Initial serum (1,3)-β-D-glucan as a predictor of mortality in proven candidaemia: findings from a retrospective study in two teaching hospitals in Italy and Brazil

Abstract

A retrospective study was conducted to assess the role of initial serum (1,3)-β-d-glucan (BDG) values in predicting mortality in proven candidaemia. The study was conducted in two large teaching hospitals in Italy and Brazil. From January 2009 to June 2014, all patients with proven candidaemia who underwent a BDG test within 96 hours before or after the first positive blood culture were included in the study. The primary end point was 28-day mortality, with the role of initial BDG being assessed by univariate and multivariate analyses. A total of 104 patients met the inclusion criteria. Overall, the crude 28-day mortality was 30% (31/104). In the final multivariate model, an initial BDG of >287 pg/mL (odds ratio (OR) 4.40, 95% confidence interval (CI) 1.56-12.39, p 0.005), haemodialysis (OR 4.33, 95% CI 1.24-15.17, p 0.022) and a Pitt score of ≥ 2 (OR 4.10, 95% CI 1.24-13.54, p 0.021) were significant predictors of 28-day mortality. The >287 pg/mL cutoff predicted 28-day mortality with 65% sensitivity and 70% specificity. Centre of enrolment (p for interaction 0.012), haemodialysis (p for interaction 0.062) and timing of BDG test of more than 24 hours before or after the positive culture (p for interaction 0.143) appeared to interact with BDG's ability to predict mortality. Although not statistically significant, the last two of these interactions might partially explain why BDG's ability to predict mortality was present only in the Italian cohort.

Keywords: Bloodstream infections; Candida spp.; candidaemia; mortality; β-d-glucan.