MCHB/EPI Miami Conference — December 7 - 9, 2005

Psychological Stress and Maternal Mental Health — Transcript

JEN-JEN CHENG: Good afternoon. From the previous two studies and existing literature, we understand now that maternal depression is a significant public health problem. Not only is it detrimental to mothers' well being, but also has adverse effect on her children's development as well as her family life. Here, I'd like to present to you how maternal depression may effect a child's development in terms of behavior outcome over ten year's period.

It has been known that women at childbearing age are at higher risk of depression. It has been found, for instance, that during postpartum period women's risk of depression is increased by three fold. A large body of existing literature has found that children of depressed mothers display problems ranging from difficult temperament, aggression, depressed mood, insecure attachment and ADHD. However, existing studies have various methodological problems that include cross section study design, which often times look at maternal depression and child outcome at the same time, thus the temper sequence is not as clear. All other problems include small clinical sample and incorrect statistical modeling for cluster data that may result is spurious significant association that is associated with under-estimation of standard error. It's good to see in the past decade there are a few longitudinal studies emerging examining the association between maternal depression and child outcome; however, most of them are limited by a short-term follow up that covers only limited periods of child development, either only during early childhood or adolescent only.

This study was conducted to explore the effects of early exposure to maternal depressive symptoms on a trajectory of child internalizing and externalizing behaviors over time. In the past, it is evident that not all children of depressed mother fair worse than other children. Actually, many of children of depressed mother exhibit competent adaptation to other circumstances, so it is also important for us to explore what makes these children resilient to the adverse environment, and fathers' role in family study has been neglected. And this study examined father's positive involvement and a fact modified in this association of interests.

The data results for this study come from the mother/child data center national longitudinal survey of youth, which I refer to as NLSY 79. The study period I chose is between 1992 to 2002, given that the maternal depressive symptom was first measured in this longitudinal study starting in 1992. At baseline, child age ran from zero to ten years old, and mothers' ages range from 27 to 34 in 1992. NLSY 79 was conducted every two years. Mother and children were both followed up every two years. And most of the survey were in-person survey, computer assisted. The study inclusion criteria include complete data on exposure, which is maternal depressive symptom in this case, and children have to have lived with the mother in 1992. And since this is a longitudinal study, the child outcome data has to have at least more than one way of data between 1992 and 2002. The results in the study sampled 5,669 mother and child diet.

The exposure for this study again is the maternal depressive symptom status in 1992. It's based on the center for epidemiological study depressive scale CDSD, and it's a binary indicator using a well-established couple of 16 to indicate depressed or not depressed was with greater than 16 being depressed. The outcome for this study include internalizing and externalizing behavior children between 1992 and 2002, with six waves of data, internalizing/externalizing behavior derived from the behavior problem index from NLSY. It is modeled after the well-established child behavior checklist, contains a listing of 28 mother reported child with behavior problem and symptom related questions in the past three months. The behavior problem index has eight subscales, and internalizing behavior includes anxiety depression and immature dependent subscale, whereas, externalizing behavior include antisocial, hyper-active, head strong and peer conflict/social withdrawal subscales. A sum score was used for these two outcomes, and they were standardized by age and gender, based on the general population. And this instrument was given to children age four to 14, again, it's mother self-report.

The fact modifier that I examined is fathers' involvement and predominantly positive involvement, I should stress. This is based on children's self-report. It's given to children ages ten to 14. There are seven questions were asked to children, basically all positive involvement related to their relationship with your father. And I'd like to point out, this is answer regardless the father is resident or non-resident, biological or stepfather. The theory behind it is the belief that as long as there's one significant adult functioning providing the social support needed in the child's life, and we have positive sign that would have a positive impact on the child's life. And it's based on four or five (inaudible) scales response category. And the seven questions where sum and a median was taken to use for this variable, as some children did not answer all seven questions and in order to maximize the sample size, the mean was used. Higher score referred to higher-level positive father's involvement.

In order to obtain unbiased estimate of the main effect, the multi-varied analysis does include potential count founders. Here, I'd like to point out; there are two groups of count founder. One is time unvarying, the other is time varying. Time varying characteristic of the subject that does not change over time, or that is only limited to the one-point-in-time measurement, and time varying characteristic that we felt like it changed over time and we could not possibly assume that it stayed constant over time.

Data analysis include descriptive (inaudible) as well as multi-level gross curve modeling to examine the continuity and discontinuity in child development. And it is also taking into consideration the cluster data structure, meaning each mother could have more than one child, being included in this study.

First, I'd like to share with you the sample characteristics of this study. As we can see, half of the sample is white Caucasian, and there is a large representation of Hispanic and African-American population. The mean age of baseline for the children is about five and a half, and this is equal proportion of boys and girls in the study. The majority of the mothers have at least a high school degree, about 24 percent of women reported living under poverty, and most of the subjects are urban residents. About 23 percent of women have indication depressive symptoms, and the mean score of internalizing/externalizing behavior at baseline for children is 104, which is close to the national mean of 100, and the mean age of the mother is about 31 years old.

I'd like to show you the first multi-varied analysis, based on gross curve model, specifically looking at the interaction effect of time and maternal depression on child internalizing behavior. We observe here the level of internalizing behavior is higher among children of depressed mother as compared to children of non-depressed mother. However, we do observe, there's a decreasing trend for children of depressed mother over time.

The second figure here shows the same interaction effect externalizing behavior of the children. And here again, we observe children of depressed mother exhibit higher level of externalizing behavior over time as compared to children of non-depressed mothers. However, we do see that the change rate of children, the change rate of externalizing behavior among children of depressed mother is lower than the other group.

And here, I'd like to sure with you in examining the interaction between maternal depression and father's positive development. Here, we observe father's positive involvement has a protective effect for both groups of children coming from either depressed or non-depressed mothers' family. And it's also apparent from this figure that the protective factor actually appears to be larger in a group of children coming from a non-depressed mothers. And a similar result was observed for internalizing behavior so I will not repeat it here. And as you see, there's a gradient here that higher level of father's involvement seems to have stronger protective effect of externalizing behavior in children.

To conclude this study, early exposure to maternal depression was found to be associated with higher level internalizing and externalizing behavior in child and adolescents. We observed the trajectory of internalizing behavior varied by individual characteristics. Significantly, within subject variation was observed. That implies that the level of problem behaviors over time differ from one child to the other, and the trajectory also differed from one child to the other. In this study, we also found that higher levels of positive fathers' involvement was inversely associated with externalizing and internalizing behavior in children over time, and the fact that maternal depression varied by levels of fathers' involvement.

There are limitations you'd like to know regarding this study. The CSD scale was not a clinical diagnosis. In other words, the mother's depression was not a diagnostic criteria-based condition. It also needs to be noted that maternal depressive symptoms was only available at baseline. As we know, chronic depression probably will have a stronger affect on the children, but the data is not available for this study. Another limitation to the study is, it has been found that a maternal report child problem behavior may be biased for maternal depressive state. Depressed mothers are more likely to report problem behavior in their children than non-depressed mothers. However, there are two reason (inaudible) study that examines this particular association, has indicated that mothers with depression are not any more likely to report children's problem behavior than mothers with no depression. And lastly, there are some unmeasured confounders that's not assessed in this study, and they include depression treatment of the mothers, as well as domestic violence presence in the family.

There are several strings in this study. It's a large nationally population based sample, and the in-person computer assisted interview enhanced the reliability of the self-report. This study is the first to examine fathers' positive involvement over time in a child's life, examining the effect of maternal depression and multiple measures were used for child's outcome, as well as several co-variants used in this study. And NLS provided a great opportunity to study association between maternal depression and child outcome because it provides various different measurement repeatedly covering childhood and adolescence. And there is also a large ethnic minority subgroup included in this study.

There several public health implications coming from the finding of this study. As we see, mothers' depressive symptom does have a long-term adverse effect on child outcome, and the cause for increased recognition and treatment of maternal depression. As we know, depression is not widely screened in this country, or elsewhere. And there's a great need for a universal screening for depression, either in pediatric or obstetric surrounding to address this symptom of public health issue. It is also found in this study that fathers play an important role in the care of their children if their spouse is depressed, so I think it is important for clinicians to educate partners of women with depression in the care of their illness. There's also a need to assess, there is a problem behaving offspring, of women with depression. And this study found there is individual variation in each child in terms of their development. So a tailored prevention program may be warranted that includes customized intervention program, prevention program to each child's specific need.

Thank you. And now we will start taking questions.