Causal Inference
Associations of genetic risk for Alzheimer’s disease with late-life social connectedness in Health Retirement Study: Evaluating reverse causation Azuna Sawada* Azuna Sawada Sawada Sawada Sawada Sawada Boston University
While social connectedness is often associated with Alzheimer’s disease (AD), such associations may reflect reverse causation, where preclinical AD symptoms change social engagement. Associations between AD genetic risk scores (AD-GRS) and social connectedness would likely reflect such reverse causation because AD-GRS predict AD but should not cause social disconnection. Thus, we examined at what age AD-GRS predicts social connectedness outcomes to identify when reverse causation emerges. We analyzed Health and Retirement Study participants aged ≥50 with genetic data and at least one wave of data on social connectedness between 2006 and 2022 (n=15,574 with 41,826 person-wave observations). AD-GRS was categorized into quintiles. Social connectedness outcomes included continuous scores for social participation (range: 18-126; mean±SD: 80.39±0.09), loneliness (8-56; 19.86±0.04), perceived neighborhood social cohesion (3-28; 8.38±0.02), and positive (8-43; 19.1±0.03) and negative (2-48; 22.4±0.07) social support. We fit mixed-effects linear regression models stratified by age group (50s, 60s, 70s, ≥80 years), adjusting for sex, race/ethnicity, and 10 genetic ancestry principal components. Those at the highest genetic AD risk (Q5) showed reduced social participation across all age groups compared to the lowest genetic risk group (Q1): 50s (β=−1.75, 95% CI: −2.64, −0.86); 60s (β=−2.33, 95% CI: −3.02, −1.63); 70s (β=−2.06, 95% CI: −2.76, −1.37); ≥80 (β=−2.68, 95% CI: −3.52, −1.84). Q5 was also associated with higher loneliness among those in their 60s (β=0.66, 95% CI: 0.25, 1.07) and with lower perceived social cohesion among those in their 60s (β=-0.51, 95% CI: -0.75, -0.27) and 70s (β=-0.26, 95% CI: -0.51, -0.01). We found no evidence of association for social support. Reverse causation may affect observational studies of social connectedness and AD, with concerns for social participation emerging as early as age 50-59.
