Developing an autoverification framework for medication orders at UNC Health.

Area of Science:

• Health Informatics
• Clinical Pharmacy
• Medication Safety

Background:

• Autoverification (AV) automates medication verification in electronic health records, bypassing pharmacist review.
• Successful implementation requires addressing safety and efficacy concerns for high-volume, low-risk medication orders.
• A replicable framework is needed to identify medications suitable for AV within hospital systems.

Purpose of the Study:

• To identify parameters for risk stratification of medications for AV.
• To develop a replicable framework model for identifying medications appropriate for AV at UNC Health.

Main Methods:

• Modified Delphi methodology was used to achieve consensus on risk stratification parameters.
• A risk stratification tool was applied retroactively to medication orders from October 2023.
• The study assessed the risk of adverse events for potentially autoverified orders.

Main Results:

• Fifty-five criteria reached consensus for the AV risk appraisal tool (AVRAT).
• Key criteria for flagging high-risk AV orders included age, kidney function, hemoglobin, platelets, body weight, and RRT.
• A proof-of-concept evaluation using AVRAT indicated 6.89% of orders posed a low risk for AV.

Conclusions:

• A proof-of-concept study successfully developed a framework for AV utilization.
• AV has the potential to reduce medication order review time in hospital systems.
• A relatively small proportion of medication orders may be eligible for AV.