Aging
Heterogeneous Treatment Effects of Contact with Friends on Dementia in a Nationally Representative Dataset Nicola Churchill* Nicola Churchill Roch Nianogo
Background: There are at least 6.9 million Americans living with Alzheimer’s disease and this number is expected to rise to about 13.8 million by 2060. Interacting with friends in mid-life can lower the likelihood of developing dementia later in life and improve cognitive performance. We intend to estimate heterogenous treatment effect from friendship social support defined as contact with friends on the risk of dementia using causal forest analysis in the Health and Retirement Study.
Methods: We estimated heterogenous treatment effects for each individual conditional on covariates. To do so, we used a causal forest approach, a type of random forest modeling that is generalized to produce predicted unit-level conditional average treatment effects. The subgroup average treatment effect for the overall population and stratified by race/ethnicity was obtained using inverse and additional weighting for doubly robust estimates.
Results: There was a 3.2% (95% CI: -0.053, -0.011) decrease in dementia cases among those who have contact with friends equal to or more than every few months versus those who had less. After stratifying by race/ethnicity, Whites had a 3.1% decrease (95% CI: -0.053, -0.009), Hispanics had an 8.4% decrease (95% CI: -0.157, -0.012), and Blacks reported no effect (7.9%, 95% CI: -0.016, 0.173).
Conclusion: Having contact with friends every few months or more frequently can have a protective effect on dementia risk. However, when stratified by race it was found the people who identified as Black no longer had a significant effect but both White and Hispanic individuals had maintained fewer dementia cases due to higher contact with friends.