Antifungal agents for preventing fungal infections in non‐neutropenic critically ill patients

Background Invasive fungal infections are important causes of morbidity and mortality among critically ill patients. Early institution of antifungal therapy is pivotal for mortality reduction. Starting a targeted antifungal therapy after culture positivity and fungi identification requires a long time. Therefore, alternative strategies (globally defined as 'untargeted antifungal treatments') for antifungal therapy institution in patients without proven microbiological evidence of fungal infections have been discussed by international guidelines. This review was originally published in 2006 and updated in 2016. This updated review provides additional evidence for the clinician dealing with suspicion of fungal infection in critically ill, non‐neutropenic patients, taking into account recent findings in this field. Objectives To assess the effects of untargeted treatment with any antifungal drug (either systemic or nonabsorbable) compared to placebo or no antifungal or any other antifungal drug (either systemic or nonabsorbable) in non‐neutropenic, critically ill adults and children. We assessed effectiveness in terms of total (all‐cause) mortality and incidence of proven invasive fungal infections as primary outcomes. Search methods We searched the following databases to February 2015: the Cochrane Central Register of Controlled Trials (CENTRAL), MEDLINE (OVID), and EMBASE (OVID). We also searched reference lists of identified studies and major reviews, s of conference proceedings, scientific meetings and clinical trials registries. We contacted experts in the field, study authors and pharmaceutical companies as part of the search strategy. Selection criteria We included randomized controlled trials (RCTs) (irrespective of language or publication status) comparing the use of untargeted treatment with any antifungal drug (either systemic or nonabsorbable) to placebo, no antifungal, or another antifungal agent in non‐neutropenic critically ill participants. Data collection and analysis Three authors independently applied selection criteria, extracted data and assessed the risk of bias. We resolved any discrepancies by discussion. We synthesized data using the random‐effects model and expressed the results as risk ratios (RR) with 95% confidence intervals. We assessed overall evidence quality using the GRADE approach. Main results We included 22 studies (total of 2761 participants). Of those 22 studies, 12 were included in the original published review and 10 w