Cross-Lagged Modeling of Cognition and Social Network Size in the Sydney Memory and Ageing Study

Abstract

Abstract Objectives This study assessed whether reciprocal relationships exist between cognitive function and the social network size of older adults, controlling for age, sex, education, medical conditions, and depressive symptoms. Methods Data were collected at biennial follow-ups over 6 years in the Sydney Memory and Ageing Study, a longitudinal cohort study including 1,037 community-based Sydney residents aged 70–90 years without dementia at baseline. We used random intercept cross-lagged panel models to investigate reciprocal associations between social network size and scores in each of 7 cognitive domains including a global score. Results Standardized models indicated that within-person deviation in expected language score predicted deviation in expected network size. Within-person deviation in prior expected social network size predicted deviation in expected executive function at year 6. Cross-lagged effects in models of both global cognition and memory, respectively, could not be attributed solely to within-person change. Discussion Findings support a co-constitutive view of cognitive function and social relationships in older age. Although both cognition and network size declined over time, slower than expected decline in language ability predicted less than expected contraction in social networks. A similar influence of network size on executive functioning indicated that relationships with friends and family outside of the home contributed significantly to the maintenance of higher order cognitive abilities in older late life. Diverse patterns of influence between cognitive domains and social network size over 6 years underscore the importance of assessing the complex and nuanced interplay between brain health and social relationships in older age.