Background:
Best practice advisories (BPAs) are integrated into the electronic health record (EHR) to implement evidence-based practices, but not all alerts are effective and can even contribute to alert fatigue and burnout. This project focuses on optimizing the admission medication reconciliation (AMR) alert, which is the most common alert UCSD Internal Medicine residents face. Medication reconciliation is a pillar of patient safety, especially during the admission process. A reconciliation completion rate of >90% within 24 hours has become a UCSD Health performance metric. This alert is triggered when a medication reconciliation is not finished within 24 hours of admission and appears every time orders are opened until the reconciliation is completed. This project aims to analyze this alert to understand its high fire rate, maximize its effectiveness, and improve patient safety and overall provider experience.
Methods:
This retrospective analysis evaluates the medication reconciliation BPA using Epic SlicerDicer. Tutorials on completing the medication reconciliation and processing the alert were created and distributed to all residents. Residents were also sent their individual medication reconciliation rates as feedback. A survey was distributed pre and post intervention to evaluate the efficacy of the tutorials. Medication reconciliation rates were analyzed 6 months pre and post tutorial for each senior resident and 1 month pre-intervention for interns. Statistical significance was determined using chi-squared and interrupted time series analyses.
Results:
Medication reconciliation rates were evaluated for 134 UCSD internal medicine residents. The >90% reconciliation rate was achieved in 57/134 residents pre intervention and significantly increased to 85/134 residents post intervention (p <0.001). The survey demonstrated a statistically significant (p <0.01) increase in resident understanding on how to complete a medication reconciliation and process the AMR alert from 50% to 90%. The interrupted time series analysis showed an increase in the proportion of medications reconciled within 24 hours at the time of intervention (+0.02) and a positive trend (slope change +0.006/month) post intervention.
Conclusions:
1. Reliable admission medication reconciliation is important for patient safety.
2. Evaluating BPAs to understand barriers to action is a key first step in optimizing them.
3. An educational intervention to improve action taken based on our AMR BPA has been effective.
Although the total number of AMR alert firing per month did not decrease post intervention (was not standardized for changes in hospital census), we did demonstrate a statistical improvement in >90% reconciliation rate and understanding of how to process the BPA. The change in proportion of medications reconciled within 24 hours improved but was not significant, which may be due to the small number of time points. Another limitation of this study includes survey response rate (n=20). For future work, we aim to further optimize this BPA by limiting the time of day during which it fires (covering night teams are unlikely to complete medication reconciliation), targeting the alert to first call providers, and streamlining the entire process. We hope to extend this approach to other high-frequency interruptive alerts, further improving efficiency, patient safety, and physician well-being.
