Direct and indirect effects of family members on mother's depression: an accelerated longitudinal design using over 25, 000 genotyped trios

Maternal depression is a heterogeneous disorder with multiple risk factors whose effects likely vary across the life course. Typically, depression models focus on a single risk factor and when genetically sensitive designs are used, only direct effects are considered. It remains unknown whether the influence of putative risk genes on maternal depression varies across development as a function of whether these genes are acting directly through mothers or indirectly through their family members. Here, we restructured longitudinal Symptom Checklist scores for depression assessed in over 25,000 genotyped mothers from the Norwegian Mother, Father and Child study (MoBa) by maternal age, resulting in nearly 150, 000 depression measurements across 40 childbearing years. Next, we predicted variation in depression across time, by jointly modelling 22 polygenic scores constructed in mothers (direct genetic effects), fathers and children (indirect genetic effects) for seven putative risk domains including: common disorders, psychiatric disorders, neurodevelopmental and obsessive disorders, anthropometric traits, risk taking, social demographics, and well- and ill-being. We found direct genetic effects on maternal depression were highest for the domains of well-being/ill-being (total b = 0.129), common mental disorders (total b = 0.043) and social demographics (total b = 0.037). In general, indirect genetic effects were lower than direct genetic effects, except for risk taking where fathers indirect genetic effect (total b = 0.016) was more than double mothers direct genetic effect (total b = 0.007). Interactions with indirect genetic effects over time were similarly small for fathers (total b= 0.012) and children (total b = 0.014), although effects were spread across most of the 7 domains for fathers and restricted to the sociodemographic domain for children. By jointly modelling trio effects our estimates are largely unbiased by assortative mating, population stratification and parental selection of child environments. We discuss study implications and novel insights gained by applying life course perspectives to within-family designs.