Social Isolation and Loneliness in Older Adults: A Longitudinal Analysis of Persistence and Bidirectional Influence

Background and Purpose: Social isolation and loneliness are recognized as significant public health concerns and a Social Work Grand Challenge. This is particularly true among older adults (≥50). Despite their frequent co-occurrence, social isolation and loneliness are conceptually distinct; social isolation refers to the objective state of having limited social contacts, while loneliness is the subjective experience of feeling alone, regardless of the quantity of social connections. Prior research has predominantly focused on these constructs individually, neglecting their interplay over time. No existing studies have examined the dynamic and reciprocal longitudinal relationship between social isolation and loneliness, thereby overlooking potential bidirectional influences.

Methods: This study uses four waves of data (2008, 2012, 2016, and 2020) from the Health and Retirement Study, a nationally representative panel survey of older adults in the United States. Social isolation was assessed through a six-item index of social engagement, while loneliness was measured using the 11-item UCLA Loneliness Scale. We used a random intercept cross-lagged panel model (RI-CLPM), which is ideal for disaggregating within-person processes from between-person variability and provides insight into how changes of one variable are associated with subsequent changes in another over time, while accounting for stable individual differences. We tested both constrained and unconstrained models to determine the best fit; for the constrained model, we imposed equality constraints on autoregressive and cross-lagged parameters across time, whereas the unconstrained model allowed these parameters to vary freely. We controlled for time invariant and time variant covariates including age, gender, depression, chronic health conditions, education, and race/ethnicity.

Results: At baseline, the average age of participants was 68.89 (SD = 9.73). Data was collected from an average of 2,881 participants per wave. Model comparison revealed that, prior to the inclusion of covariates, the constrained RI-CLPM provided a better fit than the unconstrained model. The constrained model had stable within-person autoregressive pathways for loneliness (β = 0.17, p < .001) and social isolation (β = 0.28, p < .001) across all timepoints. Those who reported higher levels of loneliness were more likely to report high levels of social isolation at a subsequent timepoint (β = 0.04, p <.01); high levels of social isolation was associated with subsequent elevated loneliness (β = 0.06, p<.01). These results were attenuated but remained statistically significant when adding covariates. While social isolation and loneliness were stable on a within-person basis, between-person differences in the random intercepts existed, suggesting significant between-person variability in baseline level of social isolation and loneliness.

Conclusions and Implications: Our longitudinal analysis, using four waves of the Health and Retirement Study and a constrained RI-CLPM, revealed within-person stability of both loneliness and social isolation over time. Additionally, the bidirectional associations observed between loneliness and social isolation highlights the complex, reciprocal relationship between these constructs. This study's findings suggest interventions addressing social isolation should simultaneously address loneliness (and vice versa), considering the interlinked and stable relationship between these two over time. Additionally, the stability of these constructs suggests that early identification and intervention are critical.