Abstract
The COVID-19 pandemic has caused disproportionate suffering among vulnerable and socioeconomically disadvantaged portions of the population. Low-income and minority populations are likely to experience disparate disease and mental health burdens. Currently, there is little evidence regarding how the experience of the early months of the US COVID-19 outbreak differed by income level, and how that related to mental health symptoms. The present study used data from a national sample of US adults (n = 5023) who completed measures related to the COVID experience, the COVID-19 Fear Scale, the Generalized Anxiety Disorder-7 (GAD-7), and the Patient Health Questionnaire-8 (PHQ-8). Multivariable regression was performed to determine whether income level (low: <$45,000 vs high: ≥$75,000) was significantly associated with COVID experience measures, PHQ-8, GAD-7, and COVID fear scores. Among the low-income group, COVID-19 had a significantly greater negative impact on: family income/employment, access to food, access to mental health treatment, and stress and discord in the family. Participants in the low-income group also had greater odds of a PHQ-8 score ≥10 (odds ratio [OR] = 1.38, 95% confidence interval [CI] 1.08, 1.77) and a GAD-7 score ≥10 (OR = 1.65, 95% CI 1.27, 2.14) compared to those in the high-income group. Study findings suggest substantial differences in how COVID-19 impacted daily life and mental health between adults living in low-income households compared to high-earning households during the early months of the pandemic.
Introduction
The COVID-19 pandemic, like the historic pandemics that preceded it, 1 –3 has caused disproportionate suffering among vulnerable and socioeconomically disadvantaged portions of the population. Even the first few months of the US COVID-19 outbreak showed a disproportionate impact, in terms of both incidence and severity of disease, among racial and ethnic minorities, and in lower-income zip codes. 4,5 Furthermore, this disproportionate burden extended beyond the clinical burden of COVID-19 infection. Low-income, rural, and minority populations faced greater difficulties accessing COVID-19 testing sites 6,7 and greater challenges in complying with the public health recommendations to avoid exposure to (and spread of) COVID-19. 8
Several studies have documented substantial increases in mental health burden and psychological distress from the pre-pandemic years to the era of COVID-19 9,10 – as much as 25 percentage points in one nationally representative survey study in the United States. 11 Other studies have shown that the mental health impact during the early months of the pandemic differed between countries, possibly related to differences in both infection rates and the mitigation measures implemented to contain disease spread. 12 Additionally, sociodemographic inequalities in mental health burden and psychological distress have been demonstrated within countries. Early in the US COVID-19 outbreak, reports of worsening mental health were tied to local COVID-19 case rates, 11 but the pattern did not persist. As in many other countries, including the United Kingdom, Canada, Austria, and Japan, 13 –17 risk factors for worsening mental health include lower income, lower educational attainment, younger age, being a member of a racial/ethnic minority, COVID-19-related job loss (or reduction in hours or pay), and food insecurity. 10,11,18 –20
A few studies have explored the various mechanisms through which the COVID-19 pandemic, including the public health measures instituted to control its spread, may have disproportionately increased the mental health burden for vulnerable groups. For example, among US adults, older individuals of color reported more “COVID stressors” (current and expected insecurities related to housing, food, and health care) that were associated with symptoms of anxiety and depression. 21 Additionally, among adults in the United Kingdom, individuals of lower socioeconomic position experienced significantly more adversities per week, particularly related to finances (job loss or reduced pay) and basic needs (access to food and medications). 22 Further, experiencing new or worsening socioeconomic risks (food insecurity, housing instability, interpersonal violence, and difficulties related to utilities or transportation) was associated with risk for symptoms of depression, anxiety, and/or post-traumatic stress. 23 However, there is little evidence thus far regarding how the experience of the early months of the US COVID-19 outbreak differed by income level, and how that related to mental health symptoms. Using survey data from a national sample of US adults, this study explores these questions.
Methods
Study design
This was a cross-sectional, observational, nationwide survey-based study. The study was approved by the Institutional Review Board at Baylor Scott and White Research Institute (#020-139) with a waiver of the requirement for written informed consent.
Study sample and data collection
Data were obtained from responses to an online questionnaire administered via the Qualtrics survey platform (Qualtrics, Inc., Seattle, WA) across all regions of the United States. Prospective participants were adults aged 18 years and older with sufficient mastery of English to complete the survey questionnaire. The current study focuses on data collected from a nationwide sample of 5023 participants who were enrolled in June 2020. 24 “Speed check” validation criteria were incorporated into the Qualtrics platform so that responses from participants filling out questionnaires at implausible speeds were automatically deleted from data collection.
The survey was distributed from June 22, 2020 to July 5, 2020. Income data were available for 95.5% (4797) of the 5023 participants who completed the survey. Participants selected their income from a range of categories. These participants were further classified into 3 different income-level categories: <$45,000, $45,000-$74,999, and ≥$75,000.
Study measures
Demographic information collected included age, body mass index, sex, race, ethnicity, marital status, education, current work status, occupation, number of people supported by household income, employment status before COVID 19, current living situation, current diagnosis of a psychological condition (ie, depression, bipolar/manic depressive disorder, panic disorder, generalized anxiety disorder, post-traumatic stress disorder, obsessive compulsive disorder, any phobia, schizophrenia, other), current smoking status, and current diagnosis of a comorbid condition (ie, chronic lung disease, diabetes, cardiovascular disease, chronic renal disease, liver disease, immunocompromised condition, cancer, neurologic/neurodevelopmental disability, traumatic brain injury, spinal cord injury, other chronic condition).
COVID Experience Measures: Fear of COVID-19 Scale (FCV-19S) is a 7-item scale that measures perceived fear of COVID-19 among the general population. The scale uses a 5-item bipolar Likert-style agreement response format that ranges from strongly disagree to strongly agree. Total scores range from 7 to 35. 25 FCV-19S has been found to be both reliable and valid in measuring fear of COVID 19 in different populations. 26 –32 Additionally, FCV-19S has been shown to be fully invariant across gender and age. 33 The internal consistency of FCV-19S in the study sample was excellent (Cronbach α = 0.92).
The Coronavirus Impact Scale 34 measures the impact of the COVID-19 pandemic on dimensions of life. A validated composite score for this measure has not been peer reviewed, thus selected items were assessed individually in this study. This analysis focused on daily routines, family income/employment, food access, medical care access, mental health care access, access to extended family and nonfamily social supports, experiences of stress related to the pandemic, and family stress/discord. These items have a Likert-style severity response format that includes no change, mild, moderate, and severe, and were further classified into moderate/severe vs mild/no change. The internal consistency of these items in the study sample was good (Cronbach α = 0.80). COVID-19 testing (test/none), testing positive (yes/no), and knowing someone who tested positive (yes/no) also were included in the analysis.
Patient Health Questionnaire-8 (PHQ-8) and Generalized Anxiety Disorder-7 (GAD-7) Scores: PHQ-8 is a brief, 8-item self-report measure of major depressive disorder, the validity/reliability of which is established in both the general and clinical populations. 35 Participants assess how frequently they were bothered by each symptom over the past 2 weeks, on a 4-point Likert-style rating scale ranging from 0 to 3 (0 = not at all, 1 = several days, 2 = more than half of the days, 3 = nearly every day). Because a cutoff score of ≥10 shows 88% sensitivity and 88% specificity in discriminating “probable” depression, 36 this threshold was used to dichotomize the sample into respondents with “probable depression” versus others.
GAD-7 is a 7-item screening tool for generalized anxiety disorder. Recipients assess how frequently they have been bothered by each of 7 symptoms over the past 2 weeks according to a 4-point rating scale ranging from 0 to 3 (0 = not at all, 1 = several days, 2 = more than half of the days, 3 = nearly every day). Scores are interpreted as follows: ≥10 = possible diagnosis of GAD, 5 = mild anxiety, 10 = moderate anxiety, 15 = severe anxiety. 37 The cutoff threshold of ≥10 has 89% sensitivity and 82% specificity in discriminating “probable” anxiety. 37 This cutoff score was used to dichotomize the study sample into respondents with probable anxiety versus others.
Statistical analysis
Continuous variables are summarized with means and standard deviations and categorical variables with counts and percentages. Significant differences in means and counts/percentages were assessed using t tests and chi-square tests, respectively. To adjust for differing demographic profiles, a propensity score for income level was calculated using the demographic variables and was included in the multivariable models. Multivariable logistic regression was performed to determine whether income level (low: <$45,000 vs high: ≥$75,000) was significantly associated with COVID experience measures and PHQ-8 and GAD-7 scores. Multivariable linear regression was used to determine whether there was a significant association between income level (low: <$45,000 vs high: ≥$75,000) and COVID fear score. A 5% α level was used to determine significance, and all analyses were performed using SAS version 9.4 (SAS Institute Inc., Cary, NC).
Results
Table 1 displays the demographic characteristics of participants for whom income data were available. Of the 4797 participants, 31.3% (1500) had an annual income <$45,000, 19.7% (945) had an annual income of $45,000–$74,999, and 49.0% (2352) had an annual income ≥$75,000. All demographic characteristics, except having any chronic condition, were statistically different between low-income (<$45,000) and high-income (≥$75,000) participants. Compared to participants in the high-income group, those in the low-income group were younger, had lower education levels, had a higher prevalence of a current diagnosis of a psychological condition, and a higher proportion were single. There was a higher percentage of health care workers among high-income participants (37.2% vs 12.3%).
Demographic Summary by Income (at Baseline), n = 4797
P value for comparisons between <$45,000 and >$75,000.
GED, general equivalency diploma.
Income and the COVID experience
After adjusting for all demographic characteristics, participants in the low-income group had greater odds of the COVID-19 pandemic having a moderately/severely negative impact on: family income and employment (odds ratio [OR] = 2.77, 95% confidence interval [CI] 2.25, 3.41), access to food (OR = 2.70, 95% CI 1.98, 3.68), access to mental health treatment (OR = 1.61, 95% CI 1.12, 2.30), and stress and discord in the family (OR = 1.35, 95% CI 1.04, 1.75) (Table 2). They had lower odds than the higher income group of reporting moderate/severe impact on changing routines, and access to extended family and nonfamily social supports. Those in the low-income group also had lower odds of receiving a COVID-19 test (Table 2).
COVID Experience by Income (at Baseline)
CI, confidence interval; OR, odds ratio.
Income and depression and anxiety scores
Table 3 shows the results for the multivariable adjusted analysis for income level and PHQ-8 and GAD-7 scores. Participants in the low-income group had 1.38 times greater odds of a PHQ-8 score ≥10 and 1.65 times greater odds of a GAD-7 score ≥10 compared to those in the high-income group.
Additional Patient-Reported Measures by Income (at Baseline)
CI, confidence interval; GAD-7, Generalized Anxiety Disorder-7; OR, odds ratio; PHQ-8, Patient Health Questionnaire-8.
Discussion
Study results show significant differences in the demographic, occupational, social, and clinical characteristics between adults living in households with an annual income <$45,000 vs. ≥$75,000. Specifically, the early experience of the COVID pandemic differed dramatically between income groups, with lower income individuals experiencing substantial negative impact concerning their finances and access to basic needs such as food and mental health treatment. By contrast, the higher income individuals experienced changing routines and difficulty accessing social support. Essentially, these results show that the COVID-19 pandemic has further amplified the role of social determinants of health, a select array of conditions in which people are born, live, work, and age that significantly affect all aspects of health. 38,39
The inequitable impact of the conditions in which people live creates persistent health disparities. Obvious examples of such conditions putting populations of low socioeconomic status at increased risk for exposure to COVID-19 include overcrowding and limited access to safe outdoor spaces. 40 Low education level, racial segregation, low social support, and poverty are all social determinants of health that have been shown to contribute to increased death rates in the United States. 41 In the specific context of the pandemic, inadequate sick leave and child care benefits in low-wage jobs make it difficult for workers to stay home when sick (or when a household member is sick), adding the emotional stressor of worry about exposing coworkers or carrying the virus home to family members if exposed by a coworker to the physical risks of COVID-19. 42 And although the paid sick leave policy guaranteed by the Coronavirus Aid, Relief, and Economic Security Act of 2020 43 was an important measure, it may have been inadequate for frontline workers frequently exposed to the virus, who required multiple, repeated quarantines, especially in multi-occupancy homes. Black households, Hispanic households, and those with lower educational attainment were particularly vulnerable to economic disruption going into the pandemic, 44 and thus were unlikely to be able to afford to stay home without paid sick leave.
In terms of mental health, the lower income group was significantly more likely to score above the threshold on depression and anxiety measures. One characteristic of low-wage work that predisposes workers to mental health distress in the context of the pandemic is that few such jobs are amenable to telecommuting. As such, when shelter-in-place orders and other public health measures intended to control the spread of COVID-19 were implemented, workers either lost their jobs (creating financial distress) or, in the case of essential worker roles, faced a dramatically increased workload of, in many cases, monotonous tasks (making workers prone to burnout). 42 In fact, changes to work situations related to COVID mitigation strategies have been identified as the single greatest occupational source of anxiety and depression. Key mitigating factors for that stress include reliance on family and/or trade unions for financial support, access to sick leave and unemployment benefits, and availability of personal protective equipment when the work carried a threat of contagion. 45
Study findings of greater mental health impact among people with lower household income is consistent with observations from the United States and several other countries that the mental health burden associated with the COVID-19 pandemic is falling disproportionately on the economically vulnerable. 10,11,13 –20 It also is worth noting that low income is a risk factor for poor mental health even outside the pandemic context, but COVID-19 has made the disparities more glaring, in some cases by widening them, in others by simply shifting the prevalence for all groups up such that neither the overall prevalence nor the differences between groups is easy to dismiss or ignore. 11 Previous research focusing on COVID-19 has similarly found that lower income is associated with greater economic impact, while higher income is associated with a greater impact on social conditions. 46 Where present study results shed new light is on what such findings mean for mental health: at least in the short term (∼3 months into the pandemic), present study results suggest that the social impact is less detrimental for mental health than the economic impact. Furthermore, the disparities observed are likely to persist or broaden further even after the immediate effects of the pandemic end: sizable income gaps have a profound impact on mental health, with low socioeconomic status being associated with greater odds of depression and likelihood of persistent depression. 47 Thus, individuals disproportionately affected by COVID are at elevated risk for prolonged post-pandemic poor mental health and accompanying diminished employment prospects. 48
Existing research provides guidance as to the policy approaches that can both help prevent the disparities that have already emerged related to the COVID-19 pandemic from broadening further and avoid such a catastrophic recurrence in future pandemics: Comparisons between US states that had different social supports available showed that greater support – specifically, having expanded the Medicaid program to decrease the proportion of the population without health care coverage, offering timely and accessible unemployment insurance, and suspending utility shutoffs – weakens the association between household income shocks and mental health. 49 Following a similar line, other research suggests that providing a universal basic income enhances community as well as individual outcomes by alleviating the impact of psychosocial stressors, which in the context of such widespread threats to economic stability and physical and mental health as the COVID-19 pandemic encompasses, serves to mitigate the inequalities that otherwise would be exacerbated and better positions society as a whole for recovery as the situation resolves or is adapted to. 50
Limitations
This study has several limitations that should be considered. As with any survey, self-selection bias is a potential concern and, because this was an online survey administered only in English, people with limited internet access and/or computer skills, and non-English speakers are likely underrepresented. Additionally, respondents included 30% general population, 40% health care workers, and 30% essential workers. As a result, income levels may not fully approximate the US population. Another limitation to consider is that study data on current or past psychological conditions are based on self-report rather than clinical diagnoses. Study measures of anxiety and depression symptoms employed screening tools, and so should not be interpreted as prevalence estimates of depression and anxiety in the population, as this risks overestimation of these conditions. 51 Lastly, household income information was collected in ranges, and so where the income fell relative to benchmarks (eg, Federal Poverty Level, median national household income) could not be calculated. The cutoffs of <$45,000 and ≥$75,000 were chosen for the low- and high-income groups, respectively, based on the ranges collected that came closest to falling below 200% of the 2020 Federal Poverty Level ($43,440 52 and a common cap on hospital financial assistance and similar programs) for a 3-person household for the low-income group, and to exceeding the 2019 national median income ($68,703 53 ) for the high-income group. The group in the middle ($45,000–$74,999) – in which number of people in the household, as well as differences in the cost of living among respondents' locations, would make substantial differences in how “livable” an income truly was and thus the resources and reserves available to that household during the pandemic – were not included in the comparisons.
Conclusions
Additional research is needed to determine whether the patterns of COVID experience and mental health impact observed here, in the survey administered in June 2020, persisted as the pandemic continued or if different trajectories emerged. Nonetheless, study findings support both anecdotal and academic reports of the how the global pandemic has amplified economic inequalities. Policy makers confronted again by the need to implement public health measures to control the spread of disease must consider the difference in COVID experience and mental health burden between income groups as they weigh options that, although they may be implemented uniformly, will produce different effects in different portions of the community. Ultimately, existing social policy is likely inadequate to support vulnerable populations because of the limited reach of many of those policies and the disproportionate hardship associated with low income and low-wage occupations.
Footnotes
Authors' Contributions
Drs. Warren and Powers are responsible for the original design of study and Drs. Hall, Sanchez, and Bennett, and Ms da Graca are responsible for the design specific to this paper. Drs. Warren, Powers, and Bennett were part of reviewing survey questions and coordinating with Qualtrics for review and receipt of survey responses. Drs. Hall and Bennett performed the data analysis and interpretation of the results. Drs. Hall, Sanchez, and Bennett, and Ms. da Graca drafted the article. All authors were responsible for critical review, revision, and final approval of the article to be published.
Author Disclosure Statement
The authors declare that there are no conflicts of interest.
Funding Information
The authors gratefully acknowledge support for this work from the Baylor Scott & White Dallas Foundation and the W.W. Caruth, Jr. Fund at Communities Foundation of Texas.