Children of Depressed Mothers 1 Year After Remission of Maternal Depression: Findings From the STAR*D-Child Study

One consistently replicated finding in psychiatry is the increased risk of psychiatric disorders, particularly major depression and anxiety disorders, in the offspring of parents with major depression (1–12). Major depression is likely a complex disorder (13, 14) with environmental stress acting as an episode trigger in vulnerable individuals (15, 16). Parents' acute depressive symptoms likely contribute to a stressful environment for children. Decreases in depressed parents' symptoms might be linked with reductions in children's symptoms and behavioral problems. The Sequenced Treatment Alternatives to Relieve Depression (STAR*D) Child study was designed to examine the relation of maternal remission from depression to children's functioning and psychopathology (17) in mothers who were being treated in STAR*D (18, 19). When STAR*D began, only two published studies had shown that reducing parental symptoms provided some immediate benefit to children. One study's clinicians had not directly treated the parents' depression (20), and the other was a pilot study of 18 depressed mothers (21). Two additional studies have since reported findings consistent with a relationship between children's clinical state and parental depression (22, 23).

In the STAR*D-Child study, we hypothesized that remission of maternal depression would be linked with reduced offspring symptoms when the mothers remitted. At the initiation of the mother's treatment, about one-third of their children had a current psychiatric disorder and one-half had a history of psychiatric disorder (24). After 3 months of antidepressant treatment, maternal remission of major depression was significantly associated with reductions in the children's diagnoses and symptoms (17).

During the year after initiation of the mothers' treatment for depression, the outcomes for offspring of mothers who remitted early (during the first 3 months after treatment initiation), later (after 3 months of treatment), or not at all over the first year were examined separately (25). A statistically significant difference was observed between groups, with a decrease in symptoms evident in children of early and late remitters, but not in children of nonremitters. These findings were limited by the fact that children of mothers who remitted after 6 and 9 months were followed for only 1–6 months after maternal remission. Furthermore, we were not able to determine the effect of maternal remission on subsequent changes in child outcomes because the study period was 1 year from the mother's treatment initiation. Because the children were followed up every 3 months for 1 year after maternal remission, or for 2 years if the mother remained depressed, this study focuses on changes in psychiatric symptoms and functioning among children of depressed mothers during the year following remission of maternal depression to determine the longer-term effects of maternal remission on subsequent changes in child outcomes.

Method

STAR*D and STAR*D-Child Studies

STAR*D (www.star-d.org) was a multisite study designed to determine the comparative effectiveness of different treatment options for outpatients 18–75 years old with nonpsychotic major depression and without a lifetime diagnosis of bipolar disorder, schizophrenia, or schizoaffective disorder (19, 26). All study participants were initially treated with citalopram. Those intolerant of or not remitting when treated with citalopram were offered other treatments, including other antidepressants, cognitive-behavioral therapy (CBT), or a combination of these, using an equipoise randomized design (27).

The STAR*D-Child study recruited 824 women, ages 25–60 years old, at seven of the 14 regional centers participating in STAR*D. Of these, 808 women (98%) were screened to ascertain whether they had any children 7–17 years old; 177 of them (22%) had children in that age range, and of these, 174 (98%) met all eligibility criteria; 151 of those eligible (87%) agreed to enter the child study. If a mother had more than one child of eligible age, one was selected randomly (17). Mothers provided separate written informed consent for themselves and for their child.

The child assessors were not involved in the mothers' treatment, were blind to mothers' remission status, and were independent of the team that treated the mothers. The mothers' clinical assessments were completed by STAR*D staff members who were not involved in the children's assessments. Baseline assessments were conducted before or within 2 weeks after the start of treatment of the mothers' depression.

Sample

A total of 151 mother-child pairs were assessed at baseline, and 127 of these completed at least one follow-up assessment (Figure 1). Because the primary aim of the study was to examine the relation of the mother's remission to child outcomes during the year following her remission, our primary analysis was restricted to mothers who remitted in the course of the study and did not suffer a subsequent relapse while enrolled in the study. Early remitters were mothers who remitted in the first 3 months (N=36), and late remitters were those who remitted between months 3 and 12 (N=28). A comparison group of nonremitters (N=16) comprised mothers who did not remit in either the first or second year of the study. We excluded from the analyses 20 mothers who relapsed, three who were late remitters (two for whom data on child or mother were unavailable after remission and one because remission occurred during the second year of treatment), and 24 nonremitters (because data on child or mother were unavailable after the first 12 months of treatment).

Figure 2 shows the analytic design of the present study. Mothers and children were followed for varying lengths of time by design. Although children of late-remitting mothers were followed for longer periods than early-remitting mothers, the length of the study period that we analyzed is the same for all children regardless of the time of the mother's remission (i.e., the 1-year period after mother's remission).

Assessments and Measures

Maternal assessments were conducted as part of a comprehensive battery of assessments for all STAR*D participants. A diagnosis of a current maternal major depression was established by clinical interview and confirmed using a symptom checklist based on DSM-IV criteria (18). The severity of depressive symptoms over the previous 7 days was assessed by a clinician using the 17-item Hamilton Rating Scale for Depression (HAM-D; 28, 29). As in STAR*D, remission was defined as having a HAM-D score ≤7, and relapse was defined as a HAM-D score ≥14 any time after having remitted. Mothers who remitted by the first 3-month evaluation were considered early remitters; those who remitted after the first 3 months of follow-up were considered late remitters, provided that they remained in remission during the subsequent year. Not all remitted mothers were observed for the entire year of follow-up because some dropped out of the study during the follow-up period.

Remission was also assessed by self-report using the 16-item Quick Inventory of Depressive Symptomatology–Self Report (QIDS-SR). When follow-up HAM-D scores were unavailable, QIDS-SR scores were imputed based on an analysis of the relation between the HAM-D and the QIDS-SR used in the STAR*D study (30–32).

The Schedule for Affective Disorders and Schizophrenia for School-Age Children–Present and Lifetime Version (K-SADS-PL) was used to assess DSM-IV psychiatric diagnoses in the children (see online supplement of Kaufman et al. [33]). A clinical assessor interviewed mothers and children separately. To minimize time, sections of the K-SADS-PL focusing on disorders highly prevalent among children of depressed parents (i.e., mood, anxiety, and disruptive disorders [1, 34]) were used. The qualifications, training, and reliability of child assessments have been reported elsewhere (24). We examined symptoms derived from the K-SADS-PL interview with the mother (about the child) and separately with the child. We interviewed the children and scored their responses before we interviewed the mothers. A mother's interview followed immediately after the completion of the child interview. The variable “child symptoms” consists of a count of the symptoms included in the child's K-SADS-PL screening interview that were currently present at the threshold or subthreshold level. We created a similar variable based on the clinical interviews with mothers about their child.

The Child Behavior Checklist (35) assessed total, internalizing, and externalizing symptoms as reported by mothers. Scores are presented as T scores ranging from 0 to 100, with higher scores indicating a greater number or severity of symptoms. Scores above 70 indicate clinical impairment.

The Children's Global Assessment Scale (36) is a clinician-based measure of global functioning with a scale range from 0 to 100. Scores above 90 indicate superior functioning, and scores below 70 indicate impaired functioning. We used a 2-week time frame based on all available child data.

Data Analysis

The relation of maternal remission status to child outcome in the year following remission was assessed with Poisson regression analysis when the outcome was the number of child symptoms at each time point, and was assessed with longitudinal mixed-effects regression when the child outcome was a continuous variable, such as the Children's Global Assessment Scale or Child Behavior Checklist score. Maternal remission status (early, late, or nonremitting) was included in the models as a three-level categorical predictor variable. All regressions were conducted within a generalized estimating equations framework to adjust for correlations between repeated observations over time. The Poisson regressions used an exchangeable correlation matrix (37), and the longitudinal mixed-effects regressions used an unstructured covariance matrix (38). We performed analyses using PROC GENMOD (for Poisson regression) and PROC MIXED (for linear regressions using longitudinal mixed-effects analysis) in SAS, version 9 (SAS Institute, Cary, N.C.). For all of the models described above, we included the following potential confounding variables in the analyses: child's age and gender; household income; mother's baseline HAM-D score, presence of an anxiety disorder or substance use disorder at baseline, and marital status; child's treatment status over the course of the study (any versus no treatment); and child's baseline score for the outcome under consideration. An interaction term representing time by maternal remission status was included in the regression analyses to formally test whether the rate of change in child symptom outcomes varied according to maternal remission status.

Missing outcome data were handled by initially assuming that the data were missing at random for the continuous variables analyzed using mixed-effects models, and missing completely at random when using Poisson regression analysis using generalized estimating equations. Under this assumption, valid analyses can be conducted using models such as the mixed-effects models based on maximum likelihood inference. Because of the large number of dropouts at the 12-month assessment in the late remission category, we performed a sensitivity analysis by rerunning all models after deleting the 12-month assessment for all participants and comparing patterns of change as represented by the beta coefficients for the 9-month period after maternal remission, and comparing results. Data on household income were missing for three mothers, and the average household income of all mothers in each maternal remission category was imputed.

Results

Baseline Characteristics of Depressed Mothers and Their Children

A significant association was observed between household income and maternal remission status (p=0.02), with early-remitting mothers having the highest household income (Table 1). Early-remitting mothers were more likely to be currently married than late- and nonremitting mothers (p<0.05). Not surprisingly, baseline HAM-D scores of depressed mothers varied significantly by maternal remission status, with early-remitting mothers having fewer symptoms (the lowest scores). A higher proportion of children of early-remitting mothers were boys, but the association fell short of statistical significance (p=0.09). Children of early-remitting mothers had significantly fewer symptoms at baseline by mothers' reports.

Child Outcomes and Maternal Remission Status

We compared child outcomes by time of remission of maternal depression adjusting for child's age and gender; household income; mother's baseline HAM-D score, presence of an anxiety disorder at baseline, presence of a substance use disorder at baseline, and marital status; and child's baseline score for the outcome under consideration and treatment status over the course of the study (Table 2). We first determined whether there were significant changes in children's outcomes among offspring of early-, late-, and nonremitting mothers separately during the year following maternal remission. Table 2 shows that there were statistically significant decreases over time in the number of symptoms among children of both early- and late-remitting mothers (as reflected in the negative beta coefficients and associated p values) but not in children of nonremitting mothers. However, formal tests of differences in the magnitude of the rate of change in child-reported symptoms (beta coefficients) among the three groups as reflected in the group-by-time of maternal remission interaction coefficient were not statistically significant. Similar patterns were observed for maternal reports of children's symptoms (Table 2).

Child Behavior Checklist total problem scores for children of early- and late-remitting mothers decreased over time, but only the decrease in early-remitting mothers was statistically significant. Symptoms in children of nonremitting mothers increased over time, although this increase was not statistically significant (Table 2). A test of the interaction revealed marginally significant differences (p=0.057) among these groups on rate of change in Child Behavior Checklist total problem scores. Pairwise comparisons revealed that time trends in the children of early- and late-remitting mothers were significantly different from those of nonremitting mothers (early versus nonremitting: t=2.83, df=151, p=0.005; late versus nonremitting: t=3.14, df=113, p=0.002), but no significant differences were observed between the early- and late-remitting mothers. When Child Behavior Checklist internalizing scores were examined separately for each maternal remission group, we found statistically significant decreases for children of early- (p=0.03) and late-remitting (p=0.05) mothers, but not for children of nonremitting mothers. A formal test of the interaction was not statistically significant.

Child Behavior Checklist externalizing problem scores mirrored the patterns of change in total problem scores, with decreases in scores in children of early- and late-remitting mothers, although only decreases in children of early-remitting mothers reached statistical significance. There was a significant increase in scores of children of nonremitting mothers. A test of interaction showed that the variation in patterns of change in Child Behavior Checklist externalizing scores among the three maternal remission categories was statistically significant (p=0.03). Pairwise comparisons revealed that the time trend among children of nonremitting mothers was marginally different from those of early-remitting mothers (t=1.92, df=152, p=0.057) and significantly different from those of late-remitting mothers (t=3.01, df=114, p=0.003), but the difference between children of early- versus late-remitting mothers was nonsignificant.

Children's Global Assessment Scale scores showed a statistically significant improvement during the year after remission among children whose mothers were early remitters, but not among children of late remitters or nonremitters. However, rates of change across the three groups were not significantly different according to formal tests of interaction.

Because many (82%) of the children of late remitters dropped out of the study before the final assessment at the end of the 12-month period following remission, we repeated the analyses over the 9-month period following remission (not shown). Beta coefficients representing rates of change in outcome measures over the 9-month period were very similar to those estimated over the 12-month period for all maternal remission categories. Further exploration of the 14 children of late remitters who dropped out of the study after the 9-month assessment showed on average greater improvement on all measures, compared with those who stayed in the study, although these differences did not reach statistical significance. In contrast, the four children of nonremitting mothers who dropped out after the 9-month assessment appeared to be doing less well on all measures compared with those who stayed in the study. These data suggest that our findings may in fact be conservative.

Children Receiving Treatment

If a child had a psychiatric diagnosis and was distressed, the interviewer offered feedback to the child and family, encouraged treatment, and provided appropriate referrals. There were no statistically significant differences in the proportion of children receiving treatment by maternal remission status, with six (17%), nine (32%), and six (38%) children of early-, late- and nonremitting mothers, respectively, receiving treatment. Of the 21 children receiving treatment, 20 received outpatient treatment and one received residential treatment. There were no significant differences in type of treatment children received (medication, psychotherapy, or both) by maternal remission status.

Discussion

During the year following remission of maternal depression, children of early- and late-remitting mothers continued to show a significant decrease in psychiatric symptoms, as evidenced by child and maternal reports of total symptom count on the K-SADS-PL. When Child Behavior Checklist symptoms were considered, there was a significant decrease in symptoms among children of early-remitting mothers, which parallels the finding based on K-SADS-PL symptoms. Although Child Behavior Checklist outcomes for late-remitting mothers did not show statistically significant decreases for total and externalizing scores when considered alone, pairwise comparisons of decreases in symptoms of children of early- and late-remitting mothers did not show statistically significant differences. These results suggest that there is a significant decrease in children's symptoms and problem behaviors in the 1-year period following maternal remission, which does not vary with length of time to maternal remission. Among children of nonremitting mothers, there were no significant changes in either symptoms or behavior scores, with the exception of Child Behavior Checklist externalizing problem scores, which increased significantly over the study period.

Child functioning as assessed with the Children's Global Assessment Scale showed significant improvement only among children of early-remitting mothers during the time they were in the study. Their mothers were less depressed at baseline and remained depressed for a shorter time, suggesting that early remission, lower illness severity, and shorter duration of maternal depressive episodes may be important in relation to their children's adaptation.

The less consistent statistically significant decreases in symptoms and functioning among children of late-remitting mothers (compared with the decreases in children of early-remitting mothers) are probably due to the heterogeneous nature of this group, which includes mothers who remitted between 3 and 12 months after initiation of treatment. In contrast, children of mothers who did not remit over the 24-month follow-up showed no significant improvement on any outcomes during the 12–24 months after initiation of mothers' treatment. In fact, there was a significant increase in children's externalizing problems among offspring of nonremitters. This group was small, and these findings should be viewed cautiously. The sample size of nonremitters by the end of the study was small to a large extent because of the STAR*D design, which was to provide treatment until the patient went into remission.

Even with a relatively small sample size, particularly of nonremitters, we observed some differences in patterns of change in child outcomes by mother's remission status. It is likely, however, that this small sample size reduced the ability to detect statistically significant differences, especially for analyses of interactions. In addition, since our findings in this phase of the study are based on symptoms and not diagnoses, the clinical significance of the findings may be limited. Other limitations should be noted. These analyses could not take into account the type of maternal treatment (psychotherapy versus medication). Because only one-third of the patients in STAR*D were open to receiving psychotherapy (39), there were too few mothers who had been treated with psychotherapy in our study to conduct meaningful comparisons. Our study included only depressed mothers. The effect of treating depressed fathers is being studied. Children were followed for varying lengths of time from study initiation based on their mother's remission status, and it is possible that children followed for longer periods were more likely to have an increase (or decrease) in symptoms or behavioral problems merely as a function of follow-up time and not of mother's remission status. Our focus was the 1-year period following mother's remission, and therefore the varying length of follow-up time is reflected only in the length of time between study initiation and mother's remission. Our results suggest that for most outcomes there is little difference between the postremission rates of improvement in children of late- versus early-remitting mothers, which implies that it is unlikely that the length of time spent in the study is the primary factor related to our findings.

It is possible that mothers who maintained remission over the 1-year postremission period did so because their children continued to improve. In mothers who did not remit, their children's continued behavior problems may have contributed to maternal depressive symptoms. There were too few mothers who relapsed in this sample to conduct a meaningful analysis of this hypothesis. However, in the year following initiation of treatment for maternal depression, we found little evidence that mothers became less depressed in reaction to positive changes experienced by their children (25). This could be because most of the mothers in this study were moderately to severely depressed and were being aggressively treated for depression.

No causal conclusions can be drawn from these observational data. Although these results are consistent with the hypothesis that treating depressed parents to remission benefits their children, we cannot attribute the mothers' remission to the treatment itself or attribute children's better outcomes to their mothers' remission. It is possible that some unmeasured third variables (e.g., genes, stress) explain both mothers' remission and children's symptom decline.

The first published study to document the relation between remission of a mother's depression and her child's clinical state appeared relatively recently (40), and around the same time a report from a randomized clinical trial (23) provided data consistent with our results. The trial compared the outcome of school-age children whose depressed mothers were randomly assigned to receive interpersonal psychotherapy (N=26) or treatment as usual (N=21) for eight sessions. The mothers and children were assessed at baseline and at 3 and 9 months. Mothers assigned to interpersonal psychotherapy had lower levels of depressive symptoms and higher functioning at both follow-up assessments. Moreover, differences were found in their children's depressive symptoms at the 9-month follow-up. Maternal improvement preceded improvement in offspring. Although the sample was small, treatment was randomized from the beginning, unlike the design of STAR*D.

Another recent study (22) was a randomized clinical trial to determine the effect of CBT in preventing the onset of depression in adolescents at risk for depression. It was found that while CBT was generally superior to usual care, adolescents with a currently depressed parent had poorer outcomes with CBT. On a similar note, a study by Brent et al. (41) showed that the presence of maternal depression at baseline predicted poorer outcomes in depressed adolescents who received CBT.

Our findings, taken together with the studies described above, indicate that remission of depression in mothers is associated with positive outcomes in their children. Continuing the mother's treatment until remission is achieved may be warranted, as even later remission of maternal depression was associated with decreases in children's symptoms. Finally, there may be benefit to the child in having the mother remain in treatment even if she does not achieve full remission immediately. The barriers to implementing programs that provide continuous treatment of depressed parents to remission are primarily related to treatment access and retention as well as to financial obstacles in our health care system (42).