Evaluation of the CRIB as an Indicator of the Performance of Neonatal Intensive Care Units Using the Software ROCNPA

Mortality rates are considered one of the most important indicators for assessing the performance of neonatal care. Increasingly, newborns with very low birth weight (<; 1500g at birth) or less than 37 weeks of gestation, contribute significantly to the neonatal mortality and morbidity. However, the mortality rates have to be adjusted in accordance to the severity of illness and, therefore, several severity indexes have been developed. The ROC (Receiver Operating Characteristic) analysis is often used to evaluate the accuracy of the performance measures in medical diagnosis. This technique allows the study of the variation of sensitivity and specificity for different cutoff values. One major advantage of using ROC curves is the possibility to compare different severity indexes, as is the case in medical diagnosis. This study aims to analyze the performance of neonatal clinical severity scale, CRIB (Clinical Risk Index for Babies) in two intensive care units of neonatology in Portugal, to evaluate its power to discriminate between infants ”dead” and ”survivors.” At the same time, we intend to compare the performance of two units, considering as relevant hypothesis that the two neonatal units have the same AUCs (Area Under Curve). To carry out this analysis, we used the software ROCNPA (Nonparametric Receiver Operating Characteristic Analysis), developed by Braga et al. [1] and which can be downloaded in a beta version http://pessoais.dps.uminho.pt/acb/englacb/.