Search

BACKGROUND Candidemia is one of the leading causes of bloodstream infections in the pediatric intensive care unit (PICU). The aim of this study was to define characteristics and risk factors for candidemia in the PICU setting and propose a predictive model to identify the patients at risk. METHODS This was a retrospective matched case-control study in the PICU during a 9-year period. Patients with candidemia were studied and matched with control patients without candidemia. Univariate analysis was performed for potential risk factors and multivariate analysis was conducted to determine the prediction score for candidemia. RESULTS Forty-two cases of candidemia were matched with 84 control patients. Candida parapsilosis was the most common (71.4%) species. Risk factors independently associated with candidemia were: the use of >2 antibiotics in a maximum period of 4 weeks before the candidemia (odds ratio [OR]: 10.59; 95% confidence interval [CI]: 2.05-54.83), a previous bacterial infection in a maximum period of 4 weeks before the candidemia (OR: 5.56; 95% CI: 1.44-21.5) and the duration of PICU stay of >10 days (OR: 4.22; 95% CI: 1.02-17.41). The proposed predictive scoring system has a sensitivity of 95.24%, specificity of 76.12%, OR 64.0, 95% CI 14.2-288.6, the positive predictive value of 66.67% and the negative predictive value of 96.97%. CONCLUSIONS Previously reported risk factors for candidemia have been confirmed and some new have been detected. The presented scoring system can help identify patients who would benefit from prophylactic antifungal therapy.