Candidemia diagnosis is based on the combination of clinical, microbiological and laboratory data. We aimed to evaluate performances and accuracy of (1,3)-β-D-glucan (BDG) at various cut-offs in internal medicine patients. An observational retrospective−prospective study was performed. Patients with at least two determinations of BDG and paired, associated blood cultures within ±48 h were considered. A total of 140 patients were included: 26 with Candida spp. blood-stream infections (BSI) and 114 without candidemia. Patients with candidemia were older and had higher BDG values, need of parenteral nutrition, higher colonization by Candida in more than one site, presence of percutaneous gastrostomy and higher Candida or Charlson scores. BDG maintained the best compromise between sensitivity, specificity and optimal negative predictive value was 150 pg/mL. BDG values at cut-off of 150 pg/mL increase the strength of association between BDG and development of candidemia (Odds Ratio—OR 5.58; CI 2.48−12.53 vs. OR 1.06; CI 1.003−1.008). Analyzing BDG > 150 pg/mL along with Candida score > 2 and Charlson score > 4, the strength of the association amongst BDG, clinical scores and development of candidemia is increased. The overall clinical evaluation with the help of scores that consider BDG values > 150 pg/mL, Candida score > 2 and Charlson score > 4 in combination seems to predict better the need of antifungal empiric treatment.
Keywords: biomarkers; candidemia; fungi; internal medicine; yeast; β-D-glucan.