Authors: Viviane Valdes (1Boston Children’s Hospital (Division of Developmental Medicine), Harvard Medical School (Department of Pediatrics), Boston, Massachusetts, USA), Linda W. Craighead (2Emory University (Department of Psychology), Atlanta, Massachusetts USA), Charles A. Nelson, III (1Boston Children’s Hospital (Division of Developmental Medicine), Harvard Medical School (Department of Pediatrics), Boston, Massachusetts, USA; 3Harvard Graduate School of Education, Cambridge, Massachusetts, USA), Michelle Bosquet Enlow (1Boston Children’s Hospital (Division of Developmental Medicine), Harvard Medical School (Department of Pediatrics), Boston, Massachusetts, USA)
Categories: Article
Source: Infancy : the official journal of the International Society on Infant Studies
Doi: 10.1111/infa.12628
Authors: Viviane Valdes, Linda W. Craighead, Charles A. Nelson, Michelle Bosquet Enlow
In the current study we identified salient parental factors for child anxiety symptoms by considering the role of stressful life events, maternal anxiety symptoms, maternal depressive symptoms, and maternal neuroticism. Families (N = 399) in an urban area in the United States were participants in a longitudinal study beginning in infancy. Mothers completed measures of stressful life events (Revised Life Events Questionnaire at all visits), maternal anxiety and depressive symptoms (State–Trait Anxiety Inventory and Beck Depression Inventory, respectively, at infancy between 5 and 12 months, at 2 years, and at 3 years), maternal neuroticism (NEO Five–Factor Inventory at infancy), and child anxiety symptoms (Child Behavior Checklist 1.5–5 at 5 years). Linear mixed models (LMMs) were used in analyses. Maternal depressive symptoms from infancy through 3 years were associated with child anxiety symptoms; other main effects modeled (stressful life events, maternal anxiety symptoms, maternal neuroticism) were not associated with child anxiety symptoms. There was a significant interaction effect between stressful life events and maternal depression. Stressful events from infancy through 5 years of age increased risk for child anxiety symptoms at 5 years if the child’s mother had a mild mood disturbance or depression, but not for children with non–depressed mothers.
There is a strong evidence base that individuals with anxiety disorders are more likely to have relatives, particularly first degree relatives, that also have the same diagnosis (Schreier et al., 2008; Steinhausen et al., 2009; Wei & Kendall, 2014). Approximately 80% of parents of youth with an anxiety disorder meet criteria for an anxiety disorder themselves (Ginsburg & Schlossberg, 2002). Children with two parents who experience anxiety disorders or with parents who have more severe presentations (e.g., early onset, multiple anxiety disorders, greater impairment) are at especially high risk (Johnson et al., 2008; Schreier et al., 2008; Wittchen et al., 2000). Data suggest that both genetic and environmental factors contribute to the co–occurrence of anxiety disorders among children and their parents, including the “shared environment” between parents and children (Gregory & Eley, 2007).
The Family Stress Model (FSM) has been proposed as a conceptual framework for understanding how environmental risk factors may interact to produce increased risk for child or adolescent “maladjustment” (Masarik & Conger, 2017). The FSM theorizes that exposure to economic hardships or stressors may lead to psychological distress (e.g., depression and anxiety symptoms) in parents which would subsequently disrupt parenting practices that in turn negatively affect child adjustment (Masarik & Conger, 2017). For example, psychological distress and economic stress may decrease engagement with supportive and sensitive parenting practices, quality time with child, enrichment activities, and increase risk for punitive or overcontrolling behaviors (Masarik & Conger, 2017).
Multiple environmental factors have been proposed to explain the intergenerational transmission of anxiety in school–aged children and adolescents. First, parents with anxiety may be more likely to communicate increased threat interpretations and expectations to their children (7–13 years of age), either through language or modeling, and may be less likely to model effective coping strategies in the context of stressful events (Burstein & Ginsburg, 2010; Platt et al., 2016). Studies have also found that stressful events in the family environment (e.g., child health or parent health problems, death of a family member, family conflict, financial problems) and internalizing symptoms in children (7 years old), adolescents (10–14 years old), and young adults (18–33 years old) are often partly explained by parental levels of stress and psychopathology (Van Oort et al., 2010; Östberg & Hagekull, 2013; Williams et al., 2005).
Parental Generalized Anxiety Disorder (GAD), specifically, has been associated with GAD, panic/phobia conditions, and comorbid anxiety/depressive conditions but not with pure depressive disorders in children (Beesdo et al., 2010; Beesdo–Baum & Knappe, 2012). Associations between parent depression and child anxiety have also been examined. Here it has been found that adolescents (14–17 years old) and young adults (20–35 years old) that have parents with major depression are at increased risk for depression themselves, and also for anxiety disorders and substance use disorder (Beesdo–Baum & Knappe, 2012; Lieb et al., 2002; Weissman et al., 2006). Further, maternal depression has been associated with increased psychological aggression (e.g., shouting, name calling) from the parent toward the child, which has been associated with an increased risk for child internalizing symptoms at 9 years of age, including anxiety (Kuckertz et al., 2018).
Parental personality characteristics and caregiving behaviors have also been examined in relation to offspring anxiety. For example, neuroticism and anxious parenting styles (e.g., holding more negative expectations, greater intrusiveness, overprotection, expressed anxiety) have been associated with higher levels of anxiety in offspring (Creswell et al., 2010, 2013). Research by Eley et al. (2015) reported significant environmental transmission of parental neuroticism to adolescent offspring, such that they have a higher risk of both neurotic traits themselves and anxiety symptoms. Other researchers have used intergenerational data, including parents, monozygotic twins, and sibling pairs to account for assortative mating (by estimating covariance between trait–associated loci in distant parts of the genome) and genetic transmission of risk from parent to offspring (Ask et al., 2021; Torvik et al., 2022; Yengo et al., 2018). They find that parental neuroticism is significantly associated with anxiety and depressive symptoms in children at 8 years of age, even after adjusting for confounding effects of genetic transmission and assortative mating (Ask et al., 2021).
Some literature has explored how the effects of stressful event exposure on child/adolescent mental health outcomes may vary due to the presence of maternal psychopathology and personality traits. Early work by Hammen et al. (1991) found that stressful events increased risk for depressive symptoms in children/adolescents (8–16 years), but only among children with mothers who had depression. They suggested that maternal depression may act as a stressor for the child and limit a mother’s ability to buffer the effects of other stressors due to maladaptive coping (Hammen et al., 1991). Subsequent research has replicated these findings for both anxiety and depression symptoms in children/adolescents (8–15 years) (Feurer et al., 2016; Henry et al., 2020). More research is needed to examine anxiety outcomes specifically in early childhood and to test potential interaction effects between stressful events and maternal anxiety/neuroticism.
The current study focuses specifically on the role of factors discussed thus far (i.e., maternal anxiety, maternal depression, maternal neuroticism, and stressful events) to test the theoretical framework provided by the Family Stress Model. However, there is a vast body of research examining factors that influence the development of anxiety disorders starting in preschool. These factors may also be applicable earlier in development and should be considered in future work. Some examples include genetic variants, child temperament, psychobiological reactivity, parenting style, parental and child trauma and maltreatment, social determinants (e.g., financial hardship), and an absence of protective factors (e.g., coping skills, positive relationship with a significant adult, reinforcement for mastery and positive future expectations). For a comprehensive review of risk and protective factors for anxiety disorders, please refer to Al–Biltagi and Sarhan (2016).
Research specifically focused on early childhood is needed, as many anxiety disorders tend to have early onset (e.g., separation anxiety, specific phobias, social anxiety, generalized anxiety disorder) (Beesdo–Baum & Knappe, 2012; Kessler et al., 2005). Despite this need, much of the existing research on anxiety disorders is focused on school–aged children and risk factors in late childhood and adolescence may not be generalizable to early childhood. Early childhood is also an important developmental period to better understand as early onset of anxiety is associated with notable developmental, psychosocial, and psychopathologic complications (Beesdo–Baum & Knappe, 2012).
Furthermore, although several potential risk and protective factors have been previously identified for anxiety disorders, much of the prior research has been cross–sectional and bivariate. That is, it examines a specific variable alone in its association with concurrent levels of child/adolescent anxiety symptoms (e.g., maternal anxiety and child anxiety). Given the multifactorial etiology of anxiety disorders, studies that examine multiple variables longitudinally are necessary to identify the most salient and specific risk factors involved in emergence of anxiety disorders in early childhood. Such research could inform the development of targeted identification and intervention approaches to treat or prevent anxiety disorders and improve access to mental health services.
The current study aims to address some of these gaps by examining maternal mental health and personality as potential correlates of anxiety symptoms in early childhood in a longitudinal study design. Specifically, we tested the following (a) Higher levels of maternal anxiety symptoms (measured in infancy between 5 and 12 months, and ages 2 and 3 years), maternal depressive symptoms (measured in infancy and ages 2 and 3 years), and maternal neuroticism (measured at infancy) will be associated with higher levels of child anxiety symptoms at age 5 years (b) Family stress exposures (measured from infancy to age 5 years) will interact with maternal anxiety symptoms, depressive symptoms, and neuroticism to predict increased risk for child anxiety symptoms at age 5 years.
Participants were recruited from a registry of local births comprising families that indicated willingness to participate in developmental research. The present study was conducted according to guidelines outlined in the Declaration of Helsinki, with written informed consent obtained from a parent or guardian for each child before assessments and data collection. All procedures involving human subjects in this study were approved by the Institutional Review Board at Boston Children’s Hospital. Families in the current analyses participated in a prospective study to examine the early development of emotion processing. Exclusion criteria included known prenatal or perinatal complications, maternal use of medications during pregnancy that may have significant impact on fetal brain development (i.e., anticonvulsants, antipsychotics, opioids), pre– or post–term birth (±3 weeks from due date), developmental delay, uncorrected vision difficulties, and neurological disorder or trauma. After enrollment, families were no longer followed, and their data were excluded from analyses if the child was diagnosed with an autism spectrum disorder, or a genetic or other condition known to influence neurodevelopment (n = 29). By design, families were enrolled in the parent study when the children were 5, 7, or 12 months old (infancy), with a smaller subsample to be followed when the children were 2, 3, and 5 years of age. All models are adjusted for random effects of age. Approximately 778 infant–mother dyads were enrolled in the baseline visit (at 5, 7, or 12 months of age); 400 child–mother dyads participated at the 5 year follow–up. Only participants with child anxiety data at the 5 year visit were included in the current analyses (n = 399). Power analyses were conducted using G*Power 3.1.9.6 software and a sample size of 400 was sufficiently powered (1 – β = 0.9988) to detect even small effect sizes (e.g., 0.1) in models including eight variables.
Questionnaires were administered online, completed by the child’s parent, primarily the child’s mother (97%), at the time points specified below. A very small minority of parental respondents identified as fathers. We elected to retain data from these families to maximize sample size and power. Moreover, these fathers likely completed the questionnaires because they were the child’s primary caregiver, which may be more relevant for the current analyses than limiting to maternal respondents only. Because the respondents were almost exclusively mothers, we describe the data in relation to maternal symptoms and scores.
Exposure to stressful life events was measured using a modified 30-item version of the Recent Life Events Questionnaire (RLEQ Brugha et al., 1985); at each assessment time point (infancy, 2, 3, 5 years). The original RLEQ was developed using an inventory of 67 possible event categories to identify common life events that a high proportion of respondents report as having marked or moderate long–term threat (as opposed to mild or no long–term threat). The modified RLEQ used in the current study focused on recent life stressors most likely to be of relevance in this sample. The measure administered included items pertaining to serious injury or illness, death of relative or friends, separation (relationship or marriage), serious problems with close friends or relatives, serious abuse/attack/threats, unemployment or job loss, financial crisis, problems with police or court appearance, burglary/mugging, miscarriage/stillbirth, moving homes (through choice or not), and housing difficulties. The parent was asked to rate whether they had experienced the event (yes/no) in the time period of interest (e.g., from pregnancy to the time of the infant assessment).
Maternal anxiety symptoms were measured using the State–Trait Anxiety Inventory (STAI) during the infancy and 2 and 3 year time points (Spielberger, 1971). The STAI includes 20 items that assess trait anxiety and an additional 20 items that assess for state anxiety. Examples of state anxiety items include “I am tense” and “I feel secure” whereas trait anxiety items include “I worry too much over something that doesn’t really matter” and “I am a steady person.” All items on the inventory are rated on a four–point scale (1–4) ranging from “almost never” to “almost always.” Internal consistency estimates for the current sample were α = 0.89 at the infancy, 2, and 3 year time point. The current study measured and used composite trait anxiety scores in analyses. Higher composite scores indicate greater levels of anxiety (Spielberger, 1989; Spielberger et al., 1983).
Maternal depressive symptoms were assessed using the Beck Depression Inventory (BDI) during the infancy, 2, and 3 year time points. The BDI is a 21-item, self–report inventory that measures characteristic attitudes and symptoms of depression (Beck et al., 1961). Each item includes statements with corresponding scores (0–3) assessing for the frequency and severity of a specific symptom or characteristic experienced over the prior 2 weeks, including the day of assessment. For example, a participant is asked to select one of the following 0 = “I do not feel sad,” 1 = “I feel sad,” 2 = “I am sad all of the time and I can’t snap out of it,” and 3 = “I am so sad and unhappy that I can’t stand it.” A composite score is determined by summing across all items, with higher scores indicating higher levels of depressive symptoms. Individuals with scores from 0 to 10 are thought to be experiencing normal ups and downs, scores from 11 to 16 are categorized as a mild mood disturbance, 17–20 as borderline clinical depression, 21–30 as moderate depression, 31–40 as severe depression, and over 40 as extreme depression (Beck et al., 1988).
Maternal personality traits were measured using the NEO Five–Factor Inventory (NEO–FFI) during the infancy time point (Costa & McCrae, 1989). The NEO–FFI is a 60-item self–report questionnaire that measures five domains of personality, including neuroticism, extraversion, openness, agreeableness, and conscientiousness. The neuroticism domain was used in this study as a potential risk factor for child anxiety. Each of the items included in the NEO–FFI is measured on a Likert scale that ranges from a score of 0 (“strongly disagree”) to 4 (“strongly agree”). The internal consistency estimate for the current sample on the neuroticism subscale was α = 0.84. A composite score can be determined across items in each domain, with higher scores indicating stronger traits (e.g., a higher neuroticism score indicates higher levels of neuroticism) (Robins et al., 2001).
Child anxiety symptoms were measured at age 5 years using the parent–report Child Behavior Checklist for ages 1.5–5 (Achenbach & Edelbrock, 1991; Achenbach & Rescorla, 2000). The CBCL 1½–5 assesses behavioral and emotional problems in young children. The parent–report version obtains parent ratings for 99 problem items that can be scored according to syndrome or DSM–oriented scales (Achenbach et al., 2001). The current analyses used the DSM–oriented “anxiety problems” scale as an indicator of child anxiety symptoms (Achenbach et al., 2001). The internal consistency estimate for the current sample on the anxiety problems scale was α = 0.69. Parents were asked to rate how true each item was for their child in the prior 6 months on a three–point scale (0 = not true, 1 = somewhat or sometimes true, 2 = very true or often true). Scores on items are summed to get a total score on validated scales and standardized t–scores are also produced, which is what is used in the current analyses. A higher score reflects a parent’s report of greater presence and severity of symptoms in their child. T–scores on the CBCL have a mean of 50 and a standard deviation of 10. Scores in the 60–63 range are considered borderline for psychopathology, and scores greater than 63 are considered clinically elevated.
Data analyses were conducted using IBM SPSS Statistics for Macintosh Version 29 (IBM Corp.). Descriptive statistics for demographic variables collected at each visit were run in addition to the main analyses. Data were evaluated for normality and multicollinearity using univariate procedures and tolerance statistics. Bivariate analyses were conducted to identify potential covariates that should be included in models. Covariates that were significantly associated with the main predictor variables and the outcome variable at p < 0.05 and that did not negatively impact model fit statistics were included in the final models. Main analyses tested the study hypotheses as
Data were analyzed using linear mixed models (LMMs) with maximum likelihood (ML) estimation (Raudenbush & Bryk, 2002). This method allows for modeling of individual changes over time by creating two–level hierarchical models that are able to nest time within individuals (Bryk & Raudenbush, 1992; Miyazaki & Raudenbush, 2000). To run mixed effect models, data were prepared into a stacked format whereby each wave of data was “stacked” vertically, and a time variable was created to differentiate between time points, as opposed to having a unique variable for each wave of data collected. Age of the child in months at each visit was used instead of study visit number to take into account different time intervals between study visits.
To establish the non–zero covariance structure that was most appropriate for the data, various repeated covariance types were tested to determine the best model fit for the data. Existing protocols for developmental data called for testing “first order autoregressive [AR (1)]” covariance structure first, as it assumes that adjacent observations for the same subject indexed by age will have errors with higher correlations than those observations that are farther apart (West, 2009). Next, the continuous dependent variable (anxiety symptoms at age 5) and independent variables (stressful experiences, maternal anxiety symptoms, maternal depressive symptoms, maternal neuroticism, and interaction terms longitudinally) were defined. Fixed effects were determined for age for longitudinal variables (interaction term created for covariate and age). Random effects for age (slope coefficient) and intercept were also be specified.
Once the AR (1) model was run, two estimates of covariance parameters defined the error covariance (a) the variance of random errors and (b) correlation of adjacent errors. Information criteria (−2 log likelihood, AIC, AICC, CAIC, BIC) were then used to compare models with different covariance structures. Existing research suggests that likelihood ratio tests are the most appropriate covariance parameters assuming fairly large sample sizes and that fixed effects are constant (West, 2009). Simulation studies of model selection criteria when selecting among competing hierarchical linear models suggests that, for complex models with larger sample sizes (as many of those in the current analysis are), BIC and CAIC are most accurate (Whittaker & Furlow, 2009). Consequently, these two ICs were used to determine the best model fit for hypotheses of interest.
The main study hypotheses were tested using linear mixed models following the protocol outlined above. For these models, the continuous dependent variable defined was anxiety symptoms at age 5 years. All other variables included in the models were included as covariates, as they were continuous variables; these variables included total number of stressful experiences (infancy, 2, 3, and 5 year time points), maternal anxiety symptoms (infancy, 2, and 3 year time points), maternal depressive symptoms (infancy, 2, and 3 year time points), and maternal neuroticism (infancy). Fixed effects for age were specified for longitudinal variables, and random effects for age were also specified. Interaction terms (i.e., stressful life events x maternal anxiety symptoms, stressful life events × maternal depressive symptoms, and familial stressors × maternal neuroticism) were run to test our second hypothesis. Additional nested models were also run to probe the interaction terms. A sensitivity analysis including stressful life events only from the infancy, 2, and 3 year time points was also conducted.
Descriptive statistics for the sample’s demographic characteristics are presented in Table 1. The sample was approximately equally distributed among males and females, predominantly non–Hispanic White, and of middle to high socioeconomic status. Descriptive statistics for the main study variables are reported in Table 2.
Maternal STAI scores were 33.75, 32.43, and 33.26, respectively, at the infancy (5–12 months), 2, and 3 year time points (a cut–off score of 40 is indicative of a potential anxiety disorder and higher scores indicate more anxiety symptoms/severity). STAI composite scores were strongly correlated across time, rs = 0.71–0.77, ps < 0.001. Mean maternal BDI scores were 5.69, 4.90, and 5.55 (cut–off score of 10 is used for depression screening and higher scores indicate more depression symptoms/severity), respectively, at the infancy (5–12 months), 2, and 3 year time points. BDI composite scores were moderately correlated across time, rs = 0.49–0.61, ps < 0.001.
Bivariate analyses of the main study variables with all demographic variables (child sex, child race, child ethnicity, maternal and paternal education, family income) were conducted to identify potential covariates. None of the demographic variables tested were associated with the variables of interest and thus were not included in the multivariate models. Linear mixed models included fixed effects for age.
A linear mixed model was run to determine which maternal variables best predicted child anxiety symptoms at age 5 years. Of the fixed effects modeled (stressful life events, maternal anxiety symptoms, maternal depressive symptoms, and maternal neuroticism), only maternal depressive symptoms (t = 2.53, p = 0.001) had a significant main effect on child anxiety symptoms (Table 3). An additional Linear Mixed Model was run to test for potential interaction effects between stressful life events and each maternal factor (anxiety symptoms, depressive symptoms, neuroticism). In this model, familial stressful life events and maternal depressive symptoms interacted (t = 2.93, p = 0.003) to predict child anxiety symptoms (Table 3). Other interaction terms tested were not significant. Table 3 presents the full models. The interaction was probed through models nested by maternal depressive symptom level (normal ups and downs vs. mild mood disturbance or depression). A greater number of stressful life events was associated with more anxiety symptoms at 5 years only among children with mothers who had mild mood disturbances or depression (t = 3.56, p < 0.001). Number of stressful life events was not associated with level of anxiety symptoms among children with mothers who had low risk for depression (normal ups and downs). The interaction effect and results from nested models are plotted in Figure 1.
An additional sensitivity analysis was conducted that included familial stressful life events only until the 3 year time point (i.e., not including events from 3 to 5 years) to parallel the data available for the maternal depressive and anxiety symptom measures (i.e., at infancy, 2, and 3 years). The pattern of results in LMMs was consistent in the main analyses and the sensitivity analyses. The only significant interaction effect observed was between stressful life events and maternal anxiety (t = 2.76, p = 0.006). No significant interaction effects were observed between stressful life events and maternal anxiety or stressful life events and maternal neuroticism.
The overall goal of the current investigation was to examine the contribution of maternal mental health and personality indicators, measured longitudinally, on child anxiety symptoms at age 5 years. Results indicated that maternal depressive symptoms (infancy to 3 years) predicted child anxiety symptoms at age 5, whereas maternal anxiety symptoms (infancy to 3 years) and parent neuroticism at infancy failed to show associations with child anxiety symptoms. Maternal depressive symptoms (infancy to 3 years) also interacted with familial stress exposures (birth to 5 years) in predicting child anxiety symptoms. A greater number of stressful life events was associated with more anxiety symptoms only for children with mothers that had mild mood disturbances or depression, but not for those with mothers that had low risk for depression (normal ups and downs). These findings suggest that, in the first 5 years of life, maternal depressive symptoms and family stress exposures may be particularly relevant for the development of child anxiety symptoms.
There is some evidence that stressful events in the family environment, and internalizing symptoms in children and adolescents may be explained by levels of parent psychopathology (Van Oort et al., 2010; Östberg & Hagekull, 2013; Williams et al., 2005). Parents with anxiety might be more likely to communicate increased threat interpretations and expectations to their children, and less likely to model effective coping strategies in the context of stressful events (Burstein & Ginsburg, 2010; Platt et al., 2016). In our longitudinal models including stressful life events, maternal anxiety symptoms, and maternal depressive symptoms, maternal anxiety and stressful life events did not significantly predict anxiety symptoms in children.
It is possible that although maternal anxiety symptoms are associated with child anxiety symptoms in school–aged children and when examined cross–sectionally or without considering other predictors, the effects of maternal anxiety symptoms on child anxiety may be better explained by other factors modeled (e.g., maternal depression, neuroticism, or stressful life events). These findings should be interpreted within the context of some important considerations. One is that prenatal maternal anxiety and depressive symptoms were not included in the current study design, and these factors could affect the current findings. Notably, in a recent systematic review, both prenatal and postnatal anxiety were found to have a small adverse effect on child emotional outcomes (Rees et al., 2019). In studies adjusting for postnatal maternal anxiety and depression, prenatal anxiety was no longer associated with child emotional problems including anxiety (Rees et al., 2019). Nonetheless, future work including multivariate modeling of prenatal and postnatal effects, as well as various domains of psychopathology, is needed to identify maternal symptoms and time points to target through preventative intervention efforts.
Another relevant consideration is that the current study used a self–report trait measure of anxiety (STAI), which asks about the individual’s feelings of anxiety generally, administered at three time points (infancy, 2, and 3 years). The STAI trait scale was used to capture trait characteristics of anxiety (vs. state characteristics which may be more transient) and scores showed stability over time in the current sample. Prior research has shown that the STAI can be used to capture approximately 84% of mothers with clinical levels of anxiety symptomology (Dennis et al., 2013). Nonetheless, future research is needed with clinical instruments and samples to determine whether the observed findings are replicable or may be specific to community samples.
Parents with higher levels of neuroticism are also more likely to model higher levels of perceived stress and be more overprotective in ways that limit opportunities to learn effective coping among school–aged children (7+ years of age), increasing risk for anxiety and depression (Burstein & Ginsburg, 2010; Edwards et al., 2010; Platt et al., 2016; Waters et al., 2012). In the current study, maternal neuroticism (personality trait) measured at the infancy visit (5–12 months) was not significantly associated with child anxiety symptoms at 5 years of age. Similarly, it is possible that maternal neuroticism contributes to anxiety and depression risk in older offspring (school–aged children, adolescents, or adults), but not in younger children. In our analyses, after adjusting for the effects of maternal anxiety and depressive symptoms, maternal neuroticism was not associated with child anxiety symptoms at age 5 years. In a comparison of monozygotic (n = 385) and dizygotic twins (n = 486), Eley et al. (2015) found evidence for direct environmental transmission of neurotic traits from parent to offspring, independent of the effects of genetic confounding. They suggested that children and adolescents likely acquire neurotic traits over time via parental modeling (Eley et al., 2015). Given this, maternal neurotic traits may not exert sufficient effects on child anxiety risk in early childhood to be observable but may start to become evident later in childhood as children witness continued modeling of this trait.
Finally, maternal depression has also been linked to increased risk for child internalizing symptoms, including anxiety, at 9 years of age (Kuckertz et al., 2018). In our statistical model that included all parent variables (maternal anxiety symptoms, depressive symptoms, neuroticism), higher levels of maternal depressive symptoms across infancy (5–12 months) to age 3 years was the best predictor of anxiety symptoms in children at 5 years of age. A greater number of maternal depressive symptoms (from infancy to 3 years) was associated with more anxiety symptoms in children at 5 years. In our model that tested for interaction effects between maternal variables and familiar stress exposures, the interaction between maternal depressive symptoms and familial stressful life events was significant in predicting child anxiety symptoms at age 5 years. A greater number of stressful life events (from infancy through 5 years) was associated with more anxiety symptoms at 5 years only for children with mothers that had mild mood disturbances or depression, but not for those with mothers that had low risk for depression (normal ups and downs).
The deleterious effects of maternal depressive symptoms on child anxiety were evident in this community sample with relatively lower rates of depression symptoms (compared to clinical samples). The effects of maternal depression symptoms on child anxiety were also observed even though the current study used a self–report measure of depression (BDI), and only captured symptoms in the past 2 weeks at three time points (infancy, 2, and 3 years). The BDI has been found to have 100% sensitivity and 99% specificity for depressive disorders when used with a threshold of 12/13 (Lasa et al., 2000). Nonetheless, research is needed to replicate the current findings using diagnostic assessments and in clinical samples to determine whether the observed associations generalize to those that have a higher burden and representation of depressive symptoms.
The current findings replicate and are consistent with research by Kuckertz et al. (2018), which found that parent depression at age 5 years predicted child internalizing symptoms at age 9 years. The current findings add to this literature by suggesting that this link may be present at an even earlier developmental period, and that maternal depression symptoms may be a particularly salient factor for child anxiety symptoms at this earlier developmental stage. The current findings further suggest that stressful life events may increase risk for child anxiety symptoms in early life particularly in the presence of maternal depression symptoms. Maternal depressive symptoms may function as a stressor for the child that augments the effects of other stressful experiences. Mothers with depression symptoms may also have difficulty facilitating the development of adaptive coping strategies, and with limited adaptive coping strategies children may have greater vulnerability for anxiety (Hammen et al., 1991). For example, depressed mothers have been shown to exhibit greater withdrawal in interaction with their children, which limits their ability to dyadically regulate stressors. Further, they may model maladaptive coping strategies, such as disengagement, which can increase anxiety over time by limiting the child’s opportunities for extinction/desensitization. Finally, children of depressed mothers may be hyper–attuned to stressors, magnifying the impact of any stress exposure on anxiety symptoms (Henry et al., 2020).
Given our findings, routine screening for maternal depressive symptoms during early childhood well child visits and targeting such symptoms with appropriate treatments (e.g., psychotherapy, parent training, medication) may help reduce risk for anxiety problems in children. Such an approach is likely to yield benefits to the parent directly and may reduce the likelihood that their children will develop anxiety. Clinically, this type of approach would require substantial collaboration between obstetricians/gynecologists, primary care providers, and pediatricians as well as a reduction in barriers to mental health services for families through clinic and policy level changes (Owens et al., 2002).
The limitations of the current study deserve consideration. First, although the sample size was large enough to provide sufficient power for the current analyses, there was some minor variation in available data across measures. Efforts were made to account for this limitation by utilizing linear mixed models for measures collected at more than one time point. Linear models using individual growth curves do not require balanced data across different waves of data collection (e.g., sample size, time intervals, missing data) (Shek & Ma, 2011).
Given the sociodemographic characteristics of the study sample, the ability to generalize the results may be limited. The sample consisted largely of middle to upper income, well–educated families living in an urban/suburban area of the northeastern United States. Research suggests that lower levels of education, subjective social status, parent employment, and socioeconomic status broadly are associated with anxiety disorders, with those in lower SES groups having an increased risk for anxiety (Letourneau et al., 2013; McLaughlin et al., 2012; Shanahan et al., 2008). Additional research in a more socioeconomically diverse sample is needed to determine whether the current findings generalize to other populations and settings.
The study’s sample was also a community sample that likely has lower rates of psychopathology than what might be found in a clinical setting. For instance, on the depression inventory used, at the infancy visit (5–12 months post–partum) 55 mothers met the BDI cut–off for mild to severe depression symptoms (14.44%), at the 2 year visit 41 mothers met the cut–off (10.3%), and at the 3 year visit 45 mothers met the cut–off (14.52%). More research in clinical settings is necessary to determine whether findings from the current community sample may hold or differ in a sample enriched for high psychopathology risk (e.g., parents with diagnosed internalizing disorders and children at high risk for developing anxiety disorders).
Strengths and weaknesses of the measures used should also be noted. Given the target sample’s age for the current study (infancy to 5 years), all of the measures relied on parent report. The measures used are considered clinically valid for the assessment of symptoms among young children (i.e., parent report of anxiety on the CBCL) and for adults (self–reports, i.e., BDI, STAI, NEO–FFI) (Deighton et al., 2014). Although these measures are well validated, one limitation of using them together is that each measure has a different time span for which it is assessing symptoms and thus may have varying levels of state/trait stability (e.g., the BDI measures depression symptoms in the past 2 weeks, whereas the STAI asks about the individual’s feelings “generally” to capture traits). Additionally, measures are administered at different time points (e.g., NEO–FFI only at infancy while the RLEQ is collected from infancy to 5 years) and do not include assessment of maternal symptoms during the prenatal period which may be important to consider in future work. In the current study, scores appeared to be relatively stable over time for measures administered repeatedly. However, given that each measure assessed for a different time span and at different visits, it is possible that findings may have differed if consistent time frames had been used. Future validation studies to standardize assessment times across measures are needed to determine if it is important to hold consistent time frames across measures of interest.
Greater maternal depressive symptoms, from infancy (5–12 months) through age 3 years, were associated with higher child anxiety symptoms at age 5 years. Moreover, maternal depressive symptoms interacted with stressful life events to predict child anxiety symptoms. That is, stressful events from infancy through 5 years increased risk for child anxiety symptoms at 5 years if the child’s mother had a mild mood disturbance or depression, but not for children with non–depressed mothers. These findings suggest that, in the first 5 years of life, parent depressive symptoms and family stressors may be particularly salient factors in the development of child anxiety symptoms. Routine screening for parent depressive symptoms and appropriate treatments (e.g., psychotherapy, parent training, medication) may help reduce risk for anxiety symptoms in children.