NICU care for critically ill newborns is expensive, fiscally and emotionally. Both public policy-makers and surrogate decision-makers deserve to be informed by the best available information about the likelihood of survival or non-survival for NICU patients. We wondered whether serial determinations of a widely accepted measure of neonatal illness severity, the SNAP score, would enhance our ability to predict survival or non-survival of ICU patients. Recognizing the ethical principle of “trial of therapy”, we concentrated particularly on patients whose NICU stay lasted >3 days.

Methods: During a 4 month study period, we determined daily SNAP scores for each patient admitted to our NICU who required mechanical ventilation for at least 3 days. We attempted to discern a pattern of progression of SNAP scores that would allow us to distinguish NICU survivors from patients who were ultimately doomed.

Results: 28 patients required mechanical ventilation in our NICU during the study period; 19 lived, 9 died 59 (55%) non-survivors died after an NICU stay of at least 3 days (average 23 days; median 18 days). As expected, SNAP on DOL 1 was significantly higher for NICU non-survivors than for infants who survived to discharge (17.9 ± 4.9 vs 11.7 ± 7.2; p < 0.05). For the non-surviving patients who lived 3 days or more, SNAP on DOL 1 was 15.0 ± 3.5. By DOL 4, SNAP had fallen significantly (11.8± 1.6; Δ SNAP = -3.2). From DOL 4 to the day of death there wasno significant change in SNAP score (12.2 ± 5.0 on day of death; Δ SNAP Day 4 to death = 0.4 ± 5.0; p = ns). The pattern of SNAP scores for survivors resembled that of non-survivors: a significant fall in the first 3 days (12.5 to 6.9; p < 0.05) followed by no significant change from DOL 4 until day of extubation (6.9 to 4.6).

Conclusions: 1) For the NICU population as a whole, SNAP on DOL 1 was a significant predictor of the likelihood of eventual demise; 2) However, for individual patients, the process of dying was not reflected by a concomitant rise in SNAP score. 3) The qualitative course of SNAP scores for individual survivors resembled that for non-survivors. 4) Algorithmic predictors of non-survival in the NICU remain elusive -- and perhaps unobtainable.