A novel application to develop illness severity scores for predicting mortality and morbidities in a Neonatal Intensive Care Unit (NICU)

Abstract

The first hundred days of child life is considered the most critical phase of its physical development and its health. Neonates who are critically sick are kept under observation in a Neonatal Intensive Care Unit (NICU) for continuous monitoring of their health conditions. Classifying infants on the first day of admission based on illness severity score would help neonatologist in understanding the prognosis and better management of neonates at NICU. Illness severity scoring systems with a Score for Neonatal Acute Physiology (SNAP) and its variants was proposed to classify neonates based on its prognosis. A novel risk stratification module was developed which allowed neonatologist to design and develop an illness severity scoring system using the Richardson defined method. The SNAP I, modified SNAP I and SNAP II for the study population at study site was validated on 230 preterm neonates diagnosed with respiratory distress syndrome (RDS), apnea and sepsis as a retrospective study. All the scores predicted mortality with statistical significance, with improved accuracy performance (+ 15 to 25%) was seen with proposed modified SNAP I. The proposed modified SNAP I score was found to be most accurate compared to SNAP I and SNAP II for predicting the mortality and also other outcome measures such as severity of respiratory distress syndrome (RDS) (70%), presence of apnea (65%) and sepsis (80%) for the study population. Results obtained were strong indication of the need for scoring system where the variables as well as their range could be altered based on study population and NICU involved for the study.