4040 — Intended and unintended effects of a VISN-wide proton pump inhibitor deprescribing intervention: A quasi-experimental study of over 4 million Veterans
Lead/Presenter: Jacob Kurlander,
COIN - Ann Arbor
All Authors: Kurlander JE (VA Ann Arbor Center for Clinical Management Research, Ann Arbor; Department of Internal Medicine, University of Michigan, Ann Arbor), Saini SD (VA Ann Arbor Center for Clinical Management Research, Ann Arbor; Department of Internal Medicine, University of Michigan, Ann Arbor), Laine L (VA Connecticut Healthcare System, West Haven; Yale School of Medicine, New Haven), Kim M (VA Ann Arbor Center for Clinical Management Research, Ann Arbor; Consulting for Statistics, Computing and Analytics Research, University of Michigan, Ann Arbor), Saffar D (VA Ann Arbor Center for Clinical Management Research, Ann Arbor), Myers A (VA Ann Arbor Center for Clinical Management Research, Ann Arbor), Holleman RG (VA Ann Arbor Center for Clinical Management Research, Ann Arbor), Roberts C (The Center for Health Equity Research and Promotion, Philadelphia), Shank M (Veterans Affairs Central Office Pharmacy Benefits Management Service, Washington), Yang YX (The Center for Health Equity Research and Promotion, Philadelphia; Department of Medicine, Perelman School of Medicine, University of Pennsylvania, Philadelphia)
Few interventions have proven effective in curbing the overuse of proton pump inhibitors (PPIs). In 2013, VISN 17 implemented a pharmacy-based multicomponent PPI deprescribing intervention, which restricted refills for Veterans without a long-term indication. We aimed to evaluate its impact, including overall and potentially inappropriate deprescribing, health care utilization, and clinical outcomes potentially associated with PPIs.
We conducted difference-in-difference analyses comparing outcomes in VISN 17 to all other VISNs (control sites) before and after the 2013 intervention roll-out. The pre- and post-intervention periods, defined a priori, each consisted of 9 consecutive 6-month intervals. We excluded the 12-month roll-out period from the analysis. In each interval, Veterans were included if they had at least 2 physician/mid-level primary care encounters in the 2 years preceding the interval. Outcomes included the percentage of Veterans dispensed a PPI (primary), % time with PPI gastroprotection in Veterans at high risk for upper gastrointestinal bleeding, the rate of upper endoscopies and primary care visits with an associated diagnosis code for an upper GI complaint/diagnosis, and the incidence proportion for new chronic kidney disease (CKD) and C. difficile infection using CDW data. We also examined the following outcomes for Veterans age > = 65 only using both CDW and CMS-linked data: incidence proportions of hospitalization for pneumonia, stroke, hip fracture, myocardial infarction (MI), and upper gastrointestinal bleeding.
In VISN 17, the number of Veterans analyzed per interval ranged from 192,607 to 250,349, and in control sites ranged from 3,775,978 to 4,360,908. The intervention in VISN 17 was associated with a 7.3% (95% confidence interval [CI] 7.0,7.6) absolute reduction in Veterans dispensed a PPI, compared to control sites, as well as an 11.3% (95% CI 10.5,12.0) reduction in time with PPI gastroprotection in Veterans at high risk for upper gastrointestinal bleeding. The intervention was not associated with changes in the rate of upper endoscopies (0.74 per 1000 Veterans, 95% CI -4.58,6.06) or primary care visits for upper GI complaints (3.18 per 1000 Veterans, 95% CI -4.02,10.37); nor was it associated with incidence of CKD for all ages (-0.05%, 95% CI -0.19,0.10), or hospitalizations for pneumonia (-0.06%, 95% CI -0.15,0.02), stroke (-0.03%, 95% CI -0.08,0.02), hip fracture (-0.021%, 95% CI -0.051,0.008), or MI (0.021%, 95% CI -0.058,0.015) in Veterans age > = 65 years. C difficile infection and upper gastrointestinal bleeding were unable to be analyzed because the data violated statistical assumptions.
The VISN 17 deprescribing intervention was associated with a clinically meaningful reduction in PPI use but also with a reduction in PPI gastroprotection in Veterans at high risk for upper gastrointestinal bleeding, an unintended consequence with the potential for patient harm. The intervention was not associated with reductions in adverse outcomes potentially associated with PPI use.
This PPI deprescribing intervention is one of the most effective ever described, with the potential for major cost savings. Efforts are underway to target the program more effectively, which will be necessary to minimize the risk of inappropriate deprescribing prior to broader dissemination.