Cognitive impairment (CI) and dementia can profoundly impact older adults and their families.  Timely detection of CI is essential for identifying treatable underlying causes and providing appropriate care management, including referral for neuropsychological evaluation. While routine primary care visits present an ideal opportunity for CI detection, up to half of cases may be missed.  There is a need for validated cognitive screening tools practical for use in time and resource-constrained primary care clinics (PCC).  MyCog is a brief (~ 8 minute) cognitive screening system that can be self-administered on a tablet during primary care visits.  Comprised of two cognitive measures (Dimensional Change Card Sort and Picture Sequence Memory) adapted from the NIH Toolbox for Assessment of Neurological and Behavioral Function®, MyCog has demonstrated satisfactory sensitivity and specificity for CI detection. MyCog automatically sends results directly to the EHR for real-time review and provides appropriate clinical decision-making support. We present initial results on implementing MyCog at PCCs within an academic health system and discuss implications for using tablet-apps to screen for CI in primary care more broadly.

The implementation of MyCog in 6 primary care clinics was assessed as part of an IRB-approved pragmatic trial evaluating cognitive screening for adults aged 65+ during Annual Wellness Visits (AWVs).   Using EHR data, we measured compliance through calculating the proportion of patients who completed MyCog out of all MyCog-eligible patients.

To better understand MyCog use and barriers to compliance, four in-person, semi-structured group interviews were conducted with clinic staff who use MyCog.  Interview notes were reviewed, grouped into themes, and used to formulate recommendations for increasing MyCog compliance.

Of the 1066 patients eligible, 641 completed MyCog (60.1%).  Group interviews revealed several clinician and patient barriers potentially addressable with design and workflow changes.  Barriers include:  Clinicians have no knowledge of patient’s progress in completing MyCog; patients don’t understand they need to complete two measures; patients are easily frustrated by the memory test and often show poor effort; patients and staff fail to tap a final finish button which exports data to the EHR; clinics cannot use MyCog with Spanish-speaking patients; clinics would like to use MyCog outside of an AWV.

MyCog has been successfully implemented at 6 clinics and administered to 60.1% of eligible patients during AWVs. Overall, clinicians and staff indicated positive experiences with MyCog, and expressed the desire to use it with Spanish-speaking patients and in other visit types. Other feedback reflected the need for efficiency in the workflow (e.g., providers don’t want to waste visit time waiting for scores to upload, etc.) and some patient frustration with the more difficult tasks. We will discuss the implications of these insights for tablet-based cognitive screening practices and demonstrate high-fidelity prototypes of improvements to the MyCog app design.