External Validation of Instruments to Improve Detection of Undiagnosed DementiaDr. Festa is an assistant professor in medicine within the Section of Geriatrics at Yale School of Medicine. Her
research applies routinely collected healthcare data to improve the identification and care of people living with
dementia (PLWD). As a sub-investigator within the NIA-funded Geriatric Emergency Care Applied Research
(GEAR) Network, she led the development of novel electronic health record–based instruments designed to
address the substantial burden associated with the underrecognition of Alzheimer's disease and related
dementias in emergency department settings. Through this Career Development Award, Dr. Festa seeks to
acquire expertise in external validation, implementation science, and embedded pragmatic clinical trial design
to evaluate whether these instruments, once validated, improve outcomes for PLWD.
Over half of people living with dementia (PLWD) lack or are unaware of a formal diagnosis. The emergency
department (ED) is a critical yet underutilized setting for identification, as it disproportionately serves
individuals who are prone to underrecognition. There is a need for automated EHR-based instruments that can
identify probable dementia and differentiate diagnosed from undiagnosed disease in this setting. This project
will externally validate two complementary instruments, COMPASS-ED and REMIND, designed to differentiate
diagnosed from undiagnosed dementia, and grade its severity within a national Veterans Health Administration
(VHA) ED cohort. This Career Development Award will provide Dr. Festa with training in external validation of
electronic health record (EHR) based instruments, implementation science, and embedded pragmatic clinical
trial (ePCT) design necessary to become an independent physician-investigator capable of leading trials that
improve disease recognition and care for PLWD within the ED. This training will support the following Specific
Aims: (1) To externally validate COMPASS-ED, a severity-stratified computational phenotype for probable
Alzheimer's disease and related dementias (ADRD), within a national VHA ED cohort, and (2) To externally
validate REMIND, a severity-stratified predictive model for undiagnosed ADRD, within the same cohort. This
work will establish the evidentiary basis for the scalable deployment of COMPASS-ED and REMIND within
multicenter ePCTs across the VHA and the Geriatric Emergency Care Applied Research (GEAR) Network,
positioning these instruments to trigger severity-appropriate interventions at the point of care in the ED. Their
prospective deployment will, in turn, enable Dr. Festa to lead trials evaluating whether interventions triggered by
these instruments improve outcomes for PLWD.