Abstract
Emerging research documents concerning mental health outcomes among essential workers at the start of the COVID-19 pandemic. However, mental health outcomes may have varied across states in the United States, as state-level policies differed. Questions also remain about the mental health of workers during the second year of the pandemic. Using nationally representative data from the U.S. Household Pulse Survey (April–July 2021), we documented the mental health of essential workers and tested whether state-level policies (e.g., mask mandates) reduced mental health disparities for essential workers. Results show that food and beverage essential workers experienced heightened anxiety and depression relative to nonessential workers. Moreover, for food and beverage workers, disparities in mental health were smaller in states with mask mandates, expanded paid leave, and higher minimum wage compared to states without these policies. The present study points to the potential for state-level policies to protect the mental health of essential workers.
Keywords
Essential workers—workers who continued to work in person and are critical to the functioning of society—experienced unique challenges and strains during the COVID-19 pandemic, and these stressful experiences had the potential to undermine well-being. Indeed, studies documented worse mental health among healthcare and other essential workers compared to nonessential workers around the world during the first months of the pandemic (e.g., Bell et al. 2021; Grooms et al. 2022; Mayer et al. 2022). An important consideration is that the mental health of essential workers may vary across states in the United States as states implemented different policies that had the potential to affect essential workers. This possibility is especially likely after the first year of the pandemic when sectors of the economy reopened amid relatively high rates of COVID-19 cases in the United States and states took different approaches to balance economic and public health concerns. For example, despite the persistent threat of the pandemic, Texas suspended a statewide mask mandate in March 2021, while other states (e.g., Vermont and California) maintained their mask mandates throughout the summer of 2021, likely offering protection to essential workers. The possible role of state-level policies in shaping the mental health of essential workers has not been tested in prior research.
In the present study, we use nationally representative data from the Household Pulse Survey (HPS; April–July 2021) to document the mental health of essential workers and to test whether state-level policies weakened the mental health consequences of essential work. We focus specifically on essential workers in healthcare, food and beverage, and manufacturing settings, in addition to other essential jobs (e.g., K–12 school, first responders, public transit), in comparison to nonessential workers. We blend two theoretical frameworks—the stress process model (Pearlin et al. 1981) and a socioecological framework of population health focused on state contexts (Montez 2020)—to consider the role of individual- and state-level factors for mental health. State policymakers took different approaches to safety measures and economic security efforts during the pandemic, resulting in a political landscape that varied substantially according to one’s state of residence. We hypothesize that essential workers who lived in states that implemented policies aimed at safety and financial security experienced better mental health outcomes compared to essential workers in states that created a riskier environment for in-person workers. Given significant heterogeneity in the essential worker population (McNicholas and Poydock 2020), the greater burden of stress among less resourced individuals (Pearlin et al. 1981), and the salience of policies for the health of less educated adults (Montez et al. 2019), we consider how linkages between essential work, state-level policies, and mental health are experienced differently depending on high versus low levels of educational attainment.
This study makes several contributions to the literature. First, we rely on a large, nationally representative sample to better understand the mental health experiences of essential workers in several different sectors (i.e., healthcare, food and beverage, manufacturing, and others) and across high and low levels of educational attainment. Second, we document rates of depression and anxiety among essential and nonessential workers at the start of the second year of the pandemic in the United States (April–July 2021) when stressful experiences likely persisted or increased for many essential workers. Third, we examine whether experiences of depression and anxiety among essential workers are conditioned by specific state-level policies. By situating the stress process model within the context of state environments, we suggest that stressful experiences—such as working an essential job during a pandemic—may be experienced differently depending on the state sociopolitical environment.
Understanding the mental health outcomes of essential workers is crucial to addressing the mental health crisis sparked by the COVID-19 pandemic. Depression and anxiety undermine well-being and erode quality of life. Moreover, prolonged distress can have long-term consequences, including subsequent health problems and premature mortality (Domingue et al. 2021; Thoits 2010; Walker, McGee, and Druss 2015); thus, elevated rates of depression and anxiety among essential workers could fuel health problems among this population in the future. An examination of the role of state-level policies points to specific areas of intervention to improve the mental health of workers in future public health crises.
Background
Essential Work and Mental Health
The category of “essential worker” emerged in the United States in response to the COVID-19 pandemic when many sectors of the economy shut down or moved online. Essential workers became those workers who continued to work in person given their “essential” or critical nature. The Department of Homeland Security defines the essential workforce as those who provide public health and safety, essential products, and other infrastructure support (Cybersecurity & Infrastructure Security Agency, N.d.). Essential workers are employed in sectors such as healthcare, grocery stores, manufacturing, food/agriculture, and transportation and comprise a large share of the workforce in the United States—up to 70 percent of all workers, according to some estimates (Blau, Koebe, and Meyerhofer 2021). Because essential workers comprise a large share of workers across numerous industries, there is considerable heterogeneity in this population. For example, doctors, nurses, and janitorial staff are all critical to the functioning of hospitals and medical care facilities. On the other hand, “nonessential” workers are typically considered to be workers whose jobs could be performed remotely (e.g., administrative positions) or those whose jobs were categorized as not critical to society in the context of the pandemic (Cybersecurity & Infrastructure Security Agency, N.d.). In the present study, we identify essential workers as individuals who worked outside of their home. We focus specifically on settings that exposed workers to a higher risk of COVID-19 and represented some of the largest sectors of the essential workforce including healthcare, food and beverage stores, and manufacturing (food and nonfood), as well as workers in other jobs deemed “essential” during the pandemic.
Importantly, prior research documents higher rates of depression and anxiety among essential workers compared to nonessential workers during the first months of the pandemic (e.g., Bell et al. 2021; Grooms et al. 2022; Mayer et al. 2022). Existing studies largely rely on smaller, nonrepresentative samples and often focused on experiences of mental health in the first months of the pandemic. Thus, we have a gap in knowledge about the experiences of essential workers in the second year of the pandemic when businesses in the United States began to reopen. The reopening of sectors of the economy in the second year of the pandemic resulted in increased interactions with the public for essential workers, necessitating an examination of the mental health outcomes of workers during this time. Moreover, because the essential workforce comprises a diverse group of workers (e.g., Blau et al. 2021; McNicholas and Poydock 2020), we fill a gap in the literature by considering differences in mental health outcomes among workers with higher and lower levels of education.
The stress process model provides a theoretical framework to guide the present study. Broadly speaking, the stress process model focuses on the sources, mediators, and consequences of stress (Pearlin et al. 1981). That is, a constellation of acute and chronic stressors can create challenges and hardships that undermine mental health, and individuals can harness coping resources such as a sense of control or social support to mitigate the adverse effects of stress (Pearlin et al. 1981). Recent research highlights the usefulness of the stress process model for understanding experiences during the COVID-19 pandemic by documenting the adverse mental health consequences of pandemic stressors (e.g., Moen 2022), including cumulative pandemic stressors (Louie, Upenieks, and Hill 2023), and the benefits of coping resources for combatting stress (e.g., Grace 2023; Grace and VanHeuvelen 2022; Louie et al. 2023).
The stress process model further emphasizes that stressors are not equally distributed in the population and tend to impinge on those with less power and privilege in society, fueling disparities in mental health (e.g., Pearlin et al. 1981; Pearlin et al. 2005; Turner et al. 1995). That is, the unequal distribution of exposure to stress contributes to status-based inequities in health and well-being. We theorize essential workers as a group with less power and privilege in society, vulnerable to an unequal burden of stress and the mental health consequences therein. Moreover, essential workers with lower levels of educational attainment may have a greater stress burden and fewer resources to cope with such stress if they experienced disadvantage due to their work and education statuses. As such, essential workers with less education may have experienced a heightened risk of depression and anxiety during the COVID-19 pandemic.
Essential workers likely experienced high levels of stress derived from several sources during the pandemic. For example, essential workers typically experienced many interactions with people outside their household, which could increase their risk of COVID-19 infection, hospitalization, and death. In fact, excess mortality in California (Chen et al. 2021) and Minnesota (Karnik et al. 2023) during the pandemic was highest in sectors generally deemed essential such as food service, transportation, construction, and manufacturing. The increased risk of exposure to COVID-19 at work may have been a source of stress for essential workers who worried about infection and/or spreading the COVID-19 virus to people in their household. Indeed, Woods, Schneider, and Harknett (2023) found that fewer workplace safety practices (e.g., masking, social distancing) were associated with higher levels of psychological distress among essential workers. Essential workers also experienced additional stress at work if they were tasked with the responsibility of enforcing mask wearing in businesses or if they experienced harassment from increasingly volatile and tense public citizens (Hammonds, Kerrissey, and Tomaskovic-Devey 2020). For example, healthcare workers, airline crewmembers, teachers, and school administrators became targets of verbal and physical harassment at work during the pandemic (Dye et al. 2020; Federal Aviation Administration, N.d.; Kurtz 2021). These pandemic-specific sources of stress may have contributed to high rates of depression and anxiety during the pandemic.
In addition to the risk derived from the pandemic, essential workers are disproportionately from disadvantaged groups and may experience a greater stress burden due to this disadvantaged status. For instance, because essential workers, on average, have lower wages than nonessential workers (Blau et al. 2021), financial strain may contribute to mental health challenges (Kahn and Pearlin 2006; Mirowsky and Ross 2001; Pearlin and Bierman 2013). Indeed, low wages may be especially pernicious during the pandemic. For example, limited financial resources may prevent workers from staying home when sick or quitting a job due to personal or family risk factors for COVID-19 complications. The stress from an inability to take time off from work may heighten adverse mental health responses. Given the constellation of stressors that likely impinged upon essential workers (e.g., COVID-19 exposure, harassment, financial strain), we test the following specific hypothesis:
We focus specifically on essential workers in healthcare, food and beverage, and manufacturing settings, in addition to other essential jobs (e.g., K–12 school, first response, public transit), in comparison to nonessential workers. We chose these settings (i.e., healthcare, food and beverage, manufacturing) due to their heightened exposure to COVID-19, excess mortality, and large share of the essential workforce. Indeed, healthcare and food-related sectors together comprise about half of essential workers (McNicholas and Poydock 2020). Healthcare workers experienced high levels of job stress, exhaustion, and possible COVID-19 exposure during the pandemic, and prior research documents their adverse mental health experiences in 2020 (e.g., Bell et al. 2021; Cabarkapa et al. 2020; Grooms et al. 2022). Food and beverage workers are an important group to study in the second year of the pandemic as restaurants and grocery stores became sites of conflict in the debate over how to reopen the country amid relatively high COVID-19 case rates. We also focus on workers in manufacturing in light of emerging data on high excess mortality within this industry (Chen et al. 2021)—a possible source of stress for workers in manufacturing settings. Moreover, a record number of workers quit their jobs in healthcare, food services, transportation, and other essential sectors in 2021 (Bureau of Labor Statistics 2022), perhaps signaling the stressful and unfavorable conditions of these jobs during the pandemic.
Essential Work and Mental Health: The Role of State Policies
We extend the stress process model by situating individual-level processes of stress and mental health within the context of the broader state environment. This approach essentially blends the stress process model with a socioecological framework of health, which acknowledges the central role of context for population health (McLeroy et al. 1988; Montez 2020; O’Campo and Dunn 2012). Prior research on the stress process has considered how neighborhood environments shape exposure to stress and the consequences of stress for mental health (e.g., Aneshensel 2009; Schieman, Pearlin, and Meersman 2006; Wheaton and Clarke 2003). That is, prior research notes the interdependence of individual- and neighborhood-level factors. Although neighborhood environments are fairly proximate determinants of health with robust effects on numerous health outcomes, we contend that more distal determinants of health should also be considered. Indeed, we argue for the need to situate the stress process model within a more macro-level context: the state. State policies have the potential to influence the health and well-being of individuals, including the mental health of essential workers. Therefore, we add to the voluminous literature on the stress process model by drawing attention to the role of state-level policies for individual-level experiences.
Conceptually, we suggest that state-level policies may support essential workers and reduce their distress during the COVID-19 pandemic. Prior international research (e.g., Kalleberg 2018; Kim et al. 2012) and emerging state-level research in the United States (e.g., Donnelly and Farina 2021; Donnelly and Schoenbachler 2021) suggest that policies have the potential to mitigate the mental and physical health consequences of stressful work experiences. For example, evidence harnessing variation in the United States shows that state-level policies reduced the mental health consequences of income loss—an acute work-related stressor—during the COVID-19 pandemic (Donnelly and Farina 2021). Recent research also finds that policies aimed to reduce financial hardship such as credit payment holidays (Sparkes, Wang, and Wels 2023) and eviction moratoria (Boen et al. 2023) can mitigate adverse mental health outcomes. We apply this framework to investigate whether state-level policies shape essential workers’ experiences of mental health during the pandemic.
States in the United States have vastly different policy contexts that can fuel geographic inequities in health and well-being. Largely driven by the devolution and preemption movements, states are increasingly polarized in their political landscape (Montez 2020). An emerging literature documents stark differences in health and mortality across states in the United States, and largely attributes this to different sociopolitical contexts across states (e.g., Homan 2019; Montez et al. 2020). In addition to variation in state-level policies prior to the COVID-19 pandemic, states had significant control over their public health and economic policies throughout the pandemic. As such, we examine the role of existing and pandemic-specific state policies for the mental health of essential workers by harnessing the substantial variation in state policies before and during the COVID-19 pandemic.
More specifically, we suggest that COVID-19-specific policies such as mask mandates and temporary expansions to paid leave had the potential to protect workers from exposure to COVID-19 by reducing the spread of transmission from customers and coworkers alike. Indeed, evidence suggests that increased access to paid sick leave reduced COVID-19 cases in the United States in 2020 (Pichler, Wen, and Ziebarth 2020). Living in a state with a mask mandate and/or paid sick leave, then, may have reduced stress for essential workers who feared becoming infected with COVID-19. Moreover, paid leave programs could have alleviated financial strain if a person contracted COVID-19 or had to care for a sick family member. This possibility is important given that essential workers tend to have lower wages than nonessential workers (Blau et al. 2021) and lost wages due to family illness may be especially harmful. Existing economic policies such as the state’s minimum wage could also serve as a buffer against economic-related stressors that essential workers faced. Because some essential workers tend to be more precariously employed and have less power and privilege (e.g., workers in the service sector), we suggest that protective state policies can reduce the mental health consequences of essential work by mitigating some of their precarity. However, prior research has not tested this possibility.
The state policy environment may be especially important for essential workers with lower levels of educational attainment. Indeed, prior research finds that adults with less education are more vulnerable to the state’s sociopolitical context, as education can act as a “personal firewall” (Montez et al. 2019). That is, the state environment is likely to affect the choices and constraints of adults with lower levels of education, whereas adults with higher levels of education can draw on their individual resources. In the present study, state policies may reduce the mental health consequences of essential work for adults with lower levels of education. For example, compared to higher waged and higher educated essential workers, essential workers with lower wages may be more reliant on policies that increase financial security during the pandemic such as paid leave policies. Taken together, the state sociopolitical environment has the potential to shape the mental health of essential workers, especially essential workers with lower levels of education. Therefore, we test the following specific hypothesis:
Method
Data and Sample
For this study, we use cross-sectional, national, and state representative data from the U.S. Census Bureau’s HPS. Respondents in the HPS were solicited via e-mail to complete a 20-minute online survey designed to better understand the social and economic experiences of households during the COVID-19 pandemic. Data collection for HPS began in April 2020 and is ongoing; however, the present study relies on data from Phase 3.1, which were collected from April 14, 2021 to July 5, 2021, because questions about essential work were not asked prior to Phase 3.1. The HPS administered the survey to a different sample every two weeks during Phase 3.1, resulting in six cross-sectional waves of data. To examine specific state policies, we merged state-level data (described in detail below) with the HPS. Data on state-level policies come from an online resource created by Raifman and colleagues (2020) and data on state partisan composition come from the National Conference of State Legislatures (2021a).
The HPS is a suitable data source for understanding the impact of essential work on mental health. As the survey was designed pragmatically for the COVID-19 context, it provides prescient data and researchers can access data quickly. As well, the HPS has a large sample size and reports respondents’ state of residence, which facilitates an examination of state-level policies. The HPS is also one of the few nationally representative surveys that contains data on both essential work settings and mental health.
Given this study’s focus on work and mental health, the analytic sample includes adults from all 50 states between the ages of 25 and 65 who worked for pay in the last seven days. We restrict the analytic sample to adults between ages 25 and 65 to focus on adults who are likely to have completed their education and are less likely to be partially or fully retired. As well, we excluded respondents who had missing information on mental health outcomes. Finally, we excluded respondents living in Washington D.C. due to the district’s unique political context. The final analytic sample includes 178,469 adults.
Measures
Mental health
The HPS contains four questions concerning the frequency of anxiety and depression symptoms. To measure anxiety, we use the validated two-item Generalized Anxiety Disorder scale (GAD-2; Kroenke et al. 2007). Respondents report how often over the past seven days they have been bothered by (1) feeling nervous, anxious, or on edge and (2) not being able to stop or control worrying. To measure depression, we use the validated two-item Patient Health Questionnaire (PHQ-2; Gilbody, Richards, and Hewitt 2007). Respondents report how often over the past seven days they have been bothered by (1) having little interest or pleasure in doing things and (2) feeling down, depressed, or hopeless. Each of these four questions has four response options: not at all (0), several days (1), more than half the days (2), and nearly every day (3). For each scale, respondents’ scores are summed. A score equal to 3 or more on the PHQ-2 is suggestive of major depressive disorder (Gilbody et al. 2007; Kroenke, Spitzer, and Williams 2003) and a score equal to 3 or more on the GAD-2 is suggestive of generalized anxiety disorder (Kroenke et al. 2007). For the present study, we use these validated cut points to create dichotomous measures for depression and anxiety. These validated cut points allow us to consider experiences of depression or anxiety that may be reaching clinical thresholds—a significant concern for population health. Moreover, this approach aligns with recent research examining mental health during the COVID-19 pandemic using data from the HPS (e.g., Boen et al. 2023; Thomeer 2023; Thomeer, Moody, and Yahirun 2023; Twenge and Joiner 2020).
Essential worker status
The HPS first asks respondents whether they have worked for pay in the past seven days. Respondents then report whether they worked outside of the home. Notably, the question does not specify or require a certain amount of time worked outside of the home (e.g., always outside the home vs sometimes outside the home). Respondents who worked outside of the home were asked to indicate the primary location where they had worked since January 1, 2021. There were 16 total response categories: healthcare; social service; preschool or daycare; K–12 school; other schools and instructional settings; first response; death care; correctional facility; food and beverage store; agriculture, forestry, fishing, or hunting; food manufacturing facility; nonfood manufacturing facility; public transit; U.S. Postal Service; other job deemed “essential” during the COVID-19 pandemic; and none of the above.
Using these questions about work, we created a theoretically relevant measure of essential work setting: healthcare, food and beverage store, manufacturing (both food and nonfood facilities), other job deemed “essential” during the COVID-19 pandemic, and nonessential worker (reference group). Nonessential workers are respondents who worked for pay in the past seven days but did not work outside of the home; we also include respondents who selected “none of the above” to the question about essential work setting. The category for other essential worker includes respondents who selected any other essential work setting other than healthcare, food and beverage, and manufacturing.
Although the category for healthcare workers includes a diverse group of workers (i.e., doctors, nurses, medical staff), we include this category because healthcare workers were a core group of essential workers with salient risks and stress exposure. We note that the measure of essential work in the HPS is self-reported. As such, this measure of work setting could reflect individual perceptions of who an essential worker is, as well as definitions of essential work that varied across states. Many states deferred to federal guidelines and often had similar definitions for the most common work settings such as healthcare and food and agriculture work.
State-level policies
We consider three policies related to the economic and health landscape of states, including policies enacted during the pandemic (i.e., mask mandates and expanded paid leave policies) and an existing economic policy (i.e., state minimum wage). Specifically, policies include whether a state had a current mask mandate as of July 1, 2021 (1 = yes), whether the state had enacted temporary COVID-19-specific paid leave expansions (1 = yes), and the state minimum wage in 2021 (range: $7.25–$14 per hour).
Covariates
All analyses account for sociodemographic characteristics that are likely to affect work status and mental health. Covariates include age (in years), gender (1 = female), race/ethnicity (non-Hispanic White, non-Hispanic Black, Hispanic, other race/ethnicity), and educational attainment. Educational attainment is categorized as lower level of education (less than high school degree, high school graduate, some college, associate degree) versus higher level of education (bachelor’s degree, graduate degree). Analyses also account for state-level COVID-19 mortality rates (per 100,000 people) for each survey week in the analytic period. In analyses of state-level policies, we include a control for the state-level political context in 2021 (democrat control, mixed control [e.g., a republican governor and democrat-controlled legislature], republican control).
Analytic Approach
We use multilevel models wherein individuals (Level 1) are nested within states (Level 2). Models include random effects for states. A likelihood ratio test indicated models were not improved with the inclusion of a random effect for essential worker. Analyses unfold in two main steps. First, we employ multilevel logistic regression models to regress anxiety and depression on essential work status (Hypothesis 1). A first model includes essential work status and covariates before adding the interaction of essential work with educational attainment in a second model.
Second, we examine to what extent state-level policies moderate associations between essential work and mental health (Hypothesis 2). To do so, we estimate multilevel logistic regression models to regress anxiety and depression on essential work status, each state-level policy measure, the interaction of essential work status with the state-level policy measure, and all sociodemographic covariates. We examine each state-level variable in separate models to consider the effects of each specific policy separately. Moreover, scholars recommend only including one Level-2 variable per 10 units (Field 2009); thus, we should not include more than five variables including state policies, state-level control variables, and cross-level interaction terms in this sample of 50 states.
Because adults with lower levels of education tend to be more vulnerable to the state environment (e.g., Montez et al. 2019), we stratify these results by low and high levels of educational attainment. The results for adults with low levels of education are presented in the main tables, whereas the results for adults with high levels of educational attainment are presented in the Supplemental Material.
Because the HPS does not provide weights at the state level, we do not apply weights to the multilevel models (Tables 2-4). Moreover, prior research finds that weighted multilevel models do not perform as well when the intraclass correlation (ICC) is small and/or when the outcome is not normally distributed; unweighted estimators perform adequately in properly specified nonlinear models (Koziol, Bovaird, and Suarez 2017). We also posit that the inclusion of covariates associated with sample selection (e.g., gender, race/ethnicity) can reduce bias without weights (Heeringa, West, and Berglund 2017). Analyses are conducted using the—melogit—package in Stata, which uses maximum likelihood estimation with numerical integration.
Results
Descriptive Results
Table 1 presents the descriptive information for the overall analytic sample and stratified by essential work status. Table 1 shows that about half of the sample worked in essential jobs at this time, with most working in other essential jobs (32 percent) or healthcare (12 percent). Regarding the mental health of the sample, Table 1 shows that rates of depression (19 percent) and anxiety (24 percent) were elevated during the study period (April–July 2021). These rates are about three times higher than a similar period in 2019 (Czeisler et al. 2020) but slightly lower than estimates from the beginning of the pandemic, when 24 percent of the country were categorized as depressed and 35 percent were categorized as having symptoms of anxiety (April–July 2020; Donnelly and Farina 2021). However, we note that the analytic sample in the present study is restricted to working adults and is not directly comparable to prior estimates based on the general population.
Weighted Descriptive Statistics for Analytic Sample (Household Pulse Survey, n = 178,219).
Note. COVID-19 = coronavirus disease-19.
Table 1 further demonstrates wide variation in mental health across different types of essential workers with food and beverage workers exhibiting higher rates of anxiety (32 percent) and depression (29 percent) than nonessential workers (24 and 19 percent, respectively). Workers in healthcare (25 percent) and other essential jobs (25 percent) also reported slightly elevated rates of anxiety relative to nonessential workers. These differences are statistically significant based on two-tailed t-tests. Table 1 also highlights differences in the composition of essential workers. For instance, women and Black adults were more likely to work in healthcare settings, White adults were more likely to work in manufacturing and “other” essential settings, and adults with lower levels of educational attainment were more likely to work in food and beverage and manufacturing settings.
Table 1 presents descriptive information on state policies. As of July 1, 2021, eight states had mask mandates in place, with 23 percent of respondents in the sample living in states with a mask mandate. Nine states had expanded paid leave during the study period, with 33 percent of the sample living in such states. The minimum wage averaged $9.82 and ranged from $7.25 to $14 per hour in 2021.
Essential Work and Mental Health
To examine whether disparities in mental health existed by work status, we regressed anxiety (Panel A) and depression (Panel B) on essential work setting (Table 2). Model 1 of Table 2 (Panel A) shows that workers in food and beverage settings (b = 0.22, p < .001) had higher log-odds of anxiety when compared to nonessential workers, net of covariates. For instance, food and beverage workers had 25 percent higher odds of exhibiting an anxiety disorder than nonessential workers (b = 0.22). On the other hand, workers in manufacturing had lower log-odds of anxiety compared to nonessential workers (b = −0.11, p < .01). Model 2 adds the interaction of essential work status with educational attainment. The positive and significant interaction term for healthcare workers suggests that the difference in anxiety between healthcare workers and nonessential workers was larger for workers with lower levels of education. Indeed, the negative main effect for healthcare workers in Model 2 (b = −0.10) indicates lower log-odds of anxiety for higher educated healthcare workers compared to higher educated nonessential workers. The magnitude of the interaction term (b = 0.17) indicates that less educated healthcare workers had slightly higher odds of anxiety relative to less educated nonessential workers. A postestimation chi-square test confirms that the difference in anxiety between less educated healthcare and nonessential workers is statistically significant.
Multilevel Logistic Regression Models Predicting Anxiety and Depression for U.S. Working Adults (Household Pulse Survey, n = 178,219).
Note. Coefficients (log-odds) presented. Standard errors in parentheses. COVID-19 = coronavirus disease-19. ICC = intraclass correlation.
p < .05. **p < .01. ***p < .001 (two-tailed tests).
Turning to depression, Model 3 of Table 2 (Panel B) shows that workers in food and beverage settings (b = 0.31, p < .001) had higher log-odds of depression when compared to nonessential workers, net of covariates. For example, food and beverage workers had 36 percent higher odds of symptoms of depression than nonessential workers (b = 0.31). On the other hand, compared to nonessential workers, healthcare workers (b = −0.12, p < .001) and workers in other essential settings (b = −0.03, p < .05) had lower log-odds of depression. When adding the interaction of essential work with educational attainment (Panel B, Model 4), the interaction term for healthcare workers was again positive and statistically significant, and the main effect for healthcare workers was negative and significant. The magnitude of the interaction term was the same as the main effect for healthcare workers, suggesting mental health protection for higher educated healthcare workers, but similar log-odds of depression for less educated healthcare and nonessential workers. Indeed, a chi-square test confirms that the difference in depression between less educated healthcare and nonessential workers is not statistically significant.
To illustrate these results, Figure 1 presents the predicted probability of anxiety (Panel A) and depression (Panel B) based on estimates in Models 2 and 4 of Table 2, respectively. Three main findings emerged. First, Figure 1 shows that adults with lower levels of education had a higher probability of anxiety and depression compared to their higher educated counterparts. Second, at lower and higher levels of educational attainment, food and beverage workers had the highest probability of anxiety and depression. Healthcare workers with lower levels of education also had heightened probability of anxiety. Third, while education seems to be protective against anxiety and depression for healthcare essential workers, this was not the case for other types of essential workers. Thus, workers with lower levels of education, particularly those in food and beverage settings, had the highest predicted probability of anxiety and depression.
Predicted probability of anxiety and depression by essential work status and educational attainment (Household Pulse Survey; n = 178,219).
Overall, findings provide partial support for Hypothesis 1 such that food and beverage workers and less educated healthcare workers had higher odds of anxiety and depression (food and beverage workers only) compared to nonessential workers. The probability of anxiety and depression was highest for food and beverage workers with lower levels of educational attainment.
Essential Work, State Policies, and Mental Health
Before testing whether state policies moderated associations between essential work and mental health, we descriptively examined the prevalence of depression and anxiety across essential work status and states. To do so, we estimated logistic regression models regressing depression and anxiety on essential work, state of residence, age, and gender. Figures 2 and 3 show the predicted probabilities of anxiety and depression, respectively, during the study period (April–July 2021) based on these estimates. For parsimony and in line with the key disparities in Table 2, we show results for nonessential workers, food and beverage workers, and other essential workers in each state. Full figures with healthcare and manufacturing workers are available upon request. We are unable to stratify the figures by educational attainment due to small cell sizes in certain states (e.g., high-educated food and beverage workers in Alaska).
Predicted probability of anxiety across states in the United States by essential worker status (Household Pulse Survey; n = 178,219).
Predicted probability of depression across states in the United States by essential worker status (Household Pulse Survey; n = 178,219).
We note three key findings from Figures 2 and 3. First, Figures 2 and 3 show that the prevalence of anxiety and depression was highest among food and beverage workers in almost every state. Second, these figures demonstrate considerable variation in rates of anxiety and depression for workers, especially food and beverage workers, across states. For example, the prevalence of anxiety (Figure 2) among food and beverage workers ranged from 47 percent in Louisiana to 9 percent in New Mexico. Similarly, the prevalence of depression (Figure 3) among food and beverage workers ranged from 50 percent in New Hampshire to 5 percent in New Mexico. Finally, the magnitude of variation in depression and anxiety across states was much smaller for nonessential workers and other essential workers compared to food and beverage workers. Overall, Figures 2 and 3 show that rates of depression and anxiety varied across states and by essential work status.
To examine whether the association between essential work and mental health was moderated by state-level policies, we iteratively tested the interaction of essential work status with each of the three state-level policies. Tables 3 and 4 present the results for workers with lower levels of education. Starting with anxiety, Table 3 indicates that the difference in anxiety between food and beverage workers and nonessential workers was smaller when the individual lived in a state that imposed a mask mandate (Model 1, b = −0.31, p < .01). Although the coefficient for expanded paid leave was large in magnitude (b = −0.17), it was not statistically significant. The association between healthcare work, manufacturing work, and other essential work with anxiety was not moderated by any of the state policies in the present study.
Multilevel Logistic Regression Models Predicting Anxiety for Adults with Low Educational Attainment (Household Pulse Survey, n = 66,583).
Note. Coefficients (log-odds) presented. Standard errors in parentheses. Models control for age, race/ethnicity, gender, job insecurity, state partisan composition, and state-level COVID-19 death rates. COVID-19 = coronavirus disease-19. ICC = intraclass correlation.
p < .05. **p < .01. ***p < .001 (two-tailed tests).
When examining the interaction of essential work and state-level policies among higher educated workers (Supplemental Material A), we only found one statistically significant interaction. The difference in anxiety between manufacturing workers and nonessential workers was smaller when the state’s minimum wage was higher (Supplemental Material A, Model 3, p < .05). Otherwise, mask mandates, paid leave, and minimum wage policies did not moderate the association between essential work and anxiety for higher educated workers.
Turning to depression, Table 4 indicates that the disparity in depression between food and beverage workers and nonessential workers was smaller for workers living in a state that imposed a mask mandate (Model 1, b = −0.27, p < .05), expanded paid leave policies during the pandemic (Model 2, b = −0.20, p < .05), and legislated higher minimum wage prior to the pandemic (Model 3, b = −0.04, p < .05). State-level policies did not moderate the association of healthcare, manufacturing, or other essential work with depression for workers with lower levels of education. Among higher educated workers (Supplemental Material B), state policies did not weaken or exacerbate differences in depression based on essential work status.
Multilevel Logistic Regression Models Predicting Depression for Adults with Low Educational Attainment (Household Pulse Survey, n = 66,583).
Note. Coefficients (log-odds) presented. Standard errors in parentheses. Models control for age, race/ethnicity, gender, job insecurity, state partisan composition, and state-level COVID-19 death rates. COVID-19 = coronavirus disease-19. ICC = intraclass correlation.
p < .05. **p < .01. ***p < .001 (two-tailed tests).
Overall, results in Tables 3 and 4 suggest that, among less educated adults, state-level policies reduced the disparity in anxiety and depression for food and beverage workers relative to nonessential workers. Findings support Hypothesis 2 when considering workers in food and beverage settings, but not those in healthcare, manufacturing, and other essential settings.
As a supplemental test, we examined three-way interactions of race/gender, essential work, and state-level policies to test whether the moderating effects of state-level policies were further conditioned by race/ethnicity or gender. Interaction terms were not statistically significant, suggesting that policies may reduce the adverse consequences of essential work (i.e., food and beverage work) in similar ways across race/ethnicity and gender. We note, however, that racial/ethnic, gender, and educational inequities in the composition of the essential workforce can contribute to inequities in mental health.
We used the estimates from Table 4 to calculate the average marginal effect of lower educated food and beverage workers compared to lower educated nonessential workers across state-level policies on the predicted probability of depression. Put differently, Figure 4 presents the difference in the probability of depression between food and beverage workers and nonessential workers by the state policy for adults with lower levels of educational attainment. Figure 4 shows that the disparity in depression between food and beverage workers and nonessential workers was much larger in states without a mask mandate, in states without a COVID-19 paid leave expansion, and in states with a minimum wage of $7.25 (the federal minimum). Overall, Figure 4 shows how state-level policies can reduce the adverse mental health consequences of working in food and beverage settings.
Average marginal effect of food and beverage work across state policies on predicted probability of depression for adults with low educational attainment (Household Pulse Survey; n = 66,583).
Discussion
Elevated rates of depression and anxiety among essential workers during the COVID-19 pandemic are of great concern (Bell et al. 2021; Cabarkapa et al. 2020; Grooms et al. 2022; Mayer et al. 2022). However, the mental health of essential workers may have varied across states, as states differentially implemented policies that had the potential to protect the mental health of essential workers. Prior research has not considered whether state-level policies shaped the mental health of essential workers during the pandemic. Using nationally representative data from the HPS from April to July 2021, we provide new insight into how specific state-level policies weakened the mental health consequences of essential work, primarily for workers in food and beverage settings. We highlight three key themes from the present study.
First, while prior research documents the mental health of essential workers in the first months of the pandemic in 2020 (e.g., Bell et al. 2021; De Boni et al. 2020; Grooms et al. 2022; Mayer et al. 2022), we build on this work with a large, representative sample to document the elevated rates of depression and anxiety among essential workers during the second year of the pandemic in the United States. We find that rates of depression and anxiety were highest among food and beverage workers relative to nonessential workers during the study period (April–July 2021). This was especially true for workers with lower levels of educational attainment. Contrary to prior research documenting worse mental health outcomes among healthcare workers at the start of the pandemic (e.g., Bell et al. 2021), we generally did not find elevated rates of depression and anxiety among healthcare workers. The one exception was that healthcare workers with lower levels of education had a higher probability of anxiety compared to nonessential workers. In fact, higher educated healthcare workers had favorable mental health outcomes relative to their nonessential counterparts. It is possible that the availability of the COVID-19 vaccine and somewhat lower COVID-19 case rates at this time (April–July 2021) alleviated some of the strain faced by healthcare workers. Moreover, healthcare workers may have benefited from organization-level mask mandates at hospitals and other healthcare offices regardless of local or state mandates.
It is worth noting that this specific time period (April–July 2021) marks a time when vaccines were increasingly available and COVID-19 cases had fallen from a winter peak in the United States. Elevated rates of depression and anxiety for certain essential workers amid this somewhat optimistic moment may be indicative of a wide range of stressors experienced by these workers. For example, in addition to COVID-19 exposure at work, sources of stress may have included new work responsibilities related to mask wearing and/or exposure to harassment from the public. Moreover, some essential workers such as those in food and beverage settings were working in lower wage, precarious jobs (Blau et al. 2021; Debus, Unger, and Probst 2021), and precarious jobs can have adverse consequences for mental health (e.g., Kalleberg 2018). Indeed, in supplemental analyses (available upon request), we found that financial strain partially explained the elevated rates of depression and anxiety among food and beverage workers. Financial strain may be particularly distressing during a pandemic when a lack of national, extensive paid sick leave forced workers to balance their health with a paycheck. These workers, then, may have experienced a “double jeopardy” of essential work and precarious work. Future research should investigate the specific factors contributing to elevated rates of depression and anxiety among some essential workers during various points of the pandemic.
A second theme concerns variation in mental health by essential work status across states. To our knowledge, no other study has documented state-level variation in the mental health of essential workers. We found considerable variation in the prevalence of anxiety and depression across states for workers, especially food and beverage essential workers. These findings align with and build on previous research that documents variation in mental health by unemployment across states (Donnelly and Farina 2021) and variation in self-rated health by part-time work status across states (Donnelly and Schoenbachler 2021). We hypothesize that this variation likely exists because of diverging state-level political landscapes over the past four decades (Montez 2020), in addition to substantial variation in the implementation of policies specific to the COVID-19 pandemic. Given robust associations between mental and physical health (e.g., Domingue et al. 2021; Thoits 2010; Walker et al. 2015), geographic disparities in mental health during the pandemic could fuel long-term geographic disparities in health and mortality.
Finally, findings suggest that the adverse mental health outcomes of food and beverage workers can be reduced when workers live in states with policies that protect and support individuals. For example, we found that mask mandates, expanded paid leave, and higher minimum wage reduced the prevalence of depression and anxiety among essential workers in food and beverage settings. This finding points to the importance of existing policies (i.e., minimum wage) in addition to pandemic-specific policies (i.e., mask mandates, expanded paid leave). Thus, a confluence of policies appears to be important for mental health, especially during a crisis like the COVID-19 pandemic. The present study, then, adds to a growing body of research documenting how policies mitigated adverse mental health outcomes during the COVID-19 pandemic (e.g., Boen et al. 2023; Donnelly and Farina 2021; Farina et al. 2023; Sparkes et al. 2023). Although mask mandates may be a source of stress for essential workers tasked with enforcement, these mandates do seem to protect the mental health of food and beverage workers perhaps because they reduce concerns about the spread of COVID-19 in businesses. Moreover, the beneficial effect of expanded paid leave policies and higher state-level minimum wage may alleviate some of the distress of financial insecurity among workers. That is, essential workers tend to have lower wages than nonessential workers (Blau et al. 2021), and the ability to draw on paid leave in the event of illness may support mental health outcomes, especially during a public health crisis. Food and beverage workers may be particularly sensitive to state policies because they tend to be precariously employed with high levels of financial strain (e.g., Blau et al. 2021), they have significant contact with the public, and they have experienced strain from managing customer ire in grocery stores and restaurants during the pandemic (Hammonds et al. 2020). These conditions help to explain why policies promoting economic security and public health were particularly important for food and beverage workers.
The findings from the present study advance the stress process model by situating experiences of stress and mental health within the context of state-level policies. That is, we argue that adverse mental health consequences of work-related stress—a key part of the original framing of the stress process model (Pearlin et al. 1981)—must be considered in light of macro-level contexts. We offer an innovative method for understanding variation in mental health during the pandemic, and future research should examine the role of state-level policies for understanding the mental health consequences of other stressful experiences brought on by the pandemic. Moreover, our findings point to state-level policies as being especially important for the mental health of essential workers with lower levels of education. This finding generally aligns with prior research on the importance of state-level sociopolitical contexts for adults with fewer economic resources (Montez et al. 2019). The stress process model emphasizes that the unequal distribution of stressors contributes to inequities in health and well-being (e.g., Pearlin et al. 1981; Pearlin et al. 2005; Turner et al. 1995), and future studies should unpack the role of state policies for marginalized populations such as racially minoritized individuals in essential jobs. Indeed, we found that state policies weakened the mental health consequences of essential work similarly across race/ethnicity and gender; thus, inequities in exposure to essential work by race/ethnicity, gender, and educational attainment may contribute to mental health inequities. State policies, then, could be important avenues for reducing mental health inequities. Taken together, we emphasize the power of state-level policies to improve the mental health of essential workers and the need for theories of stress and mental health to consider the state policy environment.
The present study provides new insights about the mental health of essential workers; however, limitations should be noted. First, the HPS has a cross-sectional study design, precluding the ability to examine changes in mental health over time, including rates of depression and anxiety prior to the start of the pandemic. Second, the HPS relies on online surveys and response rates are quite low, which raises concerns about population representativeness and accuracy (e.g., Bradley et al. 2021; Peterson et al. 2021). Third, because the HPS did not ask about essential work status before April 2021, we cannot test the effects of policies that were implemented and expired earlier in the pandemic such as shelter-in-place orders and the closure of restaurants and bars. An additional limitation is the lack of information about self-rated health given linkages between physical and mental health. Finally, the HPS measure of essential work asks respondents to report their work setting, rather than their specific occupation. Therefore, we cannot observe occupational differences within a work setting. For example, doctors, nurses, and custodial staff are all likely to be in the “healthcare” category of essential work even though their resources and experiences may differ substantially. Future research on the mental health of essential workers would benefit from more granular measures of employment and occupation. Details on industry and occupation could also allow a nuanced examination of industry-specific unionization as a possible source of worker resistance that could improve mental health. Moreover, the self-reported nature of the essential work measure makes it susceptible to different definitions of essential work in each state. Variation in the definition of essential work is most likely to affect the group of “other” essential workers in this study, as workers in healthcare and food settings were consistently identified as essential (National Conference of State Legislatures, 2021b). Additional questions about work such as the ability to telework, the amount of work hours (at home or in person), and the experience of precarious work (e.g., lack of schedule control, contract work) would help contextualize findings.
The COVID-19 pandemic created a mental health crisis (e.g., Czeisler et al. 2020; Vahratian et al. 2021), and essential workers were at an increased risk of depression and anxiety (e.g., Bell et al. 2021; Cabarkapa et al. 2020). Because essential workers comprise the majority of workers in the United States (Blau et al. 2021), a substantial number of Americans were at risk of these adverse outcomes. Moreover, the differential implementation of policies across states before and during the pandemic exacerbates the crisis, as findings from the present study indicate that state policies have the potential to reduce the adverse mental health consequences of working in essential jobs. Additional research is needed to fully understand the impact of COVID-19 on workers. For instance, much remains to be known about the extent to which the pandemic fueled existing inequities in mental health via the unequal exposure to essential work. Future research should also aim to understand the specific mechanisms linking essential work to depression/anxiety, including which characteristics of work exacerbate or protect against mental health challenges. Indeed, scholars should draw on myriad data sources and methodological approaches, such as qualitative research designs, to disentangle the complex mechanisms contributing to high rates of anxiety and depression among essential workers, especially workers in food and beverage settings. Finally, future research should continue to document the health and well-being of essential workers to understand the enduring toll of the COVID-19 pandemic on this population. Examining the mental health of workers and the role of state-level policies in mitigating depression/anxiety is a necessary endeavor to improve population health and well-being.
Supplemental Material
sj-docx-1-smh-10.1177_21568693241226979 – Supplemental material for The Mental Health of Essential Workers during the COVID-19 Pandemic: The Role of U.S. State-level Policies
Supplemental material, sj-docx-1-smh-10.1177_21568693241226979 for The Mental Health of Essential Workers during the COVID-19 Pandemic: The Role of U.S. State-level Policies by Rachel Donnelly and Adam K. Schoenbachler in Society and Mental Health
Footnotes
Acknowledgements
The authors are grateful to Dan Cornfield for his helpful comments on an earlier version of this article.
Funding
The author(s) disclosed receipt of the following financial support for research, authorship, and/or publication of this article: This work was supported, in part, by the National Institutes of Health (grant no. R03MH128649).
Supplemental Material
Supplemental material for this article is available online.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
