A scoring model of factors associated with Candida glabrata candidemia among critically ill patients

Abstract

Candida species, including Candida glabrata (CG), are common causes of bloodstream infections among intensive care unit (ICU) patients. Many CG isolates have decreased susceptibility to fluconazole. Constructing a scoring model of factors associated with CG candidemia in ICU patients that can be used if fluconazole susceptibility testing is not readily available. We identified patients with candidemia that were admitted to the ICU of the Mayo Clinic in Rochester, Minnesota from 1998 to 2006. Using patient demographical and clinical data abstracted via chart review, a multivariable logistic regression model was developed to distinguish those with CG candidemia. We identified 246 patients with candidemia including 68 CG cases. Multivariable analysis identified four independent factors associated with CG candidemia: absence of renal failure, less than 7 days in the hospital, abdominal surgery and fluconazole use. The predictive ability of the model, based on the c-statistic, was 0.727. In a large ICU cohort, a scoring model that included four risk factors, which are readily ascertainable at the bedside, was created to distinguish candidemia due to CG from other causes of candidemia. The identification of risk factors associated with CG candidemia could aid physicians in the selection of the optimal initial antifungal therapy.