Background: Hospitalized patients receiving direct oral anticoagulants (DOACs) sometimes require bridging with unfractionated heparin (UFH). Monitoring UFH with anti-Xa assays has been shown to correlate with better outcomes. However, DOACs interfere with anti-Xa assays resulting in inappropriate UFH dose adjustments that can negatively impact patient care. In 2015, we deployed an electronic health record (EHR) clinical decision support tool and a two-screen order set across multiple hospitals to assist clinicians with UFH infusion ordering in patients who took DOACs prior to admission. Despite these interventions, there have been several instances of inappropriate UFH infusion ordering and administration, resulting in serious events that compromised safe patient care.

Purpose: The goal of the initiative was to redesign, rebuild and implement a more user-friendly DOAC interference order set to avoid adverse events related to UFH infusions.

Description: We performed a root cause analysis of 70 patient safety events linked to DOAC interference with UFH infusion at eleven UPMC hospitals. The incidents were reported between June 2015 and January 2018. In 84% of events, the clinicians overlooked the clinical decision support alert and ordered the wrong heparin infusion order set which was corrected by the hospital pharmacists in 34% of patients. Nursing staff frequently sought the UFH infusion nomogram although the drug is typically infused at a fixed rate for the first 72 hours. In most of the events, the error was due to a problem with the ordering interface design: the existing two-screen order set was too difficult and confusing to use. As a result, clinicians failed to place anti-Xa monitoring orders to be initiated after the first 72 hours of UFH infusion administration. Next, we convened a multidisciplinary team to develop and implement a simplified one screen order set, requiring fewer mouse clicks. The clinical decision support alert was also overhauled. For nursing, we created an informational nomogram-like form as a reminder to avoid anti-Xa monitoring for the first 72 hours. We also instituted educational sessions for clinicians, nurses and pharmacists focused on order set changes. In the first 3 months after go-live, there was only one patient safety event report related to incorrectly ordered DOAC interference order set.

Conclusions: Commonly prescribed DOACs interfere with anti-Xa assays and can lead to improper dose adjustments and potentially negative outcomes for patients on UFH infusion. EHR user interfaces and alerts require continuous evaluation for workflow fit and clinician acceptance. Single screen order set design is essential to improve usability, patient safety and reduce medication errors.