In response to PA-17-089, we propose exploratory analyses with an existing epidemiologic heart failure (HF) cohort linked to electronic health record (EHR) data to address the combined burden of coexisting Alzheimer?s disease and related dementias (AD/ADRD) and HF. We will capitalize on the Rochester Epidemiology Project (REP) and apply novel data science methods to existing community datasets. Recent adverse trends in cardiovascular disease (CVD) mortality are due to a major increase in HF deaths, and the uncontrolled epidemic of HF compromises progress against CVD. Addressing the epidemic of HF, a disease of aging populations, requires understanding how coexisting conditions interact with HF to impact outcomes and health care utilization. AD/ADRD occupies a distinct position within coexisting conditions as HF increases the risk of AD/ADRD, in part, via shared pathways. However, this association and, in particular, which phenotype of the heterogeneous HF syndrome is associated with AD/ADRD is not well known. AD/ADRD can adversely impact the outcomes of HF, a serious chronic disease that requires effective self-management and cognitive skills. Yet, how AD/ADRD impacts health care utilization and outcomes in HF is not well understood. These knowledge gaps can be addressed by secondary analyses of existing data. Our objectives are to study, in an epidemiological cohort, the characteristics of HF that impart a higher risk of AD/ADRD, and the impact of AD/ADRD on outcomes (death, cardiovascular events, hospitalizations and admission to skilled nursing facility). Our proposal directly addresses key points of the PA, such as ?Effects of specific combinations of two or more comorbid conditions on risks for specific adverse health outcomes, impact of specific combinations of two or more chronic conditions, ?interactions among disease processes, and health outcomes in complex older patients with multiple chronic conditions.? This work will capitalize on a large epidemiological HF cohort linked to comprehensive EHR data within the rich environment of the REP, optimally suited to our proposed secondary analyses. This proposal leverages the experience of our team with the epidemiology of HF and with novel data science methods readily applicable to secondary analyses of existing EHR data sets.
Aim 1 will apply machine learning to discover HF phenotypes associated with AD/ADRD while considering cardiac function, occurrence of atrial fibrillation and other coexisting conditions.
Aim 2 will evaluate the impact of AD/ADRD on outcomes of HF. Conducting this work in a cohort within a geographically defined population and extensive longitudinal follow up will ensure that our results reflect the experience of community patients with HF. An improved understanding of the coexistence of AD/ADRD and HF will establish the prevalence of this association in the community and characterize the type of HF associated with AD/ADRD, which may allow earlier diagnosis, thereby enabling interventions to prevent or forestall early disease. Finally, improving our knowledge of how AD/ADRD affects outcomes will foster more precise management.
We propose exploratory analyses with an existing epidemiologic heart failure (HF) cohort linked to electronic health record data to address the combined burden of coexisting dementia and HF. We will capitalize on the Rochester Epidemiology Project and apply novel data science methods to existing community datasets.