Failure of Best Practice Alerts to Affect C difficile Test Utilization. (21st September 2018)

• Record Type: Journal Article
• Authors: Swimley, Kent; Olsen, Randall; Long, S Wesley; Christensen, Paul; Davis, James
• Is Part Of: American journal of clinical pathology. Volume 150(2018)Supplement 1
• Page Start: S144
• Page End: S144
• Publication Date: 2018-09-21
• Subjects: Diagnosis, Laboratory -- Periodicals; Pathology -- Periodicals; 616.07
• DOI: 10.1093/ajcp/aqy112.343
• Languages: English
• ISSNs: 0002-9173

Abstract

Abstract: Rates of C difficile infection have become a prominent marker of health care quality. Assessing true C difficile infection rates have been complicated by the advent of modern molecular diagnostics tests that are more sensitive than older methods such as ELISA. This exquisite sensitivity greatly increases the chance of detecting colonization, rather than true infection, when the tests are used inappropriately. Thus, measures to guide physician utilization of C difficile testing are necessary. Our clinical laboratory informatics team created three specific EPIC Best Practice Advisories (BPAs) to guide physician ordering of C difficile testing. In particular, we targeted three conditions: (1) attempting to order a multipathogen stool panel on patients who had been admitted more than 3 days prior (inappropriate testing because these pathogens are unlikely to be health care associated), (2) attempting to order a C difficile test on a patient with a previous positive C difficile test in the prior 14 days (inappropriate testing for cure), and (3) attempting to order a C difficile test on a patient who is currently on a laxative or had an enema in the prior 24 hours (inappropriate testing because C difficile detection is likely to represent colonization rather than infection). We placed the three BPAs into service simultaneously in December 2016, with physician notification sent by email. Although initial results indicated BPA response rates of achieving the desired behavior 30% to 50% of the time, we saw no effect on overall C difficile detection rate or test utilization. Further investigation showed that although one or more individual providers on the patient's health care team might initially respond as desired, eventually one provider would override the BPA and order the testing anyway. Thus, on a per patient basis, the success rate to stop inappropriate C difficile testing approached zero. Our BPA experience illustrates the difficulty in achieving alterations in physician behavior using Clinical Decision Support tools with multidisciplinary medical teams and underscores the need to fully investigate the data to observe outcomes in clinical informatics interventions.