Development and validation of a risk calculator of neonatal sepsis caused by Gram-negative bacteria based on hematological parameters

Abstract

RESULTS: We identified 103 cases of culture-proven neonatal sepsis that underwent a complete blood count performed at the time of clinical suspicion. Sepsis due to gramnegativas bacteria was predicted by age, anemia, mean corpuscular hemoglobin concentration, platelet count, mean platelet volume and white blood cell count. We also found statistically significant associations between female sex and neonatal sepsis due to Coagulase Negative Staphylococci [OR = 2.11, 95%CI 1.08-4.11], thrombocytopenia and late-onset neonatal sepsis [OR = 2.1, 95%CI 1.71-2.58], anemia and neonatal sepsis due to enterobacteria [OR=3.27, 95%C11.22-8.76], thrombocytopenia and neonatal sepsis due to enterobacteria [OR=3.62, 95%Cl 1.28-10.17], and thrombocytopenia with neonatal sepsis due to Coagulase Negative Staphylococci [OR = 0.23, 95%CI 0.09-0.56]. CONCLUSIONS: By employing hematological parameters measured at the time of clinical suspicion of NS, it is possible to predict the probability that the causative agent is a Gram-negative bacteria. BACKGROUND: Neonatal sepsis due to Gram-negative bacteria has a high mortality, and its empirical antibiotic treatment can lead to adverse results if it turns out to be insufficient or excessive. OBJECTIVE: The aim of this study was to develop and validate a calculator capable of estimating the probability that the causative agent of neonatal sepsis is a Gram-negative bacteria, using hematological parameters measured at the time of clinical suspicion. MATERIAL AND METHODS: In this retrospective cross-sectional study, we conducted an analysis of hospitalized patients with culture-proven neonatal sepsis at our institution between 2002 and 2019. Binary logistic regression was used to identify hematological parameters associated with neonatal sepsis due to gramnegativas bacteria, and a K-fold resampling method was used for cross-validation.