Background: Invasive candidiasis (IC) is a life-threatening ICU-acquired infection. A strong correlation between time to antifungal therapy (AFT) administration and outcome has been established. Empirical therapy benefit should be balanced with the risk of echinocandin overuse. We assessed therefore a decision rule that aimed at guiding empirical therapy.
Methods: A 45-month prospective cohort study in a teaching medical ICU. All of the patients with suspected IC (uncontrolled sepsis despite broad spectrum antibiotics without any bacterial proven infection in patients with Candida score ≥ 3 points including multifocal Candida sp. colonization) were eligible. The primary endpoint was proven IC diagnosis (i.e., candidemia) following treatment onset. Timing of AFT administration was also investigated in those latter patients. Antifungal therapy step-down and discontinuation was done according to international guidelines in patients with candidemia. Otherwise, echinocandin discontinuation was encouraged in patients without proven IC, excepting when a clinical improvement was achieved without any other explanation that antifungals initiation (i.e., probable IC). In addition, a survival multivariate analysis using a Cox model was conducted.
Results: Fifty-one patients were given an echinocandin with respect to our decision rule. Among them, candidemia was diagnosed thereafter in 9 patients. Over the same period, antifungal therapy was triggered by candidemia announcement (i.e., definite therapy) in 12 patients who did not fulfill criteria for empirical therapy before. Time elapsed from candidemia onset to echinocandin therapy initiation was shortened (0.4 [0.5] vs. 2.4 [2.8] hours; p = 0.04) when it was given empirically. In addition, 18 patients clinically improved under empirical antifungal therapy without any obvious other explanation, despite IC remained unproven. Moreover, echinocandin exposure duration was independently related to survival in those patients. Over the same period, our predefined criteria for empirical therapy were overruled in 55 cases. None of them develop IC thereafter. Finally, Our decision rule allowed IC early recognition of proven/probable IC with sensitivity, specificity, positive and negative predictive value of 69.2%, 82.1%, 69.2% and 82.1%, respectively.
Conclusion: Implementation of pragmatic guidelines for empirical AFT based on CS and fungal colonization assessment could be useful in selecting patients who really benefit from an echinocandin.