Alzheimer's disease (AD) is the sixth leading cause of death in the United States, where currently 5.8 million
people are living with AD dementias, and this number is projected to almost triple to 13.8 million by 2050. In
addition AD and related dementias (AD/ADRD) healthcare costs for people =65 years were an estimated $277
billion in 2018. Although putative risk and protective factors have been identified, published studies have been
unable to identify how to prevent and mitigate disease progression and who is most vulnerable mostly because
the existing studies are based on small sample size and lack statistical power to disentangle the effects of
different factors. Exposure to fine particulate matter (PM2.5) (a ubiquitous yet modifiable exposure dangerous to
the aging brain) has been associated with decreased cognitive function, faster cognitive decline, dementia, AD,
and mild cognitive impairment (MCI). Toxicological and human studies provide evidence of an association
between air pollution and neurodegeneration, highlighting potential biological pathways that include systemic
inflammation and oxidative stress. In response to PAR-17-054, our goal is to leverage massive nationwide
datasets (Medicare and Medicaid), coupled with advanced statistical methods, to overcome the limitations of
existing studies and clarify risk and protective factors for AD/ADRD. Specifically, in Aim 1 we will conduct
national epidemiological studies using Medicare and Medicaid claims for the period 2000-2021 for the continental
US, to estimate the association between exposure to air pollution and the time to the first AD/ADRD
hospitalization. Among enrollees that have been hospitalized for AD/ADRD or MCI we will assess whether air
pollution exposure increases risk of mortality, and/or accelerates re-hospitalization for AD/ADRD. In Aim 2 we
will apply machine learning methods to identify co-occurrence of individual-level (previous hospitalizations, race,
age, and sex), environmental (weather, green space, and noise), and SES risk (or protective) factors to
determine which population subgroups are most/least at risk for AD/ADRD hospitalization and progression
following air pollution exposure. In Aim 3 we will develop methods to overcome statistical challenges including
(1) disentangle the effects of air pollution exposure from other confounding factors by leveraging approaches for
causal inference, and (2) correct for potential outcome misclassification. We will conduct side-by-side
epidemiological analyses using traditional methods (e.g. regression) and causal inference and machine learning
approaches to understand which statistical challenges require more sophisticated approaches. To ensure
transparency and reproducibility, we will provide peer-reviewed open-source software so other investigators may
implement our methods. In summary the results of this proposal will characterize the link between air pollution
exposure and AD/ADRD hospitalization and progression, will identify the multiple modifiable risk and protective
factors that determine vulnerability in AD/ADRD, and provide the foundation for implementable actions to prevent
and reduce this enormous health burden.