In 2022 the Diabetes Prevention Program Outcomes Study (DPPOS) incorporated Alzheimer disease (AD) and AD-related dementia (ADRD) phenotyping using the National Alzheimer’s Coordinating Center Uniform Data Set version 3 (UDSv3). UDSv3 includes a standardized neurological examination that was administered across 25 clinical sites and centrally interpreted by clinicians. To harmonize with UDSv3 and enable remote, reproducible assessment, the study team developed a video-based asynchronous neurological examination (DPPOS-VANE) for the multi-site observational cohort.
DPPOS-VANE was created through iterative consensus among research clinicians and staff feedback. The intended purposes were to: identify common neurological diagnoses aside from dementia (for example, diabetic cranial neuropathies, stroke, and parkinsonism), be reproducible across sites, and be performed without direct participant contact by the examining clinician. Design goals also emphasized independence from the variable clinical examination skills of participant coordinators (PCs).
The process included the development of a scripted exam that site staff could follow, site-specific performance review, feedback cycles, and staff certification prior to central physician review of recordings. The central reading was performed by physicians who reviewed the recorded video examinations.
Recordings were made using an iPad camera. The video examination protocol comprised structured assessments of:
The recorded format allowed physicians to read the examination asynchronously, without being present during the patient encounter.
A 10-minute training video was produced to demonstrate the examination step-by-step. The training included scripts and instructions in both English and Spanish. Following the training video, sites underwent performance review and received feedback. Staff certification at each site was required before central reading began.
Site procedures emphasized adherence to the standard script so that examinations would be reproducible and interpretable centrally. The protocol was designed to be brief and to fit within routine study visit workflows.
Over two years of implementation, 1,286 DPPOS-VANE recordings were generated, resulting in 1,284 examination reviews. Of these reviewed examinations:
Overall, 1,237 examinations (96%) were delivered as planned or were still usable despite minor deviations.
Each examination was completed within approximately 10–15 minutes. Across the 25 participating sites, the mean number of completed examinations per site was 51.4, with a site range of 14 to 92 completed exams. The implementation engaged 55 research staff members across those 25 sites and involved three physician reviewers for central reading.
This multisite implementation demonstrates feasibility of a standardized video-based asynchronous neurological examination in a large observational AD/ADRD phenotyping effort. Key operational strengths reported include high adherence to the scripted exam (93% exact script use), rapid exam completion time (10–15 minutes), broad multi-site engagement (25 sites, 55 staff), and a centralized physician review workflow.
The authors present DPPOS-VANE as an efficient neurological examination model enabled by commonly used devices and as a novel paradigm for large epidemiological studies. The source does not provide quantitative measures of diagnostic accuracy, inter-rater reliability statistics, or downstream clinical diagnostic outcomes from the recorded exams; these details were not reported in the preprint.
The study protocol was reviewed and approved by the Advarra single IRB (protocol Pro00047977). The authors confirm that appropriate participant consent and oversight were obtained. Funding sources reported include multiple U01 grants from the National Institute of Diabetes and Digestive and Kidney Diseases and support from the National Institute on Aging. Conflicts of interest disclosed in the source include consulting and editorial stipends for two authors; the remaining authors reported no conflicts.