Snappe II: Analysis of accuracy and determination of the cutoff point as a death predictor in a Brazilian neonatal intensive care unit

Abstract

Objective: To analyze the accuracy of the Score for Neonatal Acute Physiology Perinatal Extension (SNAPPE II) as a death predictor, to determine the cutoff point for mortality, and to analyze the association of independent variables with death. Methods: Prospective, longitudinal, hospital-based study on newborns admitted to the Neonatal Intensive Care Unit (NICU) for the first time from November 1, 2016 to April 30, 2017. Newborns with less than 12 hours of length of stay at the NICU, out-of-hospital births, major congenital malformations, and interhospital transfer were excluded. Variables were grouped according to hierarchical framework, related to maternal characteristics (distal level), prenatal and childbirth care (intermediate level), and birth conditions (proximal level). Descriptive analyses of SNAPPE II score ranges, Receiver Operating Characteristics Curve (ROC curve) to define the cutoff point for mortality, and bivariate analysis by the Wald test and multiple logistic regression were conducted. Results: After selection, the sample consisted of 247 newborns. In this study, the SNAPPE II cutoff point for mortality was 27, with sensitivity of 84.1% and specificity of 82.4%. 61% of those with a score ≥27 died. Multiple logistic regression showed an association between death and proximal-level variables: sepsis (Odds Ratio [OR] 10.68; 95% confidence interval [95%CI] 2.82-40.48; p<0.001); SNAPPE II ≥27 (OR 5.85; 95%CI 1.90-18.05; p=0.002); birth weight 750-999 g (OR 4.15; 95%CI 1.06-16.14; p=0.040); and nonuse of surfactant (OR 0.159; 95%CI 0.04-0.53; p=0.003). Conclusions: Neonatal mortality was directly proportional to increase in SNAPPE II. Score≥27 increased the odds of dying by six times compared with neonates with lower scores. The proximal variables related to health conditions and neonatal care were associated with death.