Most NICU patients, even those at highest risk of dying, declare themselves over the first 72 hours of life; that is, patients who survive to DOL 4 are significantly more likely to survive than their birthweight-matched cohort on DOL 1. We wondered how successful an algorithm might be in predicting survival or non-survival for those infants who survived beyond 3 days of life.

Methods: We prospectively determined the survival of 270 consecutive infants admitted to our NICU who required mechanical ventilation in 1996. We noted birthweight (b.w.), gestational age, duration of mechanical ventilation (DOV) for survivors, length of stay for non-survivors (LOS), and a previously validated measure of illness severity, FiO2, for each day of life. In an attempt to highlight a population where mortality was highest and ethical concerns sharpest, we contrasted infants with b.w. < 750 gms(ultra-low-birth weight, ULBW) to infants with b.w. > 750 gms.

Results: Overall, 56 (21%) of 270 infants who required mechanical ventilation died; 46% of these non-survivors were ULBW. For ULBW infants, 26(60%) of 43 died, and 12 (46%) of these 26 non-survivors had died by DOL 3. For ventilated infants with b.w. > 750 gms, 30 (13%) of 227 died (p<0.01 vs ULBW), and 12 (40%) of these 30 non-survivors had died by DOL 3 (p=n.s. vs ULBW). From DOL 1-3, 40% of ULBW infants with FiO2 > 90% lived, not significantly different from 41% of infants with FiO2 < 90%. However, from DOL 4-7, only 8% of ULBW infants with FiO2 > 90% survived, contrasted with 52% of ULBW infants with FiO2 < 90% (p<.01). Unfortunately, FiO2 > 90% was not a sensitive indicator of non-survival, highlighting only 21% and 13% of doomed ULBW and non-ULBW infants after DOL 3 (p=ns).

Conclusions: 1) Almost half of all doomed ventilated neonates were ULBW. 2) For ULBW infants, FiO2 > 90% did not predict mortality during the first 3 days of life. 3) However, for ULBW infants alive after DOL 3, FiO2 > 90% was a powerful predictor of ultimate non-survival and may be a good marker of futile care. 4) Nevertheless, for both ULBW and non-ULBW infants after DOL 3, only a small minority of non-survivors was identified by the FiO2 > 90% criterion. Identification of doomed neonates after DOL 3 remains an elusive goal.