The present study sought to investigate the factors that are associated with vulnerability for depression in the children of depressed (n = 74) versus non-depressed (n = 37) parents. This is the first study to compare the relative value of numerous risk factors in predicting risk in the offspring of depressed parents.
Summary of Findings
Although children suffering from a psychiatric illness were explicitly excluded in this study, major differences (\(\eta_{\text{p}}^{2}\) = 0.435) in sub-clinical symptoms of depression and general psychopathology between children of depressed (HR) versus non-depressed (LR) parents highlight the increased risk of depression for this group. Confirming hypothesis (i), these differences were evident across all measures of self- and parent-reported symptoms of depression and general psychopathology (internalising and externalising symptoms) and were supported by small to moderate correlations between children’s and parents’ symptoms of depression (r = .23). In addition, HR (vs. LR) children used fewer adaptive emotion regulation strategies (\(\eta_{\text{p}}^{2}\) = 0.13), perceived fewer positive parenting strategies (\(\eta_{\text{p}}^{2}\) = 0.13) and fewer positive life events (\(\eta_{\text{p}}^{2}\) = 0.127). Interestingly, the HR children did not differ from the LR children in the maladaptive dimensions of those variables (use of maladaptive emotion regulation strategies, frequency of negative attributions, perception of negative parenting strategies, frequency and impact of negative life events). Hence, hypothesis (ii) was only met for the positive scales. Regarding the mediation analysis, maladaptive emotion regulation strategies and negative life events were identified as partial mediators in the association between parental depression and children’s depressive symptoms, confirming hypothesis (iii) partially.
Interpretation of Findings
By demonstrating significantly more symptoms of depression and general psychopathology among the HR group with a large effect size, the data replicate earlier findings that children of depressed parents are at increased risk, not only for developing depression [
5,
77], but also internalising and externalising disorders more generally [
78,
79]. Participants who met diagnostic criteria for a psychiatric disorder following a clinical interview were excluded from the study. Nevertheless, more participants in the HR (versus LR) group showed values above the clinical cut-off for the depression and psychopathology outcome variables DIKJ, YSR, and CBCL. This demonstrates the clinical significance of our findings. Pending replication in longitudinal studies, the findings might be used to increase the effectiveness of interventions.
Interestingly, there was a discrepancy between the parent- and child-rated psychopathology, supporting previous studies [
80,
81] with parents reporting greater symptom severity in their children than the children themselves (55.6% accordance rate). This supports Angold and Costellos [
80] observation of the reduced reliability of affected parents’ ratings on children’s psychopathology, due to a loss of empathy and increased worry as a result of depression. In addition, children might spare their parents with their psychological difficulties in order to avoid burdening them even more. However, the rating of symptoms of psychopathology in the parent and child questionnaires (YSR vs. CBCL) did not differ significantly (t = 0.489,
p = .626).
The data demonstrate support for the model of familial transmission by Goodman and Gotlib [
9] which proposes risk of depression in children to be mediated by a combination of exposure to less adaptive parenting strategies, difficulties in emotion regulation, negative cognitions and increased stress exposure. HR families were characterised by a combination of less positive parenting strategies, less frequent use of adaptive emotion regulation and fewer positive life events. Although previous studies have identified these factors as playing a role in the familial transmission of psychopathology [
19,
34,
82], these have largely been based on community samples or on their moderating role within samples of children of depressed parents. Few studies have directly compared these factors between children of depressed versus non-depressed parents. In this sample, the equally large effects were displayed for emotion regulation, parenting style and life events. The failure to find a contributing role for cognitive factors such as attributional style contrasts with a study by Compas et al. [
19] which found that both coping strategies and a negative cognitive style contributed independently to the regression model (DV: child’s affective symptoms). We received critical feedback from participants regarding the ASF questionnaire: children said they had difficulties imagining the scenarios that they were asked to rate (i.e. they would not happen in their lives). In addition, attributional style is just one aspect of cognitive vulnerability. Other cognitive vulnerabilities include self-esteem, hopelessness and the cognitive triad, which we were not able to measure in the current study. Another limitation is that we only had scope to measure the child’s perception of the parenting style, which could be biased by social desirability. Future research concerned primarily with the effect of parental depression on parenting style should include observational and parent-reports of parenting to achieve better data triangulation.
Interestingly, although the HR group showed an absence of numerous adaptive skills (adaptive emotion regulation) and experiences (positive life events and positive parenting), there was no evidence that they showed increased maladaptive skills (maladaptive emotion regulation) and experiences (negative life events and negative parenting). These effects are contradictory to earlier findings characterising this sample with negative emotion regulation [
23], negative thinking style [
25], negative parenting style [
37,
38] and negative life events [
46,
83]. There are several possible explanations for these findings. One reason might be that the data from the high-risk sample were collected in the assessment session of an intervention study, where most parents were concerned about passing on their disorder to their children. Consequently, these parents may have been reflecting on their behavior and trying to protect their children from negative influences (e.g. negative thinking and behavior). In addition, the majority of parents of the HR group (92.4%) had received psychological treatment for depression before entering the study and might therefore be better informed about maladaptive coping strategies and parenting than parents (in the LR group) who had not received psychotherapy.. Nevertheless, children might lack a role model who is displaying adaptive coping strategies. In this sample, children of depressed parents differed significantly from children of non-depressed parents in their report of positive (but not negative) life events. Fewer positive life events might indirectly mirror the environment that goes along with parental depression (e.g. less family activities, holidays) as well as reflecting more indirectly—similarly to other psychological skill deficits—financial problems and conflicts in marriage that go along with parental depression. Since parents with depression experience more stress (e.g., financial problems, unemployment, etc.), we expected their children to report more negative life events (e.g., divorce of parents) [
46,
83]. However, this was not the case. One reason for this might be that the assessment was based on children’s reports of life events and children were not aware of negative life events (e.g., financial, marital problems of parents) since parents might try to shield negative life events from their children in order to protect them. The fact that the sample had a high socio-economic background may also be an important factor in explaining why children reported fewer difficulties. In addition, we mainly assessed the mechanism variables via self-report measures and desirability biases may have played a role. Future assessment and interventions with this population should consider the fact that they may not only suffer from negative life events but also from a lack of positive life events.
This is the first study to examine the extent to which several potential mechanisms may account for the relationship between parental depression and child psychopathology. Although numerous factors characterised the children of depressed versus non-depressed parents (emotion regulation, attributional style, life events, parental depression), not all factors contributed significantly in the regression model which sought to account for variance in children’s own psychopathology. Both maladaptive emotion regulation strategies and negative life events (but not attributional and parenting style), significantly predicted the children’s symptoms of depression above the background variables and parental depression. If these findings are replicated in experimental studies, prevention interventions which focus on reducing the use of maladaptive emotion regulation strategies in the face of negative life events may be particularly effective.
Strengths and Weakness
The main strength of this study is the direct comparison of multiple risk factors in a sample of children of clinically depressed parents with children of non-affected parents which enables stronger conclusions about the trans-generational effect of parental depression than studies based on community samples. Although a larger sample size would have enabled us to calculate more robust statistical models e.g. structural equational modally, recruiting families with a mentally ill parent is particularly time consuming [
84] and the power was sufficient to calculate between-group differences in potential mechanisms. Furthermore, we conducted Bayes analyses to test the extent to which any null-effects were due to a lack of statistical power.
A number of clinically-relevant instruments were used to characterize the sample. Clinical interviews (with high interrater-reliability; κ = 1) were conducted with both parents and children. As such, we could be sure about the diagnostic status of parents in both groups. This also enabled us to exclude children who had a lifetime diagnosis of a psychiatric disorder and be more certain that any vulnerabilities identified were antecedents, not consequences, of depression. On top of that, child- and parent-ratings of symptoms of general psychopathology were included. Parent-ratings alone have been shown to be less valid for children’s internalising symptoms but more valid for externalising symptoms [
85,
86]. Including both ensures a more comprehensive interpretation of the data.
One limitation of the study is the representativeness of the sample. The HR group was an opportunistic sample of families who were recruited for an intervention whereas data of the LR group were collected within an experimental study. The opportunity to receive and intervention may have meant that the HR group were generally more motivated to participate than the LR group, although in this case any group differences in terms of risk of depression are likely to
underrepresent the true difference in the general population. Although there were differences in marital status and gender between the HR and LR groups, there was no evidence that these differences affected the findings. Since taking part in a time-consuming intervention (12sessions in within 6 months) with the whole family demands a high level of motivation, especially for depressed parents, the investigated sample might not be representative of children of depressed parents in general. Furthermore, the socio-economic background of all families was high and rarely had a migration background
13; nevertheless, they differed in parent gender and marital status making the groups less comparable. In addition, few parents suffered from severe depressive episode in the sample and showed great variability in their course of depression. Following this argumentation, the effects we found might be even stronger in the general population of families with a depressed parent, since the families with more risk factors due to their socio-economic background were underrepresented in our sample. Nevertheless, our sample might be quite informative for future interventions for children of depressed parents, since it represents a subgroup that enrols in such interventions.
Although missing values could be imputed, a limitation of this study is the amount of missing data. Since this study is about families with parental depression, it is not surprising that impairments of those families are mirrored in the response rates of questionnaires. Although families were encouraged to fill in the questionnaires properly and reminded to send them back soon, many parents felt stressed by just another “to do” in their daily routine.
We included the factors in the Goodman and Gotlib [
9] model that we believed to be most important in the transgenerational transmission of depression but could not include other factors such as parental cognitive style and affect, which via social learning are proposed to explain the cognitive and affective vulnerabilities in offspring. Future research focusing on
how offspring of depressed parents acquire negative cognitive and affective vulnerabilities could also inform preventive interventions.
A final limitation of the study is that the data is cross-sectional rather than longitudinal and therefore does not allow causal interpretations to be drawn about the factors which prospectively predict the onset of a depressive episode in the children of depressed parents. In order to capture developmental risks and model resilience for depression, longitudinal studies are needed. Furthermore, additional information regarding familial factors as structure, cohesion and parent–child interaction is needed.
Future Research
There are numerous avenues for future research. Although the present study combines several relevant risk factors in order to understand the transmission of depression, the data is rather exploratory. Longitudinal studies investigating representative and larger samples are needed to explore the role of risk and protective factors that were found to characterise children of depressed parents. In addition, more characteristics of the parental depression such as age of onset as well as parental cognitive style and affect (as discussed by Goodman and Gotlib, 49), and its impact on the development of a “depressotypic-style” in children are worth investigating [
7]. Furthermore, data that allows structural equation modelling in order to achieve a better understanding of impact, overlap, and interaction of risk factors including gene-environment interactions [
87] would be beneficial. Furthermore, it remains unclear whether the included risk factors have moderating or mediating effects. As inconsistent and unclear findings of the mediating and moderating roles of those risk factors have been discussed in numerous prior studies [
7,
44,
53], future research ought to investigate these specific mechanisms in order to better understand the transgenerational pathways of depression. Moreover, additional instruments for the assessment of the cognitive factors (e.g. self-esteem, self-efficacy) may be beneficial. Further experimental data for verifying the effects of risk factors on the children’s wellbeing might also be helpful for a better understanding of transmission of depression in this high-risk group.