Comparing CRIB-II and SNAPPE-II as mortality predictors for very preterm infants.

Abstract

This article compares the severity of illness scoring systems clinical risk index for babies (CRIB)-II and score for neonatal acute physiology with perinatal extension (SNAPPE)-II for discriminatory ability and goodness of fit in the same cohort of babies of less than 32 weeks gestation and aims to provide validation in the Australian population. CRIB-II and SNAPPE-II scores were collected on the same cohort of preterm infants born within a 2-year period, 2003 and 2004. The discriminatory ability of each score was assessed by the area under the receiver operator characteristic curve, and goodness of fit was assessed by the Hosmer-Lemeshow (HL) test. The outcome measure was in-hospital mortality. A multivariate logistic regression model was tested for perinatal variables that might add to the risk of in-hospital mortality. Data for both scores were available for 1607 infants. Both scores had good discriminatory ability (CRIB-II area under the curve 0.913, standard error (SE) 0.014; SNAPPE-II area under the curve 0.907, SE 0.012) and adequate goodness of fit (HL χ2 = 11.384, 8 degrees of freedom, P = 0.183 for CRIB-II; HL χ2 = 4.319, 7 degrees of freedom, P = 0.742 for SNAPPE-II). The multivariate model did not reveal other significant variables. Both severity of illness scores are ascertained during the first 12 h of life and perform similarly. Both can facilitate risk-adjusted comparisons of mortality and quality of care after the first post-natal 12 h. CRIB-II scores have the advantage of being simpler to collect and calculate.