You are here

  • Home
  • Archive
  • Volume 65, Issue 1
  • Parental psychopathology and socioeconomic position predict adolescent offspring's mental health independently and do not interact: the TRAILS study

Article Text

Article menu

Download PDFPDF

Research report
Parental psychopathology and socioeconomic position predict adolescent offspring's mental health independently and do not interact: the TRAILS study
  1. K Amone-P'Olak1,2,
  2. H Burger1,3,
  3. M Huisman1,
  4. A J Oldehinkel1,2,4,
  5. J Ormel1,2
  1. 1Interdisciplinary Centre for Psychiatric Epidemiology (ICPE), University Medical Centre Groningen, University of Groningen, Groningen, The Netherlands
  2. 2Graduate School of Behavioral and Cognitive Neurosciences (BCN), University Medical Centre Groningen, University of Groningen, Groningen, The Netherlands
  3. 3Department of Epidemiology, University Medical Centre Groningen, University of Groningen, Groningen, The Netherlands
  4. 4Department of Child and Adolescent Psychiatry, Erasmus Medical Centre, Erasmus University, Rotterdam, The Netherlands
  1. Correspondence to Dr Huibert Burger, Interdisciplinary Centre for Psychiatric Epidemiology, Department of Epidemiology, University Medical Centre Groningen, University of Groningen, PO Box 30.001, 9700 RB Groningen, The Netherlands; h.burger{at}epi.umcg.nl

Abstract

Background Familial risk factors have been implicated in the development of mental health problems in adolescents. Whether the associations between parental loading, as assessed by lifetime psychopathology, and offspring internalising and externalising problems were moderated by family socioeconomic position (SEP) was investigated. Two hypotheses of moderation were tested: (1) the “social push” hypothesis in which parental loading effects are stronger in contexts with low environmental risks and (2) the “vulnerability” hypothesis in which parental loading effects are stronger in high-risk environments.

Method In a population-based sample of 2149, familial loading and family SEP were assessed at baseline by parent reports. Offspring psychopathology was assessed by reports from multiple informants (parent, self and teachers). Multiple linear regression was used to assess the independent associations of parental loading and family SEP on offspring psychopathology and their potential interaction.

Results Both family SEP and familial loading had significant independent main effects on offspring internalising and externalising problems. However, the interaction terms were not significant and did not add any explanatory power to the model.

Conclusions Lower levels of family SEP appear not to confer additional risks for mental health problems in offspring of parents with high loading on psychopathology. During early adolescence, parental psychopathology and low family SEP seem independent risk factors for offspring mental health problems. Results do not support either the social push or vulnerability hypothesis as no evidence of interactions between parental loading and family SEP were found.

  • Parental psychopathology
  • socioeconomic position
  • adolescents
  • mental health
  • parents CG

https://doi.org/10.1136/jech.2009.092569

Statistics from Altmetric.com

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    Introduction

    Mental health problems during adolescence are frequent,1 associated with significant impairments and future psychopathology,2 3 and can disrupt schooling and career development.4 5 Parental psychopathology and low family socioeconomic position (SEP) are known risk factors for offspring mental health problems.6–9 Parental psychopathology is reported to influence offspring mental health through at least three pathways: first, through predisposition to vulnerability genes.10 11 Second, through exposure to adverse and non-supportive family environment, for instance, poor parenting—that is, parents are more irritable, harsh and show less affection towards their children.8 12 13 Third, through bias information processing—for example, daughters of depressed women interpret information more negatively and less positively than daughters of mothers who are not depressed.9 However, genetic and environmental influences may operate on familial loading. For example, life events such as divorce or unemployment may lead to psychopathology.14 Likewise, from twin and adoption studies, we know that depression has a substantial genetic component.15 16

    Low family SEP is also recognized as a risk factor for adolescent mental health problems.6 7 For example, low SEP is associated with inadequate material goods17 and residence in poor and deprived neighbourhoods where substance abuse, deviant peer behaviours and other social problems are known to flourish.18 19 Low SEP is also a source of chronic stress that negatively affects the relationship between parents and their children, as reflected in poor family functioning, child abuse, and poor rearing and health behaviours.20 21 However, most previous studies22–24 showed that family SEP accounts for a relatively small proportion of variance in adolescent mental health (usually <5%). We therefore propose that family SEP does not have a strong direct effect on adolescent mental health, but instead moderates the influence of familial and environmental risks associated with parental psychopathology on their adolescent offspring.

    Two competing hypotheses have been put forward to explain how environmental contexts such as family SEP may moderate the influence of genetic effects on their offspring. The “social push” hypothesis25 posits that genetic risks (which include familial psychopathology) are stronger in contexts with low environmental risks (eg, high SEP) and weaker in contexts with high environmental risks such as low SEP (figure 1). The theory underlying the social push hypothesis is that an adverse environment obscures genetic effects while lack of competing adverse environmental factors enables genetic effects to emerge more strongly. Thus, low family SEP would reduce the effects of familial loading on adolescent offspring's mental health while high family SEP would amplify the effects if this hypothesis is true. The vulnerability hypothesis,26–28 on the other hand, postulates that those who are genetically predisposed may be more vulnerable to high-risk environments such as low SEP than those in low-risk environments such as high SEP (figure 2). If this hypothesis is correct, low family SEP will amplify the negative effects of familial loading on offspring's mental health.

    Figure 1
    Figure 1

    Illustration of the social push hypothesis. Higher loading on parental psychopathology (X axis) will have a stronger influence on mental health (Y axis) in adolescents from low-risk environment (high socioeconomic position (SEP)) than in those from high-risk environments (low SEP). Thus, the influence of familial loading on psychopathology on adolescent mental health is reduced by high-risk environment (low family SEP).

    Figure 2
    Figure 2

    Illustration of the vulnerability hypothesis. Higher loading on parental psychopathology (X axis) will have a stronger influence on mental health (Y axis) in adolescents from high-risk environment (low SEP) than in those from low-risk environments (high SEP). Thus, the influence of familial loading on psychopathology on adolescent mental health is amplified by high-risk environment (low family SEP). SEP, family socioeconomic position; FL, familial loading.

    Few studies have considered whether familiality for adolescent offspring mental health varied with environmental contexts such as family SEP. One study with adult samples found that the association between serotonin transporter genes (5-HTT) and depression varied with low socioeconomic status.29 Nevertheless, a recent meta-analysis showed that the serotonin transporter genotype alone or in interaction with stressful life events does not increase the risk of depression.30 To our knowledge, the only study that considered an adolescent sample showed that heritability for adolescent antisocial behaviour (a proxy for genetic heritability) increased with lower socioeconomic status in line with the social push hypothesis.31

    In the present study, we assessed the moderation of the relationship between parental loading on psychopathology (as assessed by lifetime psychopathology) and offspring mental health problems by family SEP, using data from a large cohort of Dutch adolescents from a general population. Although gender is unlikely to be related to familial loading on psychopathology or SEP, it is a known correlate of internalising and externalising problems. For that reason, we assessed for its confounding effects.

    Methods

    Sample

    Subjects were participants in the TRacking Adolescents' Individual Lives Survey (TRAILS), a prospective cohort study of Dutch (pre)adolescents, aimed at explaining the development of mental health problems from preadolescence into early adulthood. TRAILS was approved by the Central Committee on Research Involving Human Subjects. Sample selection involved five municipalities in the north of the Netherlands, including urban and rural areas. The five municipalities were requested to give names and addresses of all inhabitants born between 1 October 1989 and 30 September 1990 (first two municipalities: mean age 11.29 years (SD 0.52), range 10.0–12.0 years) or 1 October 1990 and 30 September 1991 (last three municipalities: mean age 10.72 years (SD 0.37), range 10.0–11.5 years). Two birth cohorts were used to minimize the age range during the initial assessment. Of all the children approached (N=3145), 6.7% (n=211) were excluded due to mental or physical incapability or language problems, leaving a total of 2934. Finally, 76% participated in the baseline assessment (T1-n=2230, mean age 11.1 years (SD 0.56), range 10.0–12.0 years), and 96.4% (n=2149) were assessed at follow-up (T2-mean age 13.6 years (SD 0.53), range 12.0–15.0 years), held about 2.5 years after T1 (mean follow-up time 2.47 years (SD 0.48), range 0.73–3.25 years). The T1 assessment was conducted from March 2001 through July 2002, and the T2 assessment from September 2003 to December 2004. The present study includes those 2149 adolescents who participated in the T2 assessment. Responders and non-responders did not significantly differ in levels of problem behaviours or on sociodemographic variables.32 33

    Data collection

    Interviewers were university graduates extensively trained in interviewing skills, study background and interview content. At T1, the interviewers visited parents or guardians (preferably mothers, 95.6%) at their homes to administer interviews covering a wide range of topics, including SEP, mental health and lifetime parental psychopathology. Interviews were conducted and questionnaires were filled-out after a complete description of the study was given and written informed consent was obtained from participants. At T1 and T2, children filled out questionnaires at school supervised by TRAILS assistants. Teachers were asked to fill out a brief questionnaire for all TRAILS children in their class.

    Measures

    Offspring mental health problems

    Two dimensions of mental health problems were included in this study: internalising and externalising problems. These problems were assessed at T1 and T2 with the parent-rated Child Behaviour Checklist (CBCL),34 the Youth Self-Report (YSR)35 and the Teacher Checklist of Psychopathology (TCP) with a timeframe of the past 6 months (CBCL and YSR) or 2 months (TCP). The TCP is comprised of descriptions of problem behaviours similar to Achenbach's teacher report form.36 The TCP was developed to reduce the respondent burden for teachers, as each had several participants to report on. This measure contains descriptions (vignettes) of problem behaviours corresponding to the syndrome scales of the CBCL and YSR (vignettes available upon request). Response options for each description of the TCP ranged from 0 (not applicable) to 4 (very clearly or frequently applicable). The TCP vignettes correlated around 0.60 with the full teacher report form syndrome scales filled out by a small sample of teachers (Ferdinand, 2003, internal report available upon request). Next, we created broadband scales of internalising problems (anxious/depressed, withdrawn/depressed and somatic complaints) and externalising problems (aggressive and rule-breaking behaviours). The validity of these scales have been documented34–36 and reiterated in a Dutch sample.37 In our sample, the reliability statistics are as follows: CBCL-internalising (32 items, Cronbach's α=0.85), CBCL-externalising (35 items, α=0.90), YSR-internalising (31 items, α=0.87), YSR-externalising (32 items, α=0.85), TCP-internalising (3 vignettes, α=0.71) and TCP-externalising (2 vignettes, α=0.78).

    Reports from different sources are needed to reduce rater bias in the prediction of mental health problems and provide better estimates of diagnosis than those based on a single source.38 It has also been demonstrated in previous studies that a combination of parent and teacher information results in an improvement of predictive power.38 For this reason, we computed a combined estimate using the scores on mental health problems given by the children, parents and teachers at T2. To place the same weight on information from different informants, the scores on YSR, CBCL and TCP were first standardised to a zero to one scale before averaging over informants.

    Family SEP

    Family SEP was assessed at baseline using five indicators: family income, educational level of the father and the mother and occupational level of both parents using the International Standard Classification of Occupations.39 We created a SEP variable by averaging the indicators after standardisation. An overall index of SEP was considered complementary, thus providing insights into the overall family socioeconomic disadvantage of the family and exposure of children and adolescents to social conditions such as environmental inequities and hazards, psychosocial stress and material deprivation. Although individual measures of SEP may be positively correlated, the indices are not interchangeable and may tap into different causal pathways.40–42 In addition, preliminary analyses showed no substantial differences among the different indices of SEP.

    The SEP index captured 61.2% of the variance in the five items with an internal consistency (Cronbach's α) of 0.84 in the TRAILS population. Missing values (eg, when there is only one parent in the family) were accounted for in the averaging of the standardized items. Missing values did not affect the associations of the SEP variable with other variables.43 Intact families, in which both biological parents and their children live together, constituted 76.3% (n=1640) while families in which biological parents were either divorced or separated or in which there was a step parent or single parent constituted 23.7% (n=509).

    Familial loadings on psychopathology

    Lifetime parental psychopathology was assessed at T1 by means of the TRAILS Family History Interview administered to the parents. The Family History Interview assessed five dimensions of psychopathology: depression, anxiety, substance dependence, persistent antisocial behaviour and psychosis. Each dimension was introduced by a vignette (available on request) describing the main DSM-IV characteristics of the dimension, followed by a series of questions assessing lifetime occurrence, professional treatment and medication use. Biological parents were interviewed separately using a single informant (often the mother). For each dimension, we assigned each parent to one of the following categories: 0=(probably) never had an episode, 1=(probably) yes, or 2=yes and treatment and/or medication.

    The prevalence rates in mothers and fathers were as follows: depression (27% and 15%), anxiety (16% and 6%), substance dependence (3% and 7%) and antisocial behaviour (3% and 7%). We computed familial loadings for the domains of internalising and externalising disorders separately. Both disorders are effective accounts of the number of lifetime disorders within each domain reported by the biological parents. Familial loading on internalising disorders included depression and anxiety and familial loading on externalising disorders included substance dependence and antisocial behaviours.28 The empirical justification for the construction of the familial loadings is twofold (data available on request). First, factor analysis of the disorder correlation matrix, for fathers and mothers separately, yielded two factors of internalising and externalising problems that were similar to the two-dimensional structure of common mental disorders.28 Second, the pattern of associations between parental disorders and offspring psychopathology was similar for fathers and mothers, suggesting that the paternal and maternal indices could be combined without distorting important details. Corresponding to this, paternal disorders correlated weakly with maternal disorders.28 For instance, paternal and maternal depression was associated (0.18) and so were paternal and maternal antisocial behaviour (0.26). Despite these weak correlations, the prevalence rates of subscales that constituted the parental loading on psychopathology (except father's anxiety and substance dependence and mother's antisocial behaviour) were comparable to the lifetime rates of large studies such as the NEMESIS study that used direct interviewing.28 44

    In this sample, the reliability statistics for familial loading were generally moderate: Cronbach's α=0.55 for familial loading on internalising and Cronbach's α=0.51 for familial loading on externalising. Although the Cronbach α's for the familial loading variables appeared to be moderate, their predictive validity was high and specific. Familial loading on internalising problems predicted offspring internalising problems but not externalising problems, whereas familial loading on externalising problems predicted offspring externalising but not internalising problems.

    Data analyses

    Missing data

    The percentage of missing data was between 6% and 31% for mental health dimensions. Two per cent of the data were missing for family SEP, and 2–3% were missing for lifetime parental psychopathology. To minimize loss of statistical power and risk of bias, we performed multiple imputations using the NORM programme.45 Multiple imputation is the preferred method of dealing with missing data when data are not missing completely at random.46 We created five data sets to account for the uncertainty in imputed data.47 They were analysed in an identical way, and the regression coefficients and standard errors were combined using Rubin's method for multiple imputation inference.47

    Statistical analyses

    First, we calculated descriptive statistics including product-moment correlations. Second, we performed linear regression analyses to assess the strength of individual independent predictors (family SEP and familial psychopathology) on offspring internalising (adjusted for externalising problems) and externalising (adjusted for internalising problems). Third, by means of multiple linear regression analyses, we assessed the main effects of familial loading and family SEP, as continuous variables, on adolescent internalising problems in the first model. We decided to use continuous variables for two reasons: (1) to increase the statistical power to detect interaction and (2) because preliminary analyses showed no evidence of a threshold effect in the analyses with individual independent variables and in those with predictors entered simultaneously. To rule out the possibility of spurious independent effects due to random measurement error as much as possible, we conducted and performed analyses based on factor scores. Finally, the interaction term familial loading on internalising problems × family SEP was added to the main effects in the second model. A negative regression coefficient for this interaction term would plead for the social push hypothesis, while a positive regression coefficient would support the vulnerability hypothesis. We repeated similar analyses with externalising problems. To correct for shared variance between them, we adjusted offspring internalising and externalising problems for each other to obtain dimension-specific results. Preliminary results did not show gender as a confounder in either individual independent predictors or analyses with predictors entered simultaneously and was subsequently removed from further analyses.

    Results

    Average age at T1 and T2 were mean 11.1 years (SD 0.55), range 10–12 years, and mean 13.6 years (SD 0.53), range 12–15 years, respectively. Fifty-one per cent of the adolescents were women. Correlations between the variables and their means, SD and range of scores are presented in table 1. Family SEP correlated more strongly with offspring and parental loading on externalising problems than with offspring and parental loading on internalising problems. Although significant, the correlation between family SEP and familial loading on internalising and externalising psychopathology were low. Preliminary results did not show gender as a confounder in either analyses with each predictor entered separately or analyses with predictors entered simultaneously. Consequently, it was removed from subsequent analyses.

    View this table:
    Table 1

    Bivariate correlations between measures of family SEP, familial psychopathology and offspring mental health problems

    Family SEP predicted offspring externalising problems more than internalising problems, and familial loading on internalising problems had a stronger effect on offspring internalising problems than familial loading on externalising problems had on offspring externalising problems (table 2). Each regression coefficient represents the number of SD change in the outcome variable per SD change of the independent variable. For example, the regression of internalising problems on familial loading on internalising problems means that a change of 1 SD in familial loading on internalising problems is associated with a 0.133 SD change in internalising problems.

    View this table:
    Table 2

    Regression analyses of offspring mental health problems on familial loading, family socioeconomic position and their interaction terms

    The results of analyses with individual independent predictors for familial loading on psychopathology and family SEP indicated that both factors each significantly predicted offspring internalising problems (table 2). When familial loading on internalising problems and family SEP were entered into the same model simultaneously, their respective regression coefficients only slightly changed, indicating that both factors are independently associated with more offspring internalising problems (table 2). When we repeated the same analyses with externalising problems, again familial loading on externalising problems and family SEP each significantly predicted offspring externalising problems in analyses with individual independent predictors. When familial loading on externalising problems and family SEP were simultaneously entered into the same model, their respective regression coefficients only slightly changed, indicating that both factors are independently associated with more offspring externalising problems (table 2). We performed preliminary analyses with each predictor entered separately and with predictors entered simultaneously to check for curvilinear (quadratic but no higher-order polynomials) and threshold interaction terms. However, none of these yielded significant effects.

    For offspring internalising and externalising problems, the interaction terms did not yield significant results (table 2). Compared to the first model, R2=0.098, F(3, 2145)=39.75, p<0.001, when familial loading on internalising problems and family SEP were simultaneously entered into the same model, addition of the interaction term did not improve the model for internalising problems, R2=0.098 (F(4, 2144)=30.14, p < 0.001). Likewise, compared to the first model, R2=0.122, F (3, 2145)=50.57, p < 0.001, when both familial loading on externalising problems were simultaneously entered into the same model, the second model including the interaction term did not add anything beyond the main effects of family SEP and familial loading on externalising problems on offspring externalising problems, R2=0.121 (F(4, 2144)=37.90, p<0.001).

    Discussion

    Using data from a large population cohort of young Dutch adolescents, we showed that familial loading on psychopathology and family SEP independently and significantly predicted offspring mental health problems. There was no evidence of interaction between the two factors in predicting offspring internalising and externalising problems. Therefore, neither the social push nor the vulnerability hypothesis is supported by our results.

    Strengths and limitations of the study

    The findings should be interpreted in light of three limitations. First, the familial loadings for internalising and externalising disorders took into account only lifetime parental psychopathology. Information on mental health problems of other close relatives was not included.28 In addition, only one parent (often the mother) was interviewed directly, and this parent was used to obtain information on the other parent not interviewed.28 Although evidence on the disadvantage of using family history interviews as compared to direct interviews of relatives is not conclusive,16 it is generally associated with under-reporting of lifetime mental health problems in parents. However, possible misclassification of parental psychopathology has most likely been non-differential as to the outcome variable—that is, independent of the offspring mental health and thereby unlikely to have affected the associations we studied in any substantial way.

    Second, familial loading and family SEP may be associated and partly in the same causal chain. For example, those with higher loadings on psychopathology might have had truncated education or high job turnover or losses and were therefore unable to achieve high individual SEP. Nevertheless, our analyses do not support this mechanism as the effects appeared largely independent. This became evident when we entered familial loading and SEP in the same model and the corresponding coefficients did not drop substantially (table 2). It is possible that the independent effects of familial loading and family SEP on offspring mental health were due to measurement errors. To rule out the possibility of spurious independent effects due to measurement errors, we performed our analyses based on factor scores.

    Third, we did not take into account any changes in family SEP across the follow-up period. However, family SEP, especially education, is known to be stable over time, and sudden changes from one level to another within a short duration of time, like in our study, are unlikely. In addition, the different indices of SEP can provide complementary information on exposure of children and early adolescents to social conditions such as violence or environmental hazards and access to leisure activities in a family. By using an overall index of SEP, we may gain insight into the overall socioeconomic disadvantage of the family. However, using an overall index of SEP is not without limitations. Individual measures of SEP may be positively correlated but are not interchangeable and may tap into different causal pathways.40–42 Finally, the study involved only early adolescents (mean age 13.6 years). Interaction effects may be detected later in adolescents or in adulthood as the adverse effects of low SEP are reported to be cumulative.48

    The major advantages of our study include the sufficiently large population-based sample (n=2149) to enable reasonable power for detecting interaction, prospective design and use of composite measures of offspring psychopathology based on multiple informants (parent, teacher and child). Furthermore, other assets include the use of multiply imputed datasets to address the problem of missing data, particularly common in longitudinal studies with multiple informants.

    Interpretation

    Our findings are not to be interpreted as disagreeing with the evidence for gene–environment interaction.31 This is because our measure of familial loading on psychopathology might be reflecting genetic and environmental susceptibility to psychopathology. However, it is not possible, based on our data, to disentangle whether it is the genetic or the environmental aspects of familial loading or both that account for their effects on offspring mental health problems. Thus, the mechanisms underlying our findings remain unclear.

    The findings might suggest that a model in which variability in adolescent mental problems is divided into independent components of family SEP and familial psychopathology is probably very simple to account for the dynamic interactions of genetic risks (eg, high loading on familial psychopathology) and harsh environments (low SEP) in the real world. It is possible that adolescent psychopathology is less predictable in low SEP and high loading on parental psychopathology—for example, due to the influence of non-shared environment, as postulated in the seminal work of Plomin and Daniels.49 This would, in turn, lead to an increase in environmental variability and subsequent inability to detect interaction.

    Another explanation could be that the interaction model fitting is actually being conducted on the portion of adolescent psychopathology that is independent of the components of familial psychopathology and family SEP. It is difficult to disentangle the causal relationship between family SEP and familial loadings on psychopathology. Yet, it is plausible that high loading on lifetime psychopathology can influence family SEP. For example, those with higher loadings on psychopathology might have had truncated education or high job turnover or even job losses and were therefore unable to achieve high individual SEP. In this study, however, the correlations between family SEP and familial loading on internalising and externalising problems were low; therefore, collinearity was not an issue. In addition, we simultaneously entered familial psychopathology and family SEP in the analyses as predictors, effectively correcting for each other's influences on offspring mental health problems.

    Comparison with other studies is limited because few studies have considered whether familiality for adolescent mental health varies with environmental contexts such as low family SEP. In addition, many studies have used different indices of SEP, making comparisons even more difficult. Thus, explanation of our findings may be limited to the construction and measure of SEP we used. Our results contrast with findings from a Swedish longitudinal population-based twin study (TCHAD),31 which showed that family socioeconomic status modified the influence of genetic factors on antisocial problems in adolescents. Unlike in our study, antisocial problems were not adjusted for comorbid emotional factors in the TCHAD study. In addition, the TCHAD study used educational level, occupational status and neighbourhood socioeconomic conditions as indicators of socioeconomic status, while in our study an aggregate measure of SEP (family income, level of education and occupational status for both parents) was used. Furthermore, in the TCHAD study, antisocial behaviour was assessed using a questionnaire regarding property, drug-related and violent offences. It is possible that neighbourhood socioeconomic conditions (ethnic diversity, neighbourhood basic educational and unemployment levels, and neighbourhood crimes) capture the nature of these offences (property, drug-related and violent offences) more than our measure of SEP (family income, parental education and occupational levels). Last, our sample consisted of early adolescents (mean age 13.6 years (SD 0.53)) while the TCHAD study samples were aged 16–17 years.

    Regarding the significant main effects of lifetime parental psychopathology and family SEP, our findings are generally in agreement with previous studies.7 50–54 The independence of the associations of family SEP and familial loading on psychopathology with offspring mental problems suggests that different mechanisms may be operating from familial psychopathology and SEP to offspring mental health.

    Conclusions

    During early adolescence, low levels of family SEP do not confer additional risks for mental health problems in offspring from parents with high loading on psychopathology compared to parents without a history of mental illness. Familial loading on psychopathology and family SEP seem to be independent risk factors for offspring mental health problems in early adolescence. More research is needed to further disentangle the processes between low family SEP and offspring mental health, on the one hand, and the link between parental psychopathology and offspring mental ill-health on the other. For example, it is possible that child mental health problems are bidirectional—that is, previous child mental health might have influenced parental psychopathology, which, in turn, exacerbates current child mental health problems. The likelihood of other environmental factors such as school-related factors (eg, underachievement) intervening in between familial loading and family SEP on the one hand and familial loading and offspring mental health problems on the other might be an interesting area of study for future research.

    What is already known on this subject

    • Family socioeconomic position (SEP) is associated with adolescent mental health, albeit modestly, including in this study. Parental psychopathology is known to influence offspring mental health. In addition, family SEP is associated with both parental psychopathology and adolescent mental health. Theory-driven studies to understand the processes between socioeconomic position and mental health, especially in adolescents, are lacking.

    What this study adds

    • We tested two competing interaction hypotheses: (1) the “social push” hypothesis in which the effects of parental psychopathology are hypothesised to be stronger in contexts in low environmental risks (eg, high SEP) and (2) the “vulnerability” hypothesis in which the effects of parental psychopathology are hypothesised to be stronger in contexts in high risk environments (eg, low SEP).

    • Findings demonstrated independent main effects of family SEP and parental psychopathology but no effects of their interaction terms for internalising and externalising problems.

    • During early adolescence, parental psychopathology and low family SEP seem to be independent risk factors for offspring mental health problems. Results do not support the social push or vulnerability hypothesis as no evidence of interactions between parental loading and family SEP were found.

    Research implications

    • More research is needed to further disentangle the processes between low family SEP and adolescents mental health on the one hand and the link between parental psychopathology and mental ill-health in adolescents offspring on the other.

    References

    1. ↵
      World Health Organisation. Child mental health atlas: child and adolescent mental health resources: global concerns, implications for the future. Geneva: WHO, 2005.
    2. ↵
      1. Hofstra MB,
      2. Van de Ende J,
      3. Verhulst FC
      . Adolescent's self-reported problems as predictors of psychopathology in adulthood: 10 year follow up study. Br J Psychiatr 2001;179:203–9.
      OpenUrlAbstract/FREE Full Text
    3. ↵
      1. Visser JH,
      2. Van der Ende J,
      3. Koot HM,
      4. et al
      . Predictors of psychopathology in young referred to mental health services in childhood or adolescence. Br J Psychiatry 2000;177:59–65.
      OpenUrlAbstract/FREE Full Text
    4. ↵
      1. Miech RA,
      2. Caspi A,
      3. Moffit TE,
      4. et al
      . Low socioeconomic status and mental disorders: a longitudinal study of selection and causation during young adulthood. AJS 1999;104:1096–131.
      OpenUrlCrossRef
    5. ↵
      1. Due P,
      2. Lynch J,
      3. Holstein B,
      4. et al
      . Socioeconomic health inequalities among nationally representative sample of Danish adolescents: the role of different types of social relations. J Epidemiol Community Health 2003;57:692–8.
      OpenUrlAbstract/FREE Full Text
    6. ↵
      1. Loeber R,
      2. Farrington DP,
      3. Stouthamer-Loeber M,
      4. et al
      . Antisocial behaviour and mental health problems: explanatory factors in childhood and adolescence. Mawhaw, NJ: Lawrence Erlbaum, 1998.
    7. ↵
      1. Leventhal T,
      2. Brooks-Gunn J
      . The neighbourhoods they live in: effects of neighbourhood residence upon child and adolescent outcomes. Psychol Bull 2000;126:309–37.
      OpenUrlCrossRefPubMedWeb of Science
    8. ↵
      1. Marmorstein NR,
      2. Iacono WG
      . Major depression and conduct disorder in youth: associations with parental psychopathology and parent-child conflict. J Child Psychol Psychiatr 2004;45:377–86.
      OpenUrlCrossRefPubMedWeb of Science
    9. ↵
      1. Dearing KF,
      2. Gotlib IH
      . Interpretation of ambiguous information in girls at risk for depression. J Abnorm Child Psychol 2009;37:79–91.
      OpenUrlCrossRefPubMedWeb of Science
    10. ↵
      1. Goodman SH,
      2. Gotlib IH
      . Children of depressed parents: mechanisms of risks and implications for treatment. Washington, DC: American Psychological Association, 2002.
    11. ↵
      1. Silberg J,
      2. Pickle A,
      3. Rutter M,
      4. et al
      . The influence of genetic factors and life stress on depression among adolescent girls. Arch Gen Psychiatry 1999;56:225–32.
      OpenUrlCrossRefPubMedWeb of Science
    12. ↵
      1. Conger RD,
      2. Conger KJ,
      3. Elder GH,
      4. et al
      . A family process model of economic hardship and adjustment of early adolescent boys. Child Dev 1992;63:526–41.
      OpenUrlCrossRefPubMedWeb of Science
    13. ↵
      1. Conger RD,
      2. Wallace LE,
      3. Sun Y,
      4. et al
      . Economic pressure in African American families: a replication and extension of the family stress model. Dev Psychol 2002;38:179–93.
      OpenUrlCrossRefPubMedWeb of Science
    14. ↵
      1. Kendler KS,
      2. Hettema JM,
      3. Butera F,
      4. et al
      . Life events dimensions of loss, humiliation, entrapment, and danger in the prediction of onsets of major depression and generalised anxiety. Arch Gen Psychiatry 2003;60:789–96.
      OpenUrlCrossRefPubMedWeb of Science
    15. ↵
      1. Sullivan PF,
      2. Neale MC,
      3. Kendler KS
      . Genetic epidemiology of major depression: review and meta-analysis. Am J Psychiatry 2000;157:1552–62.
      OpenUrlCrossRefPubMedWeb of Science
    16. ↵
      1. Kendler KS,
      2. Prescott CA
      . A population-based twin study of lifetime major depression in men and women. Arch Gen Psychiatry 1999;56:39–44.
      OpenUrlCrossRefPubMedWeb of Science
    17. ↵
      1. Evans GW
      . The environment of childhood poverty. Am Psychol 2004;59:77–92.
      OpenUrlCrossRefPubMedWeb of Science
    18. ↵
      1. Reijneveld SA,
      2. Brugman E,
      3. Verhulst FC,
      4. et al
      . Area deprivation and child psychosocial problems: a national cross-sectional study among school-aged children. Soc Psychiatry Psychiatr Epidemiol 2005;40:18–23.
      OpenUrlCrossRefPubMedWeb of Science
    19. ↵
      1. Schneiders J,
      2. Drukker M,
      3. Van der Ende J,
      4. et al
      . Neighbourhood socioeconomic disadvantage and behavioural problems from late childhood into early adolescence. J Epidemiol Community Health 2003;57:699–703.
      OpenUrlAbstract/FREE Full Text
    20. ↵
      1. Caspi A,
      2. Taylor A,
      3. Smart M,
      4. et al
      . Can women provide reliable information about their children's fathers? Cross-informant agreement about men's lifetime antisocial behaviour. J Child Psychol Psychiatr 2001;42:915–20.
      OpenUrlCrossRefPubMedWeb of Science
    21. ↵
      1. Stansfeld S,
      2. Head J,
      3. Bartley M,
      4. et al
      . Social position, early deprivation and the development of attachment. Soc Psychiatry Psychiatr Epidemiol 2008;43:516–26.
      OpenUrlCrossRefPubMedWeb of Science
    22. ↵
      1. Achenbach TM,
      2. Verhulst FC,
      3. Baron GD,
      4. et al
      . Epidemiological comparisons of American and Dutch children: I. Behavioural/emotional problems reported by teachers for ages 4 to 16. J Am Acad Child Adol Psychiatr 1987a;26:317–25.
      OpenUrlPubMedWeb of Science
    23. ↵
      1. Achenbach TM,
      2. McConaughy SH,
      3. Howell CT
      . Child/adolescent behavioural and emotional problems: implications of cross-informant correlations for situational specificity. Psychol Bull 1987;101:213–32.
      OpenUrlCrossRefPubMedWeb of Science
    24. ↵
      1. Amone-P'Olak K,
      2. Burger H,
      3. Ormel J,
      4. et al
      . Socioeconomic position and mental health problems in pre- and early-adolescents: the TRAILS study. Soc Psychiatry Psychiatr Epidemiol 2009;44:231–8.
      OpenUrlCrossRefPubMedWeb of Science
    25. ↵
      1. Raine A
      . Biosocial studies of antisocial and violent behaviour in children and adults: a review. J Abnorm Child Psychol 2002;30:311–26.
      OpenUrlCrossRefPubMedWeb of Science
    26. ↵
      1. Shanahan MJ,
      2. Hofer SM
      . Social context in gene–environment interactions: retrospect and prospect. J Gerontol B Psychol Sci Soc Sci 2005;60B:65–76.
      OpenUrl
    27. ↵
      1. Plomin R,
      2. Rutter M
      . Child development, molecular genetics and what to do with genes once they are found. Child Dev 1998;69:1223–42.
      OpenUrlCrossRefPubMedWeb of Science
    28. ↵
      1. Ormel J,
      2. Oldehinkel AJ,
      3. Ferdinand RF,
      4. et al
      . Internalizing and externalizing problems in adolescence: general and dimension-specific effects of familial loadings and preadolescent temperament traits. Psychol Med 2005;35:1825–35.
      OpenUrlCrossRefPubMedWeb of Science
    29. ↵
      1. Brummett BH,
      2. Boyle SH,
      3. Siegler IC,
      4. et al
      . Effects of environmental stress and gender on associations among symptoms of depression and the serotonin transporter gene linked polymorphic region (5-HTTLPR). Behav Genet 2008;38:34–43.
      OpenUrlCrossRefPubMedWeb of Science
    30. ↵
      1. Risch N,
      2. Herrell R,
      3. Lehner T,
      4. et al
      . Interaction between the serotonin transporter gene (5-HTTLPR), stressful life events, and risk of depression. JAMA 2009;301:2462–71.
      OpenUrlCrossRefPubMedWeb of Science
    31. ↵
      1. Tuvblad C,
      2. Grann M,
      3. Lichtenstein P
      . Heritability for adolescent antisocial behaviour differs with socio-economic status: gene environment interaction. J Child Psychol Psychiatry 2006;47:734–43.
      OpenUrlCrossRefPubMedWeb of Science
    32. ↵
      1. De Winter AF,
      2. Oldehinkel AJ,
      3. Veenstra R,
      4. et al
      . Evaluation of non-response bias in mental health determinants and outcomes in a large sample of pre-adolescents. Eur J Epidemiol 2005;20:173–81.
      OpenUrlCrossRefPubMedWeb of Science
    33. ↵
      1. Huisman M,
      2. Oldehinkel AJ,
      3. De Winter AF,
      4. et al
      . Cohort profile: the Dutch “TRacking Adolescents' Individual Lives Survey”; TRAILS. Int J Epidemiol 2008;37:1227–35.
      OpenUrlFREE Full Text
    34. ↵
      1. Achenbach TM
      . Manual for the child behaviour checklist/4-18 and 1991 profile. Burlington, VT: University of Vermont, 1991a.
    35. ↵
      1. Achenbach TM
      . Manual for the youth self-report and 1991 profile. Burlington, VT: University of Vermont, 1991b.
    36. ↵
      1. Achenbach TM
      . Manual for the teacher's report form and 1991 profile. Burlington, VT: University of Vermont, 1991c.
    37. ↵
      1. Verhulst FC,
      2. Van der Ende J,
      3. Koot HK
      . Handleiding voor de Youth Self-Report (YSR) (Manual for the Youth Self-Report (YSR)). Rotterdam: Sophia Children's Hospital Rotterdam, 1997.
    38. ↵
      1. Verhulst FC,
      2. Koot HM,
      3. Van der Ende J
      . Differential predictive value of parents' and teachers' reports of children's problem behaviours: a longitudinal study. J Abnorm Child Psychol 1994;22:531–46.
      OpenUrlCrossRefPubMedWeb of Science
    39. ↵
      1. Ganzeboom HBG,
      2. Treiman DJ
      . International comparable measures of occupational status for the 1988 International Classification of Occupations. Soc Sci Res 1996;25:201–39.
      OpenUrlCrossRefWeb of Science
    40. ↵
      1. Geyer S,
      2. Hemström Ö,
      3. Peter R,
      4. et al
      . Education, income and occupational class cannot be used interchangeably in social epidemiology. Empirical evidence against a common practice. J Epidemiol Community Health 2006;60:804–10.
      OpenUrlAbstract/FREE Full Text
    41. ↵
      1. Lahelma E,
      2. Laaksonen M,
      3. Martikainen P,
      4. et al
      . Multiple measures of socio-economic circumstances and common mental disorders. Soc Sci Med 2006;63:1383–99.
      OpenUrlCrossRefPubMedWeb of Science
    42. ↵
      1. Shavers VL
      . Measurement of socioeconomic status in health disparities research. J Natl Med Assoc 2007;99:1013–23.
      OpenUrlPubMed
    43. ↵
      1. Veenstra R,
      2. Lindenberg S,
      3. Oldehinkel AJ,
      4. et al
      . Bullying and victimization in elementary schools: a comparison of bullies, victims, bully/victims, and uninvolved preadolescents. Dev Psychol 2005;41:672–82.
      OpenUrlCrossRefPubMedWeb of Science
    44. ↵
      1. Bijl RV,
      2. Ravelli A,
      3. van Zessen G
      . Prevalence of psychiatric disorder in the general population: results of The Netherlands Mental Health Survey and Incidence Study (NEMESIS). Soc Psychiatry Psychiatr Epidemiol 1998;33:587–95.
      OpenUrlCrossRefPubMedWeb of Science
    45. ↵
      1. Schafer JL
      . NORM: Multiple imputation of incomplete data under a normal model, version 2.Software for Windows 95/98/NT. 1999. http://www.stat.psu.edu/∼jls/misoftwa.html (accessed 18 Sep 2009).
    46. ↵
      1. Donders AR,
      2. van der Heijden GJ,
      3. Stijnen T,
      4. et al
      . Review: a gentle introduction to imputation of missing values. J Clin Epidemiol 2006;59:1087–91.
      OpenUrlCrossRefPubMedWeb of Science
    47. ↵
      1. Rubin DB
      . Multiple imputation for nonresponse in surveys. New York: J. Wiley & Sons, 1987.
    48. ↵
      1. Bartley M,
      2. Blane D,
      3. Smith GD
      1. van de Mheen H,
      2. Stronks K,
      3. Mackenbach JP
      . A life-course perspective on socio-economic inequalities in health: the influence of childhood socio-economic conditions and selection process. In: Bartley M, Blane D, Smith GD, eds. The sociology of health inequalities. Oxford, UK: Blackwell Publishers Ltd, 1998:193–216.
    49. ↵
      1. Plomin R,
      2. Daniels D
      . Why are children in the same family so different from one another? Behav Brain Sci 1987;10:1–60.
      OpenUrlCrossRefWeb of Science
    50. ↵
      1. Lieb R,
      2. Isensee B,
      3. Hofler M,
      4. et al
      . Parental major depression and the risk of depression and other mental disorders in offspring: a prospective longitudinal community study. Arch Gen Psychiatry 2002;59:365–74.
      OpenUrlCrossRefPubMedWeb of Science
    51. ↵
      1. Rutter M,
      2. Silberg J,
      3. O'Connor T,
      4. et al
      . Genetics and child psychiatry: I Empirical research findings. J Child Psychol Psychiatry 1999a;40:3–18.
      OpenUrlCrossRefPubMedWeb of Science
    52. ↵
      1. Rutter M,
      2. Silberg J,
      3. O'Connor T,
      4. et al
      . Genetics and child psychiatry: II Empirical research findings. J Child Psychol Psychiatry 1999b;40:19–55.
      OpenUrlCrossRefPubMedWeb of Science
    53. ↵
      1. Rutter M,
      2. Giller H,
      3. Hagell A
      . Antisocial behaviour by young people. Cambridge: Cambridge University Press, 1998.
    54. ↵
      1. Rhee SH,
      2. Waldman ID
      . Genetic and environmental influences on antisocial behaviour: a meta-analysis of twin and adoption studies. Psychol Bull 2002;128:490–529.
      OpenUrlCrossRefPubMedWeb of Science

    Footnotes

    • Funding TRAILS was financially supported by various grants from the Netherlands Organization for Scientific Research NWO (Medical Research Council programme grant GB-MW 940-38-011; ZonMW Brainpower grant 100-001-004; ZonMw Risk Behavior and Dependence grants 60-60600-98-018 and 60-60600-97-118; ZonMw Culture and Health grant 261-98-710; Social Sciences Council medium-sized investment grants GB-MaGW 480-01-006 and GB-MaGW 480-07-001; Social Sciences Council project grants GB-MaGW 457-03-018, GB-MaGW 452-04-314 and GB-MaGW 452-06-004; NWO large-sized investment grant 175.010.2003.005); the Sophia Foundation for Medical Research (projects 301 and 393), the Dutch Ministry of Justice (WODC) and the participating universities (University Medical Center and University of Groningen, Erasmus University Medical Center Rotterdam, University of Utrecht, Radboud Medical Center Nijmegen and Trimbos Institute, all in the Netherlands).

    • Competing interests None.

    • Patient consent Obtained.

    • Ethics approval TRAILS was approved by the Central Committee on Research Involving Human Subjects, University Medical Centre, Groningen.

    • Provenance and peer review Not commissioned; externally peer reviewed.