SNAP-II for prediction of mortality and morbidity in extremely preterm infants

Objective: To determine the specific Score of Neonatal Acute Physiology (SNAP-II) cut-off scores associated with outcomes in extremely preterm infants, and to examine its contribution to predictive models that include nonmodifiable birth predictors.

Study design: Retrospective observational study of 9240 infants born at 22-28 weeks' gestation and admitted to the Canadian Neonatal Network from 2010 to 2015. Outcomes included early and hospital mortality, composite of mortality/morbidity and individual morbidities. The SNAP-II cut-off to predict each outcome was determined using the Youden index. Additional contributions were evaluated using a base model that adjusted for gestational age, birth weight z-score and sex and by comparing the area under the curve (AUC).

Results: The mortality/morbidity rate was 63% (5859/9240). Specific SNAP-II cut-offs ranged from 12 to 20 and were associated with each adverse outcome. Adding SNAP-II cut-offs to predictive models that included birth variables significantly improved (p < .05) the prediction of early mortality (AUC 0.84 versus 0.79), hospital mortality (AUC 0.80 versus 0.78), mortality/morbidity (AUC 0.76 versus 0.75), and severe neurological injury (AUC 0.69 versus 0.66) but had little or no effect on predictive models for retinopathy of prematurity, bronchopulmonary dysplasia, necrotizing enterocolitis, and nosocomial infection.

Conclusions: SNAP-II cut-offs were independently associated with each adverse outcome and using the proposed SNAP-II cut-offs improved the performance of predictive models for certain short-term outcomes.