Background: Prediction rules for invasive Candida infection (ICI) are available for adult but not for infants and children managed in pediatric intensive care units (PICU).
Methods: Observational study in 24 PICU with prospective phase (all children admitted during 1 year) and retrospective review of ICI records. Four logistic regression models were performed using ICI by Candida spp., Candida albicans, Candida parapsilosis or Candida tropicalis as dependent variables. Scores were constructed.
Results: One hundred and twenty five ICI (47 C. albicans, 37 C. parapsilosis, 19 C. tropicalis and 22 others) and 1022 controls were included. Incidence (cases/100 PICU admissions): 4.22 (all Candida), 2.44 (C. albicans), 1.41 (C. parapsilosis), 0.19 (C. tropicalis). ICI was associated [Area under the receiver operating characteristics curve (AUC) = 0.764, 95% confidence interval (CI) = 0.719-0.809, P < 0.001] with pre-PICU hospitalization ≥ 15 days [odds ratio (OR) = 3.3; score: +3], fever (OR = 2.6; +2), thrombopenia (OR = 2.0; +1) and parenteral nutrition (OR=2.4; +2). Additionally, the following associations were noted: C. albicans ICI (AUC = 0.716, 95% CI = 0.640-0.792, P < 0.001) with chronic metabolic disease (OR = 10.7; score:+4), surgical digestive process (OR = 2.8; +1), fever (OR = 2.8; +1) and parenteral nutrition (OR = 2.3; +1); C. parapsilosis ICI (AUC = 0.808, 95% CI = 0.739-0.877, P < 0.001) with previous colonization (OR = 7.1; score:+3), tracheostomy (OR = 5.1; +2), parenteral nutrition (OR = 4.3; +2), thrombopenia (OR = 3.6; +1) and previous bacterial infection (OR = 3.0; +1) and ICI by C. tropicalis (AUC = 0.941, 95% CI=0.886-0.995, P < 0.001) with thrombopenia (OR = 53.8; score: +10), neutropenia (OR = 7.2; +1), pre-PICU hospitalization ≥ 15 days (OR = 17.2; +3) and hematologic (OR = 22.4; +4) and cardiovascular infectious processes (OR = 5.5; +1). Specificity was >90% for cut offs of 5 (all Candida), 3 (C. albicans), 3 (C. parapsilosis) and 11 (C. tropicalis).
Conclusions: Once validated, these scores may help for identification of ICI by specific species allowing adequate empiric/prophylactic treatment.