Multidimensional Typology of Social Networks of Older Adults and Its Association with Cognitive and Physical Function in Late Life

Abstract

Social networks of older adults are an important risk factor for various health outcomes including mental health, well-being, and pre-mature mortality. Despite their relevance for health in old age, measurement of social networks to date has focused largely on isolated network characteristics, and their influence has primarily been examined in relation to emotional and mental health outcomes. As a result, the multidimensional nature of social networks and their association with aging-relevant outcomes of disability, mobility, and cognitive function remains poorly understood. In this dissertation, we used data from the National Social Life, Health, and Aging Project (NSHAP), a large, nationally-representative, prospective cohort, to examine the multidimensional and heterotypic nature of older adults’ social networks and their influence on various functional health outcomes.

In the first study, we derived multidimensional social network types among older adults in the United States by applying latent class analysis to nine observed network characteristics representing the structure, function, and quality of relationships. We found that older adults can be classified into five distinct network types that differ in network size, social support, and presence of a partner: diverse, supportive network with partner; average network with partner; partner-centered network; large, supportive network without partner; and restricted, family-centered network without partner. Membership in these network types varied by age, sex, race/ethnicity, education, and income.

Next, we examined the relevance of these network types for physical functional health in a prospective analysis and found the association between network types and functional health to be somewhat equivocal. Compared to the diverse, supportive network with partner, the partner-centered network and the average network with partner had a lower risk of onset of activities of daily living disability in the short-term (5-year risk) but not the long-term (10-year risk). Network types also had an effect on mobility, such that older adults in the restricted, family-centered network without partner had significantly slower walking times than those in the diverse, supportive network with partner. These network types, however, did not have a longitudinal effect on change in mobility.

Lastly, the third aim of this dissertation was to examine the association of social network types with cognitive function in later life. We found that social network types were not associated with onset of cognitive impairment during the 10-year study. Although unadjusted analyses suggested a higher risk of cognitive impairment onset among those in one of the two restricted network types: partner-centered network and restricted, family-centered network without partner, adjustment for sociodemographic and health-related background characteristics rendered these associations insignificant.

Taken together, the results of this dissertation provide evidence of the heterogeneity of older adults’ social networks. By taking a pattern-centered approach that simultaneously considers the structural, functional, and qualitative network characteristics, we offer a more nuanced view of individuals’ social relationships in the form of network types. The overall pattern of findings did not suggest a clear link between network types and changes in functional health outcomes, with the possible exception of short-term changes in disability. Nevertheless, the social network types identified in this dissertation can inform ways to increase available social support and reduce social isolation among older adults. Consideration of social network resources, at least in the short term, may also offer opportunities for the prevention and postponement of certain functional limitations in old age.