The Impact of COVID-19 on Families with Children: Examining Sociodemographic Differences

Abstract

Objective:

The COVID-19 pandemic has exacerbated differences related to employment and family psychological health. However, empirical evidence examining COVID-linked differences concerning children and families remains scant. The current study addresses this gap by examining sociodemographic differences associated with COVID-19 upon family access to resources and family psychological health.

Method:

A telephone survey of 600 caregivers living in Mississippi was conducted from August 2020 to April 2021. Caregivers answered questions about levels of worry regarding themselves or their child contracting COVID-19 and impact of the pandemic on household income, access to resources, and family psychological health.

Results:

Multivariate models demonstrated that Black caregivers (n=273; 45.5%) had increased odds of agreeing that they worry about contracting COVID-19 (OR=2.57). Further, as caregiver reported household annual income decreased, caregivers had increased odds of agreeing that they worry about contracting COVID-19 (OR=1.16), lost job-related income (OR=1.14), and had a hard time obtaining resources (OR=1.16) due to the pandemic. No significant differences related to rural or urban residence were observed.

Conclusions:

Findings highlight the need for pragmatic responses that are attuned to differences by providing more equitable access to resources for families. Findings suggest that strategies addressing family worry, obtaining job-related income support, and helping families obtain tangible resources may positively impact family psychological health. As population changes in vaccination rates and COVID variants emerge, re-assessment of family and community impact appears indicated. Limitations and future research directions are discussed.

In the United States (U.S.), there have been over 81 million confirmed COVID-19 cases and over 990,000 deaths related to COVID-19¹. In Deep South states like Mississippi, there have been over 798,000 cases with daily infections, hospitalizations, and deaths continuing to rise^1,2. Among cases, data demonstrates that Black and Hispanic³, urban⁴, and impoverished populations⁵ have disproportionately higher incidences of COVID-19 infection and death. However, little information on how COVID-19 has impacted children and families exist.

The COVID-19 pandemic has had a profound impact on the U.S. economy⁶. Unemployment rates increased sharply in the initial months of the pandemic as businesses closed or restricted their operating hours at the pandemic’s onset due to shelter-in-place orders and/or reduced patronage⁷. The unequal impact of economic instability on communities experiencing low socioeconomic status (SES) and that have been socially marginalized in the U.S. has been well documented⁸. This unequal impact is even more stark in rural, Deep South states, such as Mississippi, where the poverty rate is the highest in the nation (19.6%)⁹ and the state unemployment rate is the highest for Black individuals (10%) compared to White individuals (4%)¹⁰. COVID-19 has also exacerbated economic insecurity in rural areas of the U.S. and in Mississippi. For example, more individuals living in rural areas reported being unemployed during the pandemic than prior to the pandemic¹¹. Similarly, Mississippi has seen a loss of 31,000 jobs from pre-pandemic (February 2020) to June 2021¹².

These differences have led many to speculate that the COVID-19 pandemic will lead to increased inequalities related to access to resources such as employment^13,14,15 and family psychological health¹⁶. Indeed, higher COVID-19-linked job loss¹⁷ and increased food insecurity^18,19 have been observed for Black and Hispanic individuals²⁰, as well as individuals without a college degree²¹. Though these differences have been examined, investigations of differences related to access to resources and family psychological health, particularly among caregivers and children, remain scant. This is increasingly important as socioeconomic functioning is robustly associated with greater psychological health difficulties for children²². Further, literature is limited regarding COVID-linked differences in access to resources and family psychological health for rural and urban families, which is increasingly necessary to understand given the association between geographic differences in COVID-related outcomes such as cases, vaccination rates, and local impact. Documentation is needed to identify communities most impacted by public health emergencies (e.g., COVID-19 pandemic) and improve access to resources for all families.

The Present Study

This study sought to address these important gaps by examining sociodemographic differences (e.g., race and ethnicity, income, and rurality) related to the impact of COVID-19 on family access to resources (e.g., income and food) and family psychological health (e.g., worry about contracting COVID-19). Using a statewide sample from a rural region of the U.S. particularly affected by COVID-19 (Mississippi), the current study is positioned to provide key information from a traditionally under-investigated population and geographical region of the U.S. It was hypothesized that Black caregivers, caregivers with low income, and rural caregivers would report greater worry about contracting COVID-19, and report greater difficulty obtaining resources, more economic disturbance, and greater psychological health difficulties due to COVID-19 compared to White caregivers, caregivers with high income, and urban caregivers, respectively.

Method

Participants and Procedures

A representative statewide sample of participants included 600 caregivers from Mississippi with at least one child under the age of 18 living in the home. The survey was administered by the Survey Research Laboratory housed in the Social Science Research Center at Mississippi State University. A dual frame random digit dialing telephone survey was administered from August 2020 to April 2021. Social Science Research Center telephone surveys have been ruled exempt by the Mississippi State University IRB. Surveys were carried out by a trained interviewer and survey logic was conducted by a computer-assisted telephone interviewing (CATI) system. The interviewer recorded answers from the participants with pre-coded responses displayed through a computer screen, with questionnaire logic performed through the CATI system. On average, the survey took 15 minutes to administer and respondents who completed the survey received a $10 electronic gift certificate to incentivize participation. The sample was stratified to equally represent urban (n=300, RUCA codes 1–4) and rural (n=300, RUCA codes 5–10) households based on Rural-Urban Community Area (RUCA) codes. The survey was administered as part of a larger project examining caregiver perceptions of barriers related to pediatric clinical trial participation.

Dependent Variables

Impact of COVID-19.

The impact of the COVID-19 pandemic was assessed by the following items: 1) “I am stressed around other people because I worry my child or I will catch the Coronavirus.”, 2) “My family has lost job-related income due to the Coronavirus.”, 3) My family has had a hard time getting needed resources such as food, toilet paper, cleaning supplies or other resources due to the Coronavirus.”, and 4) “The Coronavirus outbreak has negatively impacted my family’s psychological health.” Responses followed a 5-point Likert scale ranging from 1 (“strongly disagree”) to 5 (“strongly agree”). These questions were adapted from the ECHO COVID-19 Questionnaire–Adult Alternate Version²³.

Independent Variables

Rurality.

Caregivers were asked to provide the zip code of their current residence. Using zip code and RUCA conversion tables²⁴, participant RUCA codes ranged from 1=“Metropolitan area core” to 10=“Rural areas” (Table 1). For the purposes of this study, RUCA codes were considered in the primary analyses as a continuous variable.

Table 1.

Reported Demographic Characteristics of Children and Caregivers

	M (SD)	N	%
Mean Child Age (in years)	9.51 (5.12)
Mean Household Size	4.13 (1.39)
Relationship to Child
Biological Parent		457	76.20
Stepparent		22	3.70
Grandparent		61	10.20
Foster Parent		3	0.50
Aunt or Uncle		23	3.80
Older Sibling		17	2.80
Older Relative		9	1.50
Older Non-relative		8	1.30
Caregiver Race and Ethnicity
Non-Hispanic Black		273	45.50
Non-Hispanic White		293	48.83
Non-Hispanic American Indian/Alaska Native		8	1.33
Non-Hispanic Asian or Pacific Islander		6	1.00
Non-Hispanic Multiracial		5	0.83
Non-Hispanic (Race Refused to Answer)		3	0.50
Hispanic Black		3	0.50
Hispanic White		4	0.67
Hispanic Multiracial		4	0.67
Hispanic (Race Refused to Answer)		1	0.17
Caregiver Partnership Status
Married		323	53.80
Living with a Partner		21	3.50
Never Married		167	27.80
Divorced		45	7.50
Separated		28	4.70
Widowed		11	1.80
Refused to Answer		5	0.80
Rurality (RUCA Code)
Metro area core		148	24.70
Metro area high commuting		41	6.80
Metro area low commuting		4	0.70
Micro area core		107	17.80
Micro area high commuting		71	11.80
Micro area low commuting		18	3.00
Small town core		63	10.50
Small town high commuting		57	9.50
Small town low commuting		30	5.00
Rural areas		61	10.20
Caregiver Educational Attainment
Grades 1–8		2	0.30
Grades 9–11		36	6.00
Completed High School or GED equivalent		150	25.00
Some college or vocational program		137	22.80
Completed Associate degree		93	15.50
Completed Bachelor’s degree		102	17.00
Completed Master’s degree		60	10.00
Beyond Master’s degree		20	3.30
Annual Household Income
Less than $10,000		38	6.30
$10,000 to under $15,000		43	7.20
$15,000 to under $20,000		39	6.50
$20,000 to under $25,000		29	4.80
$25,000 to under $35,000		35	5.80
$35,000 to under $50,000		62	10.30
$50,000 to under $75,000		81	13.50
$75,000 to under $100,000		65	10.80
$100,000 to under $150,000		51	8.50
$150,000 to under $200,000		16	2.70
$200,000 or more		17	2.80
Don’t Know/Not Sure		43	7.20
Refused to Answer		81	13.50

Note. RUCA = Rural-Urban Commuting Area. Metro = Metropolitan. Micro = Micropolitan.

Race and Ethnicity.

Caregivers self-reported their racial identity and whether they considered themselves Hispanic (Table 1). This information was combined into the following categories: non-Hispanic Black (45.50%; n=273), non-Hispanic White (48.83%; n=293), non-Hispanic American Indian/Native Alaskan (1.33%; n=8), non-Hispanic Asian/Pacific Islander (1.00%; n=6), non-Hispanic Multiracial (0.83%; n=5), Hispanic Black (0.50%; n=3), Hispanic White (0.67%; n=4), and Hispanic Multiracial (0.67%; n=4). One caregiver (0.17%) identified as Hispanic and refused to provide their race. Three caregivers (0.50%) identified as non-Hispanic and refused to provide their race. Given the low frequency of responses, analyses were limited to non-Hispanic Black (hereafter, Black) and non-Hispanic White (hereafter, White) caregivers to reduce heterogeneity.

Income.

Household annual income was assessed on an 11-point scale ranging from 1=“Less than $10,000” to 11=“$200,000 or more” (Table 1). This variable was recoded where 1=“$200,000 or more” to 11=“Less than $10,000” to provide consistency in interpreting results. Given the ordinal nature of this scale, household annual income was treated as a continuous variable.

Covariates

Child age.

Caregivers were asked to provide the ages of all children living in the home (Table 1). If a caregiver reported having multiple children under the age of 18 living in the home, they were instructed to provide information for a single focal child within the home. Focal children were chosen randomly via the CATI system prior to initiating questions pertaining to the selected child.

Partnership status.

Partnership status was assessed from a list of relationship status categories including 1) married, 2) not married but living with a partner, 3) never married, 4) divorced, 5) separated, and 6) widowed (Table 1). Partnership status was treated as a categorical variable with married status as the reference group.

Educational attainment.

Caregivers were asked to provide their highest level of educational attainment on an 8-point scale ranging from 1=“Grades 1–8” to 8=“Beyond Master’s degree” (Table 1). Given the ordinal nature of this scale, caregiver educational attainment was treated as a continuous variable.

Household size.

Caregivers were asked to separately provide the number of adults (ages 18+) and children (ages 0–18) living in the home. Answers to these items were combined to calculate household size (Table 1).

Analyses

Data were analyzed using SAS Enterprise Guide version 8.3 and the oglmx package²⁵ in R version 3.5.1 (R Core Team 2021). Univariate and multivariate ordered logistic regression (OLR) models were used to examine differences within sociodemographic variables (e.g., rurality, race and ethnicity, and income) and responses to COVID-19 related questions. Each multivariate model included child age. To adjust for socioeconomic resources within the household²⁶, caregiver partnership status, caregiver educational attainment, and household size were also included in the multivariate models. Missing data were analyzed using maximum likelihood estimation in the SAS Enterprise Guide and R software packages.

The statistical assumptions of OLR models were first assessed via analysis of proportional odds. Proportional odds assume an estimate that describes the relation between categories of the response variable is the same across all categories (e.g., strongly agree versus all other categories, agree versus all other categories, and so on). Tests of proportional odds revealed that for some OLR models one set of estimates was not appropriate to describe the relation between all pairs of the dependent variables. As such, ordered generalized linear models were utilized. Specifically, marginal effects were estimated where probabilities of each independent variable were given for each category of the dependent variable²⁵. This approach was considered for all analyses; however, inferential conclusions remained the same (see supplemental tables for estimates) when comparing estimates from this approach to the standard OLR approach. Therefore, estimates from the OLR models are reported for ease of interpretation and the ordered generalized linear effects results are reported in Supplementary Tables 1–5.

Results

Worry About Contracting COVID-19

Univariate OLR models indicated that as household income decreased, caregivers had 1.26 times the odds (95% Confidence Interval [CI]: 1.19–1.35; Table 2) of strongly agreeing (versus all other categories) that they worry about themselves or their child contracting COVID-19. Results also demonstrated that Black caregivers had 3.94 times the odds (95% CI: 2.89–5.34) of strongly agreeing compared to White caregivers (Table 2). Rural-urban differences based on RUCA code was not significantly related with caregiver worry about contracting COVID-19 (OR=0.98). Similar results were demonstrated when each sociodemographic variable was considered in a multivariate model together (Table 3).

Table 2.

Univariate Ordered Logistic Regression Models of Sociodemographic Variables Predicting COVID-19-related Impact

	COVID-19-related Impact
Indicator	Worry About Contracting COVID-19		Lost Job-related Income		Difficulty Obtaining Resources		Family Psychological Health Negatively Impacted
	OR	p-value	OR	p-value	OR	p-value	OR	p-value
Income	1.26	<.001	1.20	<.001	1.24	<.001	1.07	.04
Race and Ethnicity (Black)	3.94	<.001	2.32	<.001	1.62	<.01	1.30	.08
Rurality (RUCA Code)	.98	.39	.97	.26	1.00	.87	.98	.43

Note. RUCA = Rural-Urban Commuting Area. Income = Household annual income. OR = odds ratio. Race and Ethnicity is dummy coded with White as the reference group.

Table 3.

Multivariate Ordered Logistic Regression Models of Sociodemographic Variables Predicting COVID-19-related Impact

	COVID-19-related Impact
Indicator	Worry About Contracting COVID-19		Lost Job-related Income		Difficulty Obtaining Resources		Family Psychological Health Negatively Impacted
	OR	p-value	OR	p-value	OR	p-value	OR	p-value
Income	1.16	<.01	1.14	<.01	1.16	<.001	1.08	.10
Race and Ethnicity (Black)	2.57	<.001	1.45	.07	.95	.79	1.09	.65
Rurality (RUCA Code)	1.00	.88	.97	.39	.96	.15	.97	.23
Child Age	1.00	.80	.97	.08	1.04	<.05	1.04	<.05
Living with a Partner	1.36	.47	1.41	.45	1.62	.25	.82	.65
Never Married	1.13	.64	1.51	.11	1.36	.23	1.08	.75
Divorced	1.28	.47	1.36	.37	2.19	<.05	1.70	.10
Separated	.61	.26	1.72	.21	1.52	.33	.77	.54
Widowed	1.79	.44	3.26	.16	3.39	.12	10.66	<.01
Education	.94	.38	.95	.44	0.87	<.05	1.12	.09
Household Size	1.05	.43	1.07	.28	1.07	.31	1.01	.85

Note. RUCA = Rural-Urban Commuting Area. Income = Household annual income. Education = Caregiver educational attainment. OR = odds ratio. Race and Ethnicity is dummy coded with White as the reference. Marital status is dummy coded with Married as the reference group.

Lost Job-related Income

Likewise, univariate OLR models revealed that as household income decreased, caregivers had 1.26 times the odds (95% CI: 1.18–1.34; Table 2) of strongly agreeing that the family had lost job-related income due to the pandemic. Black caregivers had 2.33 times the odds (95% CI: 1.71–3.17) of strongly agreeing compared to White caregivers (Table 2). Rurality was not significantly related with loss of job-related income due to COVID-19 (OR=0.97). The multivariate model indicated that only household income remained significant (Table 3).

Difficulty Obtaining Resources

Similarly, univariate models showed that as household income decreased, caregivers had 1.24 times the odds (95% CI: 1.16–1.32; Table 2) of strongly agreeing that they had trouble obtaining resources due to the pandemic. Again, Black caregivers had 1.62 times the odds (95% CI: 1.21–2.19) of strongly agreeing compared to White caregivers (Table 2). Rurality was not significantly associated with difficulty obtaining resources due to COVID-19 (OR=1.00). The multivariate model indicated again that household income remained significant (Table 3). Notably, child age (OR=1.04; 95% CI: 1.01–1.07), divorced caregivers (OR=2.19; 95% CI: 1.11–4.31), and caregiver educational attainment (OR=.87; 95% CI: .76–99) were significantly related to difficulty obtaining resources due to COVID-19. Specifically, as child age (in years) increased, caregivers had 1.04 times the odds of strongly agreeing to trouble obtaining resources. Divorced caregivers had 2.19 times the odds of strongly agreeing to difficulty obtaining resources compared to married caregivers. Finally, as caregiver educational attainment increased, caregivers had decreased odds (i.e., OR<1) of strongly agreeing to difficulty obtaining resources.

Family Psychological Health

As above, univariate models showed that as household income decreased, caregivers had 1.07 times the odds (95% CI: 1.01–1.13; Table 2) of strongly agreeing that their family’s psychological health had been negatively impacted by the pandemic. Race and ethnicity and RUCA code were not significantly related to COVID-19 negatively impacting family psychological health. Household income, race and ethnicity, and rurality were not significant when considered together in the multivariate model (see Table 3). Child age (OR=1.04; 95% CI: 1.01–1.07) and widowed caregivers (OR=10.66; 95% CI: 1.95–58.25) were significantly associated with COVID-19 negatively impacting family psychological health. Specifically, with an increase in child age, caregivers had 1.04 times the odds of strongly agreeing to their family’s psychological health being negatively impacted. Widowed caregivers had 10.66 times the odds of strongly agreeing to their family’s psychological health being negatively impacted compared to married caregivers.

Discussion

The COVID-19 pandemic continues to be a public health concern in the U.S. and to families in Mississippi. Differences regarding race, SES, and rurality were glaringly present prior to the pandemic and have been exacerbated by the pandemic^20,21. Currently, investigations of sociodemographic differences related to access to resources and family psychological health, specifically regarding caregivers and children are limited. Further, literature is limited regarding COVID-linked differences in access to resources and family psychological health for rural and urban families. The present study addressed this gap in the literature by examining sociodemographic differences related to the impact of COVID-19 on family access to resources and family psychological health. Findings revealed SES differences in worry about contracting COVID-19, sustaining income, and obtaining other resources. Further, Black caregivers expressed more worry about themselves or their children becoming infected with COVID-19 in comparison to White caregivers. Third, contrary to hypotheses, caregivers residing in rural and urban areas in Mississippi did not differ in their worry about contracting COVID-19, obtaining resources, economic impact, and psychological health difficulties. Lastly, caregivers in Mississippi also did not differ concerning negative family psychological impacts among the other sociodemographic variables of interest.

There are several points of explanation for this pattern of findings. As mentioned, the COVID-19 pandemic has impacted many U.S. businesses resulting in closures and subsequent job loss. Importantly, these closures and unemployment were experienced largely by low-wage workers with jobs incompatible with remote work (e.g., service and hospitality positions)²⁷. The results of this study are consistent with these patterns. Indeed, approximately 31% of the current sample reported a household annual income less than $25,000. Race and ethnicity and household income were significantly related to worry about caregivers and children contracting COVID-19. Perhaps driving this result for Black caregivers is the well documented evidence that Black populations have been disproportionately affected by COVID-19 in terms of cases, hospitalizations, and deaths³, leading to increased worry. Caregivers employed in retail and service sector jobs had frequent contact with the public and related potential for infection may have led to increased worry than caregivers with potentially higher wage jobs that were more conducive to remote work. No differences according to rurality were observed. This is inconsistent with previous studies that have showed that rural adults were less impacted by pandemic-related labor market shocks than urban adults²⁸. This may reflect that Mississippi is classified as a mostly rural and less populous state relative to other U.S. states, but the rurality differences within the state may not be sufficient to observe differences in rural and urban areas.

Practical Implications

Caregivers expressed concerns about infection and economic hardship. These concerns and losses can lead to deteriorating family psychological health and subsequent consequences, such as, internalizing, and externalizing difficulties²⁹. Further, these and other facets of socioeconomic disadvantage are also associated with other deleterious health outcomes⁸. As such, our findings suggest that strategies addressing family worry, obtaining job-related income support, and helping families obtain tangible resources may provide a buffer against these stressors and ultimately reduce subsequent health disparities. Indeed, previous research has demonstrated that social safety net program participation improves health outcomes (e.g., higher birth weights and child cognitive development) among under-resourced families³⁰. At the time these data were collected, several social policies were implemented to provide necessary support for families with income loss. Such action included increased weekly unemployment benefits, increased food benefits, funding for emergency food banks and pantries, and rent assistance.

Despite these policies, differences remained, thus it appears that as the pandemic continues, re-assessment of the impact of COVID-19 on communities and families is warranted. Particularly, this study highlights the need for a pandemic response that targets these differences by providing more equitable access to resources for all families. These resources may be critical in addressing social disadvantages before these disadvantages lead to negative consequences³¹. In the short-term, the findings here support federal and state government efforts to maintain or reinstate unemployment benefits and extend eviction protections. Long-term, investment in the social safety net (e.g., raising minimum wage, extending the Child Tax Credit) may improve economic preparedness for current and future public health emergencies.

Limitations and Future Directions

Findings should be interpreted considering several limitations. First, data for the larger study was gathered via phone survey. Recent data suggests that response rates to telephone surveys have declined due to increased automated telemarketing calls³². Despite this national trend, the survey completion rate for our study was 36% (i.e., ratio of completed surveys to completed surveys plus refusal to engage). Future research may benefit from employing a mixed methods approach with phone and mailed surveys to increase the range of representativeness, measurement of constructs (e.g., psychological functioning), and areas of economic impact (e.g., job loss). Second, the survey was administered between August 2020 and April 2021. Shortly after data collection ended, COVID-19 vaccines began to become more widely distributed across the U.S. potentially alleviating worry for many populations related to contracting the virus. Future studies should examine the impact of vaccine rollout as it relates to COVID-19-linked insecurities and sociodemographic differences. Third, these data are from only the state of Mississippi, limiting generalizability and should be replicated in other states and nationally. Further, though our population-based sample was representative of urban and rural households in Mississippi with children, analyses were limited to Black and White caregivers. Future studies should gather information from larger population-based samples that oversample other racial and ethnic groups to examine these issues among a more racially and ethnically diverse sample of families in Mississippi. Lastly, the analyses identified caregivers who are divorced or widowed and caregivers caring for older children as populations potentially adversely impacted by the pandemic. Indeed, research consistently shows that previously married individuals have poorer psychological health and fewer economic resources, on average, than their currently married counterparts^33,34. Research also suggests older children (particularly adolescents) reported increased negative affect (e.g., sadness and worry) related to COVID-19-linked financial stress³⁵. Though wide variability was observed, particularly for widowed caregivers (e.g., 95% confidence interval was 1.95–58.25), future consideration of differences based on partnership status and care of older children and the impact of public health emergencies appears warranted.

Conclusion

Findings of the current study suggest that Black caregivers and family household income were related to increased COVID-19-related concerns. Results did not demonstrate associations with rural or urban communities. The findings have implications for policies aimed at mitigating caregiver worry and reduce economic hardship brought about by the COVID-19 pandemic. Additionally, governments should extend or reinstate social policies and protections for families most impacted by the economic instability of public health emergencies. Future research should consider the influence of vaccine rollout and its impact on worry about contracting COVID-19. Nonetheless, the current study provides key information from a traditionally under-investigated population and geographical region of the U.S.