Evaluation of a Trainee-Led Project to Reduce Inappropriate Proton Pump Inhibitor Infusion in Patients With Upper Gastrointestinal Bleeding: Skip the Drips

Continuous infusion of proton pump inhibitors (PPIs) is recommended in patients with upper gastrointestinal bleeding (UGIB) for specific situations, such as before endoscopic identification of ulcers with high-risk features^1-4 (Box). Unfortunately, PPI infusions may be continued for 72 hours without indication.^2,5 Reducing the overuse of these infusions is important because, in addition to increasing the length of stay and cost, PPI overuse is associated with various complications.⁶ In July 2015, a fellow- and resident-led intervention was initiated with the goals of decreasing the inappropriate use of PPI infusions in patients with UGIB and promoting evidence-based care at lower costs for these patients.

Box Section Ref ID

This quality improvement study used the Culture, Oversight, Systems Change, Training framework. In July 2015, the electronic health record was changed to prompt physicians to select an appropriate indication for PPI infusion orders (systems change). Pharmacists tracked monthly usage of PPI infusions (oversight). Education (training) was also provided in July 2015 for internal medicine residents, hospitalists, and emergency medicine faculty and residents. Trainees recruited faculty leaders from gastroenterology, general medicine, and emergency medicine departments to champion this initiative (culture). Institutional leaders at the University of Chicago sponsored this intervention via the Choosing Wisely challenge. This study was determined to be quality improvement, rather than human participant, research by the University of Chicago Office of Clinical Effectiveness and was therefore exempted from institutional review board approval.

From July 2015 to April 2016, PPI infusion usage, indication, and endoscopic findings were prospectively collected. Data from patients admitted for UGIB in the preceding year (July 2014 to June 2015) served as historical controls.

We used a statistical process control chart method to examine the differences in inappropriate PPI infusion use between the preintervention (July 2014 to June 2015) and postintervention (July 2015 to April 2016) periods. We also used the 2-sample test of proportions (2-sided P value, with P < .05 considered statistically significant) to assess the rates of inappropriate PPI infusion use in a concurrent control group of surgical patients whose physicians did not receive the education but had access to the electronic health record changes. Time-driven, activity-based costing analysis was used to calculate total cost savings based on US Department of Labor wages for nursing and pharmacy with Stata software, version 14.0 (StataCorp LP).

The inappropriate use of PPI infusion significantly decreased from a preintervention mean of 50% (95% CI, 45.1%-54.7%) to a postintervention mean of 15% (95% CI, 10.0%-18.5%), a 35% reduction (95% CI, −44.6% to −27.1%; P < .001) (Figure). After the intervention, PPI infusions were significantly more likely to be discontinued after endoscopy (23 of 35 [65.7%]; 95% CI, 50.0%-81.4% vs 38 of 40 [92.5%]; 95% CI, 84.3%-100%; P = .004). The total cost of a single PPI infusion was calculated to be $378 per day, whereas the estimated cost of administering PPIs twice daily was $100.87, yielding a cost savings of $277 per patient per day assuming all patients required twice-daily PPI administration. Conservative estimates indicated a savings of at least $121 000 in pharmacy costs across 9 months. These changes were not observed for surgical services in which physicians had access to electronic health record changes but no education although a nonsignificant decrease in the percentage of inappropriate PPI infusions occurred 6 months after the intervention was launched (32 of 99 [32.3%]; 95% CI, 23.1%-41.5% vs 17 of 83 [20.5%]; 95% CI, 11.8%-29.2%; P = .08).

This trainee-led intervention based on the Culture, Oversight, Systems Change, Training framework yielded a 35% decrease in inappropriate PPI infusions in patients with UGIB and a substantial cost savings in a 9-month period. Access to the systems change alone did not result in significant changes, highlighting the importance of including all aspects of the framework in the intervention.

Our quality improvement study was limited by a lack of both randomization and blinding. In addition, we did not determine the number of patients who were discharged with inappropriate PPI therapy and did not explore complications associated with inappropriate PPI usage.

This trainee-led, institution-supported, interdisciplinary intervention based on the Culture, Oversight, Systems Change, Training framework resulted in high-value, guideline-based care for inpatients with UGIB. Further research will be required to determine the durability of the intervention and unnecessary complications associated with PPI infusion overuse.

Corresponding Author: Vineet M. Arora, MD, MAPP, Section of General Internal Medicine, Department of Medicine, University of Chicago, 5841 S Maryland Ave, MC 2007 W216, Chicago, IL 60637 (varora@medicine.bsd.uchicago.edu).

Accepted for Publication: July 16, 2017.

Published Online: October 9, 2017. doi:10.1001/jamainternmed.2017.4851

Author Contributions: Dr Arora had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Coronel, Bassi, Sokol, Reddy, Arora.

Acquisition, analysis, or interpretation of data: Coronel, Bassi, Donohue-Rolfe, Byrne, Reddy, Arora.

Drafting of the manuscript: Coronel, Bassi, Byrne, Arora.

Critical revision of the manuscript for important intellectual content: Coronel, Bassi, Donohue-Rolfe, Sokol, Reddy, Arora.

Statistical analysis: Coronel, Donohue-Rolfe, Arora.

Obtained funding: Coronel, Arora.

Administrative, technical, or material support: Coronel, Byrne, Sokol, Arora.

Study supervision: Coronel, Reddy, Arora.

Conflict of Interest Disclosures: None reported.

Additional Contributions: We thank Andrew Levy, MD, University of Colorado, for co-leading the intervention design and Michael Howell, MD, MPH, Center for Healthcare Delivery Science and Innovation, University of Chicago, for his administrative support. Neither was compensated for these contributions.