168. Creation and Internal Validation of a Clinical Predictive Model for Fluconazole Resistance in Patients with Candida Bloodstream Infection. (31st December 2020)

• Record Type: Journal Article
• Authors: Rauseo, Adriana M; Olsen, Margaret A; Larson, Lindsey; Stwalley, Dustin; Hsueh, Kevin; Powderly, William; Spec, Andrej
• Is Part Of: Open forum infectious diseases. Volume 7:Number 1(2020) Supplement
• Page Start: S91
• Page End: S92
• Publication Date: 2020-12-31
• Subjects: Communicable diseases -- Periodicals; Medical microbiology -- Periodicals; Infection -- Periodicals; 616.9
• DOI: 10.1093/ofid/ofaa439.212
• Languages: English
• ISSNs: 2328-8957

Abstract

Abstract: Background: IDSA guidelines on candidemia recommend fluconazole as first-line therapy in patients considered low risk for fluconazole resistant infections. However, there is currently no mechanism to determine risk of resistance, and most community hospitals cannot perform rapid sensitivity testing, leading to prolonged use of echinocandin therapy. This study aims to develop a clinical predictive model to identify patients at low risk for fluconazole resistance where first-line use of fluconazole therapy would be acceptable without requiring resistance testing. Methods: We performed a retrospective cohort analysis of all hospitalized adult patients with a positive blood culture for Candida spp. from 2013 to 2018. Fluconazole resistance was determined using Sensititre™ YeastOne™ YO9 AST Plate, with cutoffs defined for each Candida species based on Clinical and Laboratory Standards Institute performance standards for antifungal testing (M60) in all patients. Using backwards stepwise regression, we developed a multivariable logistic regression model to identify factors associated with fluconazole resistance in patients in Candida bloodstream infection, including only variables with clinical plausibility and p < 0.1 in bivariable analysis. Stepwise regression was performed on bootstrapped samples to test individual variable stability and estimate confidence intervals. We used graphs of observed vs expected values to assess model performance across the probability spectrum. Results: We identified 539 patients with Candida bloodstream infection from 2013–2018, of which 13.4% (72/539) were fluconazole resistant. Increased risk of fluconazole resistance was associated with age (1.12 [1.01, 1.24]), bacterial septicemia (2.14 [1.20, 3.79]), receipt of previous azole therapy (5.47 [2.92, 10.26]), bone marrow transplant (2.63 [1.31, 5.29]), and myelodysplastic syndrome (3.13 [1.14, 8.60]). The model predicted fluconazole sensitivity well (c-statistic 0.788) and all the variables were stable (Figure 1). Figure 1. Graph comparing observed versus expected probability of fluconazole resistance. Bars included on the top parameter of the graph indicate the number of individuals, illustrating the distribution of the sample. Conclusion: The presented model provides a potential tool for identifying the 80% of patients at low enough risk for fluconazole resistance to receive empiric therapy with azoles and reduce use of echinocandins. Disclosures: Margaret A. Olsen, PhD, MPH, Merck (Grant/Research Support)Pfizer (Consultant, Grant/Research Support) Dustin Stwalley, MA, AbbVie Inc (Shareholder)Bristol-Myers Squibb (Shareholder) Andrej Spec, MD, MSCI, Astellas (Grant/Research Support)Mayne (Consultant)Scynexis (Consultant)