Frontline work and racial disparities in social and economic pandemic stressors during the first COVID‐19 surge

Abstract Objective To assess the magnitude of racial–ethnic disparities in pandemic‐related social stressors and examine frontline work's moderating relationship on these stressors. Data Sources Employed Californians' responses to the Institute for Governmental Studies (IGS) poll from April 16–20, 2020, were analyzed. The Pandemic Stressor Scale (PSS) assessed the extent to which respondents experienced or anticipated problems resulting from the inability to pay for basic necessities, job instability, lacking paid sick leave, unavailability of childcare, and reduced wages or work hours due to COVID‐19. Study Design Mixed‐effects generalized linear models estimated (1) racial–ethnic disparities in pandemic stressors among workers during the first COVID‐19 surge, adjusting for covariates, and (2) tested the interaction between race–ethnicity and frontline worker status, which includes a subset of essential workers who must perform their job on‐site, to assess differential associations of frontline work by race–ethnicity. Data Collection The IGS poll data from employed workers (n = 4795) were linked to the 2018 Centers for Disease Control and Prevention Social Vulnerability Index at the zip code level (N = 1068). Principal Findings The average PSS score was 37.34 (SD = 30.49). Whites had the lowest PSS score (29.88, SD = 26.52), and Latinxs had the highest (50.74, SD = 32.61). In adjusted analyses, Black frontline workers reported more pandemic‐related stressors than White frontline workers (PSS = 47.73 vs. 36.96, p < 0.001). Latinxs reported more pandemic stressors irrespective of frontline worker status. However, the 5.09‐point difference between Latinx frontline and non‐frontline workers was not statistically different from the 4.6‐point disparity between White frontline and non‐frontline workers. Conclusion Latinx workers and Black frontline workers disproportionately reported pandemic‐related stressors. To reduce stress on frontline workers during crises, worker protections like paid sick leave, universal access to childcare, and improved job security are needed, particularly for those disproportionately affected by structural inequities, such as racially minoritized populations.

worker protections like paid sick leave, universal access to childcare, and improved job security are needed, particularly for those disproportionately affected by structural inequities, such as racially minoritized populations.

K E Y W O R D S
determinants of health, health equity, population health, racial/ethnic differences in health and health care, social determinants of health, socioeconomic causes of health What is known on this topic • Pandemic-related stressors are a key driver of adverse mental health outcomes among the general public during the COVID-19 pandemic.
• Work is a social-contextual factor that shapes the degree to which individuals are exposed to pandemic-related stressors, and frontline work escalates occupational demands and limited workers' agency.
• Because of pre-existing structural inequities in employment and other determinants of health, Black and Latinx individuals are disproportionately employed as essential frontline workers in precarious jobs.

What this study adds
• Black frontline workers reported more pandemic-related stressors than White frontline workers, resulting from a lack of paid sick leave, unavailability of childcare, job insecurity, reduced wages/work hours, and financial hardships.
• Frontline work did not moderate the degree to which Latinx workers experience pandemic stressors, as Latinxs reported greater pandemic-related stressors irrespective of having a frontline position.
• Our findings remained even after accounting for individual-level socioeconomic position and area social vulnerability, underscoring the intersecting roles of racism and precarious work in stressors.

| INTRODUCTION
In 2021, COVID-19 became the third leading cause of death in the United States (US) and disproportionately reduced life expectancy among Black and Latinx groups, 1 Along with disproportionate death rates among racially marginalized groups, 2 the COVID-19 pandemic has also had deleterious population-level mental health effects. 3,4 Recent studies document that pandemic-related stressors across various life domains are major drivers of adverse mental health outcomes, including depression and anxiety. 5,6 However, empirical studies assessing pandemic-related stressors have focused on the general public or people with disabilities and have not examined racial-ethnic differences in stressors. Moreover, pre-existing economic inequality, occupational characteristics, and other social determinants attributed to racial disparities in COVID-19 health outcomes may also have led to differential exposure to pandemic-related stressors. 7,8 In particular, frontline workers from marginalized racial-ethnic groups may be more vulnerable to experiencing social stressors, which are principal and understudied mechanisms of health disparities. 9 Because of structural inequities in employment and other determinants of health, certain racial and ethnic groups are disproportionately employed as essential frontline workers in precarious jobs. 10,11 Sociological studies of the stress process model have established that individuals belonging to structurally disadvantaged groups, including racialized minorities, are more likely to report adverse life events, anticipatory stressors, and chronic strains. [12][13][14] The social gradient in stress exposure stems from the unequal distribution of opportunities, risks, and resources. 15 Because racial-ethnic minority groups, like Latinx and Black individuals, have fewer flexible resources (i.e., social connections, knowledge, prestige, and power), they are more vulnerable to a broad range of stressors and are more likely to appraise an event as stressful, even after accounting for socioeconomic position. [13][14][15][16] Work, especially during the pandemic, is a critical social-contextual factor that shapes the degree to which individuals are exposed to stressors. 13 Without explicitly examining work arrangements, the role of work in racial disparities of pandemicrelated stressors may be masked because risks and hazards associated with particular jobs are interrelated with other axes of disadvantage. 17,18 Job stratification and occupational disadvantage along racialethnic lines 19 shape racially marginalized workers' income, conditions on and off the job, and the ease with which occupations are entered. 20,21 Black and Latinx workers are disproportionally employed in lower-status, precarious, and low-paying jobs, where they are more likely to face employer resistance to implementing safety measures 22 and exposures to physical hazards and psychosocial risks. 23 Work experiences during COVID-19, the types of jobs held (occupational segregation), and the distribution of risks and resources within the workplace (worksite segregation) are also patterned by race and ethnicity. 24 Given that Black and Latinx workers are substantially more likely than White and Asian workers to be employed in low-wage jobs characterized by precarious working conditions, 19 they may disproportionately experience pandemic-related stressors.
Frontline work during the pandemic escalated occupational demands and limited worker agency over where work could be completed (i.e., home vs. work settings). [25][26][27] Frontline workers, a subset of essential workers who must leave their homes to perform their job on-site, have on average lower education levels, have lower wages, and include a larger representation of disadvantaged groups. 28 The pandemic also highlighted disempowered workers who may not have another option but to work in person due to economic constraints. 18,29 As such, frontline work may moderate COVID-19 pandemic-related stressors and inflict social and economic costs on racial-ethnic minorities due to lower access to psychosocial and economic resources and fewer employer-based benefits. The lack of work flexibility, limited worker autonomy, and greater demands in jobs where marginalized racial-ethnic groups are overrepresented are potential sources of the diverging experiences for frontline workers of color compared to their White counterparts. 18 Anticipation of challenges resulting from economic shocks has been previously associated with depressive symptoms 30 ; such differential exposure to anticipatory and acute stressors can exacerbate mental health disparities. 15,31 Social and economic stress can lead to biological (e.g., hypothalamic-pituitary-adrenal axis dysregulation), psychological (e.g., distress), and unhealthy coping (e.g., sleep problems, alcohol use) responses that pose deleterious consequences for physical and mental health. 13,32,33 In the context of the COVID-19 pandemic, inequities in risk factors, such as those posed by frontline work, may interact with existing racial inequalities to create diverging experiences that exacerbate stressors for Latinx and Black workers.
To examine whether racial differences in pandemic-related stressors existed among workers during the first surge of the COVID-19 pandemic, we analyzed California voter responses to the UC Berkeley Institute for Governmental Studies (IGS) April 2020 poll (n = 4795). We hypothesize that Hypothesis 1. Latinx and Black workers will report more pandemic-related stressors because they have fewer flexible resources than White workers, even after accounting for individual-level socioeconomic position and area-level social vulnerability, and Hypothesis 2. Latinx and Black frontline workers will report more pandemic-related stressors than White

| Data
We analyzed data from the UC Berkeley IGS April 2020 poll of California registered voters. The IGS poll is a recurring web-based survey of California registered voters to assess public opinion about public policies, economic trends, and social issues. The poll assessed COVID-19 experiences and was administered in English and Spanish from April 16 to 20, 2020. Email invitations were sent to stratified random samples of California's registered voters and could be completed via cellphone or computer. The overall sample was stratified by age, gender, race-ethnicity, and language to obtain a proper balance of survey respondents across major segments of the state's registered voter population. A total of 150,000 respondents were invited to participate in the survey; 11,502 participated, yielding a 7.7% response rate. 38 This response rate is consistent with other polls of registered voters. 39 The completion rate was 76.2%, with 8795 people completing the entire survey. Post-stratification weights were applied to align the sample of registered voters to the population characteristics of the state's registered voters based on age, race-ethnicity, gender, education, California region of residence, and party affiliation.

| Dependent variable
The Pandemic Stressor Scale (PSS), a 5-item composite that assessed social and economic conditions reported as problematic because of the COVID-19 pandemic, is the study's dependent variable. Because the first surge of the COVID-19 pandemic was a societal-level economic and social shock, it disrupted multiple domains of everyday life beyond health. 40 The PSS assesses social stressors associated with economic security, family needs, and employment. Respondents were asked, "For each of the following, please indicate the degree to which each is a problem that you expect to face-or are already facing-as a result of COVID-19": (1) "Not being able to pay for basic necessities (i.e., food, medication, rent/mortgage)," (2) "Losing my job," "(3) Lacking paid sick leave," (4) "Not being able to get childcare," (5) "Reduced wages or work hours." Each item is scored using a Likert scale in which 0 corresponds to no problem at all, 33.3 to not much of a problem, 66.6 to a serious problem, and 100 to a very serious problem. Item scores were averaged to construct the composite measure (range: 0-100; α = 0.84), and higher scores indicate more pandemic-related social stressors. Exploratory factor analysis of the items yielded an eigenvalue of 2.7 with no other factors with eigenvalues greater than 1.0, and the single factor explained 93.52% of the variation, supporting the use of the composite measure. Factor loadings and Cronbach's alpha and inter-item correlations for the PSS are presented in Tables A1 and A2, respectively.

| Independent variable
Each participant self-reported their race-ethnicity by answering the following question: "Are you White or Caucasian, Latinx/Hispanic, Black or African-American, Asian/Pacific Islander, American Indian/Alaskan Native, or a member of another race (i.e., multiracial or other)?" We conceptualize race-ethnicity as social categories that shape the distribution of risks, resources, and opportunities. 41 It is also important to note that while race and ethnicity are two interrelated yet distinct constructs, for the purposes of this paper, we have combined them into a single construct because we focus on a single racial group (Black people) and a single ethnic group (Latinx people). Respondents identifying as American Indian/ Alaskan Native, or multiracial were combined into the "Other Race" category due to the small sample size. Our racial-ethnic categories include Asian/Pacific Islander (PI), Black, Latinx, Other Race, and White. We use "White" as the reference category given this group's larger sample size and historical advantage in social resources and work opportunities relative to racially minoritized groups.

| Moderating variable
Frontline work was examined as a moderator of the relationship between race-ethnicity and pandemic-related stress because of the opportunity for increased worker demand and limited autonomy during the pandemic. We focus on frontline workers rather than essential industries because of the considerable variation among local jurisdictions in the definition of essential industries compared to the relative permanency of frontline positions. Frontline workers are defined as employed individuals who must leave their homes to perform their job on-site and are in contact with other people regardless of their employment industry. Performing frontline work was assessed using a single item asking employed respondents: "Which best describes your workplace since the California statewide shelter-in-place went into effect?" Response options included: (1) "I am able to work from my home," (2) "I leave home to go to work, but my job involves some contact with other people," and

| Control variables
We controlled for demographic variables including nativity (foreignborn vs. US-born), sex (male/female), age (continuous), marital status (single, widowed/separated/divorced, and married/cohabiting), living with a person who is 65 years of age or older (yes or no), and having children in the household (yes or no). Participants' political party affiliations may influence self-reported pandemic stressors due to the pandemic's politicization 42 ; thus, we controlled for political party affiliation (Democrat, Republican, Independent, or Other).
Similar to salient studies, 43,44 we adjusted for indicators of individual-level socioeconomic position (SEP) and zip code social vulnerability to determine whether racial disparities in stressors remain.
Previous studies indicate that individual-and area-level socioeconomic disadvantage do not fully account for racial differences in an array of health and social outcomes. 43,45 As a fundamental cause, racism also influences health and social outcomes through personallymediated racism and race-based differences in flexible resources such as prestige, power, and freedom. 46 For individual-level SEP, we account for median education level (at least some college vs. bachelor's degree or higher), pre-pandemic income divided at the approximate median income for a single earner in California (<$59,999 or ≥$60,000), and work industry. We classified work industry categories as (1) health care (all occupations), (2) service-based, manual, or blue-collar (hotel and hospitality, retail, agriculture, transportation and utility, construction, manufacturing, delivery services, personal care services, restaurants/bars, janitorial, and landscaping), and (3) professional, white-collar (professional and business, informational technology, finance and accounting, government services, and education). Frontline workers in the health care industry may have unique experiences from other sectors, so this category was maintained separately. Zip code social vulnerability was assessed using the CDC's Social Vulnerability Index (SVI). 47 The SVI comprises four indices (i.e., socioeconomic status, household composition and disability, minority status and language, and housing type and transportation) to construct a composite measure of community susceptibility in the occurrence of a societal shock or emergency. 48      *p < 0.05; **p < 0.01; ***p < 0.001, two-tailed significance tests.

| Descriptive analysis
In Table A3, we show the weighted percentage of frontline and non-frontline workers reporting each of the five stressors (i.e., job insecurity, loss of wages/hours, lack of paid sick leave, unavailability of childcare, and not being able to pay for basic needs) by race-ethnicity.  Control variables were also positively associated with pandemicrelated stressors across the regression models, including nativity (being foreign-born) and having children in the household. Republican/Independent political party affiliation, a college degree or higher, and a pre-pandemic income of $60,000 or higher were negatively associated with pandemic-related stress.

| Moderation analysis
In Model 5, the interaction between performing frontline work and race-ethnicity, which represents the second partial derivative with respect to both frontline work and race-ethnicity, indicates that front- White frontline workers and White non-frontline workers.

| Sensitivity analyses
In the regression model that includes the interaction between raceethnicity and frontline work but does not adjust for SVI (shown in Table A5), the direction, magnitude, and significance of the associations were similar to those in the fully adjusted Model 5 reported in Table 2. In this sensitivity analysis, the coefficient of the interaction for Black frontline workers was statistically significant, and the coefficient of the main effect for Black workers remained insignificant.

| DISCUSSION
Differential exposure to stress is a key pathway explaining racial health disparities. 13,32 However, empirical research about the distribution of social and economic stressors, whether experienced or anticipatory, across racial-ethnic groups remains limited, 14  We found partial support for Hypothesis 1 that Latinx and Black workers would report more pandemic-related stressors than their Indian/Alaskan Native and multiracial groups as independent categories due to their small sample size. In future research, analyzing within-group differences by race-ethnicity may help us better understand the antecedents and consequences of pandemic-related stressors.

| CONCLUSION
Latinx workers and Black frontline workers were more likely to report experienced and anticipatory pandemic-related stressors during the COVID-19 first surge. Because the consequences of the pandemic may be long-lasting, examining the racial differences in social and economic stress exposure emanating from this shock is critical to creating social policies that bolster vulnerable populations against future pandemics and social crises. Given that Black frontline workers and all Latinx workers experienced and anticipated more problems due to a lack of paid sick leave, limited access to childcare, and precarious con- Timothy T. Brown https://orcid.org/0000-0002-5334-0768 Adrian Aguilera https://orcid.org/0000-0003-1773-8768 Hector P. Rodriguez https://orcid.org/0000-0002-6564-2229