Background This study investigates the long-term impact of childhood conditions on social relationships in later life. Social relationships predict health and emotional wellbeing across the life course and the long-term effect of childhood on later life health and wellbeing is well established. However, it is not known whether childhood conditions also predict social relationship characteristics in later life. Furthermore, it is important to determine whether any association between childhood and later life social outcomes is indirect (operating through the well-established relationship between childhood and adult socio-economic status and health) or direct.

Data Data are from two waves of the Survey of Health, Ageing and Retirement in Europe (SHARE), a nationally representative data on older Europeans. We use retrospectively reported childhood data from SHARELIFE and social relationship information (social network size, emotional closeness, social activities and satisfaction with activities undertaken) from Wave 4. We also include all respondents who are present in both waves (n = 19,000).

Methods In step one; we use regression analysis to determine the extent to which childhood conditions (childhood SES, cognition, and parental separation and childhood health) predict each social relationship characteristic. We include a set of country fixed effects, which accounted for any factors which were common to respondents in that particular country, and control for an additional set of demographic factors.

In step two, we determine how the total effect (coefficient sum) of the childhood variables is attenuated by adult circumstances, by adding control variables for adult characteristics (SES, behaviour and health) to the model in a stepwise manner. In this way, we investigate the potential mechanisms through which the effect of childhood conditions may operate.

Results Results illustrated that the long-run effects of different childhood conditions were not uniform. We observed a direct relationship between childhood health and emotional closeness and between parental separation and emotional closeness. After controlling for endogenous adult characteristics, the effects of childhood SES and cognition in predicting emotional closeness remained significant.

In this study, social relationships in later life were partly determined by childhood conditions, and these effects appeared to be largely independent of the established relationship between childhood environment and adult factors (such as socio-economic status and health). These results confirm that there are diverse, direct and indirect pathways linking early life conditions to later outcomes, and understanding the long arm of childhood into later social life requires consideration of each of these.