Early prediction of Candida glabrata fungemia in nonneutropenic critically ill patients

Objective: Candida species represent the fourth cause of nosocomial bloodstream infections worldwide. Because Candida glabrata has become the second most frequently identified yeast and because the rate of fluconazole-resistant C. glabrata strains reaches 10% to 15%, initial antifungal therapy based on fluconazole in nonneutropenic hemodynamically stable patients, as recommended by current guidelines, may be an ineffective option. Our aim was to determine easy-to-identify risk factors for C. glabrata fungemia likely to guide and improve initial antifungal therapy. Design: Prospective multicenter cohort study. Setting: Five French intensive care units. Patients: Consecutive nonneutropenic patients without known Candida colonization who had blood culture-confirmed fungemia over a 4-yr period. Interventions: None. Measurements and Main Results: A total of 8206 patients were screened. One hundred fifty-four patients with blood culture-confirmed fungemia constituted the cohort, of whom 48 had C. glabrata fungemia and 106 had nonglabrata fungemia. Patients' baseline characteristics and in-intensive care unit events potentially related to C. glabrata fungemia were systematically recorded. Compared with patients with nonglabrata fungemia, patients with C. glabrata fungemia were older and more severely ill, had received more antibiotics, and were more likely to have undergone surgery. The stepwise logistic regression analysis identified six independent risk factors for C. glabrata fungemia: age >60 yrs, recent abdominal surgery, interval from intensive care unit admission to first positive blood culture <= 7 days, recent use of cephalosporins, solid tumor, and absence of diabetes mellitus. The model showed satisfying goodness of fit (Hosmer-Lemeshow statistic = .26) and discrimination (c statistic = .89). Conclusions: We found six early available and easy-to-identify risk factors for C. glabrata fungemia. When these factors are present, alternatives to fluconazole for initial antifungal therapy should be considered. (Crit Care Med 2010; 38: 826-830)