Clinical predictors of candidemia in medical non-neutropenic, non-ICU patients. The CaMed score.

ABSTRACT

INTRODUCTION: Candida species are the leading cause of invasive fungal infections in hospitalised patients and are the fourth most common isolates recovered from patients with bloodstream infection. Few data exist on risk factors for candidemia in non-ICU patients. We performed a population-based case-control study to evaluate the main predictors for candidemia in non-ICU patients. METHODS AND FINDINGS: We included all non-neutropenic, non-critically ill and non-surgical adult patients with candidemia between January 2010 and June 2014. Patients with positive, non-candidal blood culture obtained at the same day (±2 days) were selected as controls. Cases and controls were matched according to hospital ward and clinical characteristics. Risk factors for candidemia were identified through a logistic regression. We included 56 candidemic and 512 bacteriemic non-candidemic patients. Most of candidemic patients (52) had received antibiotics prior to candidemia. Among them, the 30-day mortality rate was 34% (19/56). Multivariate analysis identified male sex, prior use of steroids, prior use of antibiotics, total parenteral nutrition and urinary catheterisation as independent predictors of candidemia. To develop the CaMed score, we rounded up weights of different risk factors as follows; total parenteral nutrition (+2), prior antibiotic therapy (+5), each of the other risk factors (+1). A score ≥ 7 identified patients at high risk of candidemia (P < 0.001; RR 29.805; CI 95% 10.652-83.397; sensitivity 79.2, specificity 82.6%, Youden index 0,62). CONCLUSIONS: Our set of easy independent predictors of candidemia in non-neutropenic, non-ICU, non-surgical patients provide a rationale for early initiation of antifungals and could reduce candidemia-related mortality.