Socioeconomic status and risk for child psychopathology: exploring gene-environment interaction in the presence of gene-environment correlation using extended families in the Norwegian Mother, Father and Child Birth Cohort Study

Abstract

Background: Low socioeconomic status (SES) is associated with increased risk for emotional and behavioural problems among children. Evidence from twin studies has shown that family SES moderates genetic and environmental influences on child mental health. However, it is also known that SES is itself under genetic influence and previous gene-environment interaction (G×E) studies have not incorporated the potential genetic overlap between child mental health and family SES into G×E analyses. We applied a novel approach using extended family data to investigate the moderation of aetiological influences on child emotional and behavioural problems by parental socioeconomic status in the presence of modelled gene-environment correlation.

Methods: The sample comprised >28,100 children in extended-family units drawn from the Norwegian Mother, Father and Child Cohort Study (MoBa). Mothers reported children's emotional and behavioural symptoms. Parents' income and educational attainment were obtained through linkage to administrative register data. Bivariate moderation Multiple-Children-of-Twins-and-Siblings (MCoTS) models were used to analyse relationships between offspring outcomes (emotional and behavioural symptom scores) and parental socioeconomic moderators (income rank and educational attainment).

Results: The aetiology of child emotional symptoms was moderated by maternal and paternal educational attainment. Shared environmental influences on child emotional symptoms were greater at lower levels of parents' education. The aetiology of child behavioural symptoms was moderated by maternal, but not paternal, socioeconomic factors. Genetic factors shared between maternal income and child behavioural symptoms were greater in families with lower levels maternal income. Nonshared environmental influences on child behavioural symptoms were greater in families with higher maternal income and education.

Conclusions: Parental socioeconomic indicators moderated familial influences and nonshared environmental influences on child emotional and behavioural outcomes. Maternal SES and child mental health share aetiological overlap such that shared genetic influence was greater at the lower end of the socioeconomic distribution. Our findings collectively highlight the role that family socioeconomic factors play in shaping the origins of child emotional and behavioural problems.

Keywords: Emotional problems; MoBa; behavioural problems; gene-environment interaction; socioeconomic status.

Figure 1.

Partial path diagram of the Multiple-Children-of-Twins-and-Siblings (MCoTS) model showing the addition of moderation terms to the intergenerational and child trait paths (right) to the MCoTS model (left; Figure S1). The model is shown for only one sibling in the parent generation and one child. Right figure. A1 = additive genetic effects on parental trait; C1 = shared environmental effects on parental trait; E1 = nonshared environmental effects on parental trait; A1’ = genetic effects shared between parental trait and offspring trait; C1’ = extended family effects (i.e. shared environment of the parents influences offspring trait); A2 = genetic effects specific to offspring trait; C2 = shared environmental effects on offspring trait; E2 = nonshared environmental effects on offspring trait; p = residual phenotypic association after accounting for genetic and environmental overlap. Left figure. A1’, C1’, and p are the variance components common to parent SES (the moderator) and child emotional or behavioural symptoms. A2, C2, and E2 are the variance components unique to child emotional or behavioural symptoms. β coefficients index the direction and magnitude of moderation. The total variance of the trait can be calculated by squaring and summing all the paths leading to it: $\text{Var}(\text{T}|\text{M})=(\text{a}1’+\text{a}1’*{\text{β}}_{\text{xc}}\text{M}{)}^2+(\text{p}+\text{p}*{\text{β}}_{\text{zc}}\text{M}{)}^2+(\text{c}1’+\text{c}1’*{\text{β}}_{\text{yc}}\text{M}{)}^2+(\text{a}2+\text{a}2*{\text{β}}_{\text{xu}}\text{M}{)}^2+(\text{e}2+\text{e}2*{\text{β}}_{\text{zu}}\text{M}{)}^2+(\text{c}2+\text{c}2*{\text{β}}_{\text{yu}}\text{M}{)}^2$. Note. Var = variance; T = trait; M = moderator; The loadings of the cross-paths connecting M to T consist of parts unrelated to the moderator M, i.e., a1’, p, and c1’ and parts that depend on M via weights ${\text{β}}_{\text{xc}}$, ${\text{β}}_{\text{zc}}$, and ${\text{β}}_{\text{yc}}$. The loadings of the paths unique to T consist of parts that are unrelated to M, i.e., a2, e2 and c2, and parts that depend on M via weights ${\text{β}}_{\text{xu}}$, ${\text{β}}_{\text{zu}}$, and ${\text{β}}_{\text{yu}}$.

Figure 2.

Unstandardised variance components in child emotional problems moderated by maternal income rank (a) and educational attainment (b).

Figure 3.

Unstandardised variance components in child behavioural problems moderated by maternal income rank (a) and education attainment (b).

Figure 4.

Unstandardised variance components in child emotional problems moderated by paternal income rank (a) and educational attainment (b).

Figure 5.

Unstandardised variance components in child behavioural problems moderated by paternal income rank (a) and educational attainment (b).