Parent Mental Health Before and During the COVID-19 Pandemic

Abstract

Although extant cross-sectional data suggest that parents have experienced numerous challenges (e.g., homeschooling, caregiver burden) and mental health consequences during the COVID-19 pandemic, longitudinal data are needed to confirm mental health changes relative to pre-pandemic levels and identify which specific pandemic-related changes most highly predict mental health during the pandemic. In two longitudinal subsamples (N=299 and N=175), we assessed change in anxiety, depression, and stress before and during the pandemic and whether the accumulation of pandemic-related changes predicted observed mental health changes. On average, parents reported increased depression and anxiety, but no significant changes in reported stress. Moreover, increased interpersonal conflict, difficulty managing work and caregiving responsibilities, and increased economic challenges were the types of pandemic-related changes that most strongly predicted worse mental health, highlighting that juggling caregiving responsibilities and economic concerns, along with the pandemic’s impact on interpersonal family relationships are key predictors of worsening parental mental illness symptoms.

Multiple measures were implemented to mitigate the spread of the deadly SARS-CoV-2 virus. Across the United States, stay-at-home orders were swiftly issued, and businesses, workplaces, schools, and community programs abruptly closed, introducing challenges, including increased unemployment, economic uncertainty, social isolation, and disruption to daily routines (Banerjee & Rai, 2020; Pak et al., 2020). These COVID-19 related stressors have had significant consequences for individuals’ mental health. Cross-sectional data indicated that rates of depression and anxiety in adults were elevated during the pandemic compared to historical norms (Nelson et al., 2020), a finding consistent across a narrative review of recent cross-sectional studies, research commentaries, and editorials (Rajkumar, 2020).

Beyond the stressors adults faced generally, parents confronted additional unique challenges, including working remotely with children, additional caregiving and homeschooling responsibilities, and conflicts between parenting and work. At the onset of the pandemic, approximately 3.5 million mothers of school-age children left active work (i.e., shifted into paid or unpaid leave, lost their job, or exited the workforce all together) and almost 1-in-2 mothers (45%) were not actively working (Heggeness et al., 2021). Moreover, parents reported higher levels of stress compared to adults without children (American Psychological Association, 2020). In a nationally representative survey, over 25% of U.S. parents reported “declining mental health” compared to the start of the pandemic (Patrick et al., 2020), and, in another sample, 2 of 5 parents met criteria for major depression or anxiety disorder (Lee et al., 2021). Collectively, parents voiced concerns about worsening mental health in themselves, their children, and their families, and an increased strain in family relationships (Evans et al., 2020). Although these studies suggest major mental health impacts of the pandemic for parents, many studies used retrospective or cross-sectional data, leaving open the question whether mental health changed longitudinally into the pandemic (and who was most likely to show change).

A handful of studies from various countries, including the United Kingdom, Canada, and Australia, have examined changes in parental mental health symptoms across the course of the COVID-19 pandemic and compared symptoms from before to during the pandemic. Overall, parental mental health symptoms, including anxiety, depression, and psychological distress, showed significant increases during the COVID-19 pandemic from pre-pandemic levels (Leach et al., 2023; Pierce et al., 2020; Racine et al., 2021; Wright et al., 2021) and parental stress and depression appeared to be at their worst when lockdown restrictions were high and improved as restrictions eased (Skripkauskaite et al., 2023). Although these studies prospectively assessed changes in mental health symptoms from before to during the pandemic, follow-up studies are required that address the same questions in more socioeconomically and racially diverse samples to test whether the results generalize to less advantaged and more marginalized populations.

Additionally, although average levels of parental mental health concerns may have increased, variability in the total number and type of pandemic-related changes parents experienced may have differentially impacted mental health. For example, studies documented that the total accumulation of COVID-19 related stressors (e.g., job loss, income loss, illness, caregiver burden, family conflict) were related to increased parent-reported daily negative mood and perceived stress (Brown et al., 2020; Gassman-Pines et al., 2020). Still, whether the accumulation of pandemic-related adversities is related to increased psychopathology from an earlier baseline remains unclear. Moreover, identification of which specific pandemic-related adversities most strongly predict parental mental health, can inform policies aimed at alleviating the adverse impacts. Importantly, while policy makers have focused on the pandemic’s economic impacts, media reports have highlighted the pandemic’s impact on family relationships and the balance between home and work, which may be critical factors undermining parental mental health (Igielnik, 2021). Moreover, important developmental theories, including the family stress model (Conger et al., 1994) and Bronfenbrenner’s ecological model (Bronfenbrenner & Morris, 1998), assert that broad external stressors (e.g., economic downturns) tend to impact parents and children via disruptions to proximal family relationships (e.g., more family conflict), underscoring the need to understand which types of pandemic-related changes most powerfully impact changes in parental mental health.

The current study investigates the COVID-19 pandemic’s impact on mental health in two samples of parents, one sampled from birth records with enrichment for neighborhoods with above average levels of poverty and another with an oversampling for non-marital births, resulting in a high representation of low-income families. Our study addresses three specific aims: (1) To examine change in parental mental health from before to during the pandemic; (2) Assess whether the accumulation of pandemic-related changes predicted changes in parental mental health; and (3) Examine which specific pandemic-related changes were most strongly associated with (changes in) parental mental health. We hypothesized that (1) depression, anxiety, and perceived stress would significantly increase during the pandemic; (2) greater accumulation of pandemic-related adversities would relate to greater depression, anxiety, and stress; and (3) that economic and interpersonal pandemic-related adversities, specifically, would be most highly associated with parental mental health. Importantly, we examined these aims within two well-sampled cohorts with enrichment for exposure to poverty and with substantial representation of families of color – all factors associated with being marginalized and facing a greater accumulation of stressors (and COVID-related stressors) (Abrams et al., 2022; Lantz et al., 2005; Pager & Shepherd, 2008; Ruprecht et al., 2021; Sellers et al., 2006).

Methods

Overview

Data reported in this study were drawn from two longitudinal studies: (1) the Twin Study of Behavioral and Emotional Development – Child (TBED-C), a subsample within the Michigan State University Twin Registry (MSUTR; Burt & Klump, 2019) and (2) the Future of Families and Child Wellbeing Study (FFCWS; Reichman et al., 2001), previously named the Fragile Families and Child Wellbeing Study.

Samples and Procedures

TBED-C –

Participants were recruited from the Twin Study of Behavioral and Emotional Development – Child (TBED-C), a subsample within the Michigan State University Twin Registry (MSUTR; Burt & Klump, 2019). TBED-C identified twins via birth records (within 120 miles of East Lansing, MI, including urban – e.g., Detroit, Flint, and Lansing, suburban, and rural areas). The study included a population-based sample of 528 twin families with children aged 6 – 10 years, and an “at-risk” sample of 502 twin families sampled from the same geographic region, but only from neighborhoods with above average levels of poverty (>10.5% of families in the neighborhood living below the poverty line at the time of recruitment; Burt & Klump, 2019). For the current study, we contacted 995 TBED-C families (i.e., those still living in MI). Of those, 799 families responded (90% response rate) and 693 agreed to participate during the early months of the pandemic (i.e., May-August 2020) resulting in a final participation rate of 70%. We examined our aims longitudinally within a subsample of parents (N=299) who had recently participated (between 2015–2020) in the Michigan Twin Neurogenetic Study (MTwiNS), originally recruited from the “at-risk” sample (Suarez et al., 2022; Tomlinson et al., 2020).

The pandemic-related data were collected via online surveys and phone interviews between May 2020 and January 2021; however, we restricted our analyses to parents who participated in the early months of the pandemic (i.e., May-August 2020) since the landscape of the pandemic changed dramatically into the fall and winter of 2020 and most parents participated in our study during this early period. Most parents were mothers (94.9%; M_age=48) and reflected the sampling frame/geographic region (Table S1): parents were 84% White, 7% Black, 6% other (3% missing) and highly educated (approximately 57% had at least a bachelor’s degree). Approximately half of the twins were female (53%) and they ranged in age from 11 to 22 years at the time of the COVID assessment (M=17, SD=2.5). In the current sample, 4.5% of families reported income below $20K and 15.9% of families reported income below $40K. Informed consent was obtained from all study participants. The institutional review boards at the University of Michigan and Michigan State University approved the current and past studies on this cohort.

SAND –

Participants were recruited from the Study of Adolescent Neural Development (SAND), a subsample of the Future of Families and Child Well-Being Study (FFCWS). FFCWS is a population-based sample of 4,898 children born in large cities (population over 200,000) across the United States with an oversampling (3:1) for non-marital births, which resulted in a high representation of low-income families and, depending on the geographic region, a high representation of families of color (Reichman et al., 2001). For the current study, we contacted 976 families primarily from 7 of the cities via phone calls, texts, emails, and mailings, and of those, 635 responded (65% response rate). We examined our aims longitudinally within a subsample of parents (N=175) from midwestern cities (Detroit, MI; Toledo, OH; Chicago, IL) who had also participated (between 2014 to 2017) in SAND when their children were 15–17 years old. Primary caregivers completed the pandemic-related surveys either through an online survey or phone interviews between May 2020 and September 2021; however, we restricted our analysis to parents who participated in the first year of the study (i.e., May 2020- May 2021; N = 624) – coinciding with the early months of the pandemic. Most parents were mothers (92%; M_age=47) and 51.9% identified as Black or African American, 41.5% identified as White, 6% other, and 0.5% missing (Table S1). The children of the primary caregivers ranged in age from 19 to 22 years at the time of the COVID assessment (M=20.2, SD=0.5). In the current sample, 27.1% of families reported income below $20K and 43.4% reported income below $40K. All study participants provided informed consent and study protocols for data presented in this report were approved by the University of Michigan Institutional Review Board.

Measures

Symptoms of Mental Illness –

Both samples utilized the same measures to assess symptoms of mental illness before and during the COVID-19 pandemic. Depressive symptoms were measured using the 21-item Beck Depression Inventory (BDI-II; Beck et al., 1996) (α=.87). One item related to ‘suicidal thoughts and wishes’ was removed from the BDI-II for the assessment during the pandemic. Anxiety symptoms were measured using the 21-item Beck Anxiety Inventory (BAI; Beck et al., 1993) (α=.90). Perceived stress was measured using the 14-item Perceived Stress Scale (PSS-14; Cohen et al., 1983) (α=.89).

COVID-19 Impacts –

Within both samples, we used the Epidemic-Pandemic Impacts Inventory (EPII; Grasso et al., 2020) to assess reported changes during the COVID-19 pandemic. The EPII contains 92 items assessing changes extending across several domains: Work/Employment (11 items), Education (2 items), Home Life (13 items), Social Activities (10 items), Economic (5 items), Emotional Health (8 items), Physical Health (8 items), Physical Distancing/Quarantine (8 items), Infection History (8 items; not administered in this study), and Positive Change (19 items). All domains except for “positive change” index pandemic-related adversities. In the instructions, participants were asked “Since the coronavirus disease pandemic began, what has changed for you or your family?” Participants read each statement and indicated whether the COVID-19 pandemic had impacted them in the way described (i.e., ‘yes’ = 1 or ‘no’ = 0). Examples of items reflecting change since the start of the pandemic include, “reduced work hours or furloughed,” and “separated from family or close friends.” Similar to previous studies using the EPII (Haydon & Salvatore, 2022; Hezel et al., 2022; Hussong et al., 2022; L. Williams et al., 2021), in the present analyses, we created two summary scores capturing negative pandemic-related changes (i.e., total number of negative events reported for self across domains) and positive pandemic-related changes (i.e., total number of positive changes reported for self). For the negative change sum score, we summed 57 items across the negative change subscales (α=.81), and for the positive change sum score, we summed 17 items from the Positive Change domain (α=.80). We excluded 10 items that had clear or moderate overlap with mental illness and/or symptoms of depression or anxiety, including 5 items from the Emotional Health domain (e.g., “increase in mental health problems/symptoms”, “increase in sleep problems/sleep quality”, “unable to access mental health treatment,” “not satisfied with changes in mental health treatment”, “increase in alcohol or substances), 3 items from the Physical Health domain (i.e., “less physical exercise or activity”, “overeating or eating more unhealthy food”, and “more time sitting down or being sedentary”), and 2 items from the Positive Change domain (“less alcohol or substances” and “increase in exercise or physical activity”). We also examined item-level associations as described below.

Demographics –

In both samples, parents reported on their own gender (male, female), age during COVID-19 assessment (in years), and racial/ethnic identity, a socially constructed category indexing unmeasured structural racism and unequal treatment and opportunity in the U.S. (TBED-C: White, non-White; SAND: Black, White, Other racial/ethnic group), and annual household income prior to the pandemic (Table S1). Since previous research has shown that disadvantage and/or marginalization (i.e., lower family income, minoritized identity) is associated with the accumulation of pandemic-related adversities (Abrams et al., 2022; Ruprecht et al., 2021), we controlled for parent race/ethnicity, gender, and household income.

Data Analytic Strategy

To examine our first aim, changes in parent mental health from pre-pandemic levels, we first conducted three paired-sample z-tests for each outcome (depression, anxiety, and stress) using the TBED-C (N-299) and SAND (N=175) longitudinal subsamples. To evaluate our second aim, the cumulative impact of pandemic-related changes on parent mental health, we conducted a single regression for each post-shutdown dependent variable (depression, anxiety, and stress) in the subsamples of parents with longitudinal data (TBED-C: N=299 and SAND: N=175), with the positive and negative change sum scores as predictor variables of interest. In this analysis, we controlled for pre-pandemic symptoms of mental illness and parent demographic information, including age, gender, race/ethnicity, and household income. In supplemental analyses using the larger samples with cross-sectional data (TBED-C: N=693 and SAND: N=624), we conducted the same regressions with cross-sectional data (though we could not include pre-pandemic symptoms of mental illness as they were not available in these larger single timepoint samples). Finally, to carry out our third aim, evaluating which specific pandemic-related changes were most strongly associated with parental symptoms of mental illness, we conducted regressions for each outcome, with each dichotomous item predicting continuous measures of parental mental health symptoms (depression, anxiety, and stress). Using data from the longitudinal subsamples of parents, TBED-C (N-299) and SAND (N=175), we adjusted for pre-pandemic levels of mental illness in each regression. Our primary goal for aim 3 was to describe the associations between each pandemic-related item and mental illness and to rank the effect sizes (i.e., t-values) to describe which items were most strongly associated with mental illness. In addition to the 10 items that had overlap with our outcome measures of mental health, we also excluded items due to low endorsement levels in the item-level analysis (<1% of parents endorsed these items). Therefore, 5 items for the TBED-C (“Had to move or relocate,” “Became homeless,” “Increase in physical conflict with partner or spouse,” “Increase in physical conflict with other adults in the home,” and “Unable to access clean water”) and 2 items for the SAND (“Became homeless” and “Unable to access clean water”) were not analyzed. Thus, we conducted a total of 69 regressions within the TBED-C sample and 72 regressions within the SAND sample. Though our goal was not null hypothesis significance testing, but rather to simply compare the effect sizes (i.e., which individual items had the largest association with each mental health outcome?), to address any concerns about multiple testing, we conducted false discovery rate (FDR; Benjamini & Hochberg, 1995) corrections for multiple comparisons within each mental health outcome analysis (TBED-C: 69 regressions per outcome; SAND: 72 regressions per outcome). In supplemental analyses, using R (version 4.2.1) we conducted a two-sample t-test for each dichotomous item with each continuous outcome (depression, anxiety, and stress) within the larger cross-sectional samples (TBED-C: N=693 and SAND: N=624). Additionally, because our measures of parental depression and anxiety were not normally distributed in either sample, we used the Wilcoxon Rank Sum test for analyses using these measures. Aside from the two-sample t-tests and Wilcoxon tests conducted in the cross-sectional sample, we conducted all models for our primary aims in Mplus (Version 8.10), using Maximum Likelihood estimation with robust standard errors to allow for missing data and protect against distortion of effects from violations of distributional assumptions (Falk, 2018).

Results

Change in Parents’ Mental Health from Before to During the COVID-19 Pandemic

Within the longitudinal TBED-C sample (N=299), parents showed a moderate increase (d=0.34) in depressive symptoms from before (M=7.19, SD=7.00) to during the pandemic (M=9.68, SD=7.69), B_difference=2.48, SE=0.43, 95% Confidence Interval (CI) [1.64, 3.33]. The percentage of parents meeting cut-off criteria for mild-to-moderate depression (BDI score between 14 – 28) increased from 16% pre-pandemic to 28% during the pandemic. In the TBED-C sample, neither anxiety (before: M=8.35, SD=8.38; during: M=8.60, SD=7.96, B_difference=0.25, SE=0.54, 95% CI [−0.82, 1.31], d=0.03), nor perceived stress (before: M=20.40, SD=8.27; during: M=20.21, SD=8.91, B_difference=−0.19, SE=0.62, 95% CI [−1.40, 1.02], d=−0.02) showed significant group-level change. In sensitivity analyses, we constrained the longitudinal TBED-C subsample to only include parents who had completed the pre-pandemic mental health measures within a 2-year window prior to the start of the pandemic (i.e., only parents who had pre-pandemic data between March 1, 2018 and March 1, 2020), resulting in a sample of N=259 parents. Corresponding with our findings from the larger longitudinal sample, parents showed a moderate increase (d=0.33) in depressive symptoms from before (M=7.14, SD=6.92) to during the pandemic (M=9.49, SD=7.35), B_difference=2.35, SE=0.46, 95% CI [1.46, 3.25]. The percentage of parents meeting cut-off criteria for mild-to-moderate depression increased from 17% pre-pandemic to 26% during the pandemic. Again, neither anxiety (before: M=8.54, SD=8.62; during: M=8.36, SD=7.59, B_difference=−0.18, SE=0.57, 95% CI [−1.31, 0.94], d=−0.03) nor perceived stress (before: M=20.34, SD=8.47; during: M=19.81, SD=8.76, B_difference=−0.52, SE=0.65, 95% CI [−1.81, 0.76], d=−0.06) exhibited significant group-level change.

Similar to the findings in the TBED-C sample, in the longitudinal SAND sample (N=175), there was also a small group-level increase (d=0.21) in depressive symptoms from before (M=7.87, SD=7.55) to during the pandemic (M=9.73, SD=9.67), B_difference=1.86, SE=0.63, 95% CI [0.63, 3.09]. The percentage of parents meeting cut-off criteria for moderate-to-severe depression (BDI score between 20–63) increased from 8% pre-pandemic to 16% during the pandemic. Additionally, we observed a small group-level increase (d=0.25) in anxiety symptoms from before (M=8.02, SD=9.63) to during the pandemic (M=10.64, SD=10.75), B_difference=2.62, SE=0.84, 95% CI [0.97, 4.27]. The percentage of parents meeting cut-off criteria for moderate-to-severe anxiety (BAI score between 16–63) increased from 15% pre-pandemic to 27% during the pandemic. Perceived stress did not show significant group-level change (before: M=19.07, SD=7.51; during: M=19.77, SD=9.51, B_difference=0.71, SE=0.69, 95% CI [−0.65, 2.06], d=0.08).

Pandemic-Related Changes and Parent Mental Health

In the longitudinal TBED-C subsample (N=299), while controlling for parent demographics (age, gender, race/ethnicity, income) and pre-pandemic levels of mental illness, the accumulation of pandemic-related adversities and positive changes were strongly associated with individual differences in all current measures of parent mental illness symptoms (depression: R²=.33, anxiety: R²=.25, and stress: R²=.21). Parents who reported a greater number of adversities also reported significantly more depressive and anxiety symptoms and greater perceived stress (Table 1, Table S2). Also, pandemic-related positive changes were inversely related with depressive symptoms and perceived stress but only inversely related to anxiety symptoms at a trend level (Table 1, Table S2). Results were similar within the larger TBED-C sample with cross-sectional data (N=693) with the only minor difference being that the accumulation of positive changes was also significantly inversely related to anxiety symptoms (Table S4).

Table 1.

Pandemic Related Changes Predict Parental Depression, Anxiety, and Stress

Outcome	Model	β	B	SE	95% CI
Results TBED-C (N = 299)
Depression	Negative Change Sum Score	.26	.39***	.07	[.25, .52]
	Positive Change Sum Score	−.13	−.29***	.10	[−.48, −.10]
Anxiety	Negative Change Sum Score	.31	.48***	.08	[.32, .64]
	Positive Change Sum Score	−.09	−.20	.10	[−.40, .01]
Stress	Negative Change Sum Score	.34	.59***	.10	[.39, .78]
	Positive Change Sum Score	−.20	−.52***	.14	[−.78, −.25]
Results SAND (N = 175)
Depression	Negative Change Sum Score	.34	.48***	.10	[.28, .67]
	Positive Change Sum Score	−.20	−.53***	.14	[−.81, −.26]
Anxiety	Negative Change Sum Score	.34	.52***	.15	[.23, .81]
	Positive Change Sum Score	−.19	−.55**	.21	[−.96, −.15]
Stress	Negative Change Sum Score	.19	.26**	.09	[.08, .44]
	Positive Change Sum Score	−.16	−.40*	.18	[−.75, −.05]

We found that the accumulation of negative changes and lack of positive changes was associated with greater anxiety and depressive symptoms and perceived stress among parents. One regression per outcome (depression, anxiety, and perceived stress) was conducted, including both the positive and negative change sum scores as predictor variables. All models controlled for parent demographics (age, gender, race/ethnicity, and income) and pre-pandemic levels of mental health symptoms. β = Standardized Estimate. B = Unstandardized Estimate. SE = Standard Error. CI = Confidence Interval.

^*p < .05;

^**p < .01;

^***p < .001.

TBED-C = Twin Study of Behavioral and Emotional Development – Child. SAND = The Study of Adolescent Neural Development.

Similar to the TBED-C sample, accounting for parent demographics (age, gender, race/ethnicity, income) and pre-pandemic levels of mental illness within the longitudinal SAND subsample (N=175), parents who reported a greater number of adversities and fewer positive changes also reported significantly more depressive (R²=.42) and anxiety symptoms (R²=.29) and greater perceived stress (R²=.30) (Table 1, Table S3). Much like the twin sample, these results were similar within the cross-sectional SAND sample (N=624; Table S4). In the supplement, we report results with all pandemic negative and positive change items, including those that overlap with symptoms of mental illness, included in the sum scores (Tables S4 & S5).

Lastly, in sensitivity analyses within the time constrained longitudinal TBED-C sample (i.e., those who had pre-pandemic data between March 1, 2018 and March 1, 2020; N=259) the results were highly similar: more pandemic-related adversities were significantly associated with worse depressive (B=.39, SE=.08, β =.27, 95% CI [.24, .54]) and anxiety (B=.44, SE=.09, β=.30, 95% CI [.26, .62]) symptoms and greater perceived stress (B=.56, SE=.11, β =.33, 95% CI [.35, .78]), and positive change was significantly inversely related to depressive (B=−.28, SE=.10, β=−.13, 95% CI [−.48, −.08]) and anxiety symptoms (B=−.27, SE=.11, β=−.12, 95% CI [−.48, −.06]) and perceived stress (B=−.54, SE=.15, β=−.21, 95% CI [−.83, −.25]).

Specific pandemic-related changes most strongly associated with parental mental health

Depression Symptoms.

Within the subsample with longitudinal data (N=299), accounting for the effect of previous depression symptoms, the top 15 items with the strongest association with depressive symptoms were primarily from the following domains (Fig. 1a, Table S6): Home Life (4 items), Physical Distancing/Quarantine (3 items), Emotional Health (2 items), Physical Health (1 item), Work & Employment (2 items), Positive Change (1 item), and Social Change (2 items). The most highly associated items were: 1) those reflecting interpersonal conflict and difficulty taking care of people in the home (e.g., increased parent-child conflict, increased child behavioral problems, increased conflict with other adults in the home, difficulty taking care of children and other family members) 2) changes to work life (e.g., difficulty transitioning to remote work, difficulty juggling work and taking care of people in the home), 3) increased health problems not related to disease, and 4) [less] time spent doing enjoyable activities or being with family and friends (e.g., limited physical closeness with a child or loved one due to concern of exposure, separation from close family or friends, and isolation due to symptoms of disease or health conditions that increase risk). Notably, some items that were expected to be strongly associated with depression (e.g., reduced work hours, furloughed) were not as strongly related. In the cross-sectional TBED-C sample (N=693), a similar pattern emerged: The same domains (home life, physical health, emotional health, physical distancing, and positive change) were most represented, and items related to interpersonal conflict and difficulty taking care of people in the home were most prominent (e.g., increased conflict with children and spouse) (Table S6).

Figure 1.

Top 15 Pandemic Impacts Associated with Parental Depressive Symptoms

While adjusting for previous symptoms of depression, we conducted a regression for each dichotomous item predicting depression during the pandemic and ranked the effect sizes (i.e., t-values) to evaluate which specific pandemic-related changes were most impactful on parental mental health. Here we show the top 15 items most strongly related to depression for each sample.

a. TBED-C = Twin Study of Behavioral and Emotional Development – Child

b. SAND = The Study of Adolescent Neural Development

Within the SAND subsample with longitudinal data (N=175), the 15 items most strongly associated with depressive symptoms, while accounting for previous depression symptoms, overlapped with some items that were also prominent in the TBED-C sample (Fig. 1b, Table S9), including items relating to (1) interpersonal conflict and difficulty taking care of people in the home (e.g., spending more time taking care of a family member, more inter-partner conflict, and more child sleep difficulties or nightmares), (2) physical health changes (e.g., increase in health problems not related to the disease), and (3) social changes (e.g., separation from close family and friends). There were also several items that did not overlap with the TBED-C sample, including those related to (1) economic concerns (e.g., unable to pay important bills, unable to get enough food or healthy food, unable to get needed medications, and difficulty getting places due to reduced public transit or safety concerns), (2) access to health care (e.g., unable to access medical care for serious condition, important medical procedures canceled), (3) positive change (found greater meaning in work, employment, or school), (4) home life (increase in physical conflict with a partner or spouse and with children), and (5) emotional health (spent more time on screens or devices). Within the full SAND sample (N=624), a similar pattern emerged, with items related to economic concerns, home life, and physical health being prominent items related to parental depression.

Anxiety Symptoms.

Within the TBED-C subsample with longitudinal data (N=299), controlling for previous anxiety symptoms, many of the same items related to depression were among the top 15 items with the strongest association with anxiety symptoms (Fig. 2a, Table S7), including increased physical health problems, increased interpersonal conflict in the home, increased child behavioral problems, and physical distancing/quarantine. However, some items were uniquely related to parental anxiety in this sample, including one economic item (“unable to get enough food or healthy food”), one emotional health item (“spent more time on screens or devices”), two home life items (“family or friends had to move into your home” and “increase in verbal arguments with partner”), and one physical distancing and quarantine items (“being isolated or quarantined due to possible exposure to disease”). A similar pattern emerged in the full sample of parents with cross-sectional data (N=693), such that items reflecting increased interpersonal conflict and child behavioral problems, physical health changes and isolation or quarantine were most impactful (Table S7).

Figure 2.

Top 15 Pandemic Impacts Associated with Parental Anxiety Symptoms

While adjusting for previous symptoms of anxiety, we conducted a regression for each dichotomous item predicting anxiety during the pandemic and ranked the effect sizes (i.e., t-values) to evaluate which specific pandemic-related changes were most impactful on parental mental health. Here we show the top 15 items most strongly related to anxiety for each sample.

a. TBED-C = Twin Study of Behavioral and Emotional Development – Child

b. SAND = The Study of Adolescent Neural Development

Like the TBED-C sample, in the SAND subsample with longitudinal data (N=175), when adjusting for past anxiety, many of the same items related to depressive symptoms were among the 15 items most strongly associated with anxiety symptoms (Fig. 2b, Table S10), including increased physical health problems not related to the disease, increased child behavioral problems, and more interpersonal conflict in the home and difficulty taking care of people in the home. Also, many items of the same items related to depression in the SAND subsample were also prominent in associations with anxiety, including items related to accessing healthcare services, increased economic concerns, and positive change (i.e., finding greater meaning in work, employment, or school). Two items related to physical health and healthcare were uniquely related to parental anxiety in this sample (“received less medical care than usual” and “elderly or disabled family member in home unable to get care”). Within the full SAND sample (N=624), a similar pattern emerged, with items related to physical health and healthcare, economic concerns, and interpersonal relationships being strongly associated with parental anxiety.

Perceived Stress.

While adjusting for previous perceived stress levels in the TBED-C subsample with longitudinal data (N=299), we found that many of the same items that were related to depression and anxiety were among the top 15 items related to parental perceived stress (Fig. 3a, Table S8), including increased interpersonal conflict in the home and difficulty taking care of children in the home, worsening child behavioral and sleep problems, increased physical health problems not related to the disease, work and employment difficulties, and positive changes (i.e., more time doing enjoyable activities). Still, there were items uniquely related to parental perceived stress, including one item from the Work/Employment domain, “provided supportive care to people with the disease,” one item from the Social Life domain, “unable to do enjoyable activities or hobbies,” and three items from the Positive Change domain, “more quality time with children and partner or spouse and improved relationship with family or friends.” Within the full TBED-C sample (N=693), a similar pattern emerged (Table S8), such that items reflecting interpersonal conflict and difficulty taking care of people in the home were most prominent, alongside items related to physical health, work and employment, and child emotional and behavioral problems.

Figure 3.

Top 15 Pandemic Impacts Associated with Parental Perceived Stress

While adjusting for previous stress, we conducted a regression for each dichotomous item predicting perceived stress during the pandemic and ranked the effect sizes (i.e., t-values) to evaluate which specific pandemic-related changes were most impactful on parental mental health. Here we show the top 15 items most strongly related to perceived stress for each sample.

a. TBED-C = Twin Study of Behavioral and Emotional Development – Child

b. SAND = The Study of Adolescent Neural Development

When adjusting for previous stress levels in the SAND subsample (N=175), similar items found to be associated with depression and anxiety in this sample also emerged in relation to parental perceived stress (Fig. 3b, Table S11), including economic concerns (e.g., unable to get enough food or healthy food), interpersonal conflict in the home (e.g., increased verbal arguments and physical conflict with a partner or spouse), physical health concerns and decreased access to healthcare (e.g., important medical procedure cancelled), and worsening child emotional and sleep problems. One positive change item was uniquely related to parental perceived stress in this sample, “more efficient or productive in work, employment, or school.” Within the full SAND sample (N=624), similar items emerged particularly within domains related to home life, economic concerns, and positive change (Table S11).

Discussion

In the current study, we examined associations between the impacts of the COVID-19 pandemic and parental mental health longitudinally across two samples, one sample representative of parents of adolescent twins with enrichment for families living in neighborhoods with above average levels of poverty, and another sample of parents of young adults with an oversampling for non-marital births, resulting in a high representation of low-income families. In both samples of parents, symptoms of depression showed a group-level increase from before to during the pandemic; however, we only found an increase in anxiety symptoms within the SAND. When examining cumulative endorsements of pandemic-related changes, the accumulation of adversities and the absence of positive changes were both associated with greater parental mental illness symptoms in both samples, even while controlling for covariates reflecting disadvantage and/or marginalization, which are associated with unequal exposures (e.g., family income, minoritized group membership). Lastly, we highlight the specific pandemic-related changes that were consistently strongly related to parental mental health and point out the similarities and differences between the two samples. For example, although pandemic adversities related to interpersonal relationships in the home, difficulty taking care of individuals in the home, and physical health concerns consistently appeared in the top 15 for both samples, economic difficulties and changes to physical health and healthcare access appeared to be much more prominent issues for parents in the SAND sample (i.e., impacts appearing in the top 5); whereas items relating to interpersonal relationships in the home were consistently most prominent in the TBED-C. Throughout the discussion, we consider important differences between the two samples, including: financial resources (i.e., SAND parents are lower income compared to TBED-C parents), ages of children in the home (i.e., TBED-C parents have younger children compared to SAND parents), and family make-up (i.e., TBED-C parents have twins) that may account for the differences we observed in pandemic impacts and change in mental health from before to during the pandemic. Thus, the pandemic appears to have negatively impacted parental mental health, particularly depression, and that risk is increased by the accumulation of pandemic-related adversities, particularly those reflecting greater interpersonal conflict in the home, changes to physical health and healthcare, and economic challenges. These findings can help inform systemic and individual efforts to ameliorate the pandemic’s negative impacts on families, particularly those that have experienced higher rates of exposure to adversities or stressors.

Across both samples, we found a significant group-level increase in depression, but not perceived stress; and we only found a small significant group-level increase in anxiety in the SAND but not the TBED-C. These findings are consistent with previous work from other countries which found that parental mental health problems showed significant increases during the COVID-19 pandemic from pre-pandemic levels (Leach et al., 2023; Pierce et al., 2020; Racine et al., 2021; Wright et al., 2021). Considering the null results for change in anxiety within the TBED-C, some parents with greater anxiety or stress levels prior to the pandemic may have found the social distancing measures early on (i.e., May to August 2020) to be helpful in terms of reducing exposure to events that evoke symptoms, leading to one group with decreased symptoms and another with increased symptoms, which would yield no significant group-level change in the sample. Alternatively, we may see increased levels of anxiety in the SAND but not in the TBED-C because of differences between the two samples regarding household income and economic concerns or problems related to physical health and healthcare. For example, parents in the TBED-C reported relatively higher incomes; therefore, it is possible that parents in the TBED-C with relatively greater financial resources were less burdened by economic difficulties experienced during the pandemic, whereas caregivers in the SAND, who were less able to offset financial difficulties, experienced worsening anxiety. This hypothesis is somewhat supported by our item-level analysis, which found four economic difficulty items to be among the top 15 pandemic impacts on anxiety for parents from the SAND sample, whereas these items did not appear as prominent in the TBED-C. Our item-level analysis also revealed that access to healthcare (e.g., important medical procedures cancelled, increase in health problems not related to disease, elderly or disabled family member not in the home unable to get care) was an important factor associated with anxiety for SAND parents, whereas only one physical health item appeared as important in the TBED-C. Therefore, the added burden of receiving less access to medical care than usual may have also played a role in the worsening anxiety for SAND caregivers. Future studies should examine patterns of how mental health has changed across a longer period of time, as the long-lasting pandemic has gone through many peaks and plateaus that could either lead to acclimation, alleviation of symptoms, or further deterioration of mental health.

While controlling for previous symptoms of mental illness, in both subsamples (TBED-C: N=299 and SAND: N=175) we found that the accumulation of pandemic-related adversities (and, to a lesser extent, lack of positive change) was related to worsening symptoms of parental mental illness, which is consistent with previous reports of COVID-19 related stressors on mental health in parents (Brown et al., 2020; Gassman-Pines et al., 2020) and adult populations (Haydon & Salvatore, 2022), more broadly. However, the replication of this finding within two large samples and our ability to control for pre-pandemic symptoms of mental illness strengthens our conclusions. Furthermore, in the longitudinal subsamples (TBED-C: N=299 and SAND: N=175), greater pandemic-related positive change was associated with decreased anxiety, depression, and stress, but exerted a smaller effect compared to the accumulation of adversities. Still, positive changes appeared to confer some benefits for parents within both samples.

Lastly, though the “cumulative risk” approach employed here and in other studies can help explain how the accumulation of adversities impacts mental health, it is also important to understand which specific adversities appear to be most harmful. For both the TBED-C and SAND samples, our item-level analysis revealed items reflecting increased interpersonal conflict in the home (e.g., increased parent-child conflict, more child behavioral problems, increased conflict with partner/spouse) emerged as high-impact items for all three outcomes. This finding aligns with the family stress model, which posits that a rise in parent emotional problems and coercive family functioning (e.g., parent-child conflict, harsh discipline, marital strife) may be the proximal mechanism through which increased hardship impacts families (Conger et al., 1994). Taken together, these results emphasize that interpersonal relationships are key to mental health, particularly during a period in which many interactions take place within the home.

Interestingly, for parents in the TBED-C, items related to increased family conflict were more strongly related to negative outcomes than those items related to unemployment and economic hardship; a surprising result, given the attention in the media and society to the economic and social hardships associated with the pandemic (Banerjee & Rai, 2020; Mann et al., 2020). In contrast, for the SAND, although items related to interpersonal conflict in the home also emerged as high-impact items for all three outcomes, we also found that items related to economic challenges (e.g., unable to get enough food or healthy food, unable to pay important bills like rent or utilities, difficulty getting places due to reduced public transit or safety concern) and physical health changes and access to healthcare (e.g., increase in health problems not related to disease, unable to access medical care for a serious condition, important medical procedures cancelled) were quite impactful on mental health. For the parents in the SAND, items related to economic challenges and physical health and healthcare changes often emerged within the top 5 impacts on mental illness, whereas, for parents in the TBED-C, the most prominent items were related to interpersonal conflict in the home.

Important differences between the TBED-C and SAND samples may help explain some of the observed differences regarding the most prominent items associated with parental mental illness. Notably, compared to the SAND caregivers the TBED-C parents have younger children (Child Age: TBED-C M=17 years, Range 11 – 22; SAND M=20 years, Range 19–22). The younger children in the TBED-C may have needed additional care, attention, and help with emotion regulation during the pandemic compared to the young adult children in the SAND sample, which may have contributed to the TBED-C caregivers struggling more with interpersonal conflict, child behavioral problems, and difficulty taking care of children in the home. Also, the TBED-C caregivers have twin children, which may result in more parental burden and increased conflict with the children in the home as well as with their partner or spouse. Furthermore, the SAND sample of parents have even lower income and resources compared to the TBED-C sample and also has more racial diversity with greater representation of families of color (51.8% of the SAND sample is Black compared to 6.6% in the TBED-C). Researchers report significant disparities for health outcomes and access to, use of, and quality of health care among individuals from low socioeconomic and/or marginalized backgrounds (Adler & Newman, 2002; Braveman et al., 2010; D. R. Williams & Jackson, 2005), and previous work has shown that COVID-19 disproportionally affected families experiencing economic disadvantage and individuals from marginalized backgrounds (Abrams et al., 2022; Ruprecht et al., 2021). Future studies should examine the directionality of associations between pandemic adversities related to financial and employment challenges, parental mental health, and interpersonal conflict in the home.

The present study included several methodological strengths, including two large population-based sample and incorporation of longitudinal data using the same measures to assess change in parental symptoms of mental illness before and during the COVID-19 pandemic. For each sample we utilized a strong sampling frame: in the TBED-C, we extended recruitment to families from rural, urban, and suburban communities, with oversampling for families living in disadvantaged neighborhoods. In the SAND, we oversampled for non-marital births in urban settings, resulting in high representation of low-income families. However, the current study is not without limitations: First, we only examined our aims longitudinally within the smaller subsamples of parents who had available data at a timepoint prior to the pandemic. Within the TBED-C subsample, families were sampled only from neighborhoods with above average poverty – both a strength and weakness in terms of generalizability. Second, the time frame for when pre-pandemic mental health was measured varied across the two samples. For the TBED-C subsample, data collection started in 2015 and was still ongoing when the COVID-19 pandemic hit; 86.6% of the subsample had participated between 2018–2020. This allowed us to conduct sensitivity analyses in which we constrained the TBED-C longitudinal sample to only include parents who had completed the pre-pandemic mental health measures within a 2-year window prior to the start of the pandemic. However, for the SAND subsample, data was collected when the target children were 15–17 years old (between 2014–2017). Thus, a longer time had passed between data collection before and during the pandemic for the SAND subsample. Also, the time frame during which our pandemic data was collected varied across the two samples. In the TBED-C, the tight timeframe (May-August 2020) allowed us to understand change occurring during the early pandemic, including schools’ closures and summer, but our results may not reflect chronic changes that happened since the start of the pandemic (e.g., the 2020–2021 school year in which some schools re-opened at various times). The SAND sample captures a larger period of the pandemic (May 2020-May 2021), but, as a result, makes it difficult to determine which periods of the pandemic most impacted parental mental health. Although comparing and contrasting findings across these two cohorts is useful from a replication perspective, the two cohort have several fundamental differences (e.g., sampling frame, age of children, demographics, timing of the pandemic data collection) that make drawing conclusions about why results differ challenging. That is, though many results were very similar across cohorts, helping to increase confidence in conclusions, when differences did emerge (e.g., economic items being more predictive of mental health in SAND vs. TBED-C), we can only speculate on which of the many factors between cohorts may explain these differences. Third, our analysis was restricted to two parent samples in the U.S. (i.e., TBED-C: Michigan; SAND: urban Midwest cities – Detroit, MI; Toledo, OH; Chicago, IL). Thus, our findings may not generalize to populations in other countries or other regions of the U.S. Differences between countries and States in terms of disease transmission rates, healthcare burden, population composition, government structure and stringency, and timing of stay-at-home mandates all likely play a role in shaping the experience and outcomes of local residents. Fourth, our samples include parents of adolescent twins and young adults; thus, our findings may not generalize to parents with younger children, who may have experienced more difficulty with virtual schooling and required more supervision. Lastly, although most parents were mothers across the two samples, who have experienced additional burdens throughout the pandemic, we included all caregivers in our analysis without power to examine whether parent gender moderated effects; still, all results remain consistent when analyses were restricted to mothers only.

Despite these limitations, we provide important information on how parents with adolescents and young adults were functioning during the COVID-19 pandemic following mandatory stay-at-home orders. Our results suggest that since the start of the pandemic, parents experienced worsening mental health problems, particularly depression and potentially anxiety, which was exacerbated by the accumulation of pandemic-related adversities, especially those related to increased interpersonal conflict at home, increased burden of juggling work and caregiving responsibilities, economic difficulties, and changes to physical health and healthcare. Therefore, parents need increased mental health support, as well as systemic change to support managing various work and caregiving responsibilities (or, alternatively, systemic level changes to relieve parents of the expectation that they juggle multiple full-time roles at the same time). These findings further emphasize that some parents are suffering and that pandemic-related changes have negatively impacted interpersonal interactions in families, to the detriment of parental mental health.

Availability of Data and Materials

All TBED-C and MTwiNS data for the current study will be shared publicly via the NIMH Data Archive, as mandated in our funding agreement.

Publicly available study data: https://nda.nih.gov/edit_collection.html?id=2818

Burt, S. A., & Hyde, L. W.; 2017; Mechanisms underlying resilience to neighborhood disadvantage; NIMH Data Archive; 2818

All SAND data for the current study will be shared publicly via the NIMH Data Archive, as mandated in our funding agreement.

Publicly available study data: https://nda.nih.gov/edit_collection.html?id=3306

Mitchell, C., Hyde, L. W., and Monk, C. S. 2019. Computational examination of RDoC threat and reward constructs in a representative, predominantly low-income, longitudinal sample at increased risk for internalizing disorders. NIMH Data Archive. 3306