Case Overview:

This case study reviews the adoption of an AI-assisted tool, DAX Copilot, within the pharmacy team of Legacy Health. While this tool was already in use by physicians within the medical group, clinical pharmacists were restricted from access until the initiation of this project.

Tool and Project Details:

Clinical pharmacists in ambulatory care play a crucial role in patient care, often managing complex medication regimens that require focused, personalized education and counseling. However, documenting in the electronic health record while interviewing patients can lead to distractions, fragmented interactions, and potential errors. To solve this problem, the Legacy Health team employed an AI tool that automates the note-writing process, allowing pharmacists to devote more attention to direct patient care during visits. By automating documentation tasks, the team has enhanced efficiency in closing chart notes, reduces workload and burnout associated with high patient volumes, and ultimately minimizes the time clinical pharmacists spend on charting.

The tool of this project, DAX Copilot, is an AI-assisted, automated clinical documentation tool embedded in the Epic Electronic Health Record and accessible through Epic Haiku and Canto. It listens to patient visits, recording subjective and objective information and summarizing it into customizable note formats. During the visits, clinicians use a wireless microphone app for audio input.

Key Elements of Success:

This PGY2 ambulatory care resident-led pilot study began with approval for a single DAX Copilot license for pharmacist use. The license was granted to a clinical pharmacist embedded in a primary care clinic, following advocacy from the ambulatory care clinical manager to an interprofessional executive leadership team. As project lead, the resident granted access to Epic Signal to monitor and track data. The resident then conducted data analysis with support from internal and external physician/pharmacist informaticists, and led biweekly meetings with a research team, composed of ambulatory care pharmacists, to troubleshoot issues, optimize research protocol, and plan next steps.

Impact on Outcomes:

After implementing the tool, the average time spent on notes decreased by 40%, from 65 to 39 minutes daily over three months. The study pharmacist also increased appointments by 25% and decreased time spent charting outside work hours by 20 minutes, or ~38.5%, compared to the 12-month average prior to DAX Copilot use. Patient satisfaction surveys showed overwhelmingly positive results, with all patients reporting satisfaction with their visits, feeling that the pharmacist engaged with them more than with their computer.

Role of the Pharmacy and Pharmacists:

The research team for this project is composed of ambulatory care pharmacists with assistance from physician informaticists and Epic trainers. The core research team is responsible for monitoring outcomes, refining study protocols, and using data to inform further expansion of tool access to the pharmacy team. The key stakeholders are clinical pharmacists within the medical group, patients receiving care from these pharmacists, and the medical group’s administration. The study pharmacist was allowed to integrate DAX Copilot into their workflow at their discretion to better align with real-world practice. Prior to gaining access to DAX Copilot, completion of a training module was required.

Budget & Resource Allocation:

This quality improvement pilot study was initiated following the creation of a project pitch outlining the potential financial and patient-centered benefits of expanding DAX Copilot access to clinical pharmacists. The cost of a license was justified by emphasizing the potential for increased appointments per day through improved visit efficiency and reduced note burden. The project pitch was presented to the executive leadership team and subsequently approved. 

Lessons Learned:

The primary challenges were justifying the license cost, selecting a suitable study pharmacist, and integrating new technology into existing workflows. License costs were justified through meticulous data tracking using Epic Signal, highlighting metrics like appointments per day that impact revenue. Metrics associated with burnout were monitored as reducing burnout and turnover may avoid costs associated with hiring and onboarding. Epic Signal data, along with individual pharmacist revenues and patient loads, were used to select a study pharmacist with the most potential for improvement.  Key advocates included the ambulatory care pharmacy clinical manager and the clinical vice president for medical informatics.

Future Goals & Ongoing Monitoring:

As a result of this study, five additional tool licenses were granted. This followed the completion of a project pitch and return on investment (ROI) analysis. Moving forward, our research team will work closely with our medical group’s informatics team. With their assistance, we can track data through monthly reports generated from Epic Signal. This method requires little manual input and is a significant departure from the time-intensive way data was tracked during Phase 1 of this project. Additionally, we gained access to the Nuance Command Center, allowing us to track daily encounters for study pharmacists. I hope to present these findings at an upcoming regional conference.