Progress on research to prevent or treat Alzheimer's disease and related causes of dementia (ADRD) has been disappointingly slow. The collection of diseases poses special difficulties for population research, including epidemiology. The predominant causal inference methods in epidemiology are based on fulfilling the 'back-door criterion', i.e., accounting for shared causes of exposure and outcome, are ill-matched for many problems in ADRD. Novel insights or more convincing findings may be derived when using methods based on instrumental variables (IV), i.e., identifying sources of variation in exposures of interest that are unrelated to the potential outcomes of ADRD. One major challenge in applying IV methods has been the need for large sample sizes, but this barrier is ameliorated as large data sets become commonplace. Another major challenge in applying IV methods is identifying plausible IVs, but because epidemiologists do not routinely use IV methods, they may not recognize potential IVs. I will discuss examples of IV-inspired approaches in ADRD research, with the goal of stimulating discussion of other potential IVs. I will also touch on the recent controversy related to aducanumab, the FDA-approved medication for treatment of AD.
Maria Glymour is a Professor in the Department of Epidemiology and Biostatistics at the University of California, San Francisco. She is a social epidemiologist and spends most of her time -- when not attending urgent zoom meetings -- puzzling about how we can use the inadequate samples, muddled study designs, and incomplete measures available to us to learn more about reducing social inequities in health
Zoom Seminar. Please contact laura.satterfield@duke.edu to obtain Seminar Link.