S2049 Reducing Inappropriate Proton Pump Inhibitor Prescribing in Non-Critically Ill Hospitalized Patients: A Single Center Quality Improvement Project

Introduction: Proton pump inhibitors (PPIs) are indicated for various gastrointestinal illnesses. However, they are sometimes inappropriately prescribed in non-critical hospitalized patients without a guideline-directed indication. PPI use is associated with Clostridioides difficile infection, among other side effects. PPIs also pose a significant global economic burden, with market estimates of $3.52 billion by 2028. Current stewardship guidelines outside the intensive care unit are institution-specific, highlighting the need for standardized criteria to optimize prescribing practices. We conducted a quality improvement study to reduce inappropriate PPI prescribing on admission for patients admitted to internal medicine teams at a tertiary care center by 20% within 6 months using 2 de-implementation strategies. Methods: Clinical data on inpatient PPI orders and International Classification of Diseases, Tenth Revision (ICD-10) codes billed for each admission to hospital medicine were extracted from the electronic medical record (EMR) via an information technology data pull and reviewed by a physician. Patients with ICD-10 codes for conditions where PPIs were indicated were excluded. Pre-intervention data (May 1, 2023-April 30, 2024) established baseline rates of inappropriate PPI orders. PPIs were removed from the EMR admission order set as the first intervention (May 1, 2024-July 30, 2024) and a standardized history and physical (H&P) note template was introduced to all medicine teams as the second intervention (July 31, 2024-December 31, 2024). Results: The rate of inappropriate PPI orders on admission decreased from 42.4% to 34.1% (19.6% reduction, P < 0.001) between the pre-intervention (n = 6,549) and the post-intervention group (n = 1,683). When stratified by teams, PPI orders decreased by 40.8% (P < 0.001) in non-academic teams between the pre- and post-intervention groups. C. difficile (defined by ICD-10 codes) decreased by 42.4% (P = 0.037) in the same groups. There was no difference in PPI prescriptions at discharge between groups (P = 0.228). Conclusion: Removing PPIs from the admission order sets and standardizing our H&P templates significantly reduced inappropriate PPI prescribing by 19.6%, primarily driven by changes in non-academic teams. Although C. difficile infections decreased by 42.4%, we cannot conclude that this was solely due to our interventions. Limitations include reliance on ICD-10 codes and inaccurate home medication lists. Additional de-implementation strategies could decrease PPI orders during hospitalization and at discharge.