Racial and Ethnic Disparities in Health-Related Socioeconomic Risks During the Early COVID-19 Pandemic: A National Survey of U.S. Women

Abstract

Background: Nearly half of U.S. women experienced new or worsening health-related socioeconomic risks (HRSRs) (food, housing, utilities and transportation difficulties, and interpersonal violence) early in the COVID-19 pandemic. We sought to examine racial/ethnic disparities in pandemic-related changes in HRSRs among women.

Materials and Methods: We conducted a cross-sectional survey (04/2020) of 3200 women. Pre- and early pandemic HRSRs were described by race/ethnicity. Weighted, multivariable logistic regression models generated odds of incident and worsening HRSRs by race/ethnicity.

Results: The majority of Black, East or Southeast (E/SE) Asian, and Hispanic women reported ≥1 prepandemic HRSR (51%–56% vs. 38% of White women, p < 0.001). By April 2020, 68% of Black, E/SE Asian, and Hispanic women and 55% of White women had ≥1 HRSR (p < 0.001). For most HRSRs, the odds of an incident or worsening condition were similar across racial/ethnic groups, except Black, E/SE Asian and Hispanic women had 2–3.6 times the odds of incident transportation difficulties compared with White women. E/SE Asian women also had higher odds of worsening transportation difficulties compared with White women (adjusted odds ratios = 2.5, 95% confidence interval 1.1–5.6). In the early pandemic, 1/19 Hispanic, 1/28 E/SE Asian, 1/36 Black and 1/100 White women had all 5 HRSRs (extreme health-related socioeconomic vulnerability).

Conclusions: Prepandemic racial/ethnic disparities in HRSRs persisted and prevalence rates increased for all groups early in the pandemic. Disparities in transportation difficulties widened. White women were much less likely than others to experience extreme health-related socioeconomic vulnerability. An equitable COVID-19 response requires attention to persistent and widening racial/ethnic disparities in HRSRs among women.

Introduction

Early in the COVID-19 pandemic, nearly half of all U.S. women experienced new or worsening health-related socioeconomic risks (HRSRs)—food, housing, transportation and utilities difficulties, and interpersonal violence (IPV).¹ Women with new/worsening HRSRs have two to three times higher risk of anxiety, depression, and traumatic stress compared with women without these risks.¹ Additionally, prepandemic socioeconomic vulnerability resulting from racial segregation, poverty, inability to work remotely, and limited health care access^2–5 likely contributes to higher rates of COVID-19 morbidity and mortality among U.S. women of racial/ethnic minority groups.^6–9 The effects of the pandemic on East or Southeast (E/SE) Asian women are of particular concern. Even before the World Health Organization (WHO) declaration of the pandemic on March 11, 2020, this group has been the target of pandemic-related racial scapegoating, humiliation, stigma, and violence.^10,11 Understanding racial/ethnic disparities in pandemic-related changes in HRSRs among U.S. women is essential to ensure an equitable approach to pandemic management and recovery.

Pandemic-related efforts to mitigate health disparities have included deliberate omission of race or ethnicity from clinical decision making about scarce critical care resources¹² and prioritizing vaccine delivery to socially vulnerable communities (including certain communities of color), where COVID-19 rates and severity have been highest.¹³ Many communities have also implemented policies and practices to ensure that basic or social needs—most commonly food and shelter—are being met, with a focus especially on lower income communities of color.^14,15 Importantly, unlike race or ethnicity, HRSRs are factors that can be modified with policies (e.g., expansion of the Supplemental Nutrition Assistance Program [SNAP] and moratoria on utilities' shutoff or eviction), community resources (e.g., emergency food support), and other targeted interventions. The 2010 World Health Organization Conceptual Framework for Action on the Social Determinants of Health, identifies HRSRs as markers of “differential exposures and differential vulnerability” to health inequities.¹⁶ Accordingly, the U.S. Centers for Medicare & Medicaid Services (CMS) and the National Academies of Science, Engineering, and Medicine (NASEM) recommend that HRSRs be assessed and addressed in culturally informed health promotion and disease prevention efforts.^17,18 Similarly, the U.S. Centers for Disease Control and Prevention advocate for attention to HRSRs as part of the pandemic response.^19–21

This study analyzes data from the National Women's Health COVID-19 Study¹ to examine patterns of change in HRSRs in the early pandemic by race/ethnicity. We hypothesized that Hispanic, non-Hispanic Black and non-Hispanic E/SE Asian women would have higher rates of pre- and early pandemic HRSRs compared with non-Hispanic White women, and be more likely to experience incident or worsening HRSRs. We also expected higher pandemic-related rates of IPV among E/SE Asian women compared with others. By filling gaps in our understanding of pandemic-related changes in HRSRs among women of different racial/ethnic groups, this study aims to inform an equitable approach to pandemic response and recovery efforts.

Materials and Methods

The National Women's Health COVID-19 Study was a cross-sectional survey conducted in April 10–24, 2020. The University of Chicago Institutional Review Board approved the study.

Sampling strategy and recruitment

Participants were recruited from Opinions 4 Good's (Op4G) U.S. national research panel. Additional details about the panel (350,000 U.S. residents recruited via nonprofit partner organizations, internet, and word of mouth) have been previously published.¹ To ensure sufficiently large sample sizes in sociodemographic groups of interest, a nested quota sampling strategy was used. The quota strata included age and education distributions reflecting the 2018 U.S. population of women and race, oversampling E/SE Asian women to achieve a sufficiently large sample for hypothesis testing (4% of the 2018 U.S. female population identified as E/SE Asian; 11% of our sample identified as non-Hispanic and E/SE Asian). Potential participants received a recruitment email with a one-time use survey link. Of 3,634 eligible panelists contacted, 3,200 completed the survey (88% cooperation rate).

The survey was fielded March–April 2020 to capture early effects of the COVID-19 pandemic in the United States. The sample size was designed to ensure reasonably precise estimates in major subgroups of interest (e.g., racial/ethnic, age, and geographic region groups), taking into consideration time needed to recruit and budget. Besides E/SE Asians, we did not need to oversample other racial/ethnic groups because the design yielded sufficient sample sizes for these groups.

Data collection and measures

Data were collected via self-administered, web-based surveys. Supplementary Appendix SA1 lists all survey questions used in the current analysis. Participants were categorized as non-Hispanic White (“White”), non-Hispanic Black (“Black”), non-Hispanic E/SE Asian (“E/SE Asian”), Hispanic, or other, based on their responses to two questions: “What race do you consider yourself to be? Please select one or more” (White, Black or African American, American Indian or Alaskan Native, Asian Indian, Chinese, Filipino, Japanese, Korean, Vietnamese, other Asian, Pacific Islander, or Other) and “Do you consider yourself to be Hispanic, Latino/a/x, or of Spanish origin?” (Yes or No). Women were categorized as E/SE Asian if they selected one or more of the following racial subgroups: Chinese, Filipino, Japanese, Korean, and/or Vietnamese. These five E/SE Asian subgroups were the most frequently targeted in early pandemic-related discrimination incidents.¹⁰ Women who selected “other race,” more than one race, Asian Indian, other Asians, American Indian or Alaskan Native, and Pacific Islander were categorized as “Other.” Women who selected “Hispanic, Latino/a/x, or of Spanish origin” were categorized as “Hispanic” regardless of which race they selected.

The survey retrospectively measured prepandemic (before the March 11, 2020 WHO declaration)²² HRSRs using the Accountable Health Communities (AHC) 10-question screening tool and categorized them as present or absent (Supplementary Appendix SA1).¹⁸ These risks included food insecurity, housing stability, transportation difficulties, utilities difficulties, and IPV. In addition, changes in these HRSRs “since the start of the pandemic” were assessed. Women were asked to report change in worry about “food running out before you got money to buy more” (5-point Likert scale; much more to much less); change in ease of getting transportation (5-point Likert scale; much easier to much harder); and their current housing and utility security. Change in IPV was assessed for each of four domains (physical hurt, verbal insults, threats of harm, and screaming or cursing) using the AHC tool (5-point Likert scale; much more to much less). Responses to one item for each domain were recoded as 1 to 5 (much less to much more), summed, and divided by four to generate a mean change score.

For this study, we define the early pandemic as the 2-month period beginning with the WHO declaration of the COVID-19 pandemic. Change in each HRSR was classified as: secure (risk was absent both before and during the early pandemic), incident (absent prepandemic and present during the early pandemic), persistent or improved (present prepandemic and unchanged/improved during the early pandemic), and worsening (present prepandemic and worse during the early pandemic). HRSR status in the prepandemic period was assessed retrospectively. The term “incident” denotes early pandemic HRSRs that were not present in the prepandemic period.

Covariates included sociodemographic variables (age, income, education, marital status, number of household people, number of household children, caregiver status, and region) and health (self-rated overall health and number of comorbidities).

Statistical analysEs

Pseudo design-based weights were generated using the raking ratio method to match the marginal distributions for age group, race, education, income category, and region to the 2018 U.S. population estimates. The use of pseudo-design weights with nonprobability sampling strategy is a statistical approach that has been used by other published studies.^23–26 This strategy, which pairs our nonprobability sample with data from a probability sample (American Community Survey data), is expected to reduce potential selection bias of the sampling strategy, maximize precision of estimates, and improve generalizability. All analyses were weighted.

We described sociodemographic characteristics and health of the sample, the prevalence and patterns of prepandemic HRSRs, and the incidence and worsening rates of pandemic-related change in each HRSR. Chi-squared tests were used to test for significant differences in variables among racial/ethnic groups. We calculated the proportion of women with each combination of early pandemic HRSRs. To visualize co-occurrence of these HRSRs, we created Venn diagrams for each racial/ethnic group of the 10 most frequent HRSR combinations using Displayr (Sydney, Australia). Multivariable logistic regression (adjusted for covariates) was used to model: (1) the odds of any (vs. none) prepandemic HRSR and each prepandemic HRSR, (2) the odds of any incident or worsening HRSR, (3) the odds of incident HRSRs for each HRSR type (e.g., the odds of incident food insecurity among those without prepandemic food insecurity), and (4) the odds of worsening HRSRs for each HRSR type among those with that HRSR prepandemic (e.g., odds of worsening food insecurity among those with prepandemic food insecurity). The primary predictor of interest was race/ethnicity, with non-Hispanic White as the reference group. The heterogeneity and relatively small size of the “Other” racial/ethnic group precluded meaningful interpretation; for completeness, data for this group were included in tables. Results are presented as adjusted odds ratios (aOR) with 95% confidence intervals (CIs). CIs were not adjusted for multiple testing. All analyses were conducted using Stata 16.1.

Results

Table 1 summarizes the sociodemographic characteristics and health of the sample by race/ethnicity. Overall, a majority of the sample had income above $50,000 (59%), more than a high school education (63%), were married or partnered (62%), and had no children (61%). In addition, 19% self-rated their health as fair/poor. Nearly half (46%) reported at least one medical comorbidity. There were significant differences between racial/ethnic groups for most sociodemographic characteristics as well as health. More Black women reported a household income less than $25,000 (28% vs. 20% overall) and fewer reported being married/partnered (38% vs. 62% overall). More E/SE Asian women reported household income greater than $100,000 (42% vs. 29% overall) and fewer reported fair/poor health (11% vs. 19% overall).

Table 1.

Sociodemographic Characteristics and Self-Rated Health by Race and Ethnicity

	Total	Non-Hispanic White	Non-Hispanic Black	Non-Hispanic East or Southeast Asian	Hispanic	Other	p-Value
	n = 3,176	n = 1,931	n = 402	n = 338	n = 401	n = 104
	%	%	%	%	%	%
Sociodemographic characteristics
Age							<0.001
18 to 44 years	44.6	42.4	44.1	42.1	60.5	32.4
45 to 64 years	33.0	34.2	32.4	35.3	25.3	37.4
65 and above	22.4	23.4	23.4	22.7	14.1	30.2
Income							0.002
Less than $25,000	19.6	19.0	27.5	10.1	20.3	18.2
$25,000–$49,000	21.3	20.2	22.5	19.0	23.5	29.6
$50,000–$99,999	30.0	30.9	27.1	29.0	27.4	32.1
$100,000 or more	29.1	29.9	22.9	41.9	28.8	20.2
Educational attainment							0.29
High school or less	37.2	36.3	36.7	37.5	42.7	34.3
More than high school	62.8	63.7	63.3	62.5	57.3	65.7
Marital status							<0.001
Single, divorced, or widowed	38.1	34.8	61.6	34.9	38.1	33.3
Married or partnered	62.0	65.2	38.4	65.1	61.9	66.7
Number in household							<0.001
Lives alone	15.5	15.6	23.7	10.4	12.5	10.5
Self +1	34.4	37.2	28.0	28.1	30.1	30.3
Self +2 or more	50.0	47.2	48.3	61.5	57.4	59.2
No. of household children							0.001
No children	61.1	63.6	63.2	58.1	52.7	51.3
1 child	17.4	15.8	17.8	26.2	20.2	21.2
2 or more children	21.5	20.6	19.0	15.8	27.2	27.6
Caregiver							<0.001
Yes	29.7	27.6	33.6	21.0	36.6	38.1
No	70.3	72.4	66.4	79.0	63.4	61.9
Region							<0.001
Midwest	20.7	24.3	14.3	9.5	16.9	8.7
Northeast	17.0	17.3	15.9	14.6	18.0	15.3
South	38.5	36.2	59.3	24.1	38.3	39.0
West	23.8	22.1	10.4	51.8	26.8	37.1
Self-rated health
Self-rated overall health							0.048
Excellent or Very good	43.3	43.4	43.3	44.4	45.9	32.5
Good	37.4	37.0	40.4	44.3	32.9	42.2
Fair or Poor	19.3	19.6	16.3	11.3	21.2	25.4
No. of comorbidities							<0.001
0	53.7	53.2	56.0	72.9	54.1	35.1
1	29.4	29.1	32.3	18.6	31.1	32.3
2	10.4	11.0	6.0	5.7	9.5	18.5
3 or more	6.6	6.7	5.8	2.8	5.2	14.0
Prepandemic health-related socioeconomic risks
No. of health-related socioeconomic risks							<0.001
0	56.3	61.8	43.9	49.5	44.7	49.3
1	21.6	20.9	26.9	24.2	20.4	21.4
2 or more	22.1	17.4	29.2	26.3	34.9	29.2

Racial/ethnic differences in prepandemic HRSRs

While the majority (62%) of White women reported no prepandemic HRSRs, the majority of Black, Hispanic, and E/SE Asian women reported one or more (56%, 55%, and 51%, respectively, p < 0.001) (Table 1). Prepandemic rates of food insecurity, the most common prepandemic HRSR for each racial/ethnic group, were more than 35% higher among Black (47%), E/SE Asian (45%), Hispanic (47%) versus White women (33%) (Fig. 1A). In adjusted analyses, Black, E/SE Asian, and Hispanic women had significantly higher odds of any prepandemic HRSR compared with White women (aOR 2.0, 95% CI 1.5–2.8; aOR 2.2, 95% CI 1.5–3.1; aOR 1.5, 95% CI 1.2–2.1, respectively) (Table 2), along with higher odds of food insecurity and transportation difficulties, specifically. In addition, compared with White women, Black women had significantly higher odds of prepandemic housing insecurity (aOR 1.9, 95% CI 1.3–2.8) and E/SE Asian and Hispanic women had significantly higher odds of IPV (aOR 2.0, 95% CI 1.2–3.2 and aOR 1.8, 95% CI 1.2–2.7).

FIG. 1.

Rates of prevalence (pre- and early pandemic, A) and early pandemic incidence and worsening (B) of HRSRs among U.S. women by race and ethnicity. Note: In Panel B, for each HRSR, the incidence rate was calculated as the number of women with an incident HRSR divided by the total number of women for whom that HRSR was absent prepandemic. The worsening rate was calculated as the number of women with a worsening HRSR divided by the total number of women for whom that HRSR was present prepandemic. HRSR, health-related socioeconomic risk.

Table 2.

Adjusted Odds of Prepandemic and Early Pandemic Changes in Health-Related Socioeconomic Risks Among U.S. Women by Race and Ethnicity (Non-Hispanic White Women as Reference Group)

	Non-Hispanic Black		Non-Hispanic East or Southeast Asian		Hispanic		Other
	aOR	95% CI	aOR	95% CI	aOR	95% CI	aOR	95% CI
Prepandemic HRSRs
At least one HRSRs	2.0***	1.5–2.8	2.2***	1.5–3.1	1.5**	1.2–2.1	1.3	0.8–2.0
Food insecurity	1.7**	1.2–2.2	2.3***	1.7–3.3	1.4*	1.0–1.8	1.1	0.6–1.7
Housing instability	1.9**	1.3–2.8	1.3	0.8–2.1	1.4	0.9–2.0	1.8	0.9–3.4
Transportation difficulties	1.8**	1.3–2.6	2.9***	2.0–4.4	2.0***	1.5–2.8	2.1*	1.2–3.7
Utilities difficulties	1.6*	1.1–2.4	1.5	1.0–2.5	2.0***	1.3–2.9	1.4	0.7–2.7
IPV	1.0	0.6–1.6	2.0**	1.2–3.2	1.8**	1.2–2.7	1.5	0.7–3.3
Early pandemic HRSRs
At least one incident or worsening HRSRs	1.3	1.0–1.7	1.6**	1.2–2.2	1.2	0.9–1.5	1.7	1.0–2.8
Food
Incident vs. Secure	0.7	0.4–1.2	0.8	0.5–1.3	0.8	0.5–1.3	1.4	0.7–2.7
Worsening vs. Persistent or improved	1.0	0.7–1.5	1.5	1.0–2.3	1.3	0.9–1.9	1.6	0.8–3.6
Housinga
Incident vs. Secure	1.0	0.5–1.8	0.8	0.4–1.7	1.7*	1.0–2.8	1.5	0.7–3.5
Transportation
Incident vs. Secure	2.1**	1.3–3.3	3.6***	2.2–6.0	2.0**	1.3–3.0	2.4*	1.1–5.2
Worsening vs. Persistent or improved	1.3	0.7–2.4	2.5*	1.1–5.6	0.6	0.4–1.1	0.8	0.3–2.0
IPV
Incident vs. Secure	1.0	0.7–1.5	1.2	0.7–2.1	1.3	0.8–2.0	1.6	0.8–3.1
Worsening vs. Persistent or improved	1.2	0.5–3.2	1.3	0.5–3.7	1.7	0.8–3.5	1.1	0.2–5.2

Non-Hispanic White women are the reference group. Logistic regression model covariates include: age (decades), income, educational attainment, marital status, number in household, number of household children, caregiver status, region, self-rated health, and number of comorbidities. HRSR change status was classified as: secure (risk was absent before and early pandemic), incident (absent prepandemic and present early pandemic), persistent or improved (present prepandemic and unchanged or improved early pandemic), and worsening (present prepandemic and worse early pandemic.

^a Due to small number of observations for the outcomes of worsening housing instability, incident utilities difficulties, and worsening utilities difficulties, multivariable logistic regressions were not conducted for these outcomes.

^* p < 0.05, ^**p < 0.01, ^***p < 0.001.

aOR, adjusted odds ratios; CI, confidence interval; HRSRs, health-related socioeconomic risks; IPV, interpersonal violence.

Racial/ethnic differences in early pandemic changes in HRSRs

Early pandemic prevalence rates of HRSRs were higher than prepandemic rates (with the exception of utilities difficulties among E/SE Asian women), lowest among White women and, for most conditions, highest among Hispanic women (Fig. 1A). More than two-thirds of Black (68%), E/SE Asian (68%), and Hispanic (68%) and more than half of White women (55%) had one or more HRSR in the early pandemic (p < 0.001). Racial/ethnic differences in early pandemic prevalence rates were significant for all HRSRs (p < 0.001).

Patterns of change between the prepandemic period and the early pandemic varied by HRSR type, but the rate of change was largely similar across racial/ethnic groups (Fig. 1A). Few women of any racial/ethnic group experienced incident utilities difficulties; however, rates of worsening utilities difficulties were twice as high among Hispanic and E/SE Asian women (19% and 15%) compared with White and Black women (7% and 6%) (Fig. 1B). More than 20% of women in each racial/ethnic group experienced incident food insecurity; about 1 in 10 experienced incident IPV (range: 9% of White women to 16% of Hispanic women) in the early pandemic. Rates of worsening food insecurity were very high among all racial/ethnic groups (range: 63% of White women to 72% of Hispanic women) as were rates of worsening IPV (range: 25% of E/SE Asian women to 40% of Hispanic women). Rates of worsening transportation difficulties were more than 20% higher for E/SE Asian women than for every other group (E/SE Asian: 78%, White: 60%, Black: 63%, and Hispanic: 49%).

Compared with White women, only E/SE Asian women had significantly higher odds of experiencing any incident or worsening HRSRs (aOR 1.6, 95% CI 1.2–2.2) in the early pandemic (Table 2). For each racial/ethnic group, the odds of experiencing an incident or worsening condition for each HRSR type were similar to White women with the exception of transportation. Compared with White women, women of all other racial/ethnic groups had significantly higher odds of experiencing incident transportation difficulties (Black: aOR 2.1, 95% CI 1.3–3.3; E/SE Asian: 3.6, 95% CI 2.2–6.0; and Hispanic 2.0, 95% CI 1.3–3.0). E/SE Asian women also had significantly higher odds of experiencing worsening transportation difficulties compared with White women (aOR 2.5, 95% CI 1.1–5.6).

Figure 2 illustrates patterns of co-occurring HRSRs by racial/ethnic group for the 10 most prevalent HRSR combinations during the early pandemic (73% of women had one of the 10 most prevalent HRSR combinations in their racial/ethnic group; Supplementary Appendix SA2 summarizes data for all 31 combinations that occurred at least once). Other than food insecurity, few women experienced other HRSRs in isolation. The most prevalent comorbid HRSRs among all racial/ethnic groups was combined food insecurity and transportation difficulties. Eleven percent of Black, 10% of E/SE Asian, and 8% of Hispanic women had both conditions, compared with 6% of White women. One in 36 Black women, 1 in 28 E/SE Asian women, and 1 in 19 Hispanic women had all 5 HRSRs (vs. 1 in 100 White women), an indicator of disparities in extreme health-related socioeconomic vulnerability.

FIG. 2.

Venn diagrams illustrating co-occurrence of HRSR by race and ethnicity among U.S. women with one or more HRSR in the early phase of the COVID-19 pandemic. Note: Venn diagrams represent the 10 most common patterns of comorbid HRSRs for each racial and ethnic group. Each circle is equivalent to the percentage of women included in the analysis with a given HRSR. Circles overlap when women report more than one HRSR. If a portion of a circle does not overlap with any other circles, it indicates that a proportion of women experienced that HRSR independent of all other HRSRs.

Discussion

This study of U.S. women fills gaps in knowledge about the early effects of the COVID-19 pandemic on racial/ethnic disparities in modifiable HRSRs among U.S. women. As hypothesized, HRSR rates were significantly higher among Black, E/SE Asian, and Hispanic women both before and early in the pandemic. White women were far less likely to experience extreme health-related socioeconomic vulnerability during the early pandemic. However, all groups experienced substantial early pandemic increases in HRSRs. With the exception of transportation difficulties, the pandemic did not differentially spare women of any racial/ethnic group from new/worsening HRSRs. Because these estimates were generated in the first several weeks after the WHO declaration of the pandemic, they are likely a conservative indicator of later pandemic conditions. Given the known differential impact of the pandemic on racial/ethnic minority groups in the United States—including ongoing pandemic-related violence and vicious targeting of E/SE Asian women—it is very likely that disparities in HRSRs among women have widened over time.

Food insecurity was the most prevalent HRSR both before and early in the pandemic. Rates were higher among Black, E/SE Asian, and Hispanic women. Prepandemic disparities in food insecurity have been reported by others, with prevalence rates of food insecurity among Black and Hispanic-led households twice that of White-led households.²⁷ Food insecurity data among E/SE Asian women in the United States are scarce.²⁸ In our sample, despite higher income and better self-rated health than other groups, the prepandemic food insecurity rate among E/SE Asians was nearly identical to rates among Black and Hispanic women. This finding may indicate that pandemic-related changes in HRSRs occurred among E/SE Asians earlier than other groups and before the March pandemic declaration. This finding may be explained by a surge in pandemic-related discrimination against E/SE Asians as early as January 2020.^11,29 Stigma and violent targeting could have caused E/SE Asian women to worry about or experience difficulty accessing food well before April 2020.

While the prevalence of early pandemic food insecurity was consistently higher among Black, E/SE Asian, and Hispanic women compared with White women, the adjusted odds of pandemic-related incident or worsening food insecurity were similar. This finding corroborates another study (conducted April–May 2020) that found Black, Asian, and Hispanic-led households were not significantly more likely to be food insecure in the early pandemic than White-led households when controlling for prepandemic food insecurity.³⁰ Food insecurity interventions, such as expanding/extending SNAP benefits during the pandemic, should be informed by the observation that many women with early pandemic food insecurity had no recent experience with this condition and most had comorbid HRSRs.¹ For E/SE Asian women, especially those with no prior history of food insecurity, internalization of the “model minority” myth³¹ may be a barrier to seeking assistance with food insecurity and other HRSRs.³² Safety concerns may also present a barrier for this group—prepandemic IPV rates were high among E/SE Asian women in our study. Lastly, the 2019 public charge rule,³³ which allowed for denial of lawful permanent residency based on the likelihood that applicants will rely on public cash assistance or government support, was a barrier to SNAP access for non-U.S. citizens (more likely among Hispanic and E/SE Asians). Support for culturally competent community-based organizations is likely needed to mitigate rising rates of food insecurity and comorbid HRSRs among racial/ethnic minority groups.³⁴

We found no racial/ethnic disparities in incident or worsening housing insecurity, although the odds of incident housing insecurity trended higher among Hispanic women as compared with White, Black, and E/SE Asian women. Relative stability in housing and utilities during the early pandemic may be explained by widespread implementation of moratoria on evictions, foreclosures, and utility shutoffs beginning in March 2020.^35–37 Some reports conducted between April and August 2020 indicate that Black, Asian, and Hispanic adults/households had higher rates of housing hardship during the pandemic compared with Whites.^38–40 Chun and Grinstein-Weiss reported similar rates of housing difficulties among White, Black, and Hispanic men and women in May 2020 that widened in subsequent months.⁴¹ These studies measured housing insecurity by on-time rent/mortgage payments or eviction/foreclosure events. In contrast, we measured participants' concern about a steady place to live. Our early pandemic estimates of housing insecurity were higher than those reported for White, Black, and Hispanic respondents by Chun and Grinstein-Weiss; this study did not report data for Asians.⁴¹

While we did not find statistically significant racial/ethnic group differences in incident or worsening IPV, the early pandemic prevalence rate of IPV was about twice the prepandemic prevalence. Hispanic and E/SE Asian women were twice as likely as White women to report prepandemic IPV; these groups also had the highest incidence rates in the early pandemic. Few studies have reported on IPV (a construct that includes, but is broader than, intimate partner violence). Furthermore, in spite of evidence showing higher rates of fatal police encounters among Black and Hispanic people before and during the early pandemic,⁴² few scientific studies have generated early pandemic estimates for IPV by race/ethnicity or gender. Concordant with our early pandemic estimates for the entire sample (19%) and for White women (16%), a web-based survey of women and men found an 18% overall prevalence of intimate partner violence in April 2020 and reported no difference in worsening intimate partner violence between Hispanic and non-Hispanic respondents.⁴³ Another study analyzed domestic violence calls in 14 large U.S. cities, finding a 10% increase following social distance recommendations in March 2020, but no relationship between call rate and census tract percentage of Black or Hispanic residents.⁴⁴

Because IPV includes, but is not limited to, domestic violence, these findings may not be inconsistent with our observations. It is possible that Hispanic and E/SE Asian women were especially prone in the early pandemic to violence that occurred outside the home. Furthermore, Hispanic and E/SE Asian women without U.S. citizenship may have been less likely to seek domestic violence assistance due to fear of deportation and would, therefore, have been under-represented in studies reporting IPV rates.⁴⁵ Contemporaneous with the timing of our study (April 2020), Stop Asian American and Pacific Islanders (AAPI) Hate recorded 1,497 acts of pandemic-related discrimination (e.g., verbal harassment, shunning, physical assault) against AAPIs, 76% of which were reported by people identifying specifically with the E/SE Asian groups included in our sample.¹⁰ Anti-Asian racism has persisted beyond the early pandemic, suggesting that our results are likely a conservative estimate of current conditions. By the end of 2020, Stop AAPI Hate had documented 2,808 discrimination acts, with rates two to three-fold higher among women.⁴⁶

Importantly, our findings add new evidence of racial/ethnic disparities in transportation difficulties early in the pandemic. These disparities are likely attributed, at least partly, to disinvestment in transportation infrastructure in communities of color.^47,48 Moreover, Black, Asian, and Hispanic workers are less likely to have a home vehicle, more likely to use public transit and have longer commutes than White workers in the United States.⁴⁹ Once the pandemic began, existing transportation barriers in combination with social distancing practices and reduced public transit schedules and other transit options likely contributed to exacerbated disparities in transportation difficulties.⁵⁰ In addition to higher odds of new transportation difficulties, E/SE Asian women with prepandemic transportation problems had higher odds of worsening transportation difficulties compared with White counterparts. Structural access disparities may have been compounded by fear of, or actual, discrimination or violence by other riders or drivers.¹⁰ Findings underscore the need to capture detailed data on transportation access by Asian subgroups, with special concern for E/SE Asian subgroups, who have been disproportionately stigmatized and targeted by the pandemic.

Rising transportation difficulties and disparities among women could pose several challenges to COVID-19 mitigation and recovery, including differential access to nutritious food, COVID-19 testing, vaccination, and care.³ Given high reliance of Black, Asian, and Hispanic workers on public transit,⁴⁹ response efforts should ensure that public transit adequately serves communities in high need during the pandemic. To address pandemic-related transportation difficulties, some community-based organizations have offered free, contactless delivery of food and medicine in addition to rides of necessity (e.g., medical appointments, grocery).⁵¹ Trusted organizations with transportation capabilities could help offset COVID-19 disparities especially by serving Black, E/SE Asian, and Hispanic women. Transportation assistance is more effective when offered in conjunction with other interventions addressing socioeconomic barriers.⁵²

These results should be interpreted in light of some limitations. Generalizability of the results may be limited by using a quota sampling strategy rather than a probability sample of the population, although estimates from this sample are on a par with those from probability samples.¹ Our approach allowed for rapid enrollment of a large, national sample to assess early pandemic effects and enabled oversampling of the overlooked, but important, subgroups of E/SE Asian women. Although validated measures were used whenever possible, change in HRSR status was measured using adapted survey items. Oversampling E/SE Asian subgroups resulted in fewer women in other Asian subgroups for subgroup analysis. In our sample, HRSR patterns in the heterogeneous “other” group typically reflected those of the non-White groups, suggesting disadvantage worthy of disaggregated analysis in future studies. Exclusion of non-English speakers likely also biases estimate of disparities in HRSRs toward the null. Prepandemic HRSRs were reported retrospectively. Early pandemic HRSR status may have influenced recall or reporting of prepandemic HRSR status. While the sample design accounts for bias associated with the variables used for calibration, it cannot account for all selection bias. Last, we did not collect data on occupations of study participants and therefore are not able to stratify by or control for their status as essential or other workers.

These findings from the National Women's Health COVID-19 Study identify early pandemic-related increases in HRSRs and persistent racial/ethnic disparities among U.S. women in HRSR domains critical to response and recovery efforts.⁵³ These domains are modifiable at the community level with culturally relevant community-based supports and at the policy level with interventions (e.g., state-level moratoria on utilities stoppage and eviction; expansion/extension of SNAP). Transportation difficulties may be an overlooked contributor to COVID-19 disparities among women. Ensuring equitable access to safe transportation during the pandemic could offset incidence and severity of other HRSRs—like food insecurity and IPV—and promote more equitable access to SARS-CoV-2 testing, vaccination, and other health care services for all women and their families.

Data Availability Statement

The data that support the findings of this study are available upon reasonable request and with the approval of the University of Chicago Institutional Review Board (IRB20-0489). The data are not publicly available as it contains information that could compromise the privacy of research participants. Please contact the senior author, STL, with any request for data access.

Author Disclosure Statement

Dr. Lindau is founder and co-owner of NowPow, LLC and president of MAPSCorps, 501c3. Neither the University of Chicago nor the University of Chicago Medicine is endorsing or promoting NowPow or MAPSCorps business, products, or services. Dr. Lindau and her spouse own equity in Glenbervie Health, LLC and healthcare-related stocks and mutual funds managed by third parties. Dr. Lindau and Dr. Haider are contributors to UpToDate. All remaining authors have no conflict of interest or disclosure.

Funding Information

Research reported in this publication was supported by 5R01AG064949, 5R01MD012630, R21CA226726 (S.T.L., J.A.M., E.A.P., M.V., K.E.W., and V.A.W.), and 5F31CA243220-02 (M.V.). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. Material support for this survey was provided by the Joseph B. De Lee Endowment to the University of Chicago Department of Obstetrics and Gynecology.

Supplementary Material

Supplementary Appendix SA1

Supplementary Appendix SA2