Project 4: Social isolation as a driver of AD/ADRD incidence and disparities

PROJECT SUMMARY Social isolation, defined as an objective state of limited to no social contact, impacts millions of older adults, with prevalence estimates varying from 10-25% across studies and measures. The pandemic exacerbated rates of social isolation, increasing concerns about the link between social isolation and risk of Alzheimer’s disease and

related dementias (ADRDs) and drawing attention to the need for interventions. Interventions target social isolation by enhancing social connections (e.g. matched peers, telephone warm lines, and community groups) or preventing the loss of social ties. However, major gaps separate observational research linking social isolation

to ADRDs from intervention design. Observational research has typically focused on broad, heterogeneous social isolation or engagement measures that do not clearly correspond to typical interventions. Further, observational research has not ruled out the potential influences of reverse causation, given the likely impact of

ADRDs on social disengagement. Moreover, questions about the heterogeneous impacts of social isolation on ADRDs across population groups (e.g. gender, race/ethnicity, socioeconomic status) and time (pre- and within- pandemic) have received little attention. As a result, we still have a limited understanding of how interventions

addressing social isolation might impact ADRDs-related inequities. Our primary objective is therefore to identify the effects of social isolation and hypothetical interventions which reduce social isolation on ADRD risk by carefully triangulating evidence across multiple data sources using a range of methods meant to address core

methodological challenges in this research. We will leverage a framework to emulate “target” intervention trials in observational cohorts, which can approximate the impact of hypothetical social isolation interventions on ADRD risk, while addressing bias in observational studies such as reverse causation and time-dependent

confounding. Our secondary objectives are to identify the extent to which social isolation contributes to well- characterized ADRDs inequities. We hypothesize that social isolation will be associated with ADRDs risk, but that estimates will vary from previously published results, and have a greater impact for structurally marginalized

groups, after addressing key sources of bias and/or directly emulating “target” intervention trials. To test this hypothesis, the following aims are proposed: Aim 1) Estimate the impact of a) social isolation and b) hypothetical social isolation interventions on ADRDs, triangulating across diverse data sources and study designs; Aim 2)

Evaluate population-level heterogeneity in the association between social isolation and ADRDs; and Aim 3) Quantify reductions in ADRDs achievable through interventions on social isolation for population subgroups. Substantial preliminary data is provided in this resubmission demonstrating the feasibility of our approach. This

proposal will have a significant impact in yielding the most rigorous estimates to-date of the potential impacts of real-world interventions to reduce social isolation on ADRDs risk and ADRDs inequities using multiple high- quality population-based data sources, which can guide ongoing intervention development.