Data Integration Methods for Environmental Exposures with Applications to Air Pollution and Asthma Morbidity

ABSTRACT Dementia is a major public health issue with substantial societal and economic burdens.

Alzheimer's disease (AD) contributes to about two-thirds of dementia cases and is the sixth leading cause of death in the US.

A small but growing number of studies have reported increased risks of hospitalizations among AD and Alzheimer's- related dementias (ADRD) patients following days with high ambient air pollution and high ambient temperature.

AD/ADRD patients may be more vulnerable to acute environmental insults because of multimorbidity and polypharmacy, poorer management of chronic diseases, and impaired behavioral responses to reduce exposure.

Given the projected increase in AD/ADRD prevalence, there is a pressing need to support patients and their caregivers by identifying modifiable risk factors, such as environmental exposures, that may reduce morbidity and improve care.

The overarching aim of this 1-year pilot project is to examine associations between short- term environmental exposures and emergency department (ED) visits for AD/ADRD.

We will leverage a unique patient-level database compiled under the parent project (R01ES027892), originally for examining air pollution exposure and asthma morbidity. The dataset includes over 330 million ED visit records from 10 US states, covering 33% of the US population.

We will expand the parent's project's analyses by conducting multi- city time-series analyses to estimate associations between daily AD/ADRD ED visits and short-term (up to 7 days) environmental exposures, including fine particulate matter pollution, nitrogen dioxide, ozone, and temperature (daily minimum and maximum).

We will first identify AD/ADRD ED visits using diagnosis codes, and then using patient residential ZIP code, we will assign exposure estimates with state-of-the-art air quality and meteorology products derived from the integration of monitoring measurements, satellite imagery, and numerical model simulations.

Stratified analyses will be conducted by age group, by whether the AD/ADRD ED visit led to hospitalization, and by geographical region.

In secondary analyses, we will also conduct analyses with ED visits of broad disease outcome groups (cardiovascular, respiratory, kidney-related, diabetes) to examine whether environmental exposure-related risks are higher among patients with an AD/ADRD diagnosis.

Because our ED visit database includes all ages, we will also capture patients with early-onset AD or mild dementia without prior diagnosis by clinicians.

To our knowledge, the proposed multi-city ED visit analyses will be the largest study to examine short-term associations between air pollution, temperature, and AD/ADRD ED visits in the US.

Past findings on hospitalizations, which represent severe cases, do not fully quantify impacts of AD/ADRD morbidity burden on healthcare systems.

Finally, results from our 1-year project can provide support for future studies to consider additional environmental exposures (e.g. other air pollutants, the built environment, and extreme weather events).

Additional epidemiology studies may also examine effect modification by individual-level and community-level factors that may confer enhanced vulnerability among AD/ADRD patients.