Category: Professional Posters
Purpose: Healthcare systems can reduce adverse drug events (ADE) and simultaneously improve patient safety by incorporating well-designed drug-drug interaction (DDI) warnings into clinical decision support tools (CDS). DDI alerts should be evaluated, optimized and customized in based on institution culture. An active surveillance program for DDI suppression by clinicians at Scripps Health is pending. Thus, we aimed to answer the following question: What is the etiology of the top ten DDI suppressions documented by prescribers and can they lead to patient harm?
Methods: Descriptive, multi-site, retrospective study at a five hospital system comprised of teaching and community hospitals. Evaluation of filtered DDI warnings in the Epic EHR reviewed from November 2018 to January 2019. Suppressed severe or contraindicated warnings were evaluated. First Databank AlertSpace® was used to obtain raw data on DDI suppressions which records and aggregates all DDI suppressions. Descriptive statistics were used to measure frequencies in responses to DDI alerts and to perform correlations. Measures of central tendency (mean, median, mode) quantified filtered suppressions.
Results: Total of 49,838 DDI were evaluated. Approximately 92% of alerts were overridden each month. Ondansetron interaction with QT prolonging agents was most common (44%). Prescribers reported “benefit of therapy outweighs risk” in 65% of the overridden DDI alerts. Of these, 3,152 (6%) of DDI suppressions came from a contraindicated combination of ketorolac plus nonsteroidal anti-inflammatory drugs. The latter were largely due to PRN prescribing on order sets. This represents over 10,000 contraindicated alerts overridden annually.
Conclusion: Multiple opportunities exist to optimize CDS tools and minimize alert fatigue. Current suppression of stock DDI alerts appear low. Of the 50% of contraindicated alerts that have potential for patient harm, re-evaluation of order set compliance is key to minimize alert fatigue. Clinicians must consider the complete clinical presentation of each patient given the DDI software’s inability to account for patient specific factors.