Parent Mental Health and Child Behavior during the COVID-19 Pandemic

Abstract

1 Introduction Children’s behavioral functioning is associated with many important outcomes, including concurrent and future mental health, academic success, and social well-being (Duncan et al., 2007, Liu et al., 2011, Reardon and Portilla, 2016). Child behavior encompasses both internalizing (i.e., inward behaviors) and externalizing behaviors (i.e., outward behaviors) (Achenbach et al., 2016), and behavior problems in early childhood may place children at risk for later mental health problems, academic problems, and social problems (van Lier et al., 2012, Weeks et al., 2016). Thus, understanding sources of variability in child behavior is crucial for developing strategies aimed at equipping children with the behavioral resources necessary to thrive. Recent research has reported increases in child behavior problems during the ongoing COVID-19 pandemic, highlighting the threat the pandemic poses to children (Colizzi et al., 2020, Patrick et al., 2020, Saurabh and Ranjan, 2020, Whittle et al., 2020). In this study, our goal is to examine whether change in child behavior during the pandemic varies as a function of two important risk factors: parental mental health and preterm birth. Two factors previously associated with increases in child behavior problems include: 1) parental mental health difficulties (Slomian et al., 2019), which have been shown to increase during environmental stressors (e.g., pandemics, natural disasters) (Brock et al., 2015, Harville et al., 2010), and 2) preterm birth (birth occurring prior to 37 weeks gestation) (Allotey et al., 2018). Up to 1 in 10 children in the United States are born prematurely, and it is estimated that the annual societal economic impact of preterm birth is $26.2 billion due to health and developmental problems, including behavior problems (Institute of Medicine (US) Committee on Understanding Premature Birth and Assuring Healthy Outcomes, 2007). Moreover, not only are parental mental health difficulties more common among parents of preterm children (Vigod et al., 2010), but preterm children may also be more sensitive to environmental stressors, like parental stress, than full-term children (Gueron‐Sela et al., 2015). Per the differential susceptibility hypothesis (Belsky et al., 2007), preterm children may be more susceptible than their full-term counterparts to both the consequences of negative environmental exposures and to the benefits of positive ones (Gueron‐Sela et al., 2015; Shah et al., 2013). Taken together, children of parents with mental health difficulties and children born prematurely might be particularly vulnerable to increases in child behavior problems during the pandemic. However, several important gaps in the literature still exist. Researchers have documented concurrent change in parental mental health and child behavior during the current pandemic (Patrick et al., 2020, Whittle et al., 2020). However, whether and how a global stressor relates to changes in child behavior via changes in parental mental health remains unclear. Relatedly, most literature focuses on the role of parental mental health disturbances. Less is known about the role of parental well-being. That is, most literature focuses on the presence vs. absence of unpleasant emotional experiences but largely ignores the presence vs. absence of pleasant emotional experiences. If preterm children are indeed more sensitive than full-term children to both negative and positive exposures, it is important to examine the continuum of positive mood in addition to negative mood. Furthermore, evidence for the differential susceptibility hypothesis is mixed. While some studies have found support for it (Gueron‐Sela et al., 2015), others have not (Hadfield et al., 2017), perhaps in part due to variations in the characteristics (e.g., chronological age, gestational age) of the children included. In the current study, we examined how change in parental mental health during the pandemic relates to change in child behavior and whether preterm children are more behaviorally susceptible than full-term children to changes in parental mental health. Specifically, we examined the following questions: 1. How has parental depression, anxiety, and well-being changed since the COVID-19 pandemic began? Based on previous findings, we hypothesize that parental depression and anxiety symptoms will have increased. However, due to limited prior literature, we do not have specific hypotheses about changes in well-being. 2. How have children’s internalizing and externalizing symptoms changed since the COVID-19 pandemic began? Based on previous findings, we hypothesize that children’s internalizing and externalizing symptoms will have increased. 3. Is change in parental depression, anxiety, and well-being associated with change in children’s internalizing and externalizing symptoms? Does prematurity moderate that relation? Based on previous findings, we expect increases in parental depression and anxiety symptoms to be associated with increases in children’s internalizing and externalizing symptoms. However, based on mixed findings, our questions about the role of parental well-being and the role of prematurity will be exploratory. Elucidating relations between parental mental health, child behavior, and prematurity has the potential to inform efforts aimed at supporting child behavior , especially for children who have biological risk factors. If preterm children are indeed behaviorally more susceptible to changes in parental mental health during the ongoing pandemic, our findings can inform prevention and intervention efforts aimed at supporting preterm children and their families. Additionally, they will inform theoretical discussions on the differential susceptibility of children born prematurely. 2 Methods 2.1 Participants and Procedure Participants in the current study included a subset of participants from an ongoing study, who were recruited via electronic health records at a local university hospital. The larger ongoing study, hereafter referred to as T1, aims to examine the role of prematurity in children’s outcomes. Parent-child dyads were eligible for T1 if the child was between the ages of 3 and 5 years old, and had no history of a genetic syndrome, birth defect, or intellectual/ developmental disability. Parents were invited to participate via phone, email, social media, and word-of-mouth. Enrollment for T1 began in June 2019. At the time of their T1 laboratory visit, 91.2% of dyads were from [Midwestern State]. Parent-child dyads attended a laboratory visit, during which parents completed consent forms, children completed assessments, and parents completed questionnaires. At their T1 visit, parents indicated whether they would like to be informed of future studies. In mid-March of 2020, we paused data collection due to COVID-19-related restrictions. In April 2020, we invited parents who completed T1 and indicated they’d like to hear about future studies to participate in a “COVID-19 Follow-Up” study, hereafter referred to as T2. For T2, we contacted parents via phone and email. Of note, only parent-child dyads who completed T1 were eligible for T2. This study received IRB approval from the University of [State] (IRB ID#201805712, initially approved on October 30, 2018). For T2, parents completed online questionnaires between April 2020 and October 2020. An average of 273 days elapsed (SD = 93 days, range = 77-442 days) between T1 and T2. A total of 48 parents completed T2, out of the 104 parents that had completed T1. Of note, parents of two sets of twins provided data for T2; however, we only included data for one twin per set in the final T2 dataset. Most parents at both T1 and T2 were Non-Hispanic White (87.1%, 87.5%, respectively), female (91.2%, 93.8%, respectively), and had at least a bachelor’s degree (77.6%, 79.2%, respectively) (Table 1 ). Of the parents that participated in T2, 86% were from [Midwestern State]. Other states included Ohio, Illinois, and Texas. We did not find any significant differences on T1 demographic characteristics, T1 child behavior scores, or T1 parental mental health scores between those dyads that participated in T2 and those that did not. Moreover, parents of preterm children and parents of full-term children did not differ from each other in their parental mental health scores at T1 or T2, and preterm children and full-term children did not differ from each other on their child behavior scores at T1 or T2 (Table 2 ). Table 1 Demographic Information for the Study Population T1 T2 M(SD) M(SD) Parent Age (years) 36.32 (5.01) 37.44 (4.51) Child Age (years) 4.60 (0.62) 5.34 (0.84) Child Gestational Age (weeks)- Full-Term 39.45 (1.39) 39.32 (1.09) Child Gestation Age (weeks)- Preterm 30.33 (4.58) 30.40 (4.49) Household Income (USD) 115,761 (74,984) 118,744 (64,032) N(%) N(%) Parent Gender Female 93 (91.2%) 45 (93.8%) Male 8 (7.8%) 2 (4.2%) Other 1 (1.0%) 1 (2.1%) Parent Race/Ethnicity Non-Hispanic White 88 (87.1%) 42 (87.5%) Hispanic White 3 (3.0%) 0 (0.0%) Non-Hispanic Black 6 (5.9%) 2 (4.2%) Non-Hispanic Asian 4 (4.0%) 4 (8.3%) Parent Education High School 3 (2.9%) 0 (0%) Some college 6 (5.8%) 3 (6.3%) Associate’s Degree 14 (13.6%) 7 (14.6%) Bachelor’s Degree 43 (41.7%) 19 (39.6%) Graduate Degree 37 (35.9%) 19 (39.6%) Child Gender Female 50 (48.5%) 22 (45.8%) Child Race/ Ethnicity Non-Hispanic White 86 (82.7%) 41 (85.4%) Hispanic White 5 (4.8%) 1 (2.1%) Non-Hispanic Black 5 (4.8%) 2 (4.2%) Hispanic Black 2 (1.9%) 0 (0.0%) Non-Hispanic Asian 5 (4.8%) 4 (8.3%) Non-Hispanic Other 1 (1.0%) 0 (0.0%) Child Premature (<37 weeks gestation) Yes 36 (35.0%) 20 (41.7%) Note: For T1, N=104; For T2, N=48 Table 2 T1 and T2 IDAS-II and CBCL Scores for Preterm and Full-Term Children T1 Preterm Full-Term t-test CBCL- Internalizing 9.72 6.67 t(43)=-1.39, p=0.17 CBCL- Externalizing 12.22 9.85 t(43)=-0.81, p =0.42 IDAS- General Depression 35.61 34.36 t(41)=-0.38, p =0.71 IDAS- Social Anxiety 8.75 8.07 t(46)=-0.69, p =0.49 IDAS- Well Being 24.00 26.11 t(44)=1.15, p =0.26 T2 Preterm Full-Term t-test CBCL- Internalizing 9.33 10.00 t(44)=0.24, p =0.81 CBCL- Externalizing 11.94 13.14 t(43)=0.38, p =0.71 IDAS- General Depression 36.29 39.96 t(40)=1.02, p =0.31 IDAS- Social Anxiety 7.53 8.30 t(42)=0.99, p =0.33 IDAS- Well Being 19.41 21.19 t(42)=0.95, p =0.35 2.2 Measures Timepoint. Timepoint refers to T1 (pre-pandemic) and T2 (during pandemic). Timepoint was dummy-coded with T1 as the reference point. Days Elapsed. Days elapsed refers to the number of days between when participants completed T1 and when they completed T2. Prematurity. Parents reported their child’s gestational age at T1. We considered those born at <37 weeks gestation preterm and those born at ≥37 weeks gestation full-term. Parental Mental Health. Inventory of Depression and Anxiety Symptoms, Expanded Version (IDAS-II). Parents completed the IDAS-II (Watson et al., 2007, Watson et al., 2012) at both T1 and T2. The IDAS-II is a 99-item inventory. Parents reported on the extent to which they experienced various depression and anxiety symptoms during the previous two weeks, using a 5-point Likert scale. We used three subscales from the IDAS-II to reflect parental mental health: 1) General Depression, 2) Social Anxiety, and 3) Well-Being. It should be noted that this was not a clinical sample (Stasik-O’Brien et al., 2019). Using the screening cut-offs from Stasik-O’Brien et al. (2019), 14% of parents at T1 and 16.7% of parents at T2 scored above the screening cut-off for Major Depressive Disorder on the IDAS-II General Depression Scale; 35.4% of parents at T1 and 22.2% parents at T2 scored above the screening cut-off for Social Phobia on the IDAS-II Social Anxiety Scale. Child Behavior. Child Behavior Checklist (CBCL). Parents completed the CBCL (Achenbach & Rescorla, 2000) at both T1 and T2. The CBCL is a 100-item inventory. Parents reported on how true statements were of their child’s behavior, using a 3-point Likert scale. We used two subscales from the CBCL to reflect child behavior problems: 1) Internalizing and 2) Externalizing. It should be noted that this was not a clinical sample. Using CBCL t-scores, 24.4% of children at T1 and 22.2% of children at T2 scored in the borderline clinical to clinical range for internalizing symptoms; 14% of children at T1 and 20.5% of children at T2 scored in the borderline clinical to clinical range for externalizing symptoms. As this was a non-clinical, community sample, we used raw scores for analyses involving the Internalizing and Externalizing subscales to preserve the full range of variability in scores (Thurber & Sheehan, 2012). 3 Analytic Plan To address Research Question 1, we examined change in parental mental health between T1 and T2 with multi-level mixed models fit using the lmer function in the Lme4 package (Bates et al., 2014) in R (R Core Team, 2018). Using AIC to compare models, nested models were evaluated to determine the maximal random and fixed effects structure supported by the data for each parental mental health domain. To address Research Question 2, we followed the same procedure to examine change in child externalizing and internalizing behavior between T1 and T2. The most complex models evaluated for both parental mental health and child behavior included the following predictors: timepoint, days elapsed, prematurity, and a random intercept for participant. The most complex models for child behavior also included child age at T1. The interaction between prematurity and parent or child variables was added to the model after the most parsimonious model was selected. To address Research Question 3, we ran multiple linear regressions to determine whether change in parental mental health was associated with change in child behavior. The most complex models evaluated included the following predictors: change in parental mental health, parental mental health at T1, days elapsed, and prematurity. Given the small sample size, the interaction between prematurity and change in parental mental health was added to the model after the most parsimonious model was selected. All results remained the same when we used a continuous measure of gestational age (in weeks) instead of categorical prematurity. To account for the possibility that the relation between parental mental health and child behavior is bidirectional, we also conducted exploratory analyses in which change in child behavior was associated with change in parental mental health. 4 Results 4.1 How has parental depression, anxiety, and well-being changed since the COVID-19 pandemic began? The best fitting models for parental depression symptoms, parental anxiety symptoms, and parental well-being included a random intercept for participant and fixed effects for timepoint and prematurity. Timepoint was significantly associated with change in parental depression symptoms, t(38.61) = 2.58, p = .01, and in parental well-being, t(41.03) = -6.49, p < .001, but not in parental anxiety symptoms, t(44.63) = -0.38, n.s. Parental depression symptoms were significantly greater at T2 than they were at T1, and parental well-being was significantly lower at T2 than at T1. Prematurity was not significantly associated with change in parental depression symptoms, anxiety symptoms, or well-being. Moreover, an interaction factor between timepoint and prematurity did not improve the model fit and was therefore excluded. 4.2 How have children’s internalizing and externalizing symptoms changed since the COVID-19 pandemic began? The best fitting model for children’s internalizing symptoms and externalizing symptoms included a random intercept for participant and fixed effects for timepoint and prematurity. Results suggested that timepoint was significantly associated with change in children’s internalizing symptoms, t(39.53) = 2.21, p = .03, and children’s externalizing symptoms, t(40.91) = 2.08, p = .04. Children’s internalizing symptoms and externalizing symptoms were significantly greater at T2 than they were at T1. Child age at T1 improved the model fit for children’s internalizing symptoms but not externalizing symptoms, with child age at T1 positively associated with internalizing symptoms. Prematurity was not significantly associated with children’s internalizing symptoms or externalizing symptoms. Moreover, an interaction factor between timepoint and prematurity did not improve the model fit for children’s internalizing symptoms or externalizing symptoms and was therefore excluded from both models. Is change in parental mental health disturbances and parental well-being associated with change in child behavior? If so, does prematurity moderate that relation? Change in parental depression symptoms was positively associated with change in children’s internalizing symptoms and externalizing symptoms. Neither change in parental anxiety symptoms nor change in parental well-being was associated with change in children’s internalizing symptoms or externalizing symptoms (Table 3 ). Table 3 Linear Regressions Predicting Child Behavior Change Estimate SE 95 CI% LL 95 CI% UL t-value p-value Adjusted R2 Child Externalizing Change IDAS GD Change 0.40 0.17 0.05 0.75 2.32 0.03* 0.12 IDAS WB Change -0.09 0.27 -0.63 0.45 -0.34 0.74 -0.05 IDAS SA Change -0.31 0.48 -1.30 0.67 -0.65 0.52 0.13 Child Internalizing Change IDAS GD Change 0.49 0.11 0.27 0.71 4.54 < 0.001* 0.37 IDAS WB Change -0.01 0.20 -0.41 0.39 -0.05 0.96 -0.07 IDAS SA Change 0.06 0.40 -0.75 0.87 0.16 0.88 -0.07 Note: *p<.05. Statistics are from six different models for, each including an IDAS subscale and child behavior subscale. Each model includes the following covariates: parental mental health at T1 and prematurity. Days elapsed and the interaction factor between parental mental health and prematurity did not improve the model, so they were not included as covariates. Parental mental health at T1 did not emerge as a significant predictor in any model for children’s internalizing symptoms or externalizing symptoms. Prematurity did not moderate the effect of change in parental depression symptoms, anxiety symptoms, or well-being on children’s internalizing symptoms or externalizing symptoms. Figure 1, Figure 2 represent individual trajectories for change in children’s internalizing symptoms and change in externalizing symptoms, respectively. On each figure, for visualization purposes only, we divided the children into two groups using a median split on change in parental mental health: high versus low change on the three IDAS-II subscales. Consistent with the analyses, the figures show that children whose parents exhibited above median change on parental depression symptoms exhibited greater change in internalizing and externalizing symptoms (light grey line), compared to children whose parents exhibited below median change on parental depression symptoms (dark grey line). Figure 1 Individual trajectories of change in children’s internalizing symptoms as a function of change in parental mental health: a) parental depression symptoms b) parental anxiety symptoms, c) parental well-being.Note: High and low change groups are defined by a median split on each parent IDAS-II scale measuring parental mental health. Light grey represents change in children’s internalizing symptoms for children whose parents scored above the median. Dark grey represents change in children’s internalizing symptoms for children whose parents scored below the median. The solid line represents the Loess curve fit on the observation. Figure 2 Individual trajectories of change in children’s externalizing symptoms as a function of change in parental mental health: a) parental depression symptoms b) parental anxiety symptoms, c) parental well-being. Note: High and low change groups are defined by a median split on each parent IDAS-II scale. Light grey represents change in children’s externalizing symptoms for children whose parents scored above the median. Dark grey represents change in children’s externalizing symptoms for children whose parents scored below the median. The solid line represents the Loess curve fit on the observation. To account for the possibility that the relation between parental mental health and child behavior is bidirectional, we also conducted exploratory analyses in which child behavior predicted parental mental health. We found that, when accounting for child externalizing symptoms at T1, change in child externalizing symptoms was associated with change in parental depression and change in parental anxiety but not parental well-being. Moreover, we found that, even when accounting for child internalizing symptoms at T1, change in child internalizing symptoms was associated with change in parental Depression but not parental anxiety or parental well-being. 5 Discussion In this study, we examined whether parental mental health and child behavior changed during the COVID-19 pandemic. Importantly, we also examined how changes in parental mental health related to changes in child behavior and whether preterm children were more behaviorally susceptible than full-term children to changes in parental mental health. In line with our hypotheses, parental depression symptoms, children’s internalizing symptoms, and children’s externalizing symptoms significantly increased, and parental well-being significantly decreased during the pandemic. Notably, for these variables, adding timepoint improved the model fit but adding days elapsed did not, suggesting that change in these variables was not necessarily related to the mere passage of time. In other words, although we cannot conclude that the pandemic is what caused a decline in parental mental health and child behavior, our results suggest that parental mental health and child behavior declined between T1 (pre-pandemic; June 2019-March 2020) and T2 (during pandemic; April 2020 to October 2020) even when accounting for the mere passage of time. Recent studies with larger samples have reported concurrent changes in parental mental health and child behavior during the pandemic (Patrick et al., 2020). Our study expanded upon this work, and our results suggest that the pandemic might impact child behavior via parental mental health. Moreover, our results suggested that increases in specifically parental depression symptoms, but not increases in parental anxiety symptoms or reductions in parental well-being, were associated with increases in child behavior problems. These findings contribute to the literature on identifying the specific aspects of parental mental health that most strongly relate to change in child behavior. Relations remained robust even when accounting for parental depression symptoms at T1; that is, the baseline severity of parental depression symptoms at T1 did not account for the relation between change in parental depression symptoms and change in child behavior. Moreover, few parents scored above the screening cut-off on the IDAS-II General Depression Scale. Thus, even changes in subclinical levels of symptoms were associated with child behavior, regardless of baseline differences in parental mental health. Why might changes in parental depression symptoms be related to change in child behavior? First, increases in parental depression symptoms may have led to changes in parenting behavior, in turn leading to increases in children’s internalizing symptoms and children’s externalizing symptoms (Network, 1999, Reising et al., 2013, Slomian et al., 2019, Sohr-Preston and Scaramella, 2006). Indeed, some literature during the pandemic has already reported changes in parenting behavior, including increases in the use of harsh parenting and reductions in parent-child relationship closeness (Chung et al., 2020). It is also possible that parental depression symptoms increased as a result of increasing child behavior problems rather than vice versa (Herring et al., 2006) and that the relation between parental depression symptoms and child symptoms was bidirectional (Neece et al., 2012). Indeed, our results suggested that the relation between parental depression symptoms and child symptoms may be bidirectional; child internalizing and externalizing symptoms were associated with parental depression symptoms in an exploratory regression model. Finally, parents with more mental health problems may rate their children’s behavior more negatively than parents with fewer mental health problems. However, research that has accounted for parental reporting errors has still suggested a positive association between parental mental health disturbances and child behavior problems (Vallotton et al., 2016). Regardless of the specific mechanism, our findings suggest that parental mental health and child behavioral problems changed together during the pandemic. Finally, in our study, inconsistent with the differential susceptibility hypothesis, prematurity did not moderate change in parental mental health, change in child behavior, or the effect of change in parental mental health on change in child behavior. The children in our study were older than the children in other studies that found support for the differential susceptibility hypothesis (Gueron‐Sela et al., 2015). Perhaps by the time preterm children are preschool age, the strength of their susceptibility to environmental factors dissipates. Although many studies have documented prematurity-related behavioral difficulties well into childhood (Arpi & Ferrari, 2013), other studies have similarly failed to find support for an interaction between parental mental health and child outcomes in older preterm children (Hadfield et al., 2017). It is also possible that, because the majority of our sample was from a higher socioeconomic status, they benefited from resources that buffered the effects of prematurity on parental mental health and child behavior (ElHassan et al., 2018). Finally, our sample was small, and we may have been underpowered to detect statistically significant interaction effects. Although we found similar results when using continuous gestational age rather than binary categorical prematurity, some studies have reported significant interactions when using more than two categories of prematurity (e.g., very early preterm, early preterm, moderate preterm, late preterm, and term) (ElHassan et al., 2018). We did not have the power to parse our preterm sample beyond two categories. Our study has limitations that should be addressed in future studies. First, the subsample that participated in T2 was small, although the response rate was similar to other, larger studies on the effect of the pandemic (Patrick et al., 2020). Second, our sample was predominantly White and of a high socioeconomic status, limiting the generalizability of our findings. However, of note, the fact that we observed pandemic-related changes in parental mental health and child behavior even in a relatively advantaged population highlights the widespread influence of the pandemic. Third, there was variability in days elapsed between T1 and T2. The stressors associated with the pandemic have been dynamic, so the time at which the parent completed questionnaires may have influenced their answers. Due to the small sample for T2 and the variability in days elapsed between T1 and T2, we were not able to account for the various aspects of the pandemic that may have influenced answers (e.g., school closures, mask mandates). Future studies should pinpoint which aspects of the pandemic may be most strongly associated with changes in parental mental health and child behavior. Finally, we relied on self-report data for both parent and child measures; however, we had data prior to the pandemic, so we were able to avoid recall bias. To our knowledge, this was the first study to examine the relation between parental mental health and child behavior in a group of children with biological risk during a global stressor. We found increases in parental depression symptoms, child internalizing symptoms, and child externalizing symptoms and decreases in parental well-being. Increases in parental depression symptoms across time were associated with increases in child internalizing and externalizing symptoms across time. This relation was not moderated by prematurity. Our findings have the potential to inform prevention and intervention efforts aimed at equipping children with the behavioral resources necessary to thrive. Uncited references Achenbach, xxxx, Gueron-Sela et al., 2015, Institute of Medicine (US), 2007. CRediT authorship contribution statement Francesca Scheiber: Conceptualization, Methodology, Formal analysis, Investigation, Data curation, Writing – original draft, Visualization, Funding acquisition. Paige M. Nelson: Conceptualization, Methodology, Formal analysis, Investigation, Data curation, Writing – review & editing, Visualization, Funding acquisition. Allison Momany: Conceptualization, Methodology, Formal analysis, Investigation, Data curation, Writing – review & editing, Visualization, Funding acquisition. Kelli K. Ryckman: . Ö. Ece Demir-Lira: Conceptualization, Methodology, Formal analysis, Writing – review & editing, Visualization, Funding acquisition, Supervision. Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.