune system	−0.049	0.028	−1.770	0.077	−0.103	0.005
Chronic lung disease	−0.006	0.006	−1.050	0.295	−0.018	0.005
Disparities in COVID- 19 treatment
No barriers to testing	0.099	0.062	1.590	0.113	−0.023	0.220
Do not want testing	−0.078	0.032	−2.420	0.016	−0.141	−0.015
Cost of testing	−0.024	0.015	−1.660	0.097	−0.053	0.004
Unable to get to a testing facility	0.004	0.030	0.150	0.884	−0.054	0.063
Testing was unavailable	−0.065	0.060	−1.080	0.281	−0.183	0.053
Do not know where to get tested	0.063	0.059	1.060	0.291	−0.054	0.179
Do not want to miss work	−0.033	0.018	−1.780	0.076	−0.069	0.003
Too long to receive results	0.057	0.054	1.050	0.296	−0.050	0.164
Told that I do not need a test	−0.083	0.027	−3.040	0.002	−0.136	−0.029
Something else	0.006	0.073	0.090	0.930	−0.136	0.149

^a Each racial/ethnic minority group was matched with non-Hispanic Whites to achieve a better balance in confounders, including age, sex, education, marital status, employment status, household income, and poverty status. The 1:3 propensity score matching was performed by using the nearest-neighbor method. The “teffects psmatch” command was used for propensity score matching in STATA SE version 16. ^b Average treatment effect for the treated (ATET), standard errors (SE), Z-statistic, p-value and 95% confidence intervals (CI) for each race/ethnicity variable were reported in this table.

and

Table A4. Average effects of race/ethnicity on COVID-19 exposure, susceptibility, and treatment among non-Hispanic Others ^{a, b}.

	ATET	SE	Z	p-Value	Lower 95% CI	Upper 95% CI
Disparities in COVID-19 Exposure
Living in an urban area	−0.003	0.149	−0.020	0.982	−0.295	0.288
Living in an apartment	0.328	0.146	2.240	0.025	0.041	0.614
Practiced social distancing	−0.080	0.147	−0.540	0.586	−0.369	0.208
Not practiced—do not want	−0.135	0.162	−0.830	0.404	−0.452	0.182
Not practiced—work required	0.002	0.196	0.010	0.991	−0.382	0.386
Not practiced—not get paid	0.159	0.166	0.960	0.339	−0.167	0.484
Not practiced—no sick leave	0.315	0.071	4.460	0.000	0.177	0.454
Not practiced—lose job	0.150	0.104	1.450	0.148	−0.053	0.354
Not practiced—public transportation	−0.032	0.046	−0.680	0.494	−0.122	0.059
Not practiced—family care	0.286	0.113	2.540	0.011	0.065	0.507
Disparities in COVID-19 susceptibility
Heart disease	−0.019	0.018	−1.040	0.297	−0.053	0.016
High blood pressure	−0.240	0.077	−3.130	0.002	−0.390	−0.090
Diabetes	−0.112	0.036	−3.070	0.002	−0.184	−0.041
Asthma	−0.118	0.065	−1.810	0.071	−0.245	0.010
Cancer	0.027	0.089	0.300	0.764	−0.148	0.202
Suppressed immune system	−0.070	0.121	−0.570	0.566	−0.307	0.168
Chronic lung disease	0.000	0.011	0.000	1.000	−0.022	0.022
Disparities in COVID-19 treatment
No barriers to testing	0.107	0.163	0.660	0.512	−0.212	0.426
Do not want testing	−0.066	0.033	−2.020	0.043	−0.130	−0.002
Cost of testing	−0.046	0.032	−1.470	0.142	−0.108	0.016
Unable to get to a testing facility	−0.046	0.043	−1.080	0.279	−0.130	0.038
Testing was unavailable	0.156	0.153	1.020	0.308	−0.144	0.457
Do not know where to get tested	−0.094	0.065	−1.460	0.144	−0.221	0.032
Do not want to miss work	−0.019	0.019	−0.990	0.323	−0.055	0.018
Too long to receive results	0.172	0.133	1.290	0.196	−0.089	0.433
Told that I do not need a test	0.080	0.022	3.700	0.000	0.037	0.122
Something else	−0.062	0.043	−1.460	0.144	−0.146	0.021

^a Each racial/ethnic minority group was matched with non-Hispanic Whites to achieve a better balance in confounders, including age, sex, education, marital status, employment status, household income, and poverty status. The 1:3 propensity score matching was performed by using the nearest-neighbor method. The “teffects psmatch” command was used for propensity score matching in STATA SE version 16. ^b Average treatment effect for the treated (ATET), standard errors (SE), Z-statistic, p-value and 95% confidence intervals (CI) for each race/ethnicity variable were reported in this table.

provides more explanation on the conceptual framework for COVID-19 exposure, susceptibility, and treatment among racial and ethnic minorities.

2.4. Analysis

Compared to racial and ethnic minorities, non-Hispanic White population exhibit different demographic characteristics and socioeconomic characteristics which can affects COVID-19 exposure, susceptibility, and treatment measures. The average effect of race and ethnicity on COVID-19 exposure, susceptibility, and treatment for racial and ethnic minorities was calculated by using propensity score matching to achieve a better balance in the covariates between each racial/ethnic minority group and the matched NHW respondents. Nearest-neighbor 1:3 matching was performed to improve balance in age, sex, education, marital status, employment status, household income, and poverty status of the respondents. Propensity score matching was selected over propensity score stratification and covariate adjustment because it provides more unbiased estimates than other methods when assessing binary outcome measures [18]. In addition, Pearson chi-square tests were used to examine the differences in COVID-19 exposure, susceptibility, and treatment variables across racial and ethnic groups. This study was reviewed and approved as exempt research by the Institutional Review Board of Authors’ University (STUDY00016682).

3. Results

3.1. Descriptive Statistics

Table 2. Descriptive statistics stratified by race and ethnicity.

Measures	Non-Hispanic White Respondents (n = 947)		Non-Hispanic Black Respondents (n = 34)		Hispanic Respondents (n = 23)		Non-Hispanic Asian Respondents (n = 20)		Non-Hispanic Other Respondents a (n = 19)		p-Value c
	No. b	% b	No. b	% b	No. b	% b	No. b	% b	No. b	% b
Age
18–34	237	25.03	15	44.12	15	65.22	12	60.00	12	63.16	<0.01
35–64	473	49.95	13	38.24	7	30.43	8	40.00	5	26.32	0.05
65 and older	237	25.03	6	17.65	1	4.35	0	0.00	2	10.53	<0.01
Gender
Male	452	47.73	19	55.88	13	56.52	11	55.00	5	26.32	0.24
Female	495	52.27	15	44.12	10	43.48	9	45.00	14	73.68	0.24
Education
Highschool or less	220	23.31	7	20.59	6	27.27	2	10.00	4	21.05	0.67
Some college	306	32.42	14	41.18	8	36.36	5	25.00	6	31.58	0.77
College	241	25.53	12	35.29	6	27.27	10	50.00	5	26.32	0.11
Master or more	177	18.75	1	2.94	2	9.09	3	15.00	4	21.05	0.14
Marital status
Married	548	57.93	11	32.35	13	56.52	11	55.00	9	47.37	0.05
Unmarried	398	42.07	23	67.65	10	43.48	9	45.00	10	52.63	0.05
Employment status
Employed	486	51.43	16	47.06	11	47.83	15	75.00	9	47.37	0.29
Unemployed	459	48.57	18	52.94	12	52.17	5	25.00	10	52.63	0.29
Income
<30 K	188	20.82	13	40.63	5	23.81	6	30.00	9	52.94	<0.01
30–60 K	261	28.90	10	31.25	8	38.10	4	20.00	5	29.41	0.79
60–100 K	242	26.80	4	12.50	5	23.81	6	30.00	1	5.88	0.13
≥100 K	212	23.48	5	15.63	3	14.29	4	20.00	2	11.76	0.51
Poverty status
<200% poverty line	274	30.34	15	46.88	8	38.10	7	35.00	12	70.59	<0.01
≥200% poverty line	629	69.66	17	53.13	13	61.90	13	65.00	5	29.41	<0.01

^a Non-Hispanic Other respondents include American Indians, Alaska Natives, Native Hawaiians, Other Pacific Islanders, and multi-race respondents. ^b Frequency counts and percentages are reported for all categorical variables. ^c p-value shows the difference across racial and ethnic groups from Pearson chi-square tests.

shows that 1043 Pennsylvanians, including 947 NHW, 34 NHB, 23 Hispanic, 20 NHA, and 19 NHO respondents, completed the survey between 23 February and 20 April 2021. Table 2 compares the demographic and socioeconomic characteristics across racial/ethnic groups. Racial and ethnic minority respondents were more likely to be younger than NHW respondents. A total of 65.22% of Hispanics were between 18 and 34 years old compared to 25.03% of NHWs. A total of 25.03% of NHWs were 65 years and older, compared to no observation from NHAs. A total of 57.93% of NHWs were married, compared to 32.35% of NHBs. Racial and ethnic minorities were more likely to live in households with lower incomes than NHWs. A total of 52.94% of NHOs lived in households with incomes below 30 K, compared to 20.82% of NHWs. A total of 70.59% of NHOs lived in households that were below 200% of the federal poverty line, compared to 30.34% of NHWs.

3.2. Bivariate Analyses of Racial and Ethnic COVID-19 Disparities

Table 3. Racial and ethnic disparities in exposure, susceptibility, and treatment to COVID-19.

	Non-Hispanic White Respondents (n = 947)		Non-Hispanic Black Respondents (n = 34)		Hispanic Respondents (n = 23)		Non-Hispanic Asian Respondents (n = 20)		Non-Hispanic Other Respondents a (n = 19)		p-Value c
	No. b	% b	No. b	% b	No. b	% b	No. b	% b	No. b	% b
Disparities in COVID-19 exposure
Living in an urban area	685	72.33	31	91.18	15	65.22	19	95.00	10	52.63	<0.01
Living in an apartment	157	16.63	16	48.48	9	42.86	5	25.00	9	47.37	<0.01
Practiced social distancing	554	58.50	16	47.06	14	60.87	13	65.00	8	42.11	0.38
Not practiced—do not want	105	11.09	1	2.94	2	8.70	0	0.00	2	10.53	0.21
Not practiced—work required	147	15.52	4	11.76	7	30.43	3	15.00	4	21.05	0.27
Not practiced—not paid	146	15.42	7	20.59	5	21.74	2	21.74	5	26.32	0.89
Not practiced—no sick leave	83	8.76	2	5.88	4	17.39	1	5.00	4	21.05	0.49
Not practiced—lose job	123	12.99	5	14.71	4	17.39	2	10.00	5	26.32	0.82
Not practiced—public transportation	52	5.49	6	17.65	2	8.70	3	15.00	2	10.53	0.07
Not practiced—family care	113	11.93	3	8.82	3	13.04	3	15.00	5	26.32	0.60
Disparities in COVID-19 susceptibility
Heart disease	58	6.22	3	8.82	0	0.00	0	0.00	0	0.00	0.35
High blood pressure	309	33.12	12	35.29	5	23.81	1	5.00	1	5.26	<0.01
Diabetes	123	13.18	6	17.65	2	9.52	2	10.00	1	5.26	0.73
Asthma	140	15.01	5	14.71	1	4.76	1	5.00	5	26.32	0.26
Cancer	63	6.75	1	2.94	0	0.00	0	0.00	2	10.53	0.38
Suppressed immune system	58	6.22	0	0.00	1	4.76	0	0.00	2	10.53	0.37
Chronic lung disease	65	6.97	2	5.88	0	0.00	0	0.00	0	0.00	0.34
Disparities in COVID-19 treatment
No barriers to testing	652	68.85	21	61.76	20	86.96	13	65.00	11	57.89	0.24
Do not want testing	72	7.60	2	5.88	1	4.35	0	0.00	0	0.00	0.46
Cost of testing	38	4.01	0	0.00	1	4.35	0	0.00	1	5.26	0.67
Unable to get to a testing facility	29	3.06	4	11.76	0	0.00	1	5.00	3	15.79	<0.01
Testing was unavailable	76	8.03	7	20.59	2	8.70	2	10.00	3	15.79	0.10
Do not know where to get tested	40	4.22	2	5.88	0	0.00	2	10.00	1	5.26	0.58
Do not want to miss work	24	2.53	0	0.00	0	0.00	0	0.00	1	5.26	0.62
Too long to receive results	28	2.96	1	2.94	1	4.35	2	10.00	2	10.53	0.17
Told that I do not need a test	52	5.49	1	2.94	1	4.35	0	0.00	2	10.53	0.63
Something else	44	4.65	2	5.88	2	8.70	3	15.00	0	0.00	0.18

^a Non-Hispanic Other respondents include American Indians, Alaska Natives, Native Hawaiians, Other Pacific Islanders, and multi-race respondents. ^b Frequency counts and percentages are reported for all categorical variables. ^c p-value shows the difference across racial and ethnic groups from Pearson chi-square tests.

shows the results of bivariate analyses of COVID-19 exposure, susceptibility, and treatment across racial/ethnic groups. A total of 95% of NHAs were living in an urban area, compared to 72.33% of NHWs. A total of 48.48% of NHBs were living in an apartment or group quarters, compared to 16.63% of NHWs. Reliance on public transportation as the only barrier to social distancing showed at least a marginally significant difference across racial/ethnic groups. A total of 32.63% of NHWs reported having high blood pressure, compared to 5% of NHAs. A total of 15.79% of NHOs were unable to get to a COVID-19 testing facility, compared to 3.06% of NHWs.

3.3. Score Matching Analyses of Racial and Ethnic COVID-19 Disparities

3.3.1. Disparities in COVID-19 Exposure

Table 4. Average effects of race and ethnicity on COVID-19 exposure, susceptibility, and treatment for racial and ethnic minorities ^a.

	Non-Hispanic Black Respondents		Hispanic Respondents		Non-Hispanic Asian Respondents		Non-Hispanic Other Respondents b
	ATET c	CI c	ATET c	CI c	ATET c	CI c	ATET c	CI c
Disparities in COVID-19 exposure
Living in an urban area	0.203 **	[0.040, 0.367]	−0.070	[−0.328, 0.188]	0.217 ***	[0.133, 0.301]	−0.003	[−0.295, 0.288]
Living in an apartment	0.157	[−0.088, 0.403]	0.207 *	[−0.025, 0.439]	0.24 **	[0.011, 0.469]	0.328 **	[0.041, 0.614]
Practiced social distancing	−0.103	[−0.305, 0.098]	−0.096	[−0.315, 0.124]	0.139 ***	[0.082, 0.196]	−0.080	[−0.369, 0.208]
Not practiced—do not want	−0.295 ***	[−0.399, −0.190]	−0.077	[−0.449, 0.295]	−0.338 ***	[−0.443, −0.232]	−0.135	[−0.452, 0.182]
Not practiced—work required	−0.075	[−0.378, 0.229]	0.295 **	[0.054, 0.537]	0.007	[−0.286, 0.299]	0.002	[−0.382, 0.386]
Not practiced—not paid	0.014	[−0.247, 0.274]	−0.138	[−0.453, 0.177]	−0.323 *	[−0.669, 0.023]	0.159	[−0.167, 0.484]
Not practiced—no sick leave	−0.093	[−0.439, 0.253]	0.114	[−0.325, 0.553]	−0.125	[−0.549, 0.299]	0.315 ***	[0.177, 0.454]
Not practiced—lose job	−0.003	[−0.268, 0.263]	0.019	[−0.486, 0.524]	−0.274	[−0.684, 0.135]	0.150	[−0.053, 0.354]
Not practiced—public transportation	0.266 **	[0.030, 0.502]	0.143	[−0.227, 0.513]	0.042	[−0.280, 0.363]	−0.032	[−0.122, 0.059]
Not practiced—family care	0.026	[−0.281, 0.333]	0.060	[−0.348, 0.467]	−0.075	[−0.434, 0.284]	0.286 **	[0.065, 0.507]
Disparities in COVID-19 susceptibility
Heart disease	−0.011	[−0.034, 0.012]	−0.018	[−0.041, 0.005]	−0.009	[−0.027, 0.009]	−0.019	[−0.053, 0.016]
High blood pressure	0.066	[−0.118, 0.25]	−0.008	[−0.224, 0.207]	−0.122 **	[−0.231, −0.013]	−0.240 ***	[−0.390, −0.090]
Diabetes	0.099 ***	[0.026, 0.173]	0.015	[−0.133, 0.162]	−0.063	[−0.246, 0.120]	−0.112 ***	[−0.184, −0.041]
Asthma	0.003	[−0.119, 0.125]	−0.095	[−0.239, 0.049]	−0.086	[−0.244, 0.071]	−0.118 *	[−0.245, 0.010]
Cancer	−0.034 *	[−0.072, 0.004]	−0.013	[−0.039, 0.013]	−0.034 **	[−0.067, 0.000]	0.027	[−0.148, 0.202]
Suppressed immune system	−0.029	[−0.066, 0.008]	−0.034	[−0.085, 0.017]	−0.049 *	[−0.103, 0.005]	−0.070	[−0.307, 0.168]
Chronic lung disease	−0.015	[−0.039, 0.009]	−0.032	[−0.077, 0.013]	−0.006	[−0.018, 0.005]	0.000	[−0.022, 0.022]
Disparities in COVID-19 treatment
No barriers to testing	−0.069	[−0.293, 0.155]	0.386 **	[0.176, 0.596]	0.099	[−0.023, 0.22]	0.107	[−0.212, 0.426]
Do not want testing	−0.025	[−0.125, 0.074]	−0.103 ***	−0.175, −0.031]	−0.078 **	[−0.141, −0.015]	−0.066 **	[−0.130, −0.002]
Cost of testing	−0.049 **	[−0.091, −0.007]	−0.107 **	[−0.206, −0.009]	−0.024 *	[−0.053, 0.004]	−0.046	[−0.108, 0.016]
Unable to get to a testing facility	0.054 ***	[0.025, 0.084]	−0.086 *	[−0.187, 0.014]	0.004	[−0.054, 0.063]	−0.046	[−0.130, 0.038]
Testing was unavailable	0.179 ***	[0.111, 0.247]	−0.126 ***	[−0.220, −0.032]	−0.065	[−0.183, 0.053]	0.156	[−0.144, 0.457]
Do not know where to get tested	−0.108 ***	[−0.189, −0.027]	−0.030	[−0.081, 0.021]	0.063	[−0.054, 0.179]	−0.094	[−0.221, 0.032]
Do not want to miss work	−0.026	[−0.063, 0.011]	−0.062 **	[−0.124, 0.000]	−0.033 *	[−0.069, 0.003]	−0.019	[−0.055, 0.018]
Too long to receive results	−0.034	[−0.129, 0.06]	−0.076 *	[−0.157, 0.005]	0.057	[−0.050, 0.164]	0.172	[−0.089, 0.433]
Told that I do not need a test	−0.042	[−0.149, 0.066]	−0.130 ***	[−0.221, −0.038]	−0.083 ***	[−0.136, −0.029]	0.080 ***	[0.037, 0.122]
Something else	−0.053 **	[−0.098, −0.007]	−0.074	[−0.288, 0.139]	0.006	[−0.136, 0.149]	−0.062	[−0.146, 0.021]

^a Each racial/ethnic minority group was matched with non-Hispanic Whites to achieve a better balance in confounders, including age, sex, education, marital status, employment status, household income, and poverty status. A 1:3 propensity score matching was performed by using the nearest-neighbor method. ^b Non-Hispanic Other respondents include American Indians or Alaska Natives, Native Hawaiians and Other Pacific Islanders, and multi-race respondents. ^c Average treatment effect for the treated (ATET) and 95% confidence intervals (CI) for race/ethnicity variables were reported in this table. * p < 0.1 percent; ** p < 0.05 percent; *** p < 0.01 percent.

reports the average effects of race and ethnicity on COVID-19 exposure, susceptibility, and treatment for racial and ethnic minorities compared to NHWs by using propensity score matching. After matching, the probability of living in an urban area was 20.3% higher among NHBs and 21.7% higher among NHAs than NHWs. The probability of living in an apartment or group quarters was 24% higher among NHAs, 32.8% higher among NHOs than NHWs. The probability of “do not want to practice social distancing” was 29.5% lower among NHBs, 33.8% lower among NHAs than NHWs. Hispanics were 29.5 % more likely to be unable to practice social distancing due to work requirement than NHWs. The probability of being unable to practice social distancing due to lack of sick leave from work was 31.5% higher among NHOs than NHWs. The probability of being unable to practice social distancing due to public transportation was 30.8% higher among NHBs than NHWs. The probability of being unable to practice social distancing due to family care was 28.6% higher among NHOs than NHWs.

3.3.2. Disparities in COVID-19 Susceptibility

The probability of having hypertension was 12.2% lower among NHAs, and 24% lower among NHOs than NHWs. The probability of having diabetes was 9.9% higher among NHBs, and 11.2% lower among NHOs than NHWs. The probability of having cancer was 3.4% lower among NHAs than NHWs.

3.3.3. Disparities in COVID-19 Treatment

The probability of reporting no barrier to testing was 38.6% higher among Hispanics than NHWs. The probability of reporting “Do not want to test” was 10.3% lower among Hispanics, 7.8% lower among NHAs, and 6.6% lower among NHOs than NHWs. The probability of reporting cost as a barrier to testing was 4.9% lower among NHBs, and 10.7% lower among Hispanics than NHWs. NHBs were 5.4% more likely to be unable to get to a testing facility than NHWs. The probability of reporting “Testing was unavailable” was 17.9% higher among NHBs, and 12.6% lower among Hispanics than NHWs. The probability of reporting “Do not want to miss work” was 6.2% lower among Hispanics than NHWs. The probability of reporting “Told that I do not need a test” was 13% lower among Hispanics, 8.3% lower among NHAs, and 8% higher among NHOs than NHWs. The probability of reporting “Something else” was 5.3% lower among NHBs than NHWs. Appendix A, Table A1, Table A2, Table A3 and Table A4 reports the average effects of race and ethnicity on COVID-19 exposure, susceptibility, and treatment for racial and ethnic minorities in detail.

4. Discussion

4.1. Racial and Ethnic Disparities in COVID-19 Exposure, Susceptibility, and Treatment

Propensity score matching analyses were able to show statistically significant differences in COVID-19 exposure when each racial/ethnic minority group was matched with the NHW population with a better balance in the covariates. The findings of this study suggest that racial and ethnic minorities in Pennsylvania were more exposed to COVID-19 than the NHW population, and these disparities closely relate to where racial and ethnic minorities spend time on a regular basis. Populations living in an urban area and/or an apartment/group quarters were more likely to be exposed to COVID-19, and persistent residential segregation by race and ethnicity has aggravated this problem among racial and ethnic minorities [19]. NHB population disproportionately accounts for low-income working families living in urban areas, and even during the pandemic they must rely on public transportation, which increases their exposure to COVID-19 [20]. Poverty and income inequality contribute to work environments that prevent many Hispanics from practicing social distancing, which may lead to a higher risk of COVID-19 exposure [21]. NHA populations also tend to live in overcrowded living environments but were less likely to report “Do not want to practice social distancing” than NHWs, which confirms previous studies that found that NHAs have exhibited more social distancing than other groups [22,23]. NHO population is less likely to live in urban areas but more likely to be exposed to COVID-19 than the NHW population because they tend to live with extended family members in overcrowded living environments and need to take care of significantly larger households [22,24].

Different from the existing studies [22], this study found that NHB respondents did not report work requirements as a barrier to social distancing, compared to NHW respondents. Pennsylvania has a relatively large rural population, and racial and ethnic minorities do not disproportionately represent essential workers in rural Pennsylvania, which may explain why the probability of reporting work requirements as a barrier to social distancing was not significantly different among NHB respondents compared to NHW respondents.

We did not find strong evidence of racial/ethnic disparities in COVID-19 susceptibility with increased risk for severe illness from COVID-19. Bivariate analyses show that higher percentages of NHBs reported having heart disease, and high blood pressure than NHWs, but these differences between NHBs and NHWs were not statistically significant. Diabetes was the only condition which NHBs were significantly more likely to report compared to NHWs. After matching, the probability of having high blood pressure, cancer, or a suppressed immune system was lower for racial/ethnic minority respondents than for NHW respondents, which is different from the previous studies [25]. Racial/ethnic minority respondents in our survey were younger than NHW respondents, as shown in Table 2, so the matched NHWs were younger than the unmatched NHW respondents. Indifferent prevalences of heart disease, and hypertension between racial/ethnic minority respondents and NHW respondents after matching may be driven by the younger age of the treatment and comparison groups, because the prevalence of the high-risk diseases increases with age [26,27,28].

We also did not find strong evidence of racial/ethnic disparities in access to COVID-19 treatment across racial and ethnic minority groups. NHB respondents reported that they were more likely to be unable to get to a testing facility and testing was unable even when they make it. NHO respondents reported that they were more likely to be told that they “do not need a test” than NHWs, but the effect size was relatively small. During the early stages of the pandemic, most states struggled to create, implement, and finance COVID-19 testing sites, resulting in racial and ethnic minorities bearing a disproportionate burden of COVID-19 due to a lack of available COVID-19 testing. This survey was conducted after the multiple waves of COVID-19, and that may explain why this study did not find strong evidence of racial and ethnic disparities in COVID-19 treatment across racial and ethnic minority groups.

4.2. Limitations of the Study

The results of this study should be interpreted with caution for the following reasons. Firstly, this study was conducted during the early phase of the COVID-19 pandemic, when the Delta variant, which was characterized by higher severity but lower transmissibility compared to Omicron and its subvariants, was predominant [29,30]. At that time, standardized measures for assessing COVID-19 exposure were not yet established, and considerable uncertainty surrounding the pandemic persisted. In addition, standardized exposure measures such as contact with confirmed cases, time in high-risk settings, and mask compliance were still lacking, and considerable uncertainty surrounded the situation [31]. COVID-19 exposure variables in this study are proxy variables for the actual COVID-19 exposure. Though these exposure variables are expected to correlate with the actual COVID-19 exposure, the correlation between the proxy and actual measures could be different across racial/ethnic groups and the contexts. Secondly, this study employed a non-probability-based sampling method, and the results of this study are not fully representative of the general Pennsylvania population in terms of race and ethnicity. Yet, by adopting a propensity score matching approach, we were able to measure the disparities in COVID-19 exposure, susceptibility, and treatment between racial/ethnic minority respondents and the matched NHW respondents. Thirdly, the conceptual framework that this study adopted was based on pandemic influenza, which is significantly different from COVID-19 in terms of contagiousness and transmission methods. Compared to influenza A and B, COVID-19 demonstrates greater transmissibility, a longer incubation and infectious period, more severe clinical symptoms, and a higher mortality rate. Nevertheless, both illnesses are contagious respiratory infections transmitted primarily through aerosols and share common symptoms such as fever, cough, and shortness of breath [32]. Fourthly, this study does not analyze other measures of COVID-19 racial and ethnic disparities in hospitalization and admission to intensive care units. Fifthly, other psychological and social factors such as trust in medicine can influence COVID-19 treatment behavior. This study was not designed to delve into these factors but rather to focus on racial and ethnic disparities in COVID-19 treatment. Lastly, there can be confounding issues such as recall bias, reporter bias, survivor bias, selection bias, or differential healthcare access affecting diagnoses behind the lower rates of COVID-19 susceptibility measures among racial and ethnic minorities compared to the non-Hispanic White respondents.

5. Conclusions

This study adopted a survey instrument based on a consistent conceptual framework to investigate racial and ethnic disparities in COVID-19 exposure, susceptibility, and treatment in Pennsylvania. The results of this study have several policy implications for addressing COVID-19 racial and ethnic disparities. Living in overcrowded living conditions, high reliance on public transportation, and overrepresentation in essential workers are largely the consequences of structural inequality [33]. It is evident that eliminating structural inequality is the fundamental solution to eradicating COVID-19 racial and ethnic disparities, but innovative and targeted approaches to reducing disparities can also provide short-term relief and support for racial and ethnic minorities. Establishing facilities for quarantine can be helpful for racial and ethnic minorities living in overcrowded housing, and mandating/recommending facial coverings or 6 feet of distancing in public transportation and essential businesses can minimize COVID-19 exposure for those who rely on public transportation and are required to work outside the home.