The association between weight and depressive symptoms is well established, but the direction of effects remains unclear. Most studies rely on body mass index (BMI) as the sole weight indicator, with few examining the aetiology of the association between weight indicators and depressive symptoms.

We analysed data from the Twins Early Development Study (TEDS) and UK Adult Twin Registry (TwinsUK) (7658 and 2775 twin pairs, respectively). A phenotypic cross-lagged panel model assessed the directionality between BMI and depressive symptoms at ages 12, 16, and 21 years in TEDS. Bivariate correlations tested the phenotypic association between a range of weight indicators and depressive symptoms in TwinsUK. In both samples, structural equation modelling of twin data investigated genetic and environmental influences between weight indicators and depression. Sensitivity analyses included two-wave phenotypic cross-lagged panel models and the exclusion of those with a BMI <18.5.

Within TEDS, the relationship between BMI and depression was bidirectional between ages 12 and 16 with a stronger influence of earlier BMI on later depression. The associations were unidirectional thereafter with depression at 16 influencing BMI at 21. Small genetic correlations were found between BMI and depression at ages 16 and 21, but not at 12. Within TwinsUK, depression was weakly correlated with weight indicators; therefore, it was not possible to generate precise estimates of genetic or environmental correlations.

The directionality of the relationship between BMI and depression appears to be developmentally sensitive. Further research with larger genetically informative samples is needed to estimate the aetiological influence on these associations.

Cognitive ability and problem behaviour (externalising and internalising problems) are variable and inter-related in children. However, it is not known if they mutually influence one another, if difficulties in one cause difficulties in the other, or if they are related only because they share causes.

Random-intercept cross-lagged models adjusted for confounding were fitted to explore this in 17,318 (51% male) children of the UK’s Millennium Cohort Study at ages 3, 5, 7, 11 and 14 years. Externalising and internalising problems were assessed using the parent-reported Strengths and Difficulties Questionnaire. Cognitive ability was measured using standardised scores of age-appropriate validated cognitive ability assessments. Where multiple cognitive assessments were available a single score was derived using principal components analysis.

There was much evidence for cross-domain longitudinal effects in childhood, especially for cognitive ability (on both internalising and externalising problems and in both males and females) and externalising problems (on internalising problems in both genders and cognitive ability in males). Bidirectional effects were childhood-limited, gender-specific and less consistent. The consistent bidirectional associations were, in males, between externalising problems and cognitive ability, and, in females, between externalising and internalising problems (although the effects of internalising problems were weak). In adolescence, only externalising problems had cross-domain effects such that, in both genders, they were associated with lower cognitive ability in subsequent measurements and increased levels of internalising problems.

In either childhood or adolescence, reducing behavioural problems could have both emotional and cognitive benefits. In childhood, improving cognitive skills could reduce both emotional and behavioural problems.

Many studies of various stress reactive phenotypes suggest that 5-HTTLPR short allele carriers (S-carriers) are characterised by the stable trait of negative affectivity that is converted to psychopathology only under conditions of stress. In this study, we examined the moderating role of the 5-HTTLPR on the relationship between two objective chronic risk factors, i.e. socioeconomic status (SES) and family structure, and internalising symptoms across adolescence.

A multigroup path analysis was employed in a general adolescent population sample of a 5-year follow-up study.

Internalising problems were significantly more stable in the S-carriers. The focus on the main dimensions of internalising problems, i.e. anxiety and depression, revealed two different developmental patterns. In the S-carriers Anxiety problems seemed to be more stable and to predict a possible evolution towards the development of Depressive problems. In the long allele homozygotes (LL-subjects) the anxiety trait was significantly less stable, and, in late-adolescence, seemed to be significantly predicted by SES, suggesting a possible gene–environment interaction (G × E). Family structure seemed to play a role in a G × E perspective only until early-adolescence, while during late-adolescence SES seemed to play a pivotal role in interaction with 5-HTTLPR, with the S-allele playing a protective role.

Future models of the developmental link between environmental adversities and internalising behaviour therefore need to consider that the effect of G × E interaction, may be associated with internalising behaviour via different mechanisms during different time frames and that shifts in the strength of this effect should be expected across development.