893. Clinical Risk Score for Prediction of Invasive Candidiasis to Guide Empiric Echinocandin Therapy

Abstract

Abstract Background Current treatment guidelines for invasive candidiasis (IC) largely recommend an echinocandin as initial therapy. Echinocandins have demonstrated non-inferiority to previously utilized agents for the treatment of IC with low toxicity, few drug-drug interactions, and activity against azole-resistant Candida species. Although echinocandins are suggested empirically when IC is suspected, recommendations for empiric therapy based upon IC risk factors do not exist. The primary objective is to develop a risk score to predict probability of IC to help triage when empiric therapy with an echinocandin is likely warranted. Methods This evaluation is a retrospective, case-control design of hospitalized adults that received at least one dose of micafungin between July 1, 2020 and June 30, 2021 for proven or suspected IC at a large academic medical center. Descriptive statistics were used to describe all variables collected. A multivariable logistic regression analysis was used to determine the predictability of identified risk factors on IC. The predictability of the model was quantified using a nomogram which sums the odds of experiencing IC based on the presence of selected predictors. The sum is correlated to a linear predictor which corresponds to the log odds of experiencing IC. The exponential on the log odds gives the predicted probability of IC. Results Three hundred eighteen patients that received at least one dose of micafungin during the time frame for proven or suspected IC were included. IC was confirmed in 105 patients (33%) with endovascular source being the most common (56.2%). The most common Candida species isolated was C. albicans (47.6%). Independent risk factors for development of IC included anti-anaerobic antimicrobial therapy (adjusted odds ratio [aOR], 0.4611; 95% confidence interval [CI], 0.232, 0.914), parenteral nutrition (aOR, 2.115; 95% CI, 1.019, 4.388), and critical illness (aOR, 0.511; 95% CI, 0.315, 0.829). Patients with prediction scores of greater than 50 had an estimated probability of IC greater than 20%. Conclusion A prediction score can be a useful tool to estimate the patient-specific risk of IC development. Integration of a risk score for IC into clinical workflow may improve empiric antifungal prescribing and reduce echinocandin utilization. Disclosures All Authors: No reported disclosures.