2363. Implementation of a Multi-Step Diagnostic Algorithm for C. difficile Infection

Abstract

BackgroundThe diagnosis of C. difficile infection (CDI) in the hospital is challenging asymptomatic colonization rates vary between 3% and 26%. Guidelines recommend multistep testing for CDI diagnosis. On July 1, 2018 a two-step testing algorithm was implemented at our institution. Positive nucleic acid amplification test (NAAT) results reflexed to a toxin enzyme immunoassay (EIA) test. The EIA test result was then used for NHSN reporting; however, both test results were visible to the clinician. Updated guidance on the interpretation of the test and treatment of CDI was released to the medical staff in July. We compared the incidence of CDI lab ID events per 1000 patient-days and the rate of C. difficile antibiotic starts before and after the implementation of the testing algorithm.MethodsA retrospective observational study was performed at an 800 bed regional medical center. CDI lab ID events between January 1 and December 31, 2018 were reviewed. Antibiotic initiation of intravenous (IV) and oral (PO) metronidazole and PO vancomycin was collected for all hospitalized patients diagnosed with C. difficile. The incidence of hospital onset (HO) and community-onset (CO) lab ID events as well as the rate of antibiotic starts were compared before and after implementation of the algorithm using a two-sided z test for proportions with an alpha of 0.05.ResultsThe incidence of HO and CO lab ID events per 1000 patient-days decreased significantly from 0.56 to 0.16 (P < 0.0001) and 1.18 to 0.3 (P < 0.0001) after implementation of the testing algorithm (Figure 1). The CDI SIR decreased from 0.729 to 0.322, (P = 0.0048). The rate of antibiotic starts per 1,000 patient-days for IV and PO Metronidazole decreased significantly from 1.1 to 0.45 (P < 0.0001) and 0.86 to 0.35 (P < 0.0001), respectively. PO Vancomycinstarts decreased from 1.51 to 1.23 (P = 0.11) (Table 1).ConclusionA two-step algorithm for diagnosing CDI decreases the overall number of HO and CO C. difficile lab ID events and decreases overall antimicrobial use for CDI. Disclosures All authors: No reported disclosures.