Copyright 2004 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.2004
Few studies have examined the effects on children of maternal mental health symptoms other than depressive symptoms or have examined the joint effects of mothers' and fathers' mental health symptoms.
To examine whether the father's mental health symptoms may modify the association between the mother's mental health and the child's behavioral and emotional health.
Cross-sectional data from a national longitudinal survey of families provided information on 822 children aged 3 to 12 years who were living with both parents. The main child outcomes were the Behavior Problems Index–Externalizing (BPI-EXT) and–Internalizing (BPI-INT) subscales. The mother's and father's mental health were each assessed by self-report using the K10, a new, validated 10-item screen for serious mental illness, including mood or anxiety disorder. Parents with scores in the upper quartile were considered to be in poorer mental health and those with scores in the other 3 quartiles were considered to be in better mental health.
Adjusted for covariates, having both parents in poorer mental health was associated with a 1-SD increase in the children's BPI-EXT scores (β coefficient, 5.2; SE, 0.9; P<.001) compared with neither parent reporting poorer mental health. This effect was substantially weakened if the mother was in poorer mental health but the father was not (β coefficient, 1.8; SE, 0.5; P<.01). There was no statistically significant effect if only the father was in poorer mental health (β coefficient, 0.1; SE, 0.6; P = .88). The risk of a child having a high BPI-EXT score (≥90th percentile for the cohort) was elevated if both parents reported poorer mental health (odds ratio, 9.2; 95% confidence interval [CI], 4.8-17.8), but was less elevated if only the mother reported poorer mental health (odds ratio, 2.3; 95% CI, 1.1-4.9), and was not elevated if only the father reported poorer mental health (odds ratio, 0.6; 95% CI, 0.2-1.9). Similar patterns emerged for children's BPI-INT scores.
A father in better mental health may buffer the influence of a mother's poorer mental health on a child's behavioral and emotional problems, and these problems seem to be most severe for children who have 2 parents with poorer mental health. The form and intensity of pediatric approaches to mothers with poorer mental health may need to consider the mental health of fathers.
Maternal depression is known to have a broad influence on child health and well-being. Children of mothers who are depressed or who have depressive symptoms are at increased risk for developmental delay,1 behavioral problems,2 depression,3 asthma morbidity,4 and injuries.5 Depressed mothers are less likely to engage in preventive parenting practices6 and are more likely to use child health care services.7 Though research initially focused on postpartum depression, it is clear that maternal depressive symptoms often persist after the postpartum period,8 and this persistence further increases the effect on children's health.9 As a result, the pediatric role in identifying and addressing maternal depressive symptoms has received increasing attention.10- 13
A notable limitation in the literature has been to focus on the mother's mental health to the exclusion of research on the father's mental health. A father may help attenuate the adverse effects of maternal depression on child health by increasing his role as a caretaker of the child and by providing additional support to the child's mother.14,15 Alternatively, if both parents are in poor mental health, the child may be at particularly high risk for poor outcomes.16
Beyond the limited attention to the fathers' mental health, few studies have evaluated the influence on children of adult mental health symptoms other than those from depression. For example, adult anxiety disorders are common, often comorbid with depression, and associated with substantial functional impairment.17 A pediatric focus in research and practice on the mother's mental health, and, in particular, on maternal depressive symptoms, may limit our understanding of how the mental health of both parents affects children and may narrow our approaches to intervention.
The aim of this study was to examine whether the father's mental health symptoms may modify the association between maternal mental health and child behavioral and emotional health. Specifically, we hypothesized that fathers in better mental health would substantially reduce or attenuate the adverse effects of poorer maternal mental health on child well-being. We examined this hypothesis in a national, population-based survey that administered a validated screen for serious adult mental illness, including depression and anxiety disorders,18,19 to both mothers and fathers.
The study was conducted using data from the Panel Study of Income Dynamics. The Panel Study of Income Dynamics is a longitudinal survey of a representative sample of US families begun in 1968. In 1997, a Child Development Supplement was fielded to assess the effects of family sociodemographic factors on children's health and development.20 The sample in this study consisted of the 1786 children aged 3 to 12 years living with both their mother and father (or mother's boyfriend). Of these children, 20 had incomplete outcome data from the in-home child survey. An additional 944 children had at least 1 parent who completed the in-home child questionnaire but did not mail back the parent mental health assessment. The final sample size was 822 children in 605 families.
Our measure of child well-being was the Behavior Problems Index (BPI). The BPI is a symptom checklist (Table 1), originally developed for use in the National Health Interview Survey with 28 items drawn largely from the Achenbach Child Behavior Checklist.21,22 The items generate 2 subscales—one describing externalizing behaviors (BPI-EXT) and the other describing internalizing behaviors (BPI-INT). Both the mother and father rated the child's behavior. The parent was asked whether a problem behavior was "often" (3), "sometimes" (2), or "not" (1) true of their child. The score for each scale was the sum of the item responses with a range of 16 to 48 for the BPI-EXT and 13 to 39 for the BPI-INT. The internal consistency of these scales is reasonably high (Cronbach α, 0.86 and 0.81, respectively).20,23 As there are no established diagnostic or therapeutic thresholds for these scales, the BPI-EXT and BPI-INT were analyzed as continuous variables with higher scores indicating worse symptoms. For the purpose of clinical interpretation, however, a dichotomous variable was also created to identify those children scoring at or above the 90th percentile for this sample.
Mental health symptoms in both mothers and fathers were assessed by their own self-report using the K10. The K10 is a 10-item scale that measures adult nonspecific psychological distress (Table 1).18 This brief scale was developed for use in large population-based surveys and is used in the National Health Interview Survey. The K10 scale has been shown to have construct and predictive validity in identifying those at risk for either mood or anxiety disorders, as established by Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition criteria.24 Importantly, the K10 is a global screen that contains items about symptoms of depression and anxiety and high scores can result from either or both types of symptoms. However, the responses do not distinguish between whether individuals are primarily experiencing anxiety, depression, or both. The items ask how frequently the respondent experienced symptoms in the past 30 days, ranging from "all the time" (5) to "not at all" (1). Cronbach α for the 10 items is 0.9319 (Cronbach α is 0.82 in our sample). The K10 scores have been examined as a dichotomous, categorical, and continuous outcome.19,24
Covariates included the sex, age, and race/ethnicity of the child, the educational level and ages of the parents, and the household size and annual income in quintiles (bottom quintile, <$26 000; top quintile, >$83 000). The current smoking status (yes/no) of both parents was also assessed. Parents who responded that they drink alcoholic beverages were asked 3 additional screening questions for problematic drinking—ever felt that you ought to cut down on your drinking, ever felt guilty about your drinking, ever had a drink first thing in the morning.
We focused on the mothers' rather than the fathers' ratings of their child's behavior to enhance the comparability of our findings to prior research that has preponderantly used maternal ratings of child behavior. However, we also briefly report our findings based on the fathers' ratings of their child's behavior. Mothers' and fathers' mental health was evaluated as a continuous measurement (K10 scores), a categorical measurement (quartiles of K10 scores), and a dichotomous measurement (highest quartile vs lower 3 quartiles.) In describing the results, we label a parent in the highest quartile as having poorer mental health and those in the other 3 quartiles as having better mental health. The correlations between mothers' K10 scores, fathers' K10 scores, and children's BPI scores (EXT and INT) were analyzed using Pearson correlation coefficients. In bivariate analyses, we used analysis of variance (ANOVA) to compare mean BPI scores by quartile of parental K10 score and by categories of other covariates (eg, maternal education). To examine potential effect modification, we stratified by maternal K10 quartiles and, within each stratum, used ANOVA to examine the association between the fathers' K10 quartiles and the children's mean BPI scores.
We conducted multivariate linear regression using child BPI-EXT and BPI-INT scores as dependent continuous variables. To assess interaction between the mothers' and fathers' mental health in the regression models, we followed the approach suggested by Rothman25 creating a single parental mental health variable with the following 4 mutually exclusive categories: both parents in the highest quartile of symptoms, only the mother in the highest quartile, only the father in the highest quartile, and neither parent in the highest quartile. For this main analysis, we also used logistic regression to predict the probability that a child scored at or above the 90th percentile for the BPI-EXT and BPI-INT. We then examined whether the parental health effects differed according to the child's sex by running separate models for boys and girls. Finally, though limited to parent-reported data, we sought to examine whether mothers in poorer mental health might overestimate their child's behavior problems.26 We assessed whether children rated in the top BPI-EXT quartile by mothers in poorer mental health were less likely to have other possibly corroborating outcomes reported ("a doctor or health professional ever said that the child had hyperactivity, ADHD [attention-deficit/hyperactivity disorder], or ADD [attention-deficit disorder], and "the child has ever been suspended or expelled from school") than those children rated in the top quartile by healthy mothers. No parallel items were available to examine maternal mental health and child BPI-INT scores.
All covariates significant at P<.10 in bivariate analyses were included in multivariate analyses. Up to 2 children per household were sampled. To account for the fact that some parents reported on 2 children and that these 2 children could not be considered fully independent observations, we used Proc Genmod in SAS (version 8.02; SAS Institute, Cary, NC), which accounts for potentially correlated observations. We also reexamined the main findings using 2 alternative approaches. One approach used a single randomly selected child per household and the second examined those families reporting on only 1 child separately from those families reporting on 2 children. The findings did not change using these alternative approaches.
The mean (SD) age of the children was 8.2 (2.9) years and mean (SD) maternal age at delivery was 27.6 (5.2) years. Forty-four percent of the mothers and 47% of the fathers had a high school diploma or less education (Table 2). The mean (SD) BPI-EXT and BPI-INT scores were 22.5 (5.2) and 15.7 (3.1), respectively. Child BPI-EXT and BPI-INT were significantly correlated (r = 0.61, P<.001). Mothers' and fathers' K10 scores, examined as continuous variables, were correlated with each other (r = 0.35), and each was correlated with the children's BPI-EXT (mothers' K10 score, r = 0.31; fathers' K10 score, r = 0.18, respectively) and BPI-INT (mothers' K10 score, r = 0.31; fathers' K10 score, r = 0.23, respectively) scores. All comparisons were statistically significant (P<.001).
When mothers' and fathers' K10 scores were examined as quartiles, those children with mothers in the highest quartile were more likely than other children to also have a father in the highest quartile (31.7% vs 18.6%, P>.001). Higher maternal and paternal K10 scores were significantly associated with increased children's BPI-EXT and BPI-INT scores (Table 2). Boys had higher BPI-EXT scores than girls, and children had higher scores if they had mothers with less education, if they lived in lower-income households, or if either parent smoked. In contrast, none of the covariates examined were significantly associated with the children's BPI-INT scores. Parents who did not complete their own mental health assessments had significantly lower family income and lower educational attainment. However, there were no significant differences between the BPI scores of those children whose parents did and those children whose parents did not mail back their own parental mental health assessments (data not shown).
Figure 1 shows evidence for potential effect modification. Among children who have a mother in poorer mental health (ie, in the fourth or highest K10 quartile), BPI-EXT scores were significantly lower if the father had better mental health. Similarly, among children with a mother in poorer mental health, BPI-INT scores were also lower if the father had better mental health (Figure 2).
Association of mothers' and fathers' mental health symptoms using the Children's Behavior Problem Index–Externalizing (BPI-EXT) scale. The analysis of variance compares the mean BPI scores across paternal mental health quartiles within each maternal mental health quartile with the first quartile being the lowest K10 score and the fourth quartile the highest K10 score.
Association of mothers' and fathers' mental health symptoms using the Children's Behavior Problem Index–Internalizing (BPI-INT) subscale. The analysis of variance compares the mean BPI scores across paternal mental health quartiles within each maternal mental health quartile with the first quartile being the lowest and the fourth quartile the highest K10 score.
In adjusted analyses, poorer maternal mental health was associated with a 1-SD increase in children's BPI-EXT scores (β coefficient, 5.2; SE, 0.9; P<.001) if the father was also in poorer mental health (Table 3). However, the effect of poorer maternal mental health was substantially weakened (β coefficient, 1.8; SE, 0.5; <.001) if the father was in better mental health. Fathers' poorer mental health was not associated with the child's BPI-EXT scores when the mother was in better mental health. A similar pattern emerged for children's BPI-INT scores, although fathers' symptoms alone were modestly associated with increased BPI-INT scores.
These findings are mirrored in logistic regression analyses that examined the probability that a child had a BPI score at or above the 90th percentile for the sample. Compared with neither parent reporting poorer mental health, the risk of a child having a high BPI-EXT score was elevated if both parents reported poorer mental health (odds ratio [OR], 9.2; 95% confidence interval [CI], 4.8-17.8), less elevated if only the mother reported poorer mental health and the father did not (OR, 2.3; 95% CI, 1.1-4.9), and not elevated if only the father reported poorer mental health (OR, 0.6; 95% CI, 0.2-1.9). Similarly, the risk of a child having a high BPI-INT score was elevated if both parents reported poorer mental health (OR, 8.1; 95% CI, 3.5-18.7), less so if only the mother reported poorer mental (OR, 3.1; 95% CI 1.7-5.8), and not elevated if the only the father reported poorer mental health (OR 2.0, 95% CI 0.9-4.4).
The same multivariate models showed a similar pattern of results when we used the father's report of the child's BPI-EXT and BPI-INT scores as the dependent variables. Children whose mother and father both had poorer mental health had significantly elevated BPI-EXT (β coefficient, 4.8; SE, 0.8; P<.001) and BPI-INT (β coefficient, 3.5; SE, 0.6; P<.001) scores compared with children for whom both parents were in better mental health. If the mother had poorer mental health but the father was in better mental health, the child's BPI-EXT and BPI-INT scores remained modestly, but significantly, increased (β coefficient, 1.1; SE, 0.5; P<.02; and β coefficient, 0.8; SE, 0.3; P<.01, respectively).
We examined potential effect modification by the child's sex (Table 4). The combined effect of the mother's and father's symptoms was particularly strong for boys, for both the BPI-EXT and BPI-INT scores (Table 4). Maternal symptoms alone had a stronger association with girls' BPI-EXT scores than boys' BPI-EXT scores; while fathers' symptoms alone appeared to influence more strongly boys' BPI-INT scores than girls' BPI-INT scores. These patterns were similar when the fathers' ratings of BPI were used.
We did not find evidence to suggest that mothers with poorer mental health overestimated their child's behavior problems. Compared with children rated in the top BPI-EXT quartile by mothers in better mental health, children rated in the top BPI-EXT quartile by mothers in poorer mental health were as likely to have received a physician diagnosis of attention-deficit/hyperactivity disorder (10.1% vs 12.8%, P = .53) and to have been suspended or expelled from school (10.3% vs 10.8%, P = .93).
While the mental health of mothers has long been recognized to affect children's well-being,3,27 the findings of this study suggest that those effects can vary depending on the mental health of the father. Specifically, we found that the adverse effects of a mother in poorer mental health on a child's behavioral and emotional problems were substantially reduced when a father reported better mental health. When both the father and mother reported poorer mental health, the influence on a child's behavioral problems was strong, particularly for boys. These findings raise several challenging questions regarding how clinicians should address the mental health of both parents and its influence on the child's health and well-being.
Prior research has established the influence of maternal depression on the child's behavioral and emotional health3,27,28 and its relevance for pediatric practice.29,30 Several studies have analyzed the independent effects on children of mothers' and fathers' mental health, but few have examined the more clinically relevant question of how the mental health of parents jointly influences child outcomes.16,31,32 The largest study of the joint effects of mothers' and fathers' mental health on children was a population-based study of twins and their parents.16 Similar to our results, the combination of maternal and paternal depression was associated with the largest increase in children's depressive symptoms. Mothers' depression alone was still modestly associated with increased child depressive symptoms, but fathers' depression alone was not. The effects of maternal depression were increased if there was a comorbid maternal or paternal anxiety disorder.16 Foley et al16 found no association between parental depression and children's externalizing symptoms. Their use of a sample of twins and their focus on depression may, in part, explain the differences.
Importantly, our study uses a new, validated measure of adult mental health that expands the prior focus on adult depressive symptoms to include anxiety symptoms as well. Adult anxiety disorders are prevalent and their influence on adult functioning is comparable to that of major depression.17 Parental anxiety disorders increase the likelihood of childhood externalizing disorders31 and anxiety disorder.33,34 Matthey et al35 note that while an anxiety component was noted early on in research on postpartum maternal mental health, little follow-up work has been done. Parental anxiety may occur with greater frequency, intensity, and associated morbidity than previously appreciated. Future research with the K10 and its shorter 6-item version may prove that they are useful instruments for clinical settings because they have the potential to efficiently screen for both mood and anxiety disorders causing functional impairment.
There are several potential limitations to the study. The cross-sectional design limits our ability to make inferences about directionality or causality in the associations between parental mental health and child behavioral and emotional health. Difficult temperament or behavior in young children may increase mental health symptoms in parents. Parental depression may increase the likelihood that parents view their children's behavior more negatively, however, this explanation seems less likely given the similarity in findings when using the behavior ratings of either mothers or fathers. Further, the relationship between the child's high BPI-EXT scores and school suspension or expulsion, for example, did not differ by maternal mental health status. Formal clinical psychiatric diagnoses were unavailable for the parents or children, and while the K10 asked about both mood and anxiety symptoms, it is not designed to distinguish between mood and anxiety disorders. Nevertheless, the K10 provided important information because it produces a score consistent with the continuum of clinically relevant symptoms underlying these disorders. Our goal was to understand the joint contribution of mothers' and fathers' mental health symptoms along this continuum rather than to establish thresholds for direct intervention. Our sample was limited to those children with complete responses from both parents. Parents who did not complete their own mental health assessments had a lower family income and a lower educational level on average though their children's reported behavior was not significantly different from children whose parents had completed their own assessments. Finally, this analysis focused on children living with both parents in one home and did not examine children living in other types of households. Further research remains to be done on children living with single parents or in homes where other adult caregivers are present.
This study did not explore the mechanisms underlying the joint influence of mothers' and fathers' mental health on the child's well-being. While a genetic basis for the findings is possible, altered parenting likely has a significant mediating role.15 For example, a recent meta-analysis found that maternal depression and psychological distress were associated with increased negative and coercive parenting behaviors and disengagement from the child.36 A father may then attenuate the influence of a mother's depression with increased caring behavior directed to the children.37,38 Alternatively, a healthy father may offer support directly to the affected mother.14 Future work may explicate these mechanisms by examining specific measures of mothers' and fathers' role functioning. From a pediatric perspective, however, the need to identify and refer affected parents will remain the same.
The important role of fathers' mental health on children's well-being raises several challenging questions for pediatric practice. Pediatricians and parents recognize maternal depression as an issue relevant to a child's well-being and recent pediatric care guidelines suggest that maternal depression is an appropriate topic for discussion in pediatric care.13 Fathers are generally less directly engaged than mothers in their children's health care.39 While some fathers bring children for health care, no pediatric studies have focused on how best to screen fathers directly for mental health problems. Although pediatricians regularly ask mothers about the family's medical history, to our knowledge, there have been no studies on the accuracy of maternal reports on fathers' mental health. Further, resources to treat adult mental health problems are limited in many areas. Given these constraints, pediatric practice recommendations to routinely assess or to address the father's mental health may be premature. However, pediatricians encountering a mother with poor mental health should inquire how well the father functions as a support for the mother and the child. The yield from such targeted inquiries should be increased, as children whose mothers were in poorer mental health were more likely to have fathers who were in poorer mental health. Indications that the father is not functioning as a significant parental resource should be added incentive to intervene promptly and intensively on behalf of the mother.
Most broadly, these findings emphasize that the form and intensity of pediatric interventions directed at mothers may need to be shaped by the child's family context. Maternal mental health, smoking, diet, and prevention practices are all recognized as relevant to child health outcomes. Yet change in the mother's health status or behavior is often strongly associated with the father's status. For example, the strongest predictor of maternal smoking relapse after pregnancy is the father's smoking status.40 In this study, fathers' mental health was a powerful modifier of the influences of mothers' mental health problems on children. Thus, regardless of the progress we make in identifying maternal health problems, whether mothers can access their own health care, adhere to treatment, or change their health behavior may depend on the well-being of the child's father or other adult caregivers in the household.
Maternal depressive symptoms are associated with behavioral and emotional problems in their children. This study adds to the literature by considering the influences on children of both anxiety and depressive symptoms among parents and by examining the joint effects of mothers' and fathers' mental health symptoms. We found that a father reported to be in better mental health may buffer the influence of a mother's poorer mental health on a child's behavioral and emotional problems, and that these problems appear most severe for children who have 2 parents with poorer mental health. These findings can help shape the form and intensity of pediatric interventions for mothers with poorer mental health.
Correspondence: Robert S. Kahn, MD, Division of General and Community Pediatrics, Cincinnati Children's Hospital Medical Center, 3333 Burnet Ave, ML 7035, Cincinnati, OH 45229-3039 (email@example.com).
Accepted for publication February 19, 2004.
This study was supported in part by grants K23 HD40362-01 (Dr Kahn) and R01-HD41141 (Drs Kahn and Whitaker) from the National Institutes of Health, Bethesda, Md.
Kahn RS, Brandt D, Whitaker RC. Combined Effect of Mothers' and Fathers' Mental Health Symptoms on Children's Behavioral and Emotional Well-being. Arch Pediatr Adolesc Med. 2004;158(8):721-729. doi:10.1001/archpedi.158.8.721