Improving causal inference in Alzheimer's Disease prevention research on modifiable risk factors: the Triangulation of Innovative Methods to EndAD (TIME-AD) project

Boston University Medical Campus, Boston MA

Investigators

Abstract

Research on Alzheimer’s Disease and Alzheimer’s Disease Related Dementias (AD/ADRD) has identified several promising risk factors which could guide strategies to prevent up to 40% of AD/ADRD. Nearly all prior evidence relies on observational data, which is prone to biases from unmeasured confounding, reverse causation, selective survival, and measurement error. The Triangulation of Innovative Methods to End AD (TIME-AD) program addresses these challenges by using an evidence triangulation framework for strengthening causal inference in observational data. This framework systematically evaluates biases, planning complementary analysis approaches with different data sources and study designs to rule out alternative interpretations for the association of each risk factor and AD/ADRD: a) doubly robust observational methods combining propensity score models with outcome models; b) instrumental variables (IV) methods using genetic and policy variations that introduce random variation in exposure; and c) quantitative bias analysis to characterize uncertainty. Project 1 will address the effects of alcohol use across the lifecourse on cognitive aging and AD/ADRD risk. Project 2 will evaluate the effects of depression and depression treatment on AD/ADRD risk and possible direct and modifying roles of chronic pain. Project 3 will assess whether AD/ADRD risk may be reduced by prevention or treatment of vision or hearing impairments. Project 4 will assess the impacts of social isolation, focusing on components of social isolation that are modifiable with existing interventions. The current proposals will extend our current knowledge by focusing on causation, including large, heterogenous samples, and rigorously evaluating heterogeneity across populations. In addition to an Administrative Core, projects will be supported by a Cognitive Outcomes, Exposure Variables, and Covariates Data Core, which will help the intensive data management involved in constructing analytic data sets and foster harmonized measures and coordinated analyses, drawing on multiple data sources to support evidence triangulation. A Genetic and Policy Data Core will bring specialized expertise on genetics, policies, and IV analysis, providing code to construct, validate, and implement IV analyses. An Analytics Core will develop and share reusable analytic code and support implementation of the most up-to- date methodology; and a Maximizing Population Impact Core will work with all components to maximize the relevance of our findings to improve outcomes across all older adults and for populations typically at elevated risk of AD/ADRD. The Maximizing Population Impact Core will ensure that research implementation is guided by potential applications of the evidence and that the findings of each study are broadly disseminated to constituencies who can act on the evidence. The synergies created by shared cores with expertise in data, analysis, and dissemination ensure that TIME-AD will have high impact.