Accounting for pre-baseline selective survival in cross-national studies of the exposome in Alzheimer’s disease and related dementias: a novel bias assessment tool

PROJECT SUMMARY Features of the exposome are likely key to the etiology of Alzheimer’s disease and related dementias (ADRD), and cross-national exposome studies of ADRD have huge promise in this area. Cross-nationally harmonized cohort studies of aging such as the US Health and Retirement Study (HRS) International Partner Studies and

their embedded Harmonized Cognitive Assessment Protocols (HCAPs) are prime potential data resources for exposome-ADRD research. These cohorts include population-based study samples of the appropriate age range for ADRD incidence, and have a wealth of interview, physical, cognitive, and biomarker assessments

alongside administrative and geographic data linkages. The international coverage and harmonized design of the HCAP studies allow for novel triangulation of exposome data across diverse country contexts. The HCAP data have great potential to strengthen causal inference and broaden the global evidence base on features of

the exposome and diversity in the mechanisms through which it may impact ADRD. However, exposome research in the HCAP network is limited due to a common bias in cohort studies of aging that is unresolved by research efforts to date: selective survival in the target population that occurs prior to the study baseline. This

bias is difficult to quantify, as most studies lack information on population members who would have been eligible for the study, but who died prior to enrollment. In the United States, mortality rates of 20-30% are not uncommon in studies of aging, and, crucially for efforts to triangulate associational data across populations,

age-specific mortality varies widely cross-nationally. Many features of the exposome are associated with survival, making this form of selection bias a salient yet under-studied threat to the validity of exposome-ADRD research. Indeed, our preliminary data indicate considerable selection bias that is differential across HCAP

studies. Our current objective is to address this bias by innovatively quantifying plausible magnitudes of pre- baseline selection bias in exposome-ADRD research using the HCAP network data, and to develop a user- friendly web-based app that will allow other researchers to understand the impacts of survival bias in their own

research on the exposome in relation to ADRD. We will first conduct country-specific simulation studies using life table data on age-specific mortality and cognitive data from our harmonized HCAP scores to generate plausible estimates for the magnitudes of pre-baseline selection bias that could affect results of exposome-

ADRD analyses in the HCAP network. We will conduct simulations for three key exposome factors thought to be salient to ADRD, with plausibly differential associations with mortality across HCAP countries: main lifetime occupational skill level, later-life exposure to air pollution, and later-life exposure to type II diabetes. We will

then build a user-friendly and comprehensive web-based selection bias analysis tool to be hosted on the Gateway to Global Aging website, which will be an important research resource to help elucidate the role of the exposome in ADRD outcomes and disparities across the HCAP network of studies.