Methods: Data and Sample: Nine waves of data (2011-2019) from the National Health and Aging Trends Study (NHATS), which began collecting data annually from a representative sample of Medicare beneficiaries in 2011, were used (N=5,142, observations=23,529).
Measures: Social isolation was assessed by items indicating social participation, living arraignments and social networks; and was classified into three categories: severe isolation, some isolation, and social integration. Within-person changes in marital status (married vs. unmarried), residence (relocate vs. not), activities of daily living limitations (ADL; yes vs. not) and instrumental activities of daily living limitations (IADL; yes vs. no) during the observation period were the independent variables of interest.
Data Analysis: Two-level multinomial logistic regression with random intercepts was conducted. To partition within- and between-person variance, within-person changes and between-person differences in marriage, relocation, ADL and IADL limitations were used to predict changes in social isolation, controlling for time and age. The analysis was weighted.
Results: The relative risk for some isolation and severe isolation, relative to social integration, would be expected to increase more than 4 (relative risk ratio [rrr] = 4.47, p<.001) and 42 (rrr = 42.47, p<.001) times, respectively, when transitioning from being married to unmarried. No sex difference was detected. Relocation increases the relative risk of some isolation for both sexes (rrr = 1.16). But relocation significantly lowers the risk of severe isolation for men (rrr = .77) and is not associated with severe isolation risk for women. Transition from no to having ADL limitations is associated with a greater risk of some isolation for women (rrr = 1.32) than for men (rrr = 1.22), and increases the risk of severe isolation similarly for both sexes. Change from no to having IADL limitation increases the relative risk of some isolation for both sexes, but increases the risk of severe isolation for women only.
Conclusions and Implications: The findings suggest that common life transitions in later life, including marital dissolution, relocation and becoming disabled, increase the risk of social isolation among older adults. Marital dissolution has the largest effects, which suggests that it is critical to help older adults connect with others and engage in social activities after they lose their spouse/partner. Relocation helps older men but not older women lower the risk of severe isolation, although relocation is a risk factor for some isolation for both sexes. The risk of social isolation associated with disability transitions seems to be gender-patterned: older women are more vulnerable to social isolation when becoming disabled. Gender-specific interventions may be appropriate in helping older men and women to stay socially connected in the face of relocation and health transitions.