Abstract 29

Aims: Illness severity scores are essential for NICU outcome comparisons. Validated scores are CRIB (clinical risk index for babies), SNAP(score for neonatal acute physiology), and SNAP-PE (SNAP perinatal extension). CRIB (6 items) is not validated for infants > 1,500g BW, while calculation of SNAP (34 items) and SNAP-PE (37 items) are time-consuming. To overcome these difficulties, we developed and evaluated NAPPI (neonatal acute physiology parameters index)(12 items). Methods: Data and variables to calculate SNAP, SNAP-PE and CRIB were collected from 10,213 infants in 17 Canadian NICUs in 1996. Subjects were randomly divided into estimation (n=6712) and validation (n=3501) sets, and the method of Pollack et al. was used to develop NAPPI from admission day SNAP variables. BW and Apgar were added to construct NAPPI-ES (NAPPI extended score) as a SNAP-PE equivalent. The scores were compared with SNAP, SNAP-PE and CRIB. Model discrimination was evaluated using area under the receiver operating curve(ROC), and calibration by goodness-of-fit tests. Time-to-conduct trials were conducted by a single individual on 10 sequential NICU admissions at Children's & Women's Health Center of B.C. Results: NAPPI correctly classified groups with increasing probability of death, irrespective of diagnosis. NAPPI was calculated in almost half the time required for a SNAP score, is comparable to SNAP and better than CRIB (p< 0.0001) in predicting mortality for all birth weights. NAPPI-ES is equivalent to SNAP-PE (see table), and is better than NAPPI (p< 0.05, one tailed) in predicting mortality.

Table 1