Objective: To determine whether the Score for Neonatal Acute Physiology, Version II (SNAP-II), improved prediction of severe (> or = grade III) intraventricular hemorrhage (IVH) and chronic lung disease (CLD) when compared to models using gestational age (GA) and traditional risk factors (e.g., Apgar score, small-for-gestational-age, sex, outborn status).
Study design: We examined 4226 infants < or = 32 weeks' GA admitted to 17 Canadian neonatal intensive care units between 1996 and 1997. We compared prediction models for severe IVH and CLD, with and without SNAP-II.
Results: SNAP-II was a significant and independent predictor of severe IVH and CLD. Addition of SNAP-II to models using GA and traditional risk variables significantly (p<0.05) improved model prediction (AUC 0.8 for severe IVH; 0.83 for CLD). Models were well calibrated (p>0.05 for Hosmer-Lemeshow goodness of fit test).
Conclusion: Addition of SNAP-II to models using GA and traditional risk factors significantly improves prediction of severe IVH and CLD.