A function to predict the risk for death in the neonatal sepsis

Abstract

This was a prospective study of 43 septic neonates at the NICU of the School of Medicine of Botucatu, São Paulo State University. Clinical and laboratory data of sepsis were analyzed based on outcome divided into two groups, survival and death. We calculated the discriminatory power of the relevant variables for the diagnosis of sepsis in each group, and using software for Discriminant Analysis, a function was proposed. There were 43 septic cases with 31 survivals and 12 deaths. The variables that had the highest discriminatory power were: n(o) of compromised systems, the SNAP, FiO2, and (A-a)O2. The study of these and others variables, such as birth weight, n(o) of risk factors, and pH using a Linear Discriminant Function(LDF) allowed us to identify the high-risk neonates for death with a low error rate (8.33%). The LDF was: F = 0.00043 (birth weight) + 0.30367 (n(o) of risk factors) - 0.1171 (n(o) of compromised systems) + 0.33223 (SNAP) + 2.27972 (pH) - 14.96511 (FiO2) + 0.01814 ((A-a)O2). If F > 22.77 there was high risk of death. This study suggests that the LDF at the onset of sepsis is useful for the early identification of the high-risk neonates that need special clinical and laboratory surveillance.