Intravenous Proton Pump Inhibitor Orders Can Be Decreased with an Electronic Alert

Abstract

Purpose: Proton pump inhibitors (PPIs) are highly effective via oral or intravenous (IV) routes. Intravenous PPIs are indicated for acute gastrointestinal hemorrhage and inability to take oral medications, but are widely overused. Electronic order alerts can alter prescribing behavior, yet their effectiveness in decreasing IV PPI orders is unknown. Therefore, we conducted a retrospective study of an electronic alert for IV PPI orders. Methods: On October 21, 2011, our institution introduced an electronic alert that provides information on IV versus oral orders for PPIs. The alert is a pop-up box informing the provider that oral PPIs cost one tenth as much as IV PPIs yet are equally bioavailable; it appears when a provider initiates an order for an IV PPI. We retrieved all PPI orders for adult inpatients from one year before the alert (pre-alert) until one year after the alert (post-alert), excluding PPIs via continuous infusion which do not trigger the alert. We parsed electronic medical records for variables related to the PPI order, the prescriber, and the patient including diet. Our primary outcome was a difference in the proportion of all completed PPI orders that were intravenous pre-alert compared to post-alert. Multivariable logistic regression modeling was used to test predictors for overriding the alert. Results: During two years centered on October 21, 2011 there were 65,893 completed orders for PPIs including 17,297 (26.3%) completed IV PPI orders. During the year post-alert, 815 (10.1%) IV PPI order attempts were not completed after triggering of the alert. Among all PPI orders, the proportion of IV PPI orders significantly decreased from pre-alert to post-alert (30.0 vs 25.6% respectively, chi-squared p<0.01). Post-alert, significant predictors for overriding the alert were non-physician prescriber (physician assistants and nurse practitioners, OR 1.47, 95% CI 1.27-1.88) and whether the order for the IV PPI was part of an order set (OR 1.48, 95% CI 1.22-1.79). Conclusion: An electronic alert for IV PPI orders effectively decreased the proportion of PPIs ordered intravenously, although 90% of alerts were overridden. Significant predictors for overriding the alert were non-physician prescribers and whether the IV PPI order was part of an order set.Figure