Abstracts

Rationale:
A learning healthcare system (LHS) improves the quality of care provided to patients by integrating existing evidence and new knowledge generated by the healthcare system. Our institution was an early member of the Epilepsy Learning Healthcare System (ELHS) and created a model for site participation.

Methods:
We created a multidisciplinary site team including clinicians (physicians, nurses, nurse practitioners), a quality improvement (QI) expert, a patient family partner, an electronic health record (EHR) analyst, and data experts. Our primary tool for data collection is a Common Data Elements (CDE)-based form in our EHR into which clinicians enter standardized data elements of epilepsy history during or after visits. These data have been used to contribute site level information to the ELHS organization and to inform institutional QI initiatives, including standardized provider documentation of seizure frequency/type and screening patients for medication barriers. To facilitate these initiatives, we have created tools including: (1) A dashboard to track provider documentation of seizure frequency with the CDE-based form, and (2) An electronic patient questionnaire about medication barriers administered at follow up visits. For patients who endorse specific medication barriers, patient educational materials addressing the specific barriers are automatically included in the after visit summary. Data from these efforts are reviewed by the site team to better understand our patient population and inform QI efforts at monthly meetings of our QI Program.

Results:
Our site has contributed data from 4,451 unique individuals and 18,905 clinic visits to ELHS using our CDE-based EHR form. The creation of our seizure documentation dashboard allows us to understand provider uptake of our standardized system for seizure documentation. Since September 2019, we have improved our rate of seizure frequency documentation using the CDE-based EHR form for office visits of epilepsy patients from 19% to 32% through clinician education efforts and EHR modifications (Figure 1). Medication barrier patient questionnaires have been completed at a median of 138 visits per month (IQR 100-167) since implementation in July 2021. Medication barriers are endorsed at fifty percent of office visits. The most commonly cited barriers include side effects (26%), incomplete seizure control (17%), forgetting medication (11%), and taste of medication (10%) (Figure 2).

Conclusions:
We have created a sustainable model for site participation in an epilepsy-specific LHS. Key components include the formation of a multidisciplinary team, EHR and data expert support, use of a QI framework, and EHR tools for efficient patient and provider data entry as well as QI interventions. Using a LHS approach, we better understand local provider practice habits surrounding seizure documentation as well as barriers faced by our patient population. Such knowledge will inform improvement efforts both locally and nationally.

Funding: None