1033. Do Automated Drug Alerts Improve Providers' Knowledge of Drug Interactions
Peter A Glassman, MBBS, MSc, VA Greater Los Angeles Healthcare System and UCLA, B Simon, VA Greater Los Angeles HSR&D Center of Excellence, P Belperio, VA Greater Los Angeles HSR&D Center of Excellence, A Lanto, VA Greater Los Angeles HSR&D Center of Excellence

Objectives: To assess whether exposure to automated drug alerts, such as in order entry, improve clinicians’ knowledge of drug interactions over time.

Methods: Two cross-sectional mail surveys, conducted in 2000 and 2002 with clinicians at a large multi-site healthcare system. Survey included items to assess recognition of drug interactions. Response rates were 64% (168/263) and 55% (145/265), respectively.

Results: In the 2002 survey, clinicians saw an average 7.6 drug alerts per week, with 17 % leading to a change in management; 87% stated that alerts helped them learn about drug interactions. Overall, clinicians correctly identified 53% of interacting drug pairs in 2000 and 54% in 2002; for interacting drug-disease pairs 62% and 60%, respectively were correctly identified. Severe, potentially life-threatening drug-drug interactions were identified correctly 54% of the time in 2000 and 58% in 2002. Severe, potentially life-threatening drug-disease interactions were identified correctly 53% of the time in 2000 and 55% in 2002. Signal to noise ratio remained a problem in both time periods, with 55% reporting this as an important barrier in 2000 and 49% in 2002.

Conclusions: Exposure to drug interaction alerts did not appear to substantially increase knowledge of selected drug interactions over time. Signal to noise ratio (lack of specificity) appears to be an important barrier, at least in a system where only 17% of alerts lead to a therapeutic change.

Impact: Automated drug alerts may have nominal impact on learning and hence probably influence prescribing by directed real-time warnings. Research into what improves specificity and clinical relevance is imperative.