Multi-morbidity 3-City Alzheimer's Disease EHR Study (M3AD Study)

The overall goal of our proposal is to place patients impacted by Alzheimer’s Disease (AD) and Related Dementia (AD/ADRD), rather than the disease, at the center of the research in order to better tailor risk prediction, enable individualized prevention, and improve clinical outcomes and management of these patients in their layered complexity. Our point of departure is the recognition that chronic diseases, including AD/ADRD, do not exist solely as isolated entities. Instead, we acknowledge that their shared organic substratum and common sets of risk and protective factors contribute to determining concomitant trajectories of multiple diseases, mutually influencing each other’s course and natural history, in ways yet unexplored. The use of large longitudinal datasets such as electronic health records (EHRs) thus becomes a key novel asset with potential to advance understanding and management of dynamic personalized risk in the layered complexity involving variations through risk factors, and concomitant diseases. Thus, we propose to federate a multiethnic 3-city EHR consortium (New York Presbyterian’s: 6 million patients (33,000 with AD/ADRD;23% Hispanic); University of Chicago: 2 million patients (11,000 with AD/ADRD; 60% Black); and University of Miami:1.4 million patients (13,000 with AD/ADRD; 50% Hispanic). Further, mindful that EHRs often overlook social variables (e.g., race/ethnicity, education, income), known to alter dementia risk, we will embed these EHRs in census-track level social determinants of health data to pursue the following specific aims: Aim 1 –Determine and quantify whether and how: trajectories of multimorbidity (timing, order of incidence, level of control/management, evolution) predict the incidence, timing, and progression of AD/ADRD, over years of longitudinal EHRs; and whether the relationships between those trajectories and AD/ADRD are modified by gender, race/ethnicity, age, education, place of birth, and other socioeconomic factors (1.a); and changes in specific multimorbidity risk factors (smoking, weight) and manageable aspects of care (blood pressure control, glycemic control) impact the incidence, timing, and progression of AD/ADRD (1.b). Aim 2 –Determine and quantify whether and how complexity patterns of multimorbidity, complexity of care management, and patient complexity predict (2.a) and impact (2.b) the incidence, timing, and progression of AD/ADRD over years of longitudinal EHRs, and whether those factors are affected by gender, race/ethnicity, age, education, place of birth, and other socioeconomic factors. Aim 3 – Determine the supplementary value of “enriched” EHR with research items already collected for participants concomitantly enrolled in research cohorts, to refining identified trajectories and complexity patterns. We have assembled an interdisciplinary complementary network of innovative researchers and will use machine learning and novel dynamic predictive and causal inference methods to identify accelerators or decelerators of AD/ADRD. This work will inform strategies to tailor risk prediction, and complex clinical management. It will also build a multiethnic harmonized dynamic platform ready for real-world evaluation of future treatments of AD/ADRD.