Children from socioeconomically disadvantaged backgrounds exhibit higher rates of emotional and behavioral difficulties compared to their more advantaged peers. These disparities are evident across various emotional and behavioral conditions, including depression, anxiety, and attention-deficit hyperactivity disorder (ADHD), and are often linked to parental education, household income, and other socioeconomic indicators. The complex interplay of factors makes it challenging to isolate the independent effects of socioeconomic factors. Parental depression, for example, can simultaneously impact family socioeconomic status and children's mental health through different pathways. Furthermore, the genetic inheritance from parents contributes significantly to educational attainment and potentially to the observed associations between parental education and children's outcomes. Therefore, disentangling the impact of parental influence through environment from direct genetic effects requires sophisticated methodological approaches. Previous studies have employed family-based designs, such as the children-of-twins method, and molecular genetic data to investigate these relationships, yielding mixed results regarding the specific role of parental education and its influence on children’s traits. Mendelian randomization (MR), which leverages genetic variants associated with an exposure as instrumental variables, provides a powerful tool to address this challenge, mitigating the biases associated with traditional observational studies. This study uses a within-family MR design in a large sample from the Norwegian Mother, Father, and Child Cohort Study (MoBa) to investigate the independent and combined effects of maternal and paternal educational attainment on children's depressive, anxiety, and ADHD traits at age 8, while accounting for direct genetic effects.

Existing research consistently demonstrates a link between socioeconomic disadvantage and increased likelihood of emotional and behavioral problems in children and adolescents. This association has been observed across various socioeconomic indicators, including parental education, household income, and housing tenure. Studies highlight that interventions targeting children's socioeconomic circumstances could potentially improve their emotional and behavioral well-being. However, determining which specific aspects of the environment to target remains challenging. The high correlation between socioeconomic factors like parental education and income complicates the separation of their individual effects. Furthermore, the potential confounding influence of parental mental health, which can affect both family socioeconomic status and children's mental health through separate pathways, needs careful consideration. In addition to shaping a child's environment, parents contribute 50% of their genome. Twin studies and genome-wide association studies (GWAS) have revealed the substantial genetic influence on educational attainment, suggesting that genetic inheritance might also play a role in the observed correlation between parental education and children's outcomes. Prior research has employed family-based approaches, such as the children-of-twins method, to address these complex interactions. Some studies utilizing molecular genetic data on multiple family members have suggested parental influence on depressive and ADHD traits but not anxiety traits, but these studies did not isolate specific parental traits like education. Other work using Mendelian randomization (MR), a method that employs genetic variants as instrumental variables, has investigated these intergenerational relationships. However, separating the influence of parents' education from the effects of inherited variants in children requires large datasets encompassing both parents and children's genetic information, limiting prior MR studies in this area. The current study aims to address these gaps by utilizing within-family MR within the MoBa cohort.

This study employed a within-family Mendelian randomization (within-family MR) design using data from the Norwegian Mother, Father, and Child Cohort Study (MoBa). The sample comprised 40,879 mother-father-child trios, with genetic data available for all three members. The study's primary objective was to evaluate the impact of maternal and paternal educational attainment on children's depressive, anxiety, and ADHD traits at age 8. Parental education was instrumented using a polygenic index (PGI) encompassing genetic variants previously associated with educational attainment. This approach minimizes confounding and reverse causation. To account for direct genetic effects, the children's own polygenic indexes for educational attainment were included as covariates in the analyses. The study included both mothers' and fathers' genotypes, allowing for the estimation of conditional effects of both parents' education. Additionally, the methodology provided an unconfounded estimate of the direct impact of genetic variants on children's traits. To address potential bias due to horizontal pleiotropy, where genetic variants in a PGI might influence the outcome through pathways not involving the exposure, the study performed additional two-sample Mendelian randomization analyses. These analyses compared associations of individual genetic variants with both exposures and outcomes to assess horizontal pleiotropy. Descriptive statistics were calculated for the study sample, including details on parental education, age, and mental health traits, as well as children's mental health traits at age 8. Multiple imputation by chained equations was used to handle missing data in the phenotypic data. The education PGI served as a strong instrument, as indicated by the first-stage F-statistic and the conditional R². Statistical analyses involved multivariable regression models, initially adjusting for basic covariates like child's sex and birth year. Subsequently, the models were adjusted for potential confounders including parental mental health traits, smoking status, and maternal parity. Mendelian randomization models were implemented using parents' PGIs to instrument parental education. Within-family MR models further controlled for children's own education PGIs to account for direct genetic effects. Sensitivity analyses were conducted, including sex-specific models, square root-transformed outcomes, and complete-case analysis, and two-sample MR analyses were performed using all SNPs in the PGI, and a subset of independent SNPs. The analyses were conducted in Stata, using robust standard errors and appropriate statistical methods for handling instrumental variables and multiple imputation.

Initial non-genetic analyses revealed small but significant associations between parental education and reduced depressive and ADHD traits in children. However, these associations attenuated after adjusting for potential confounders. Within-family Mendelian randomization models, which are less susceptible to confounding, showed limited evidence that parental education causally impacted children's depressive, anxiety, or ADHD traits. Confidence intervals suggested that effects larger than 0.08 standard deviations per year of parental education were unlikely. In contrast, children's own polygenic scores for educational attainment were negatively associated with their depressive and ADHD traits, independent of their parents' PGIs. This effect, though small (0.04 S.D. decrease in depressive traits and 0.08 S.D. decrease in ADHD traits per one S.D. increase in the PGI), suggests that the correlation between parental education and children's traits might be driven more by direct genetic effects rather than genetic nurture. Sensitivity analyses generally supported the main findings. Two-sample Mendelian randomization results using inverse variance weighted (IVW) regression showed some indication of associations between maternal education and children's traits, but other methods provided less consistent results, suggesting possible type 1 errors and the need for caution in interpreting these findings. Complete-case analyses, while less precise due to the reduced sample size, yielded some positive associations between maternal education and children's depressive and ADHD traits, which again requires careful consideration in the context of the overall findings.

This large-scale study utilizing within-family Mendelian randomization provided robust evidence that any causal effects of parental education on children's depressive, anxiety, and ADHD traits at age 8 are modest, at least within the relatively equitable socioeconomic context of Norway, where most young adults participate in tertiary education. The observed associations between parental education and children's mental health traits in non-genetic models appear to be largely explained by direct genetic effects rather than genetic nurture effects. The study's findings contrast with some previous research reporting robust associations between parental education and children's mental health in other countries, highlighting potential contextual differences in these relationships. This discrepancy may be attributed to factors such as variations in national contexts, methodologies, and the representativeness of the study populations. The significant negative association between children's own polygenic scores for educational attainment and mental health traits suggests that genetic predispositions influencing later educational success might have a protective effect on mental health in early childhood. This protective effect likely operates through pathways related to other phenotypes linked to educational attainment, such as cognitive and non-cognitive traits, and warrants further investigation. The study underscores the potential importance of other aspects of the family environment and the need to explore specific parental characteristics beyond educational attainment in understanding children's mental health.

This study, using a large sample of Norwegian mother-father-child trios and advanced genetic methods, found limited evidence for a causal relationship between parental educational attainment and children's mental health traits at age 8. The observed correlations appear to be largely explained by direct genetic effects. Children's own genetic predisposition towards higher educational attainment showed a small protective effect on depression and ADHD traits. Further research is needed in more diverse populations with greater socioeconomic inequalities to better understand the complex interplay of genetic and environmental factors influencing children's mental health.

The study's generalizability might be limited due to the relatively high level of education and better health among the participants compared to the general Norwegian population, potentially reflecting selection bias. The significant proportion of missing data in the outcome variables and reliance on parental reports could have influenced the findings. While multiple imputation was used to address missing data, this method introduces uncertainty. Future research should employ larger, more representative samples with complete data, focusing on different age groups and countries with varying levels of socioeconomic inequality. Additionally, using independent assessments of children's mental health traits might help to reduce potential bias.