Objectives: To reconceptualize hospital length of stay (LOS) among NICU survivors using a model of gestational age and normative maturation in VLBW infants; to apply this model in an examination of risks for longer LOS and inter-hospital differences in LOS.

Methods: All <1500 gm first admission survivors to 7 NICUs were abstracted for 22 months and followed to discharge (n=1353). We calculated post-conceptional age at discharge (PCAD) based on discharge to home. Covariates in models predicting PCAD include: gestational age, Score for Neonatal Acute Physiology (SNAP), and small for gestational age (SGA). In a second model, variables included those above plus diagnosis of BPD(supplemental oxygen at 28 days). Hospital indicators were added in a third model.

Results: The average infant's PCAD was 36-40 weeks (SD averages± 3.3 for infants w/ BPD and ± 2.0 w/o BPD), with a LOS penalty for very early gestation (see Table of selected gestational ages). Independent of gestational age, multiple regressions indicate SGA adds 2.2 wks (P<0.001), and each 10 pts on SNAP scale adds 1.2 weeks (P<0.001) to PCAD. With BPD controlled, SGA and SNAP scores still independently predict PCAD; BPD adds 2.4 weeks (P<0.001). Independent of all covariates, two hospitals indicated excess PCAD (2.6 weeks; P<0.001) and 0.7 weeks; P<0.001).

Table 1
Full size table

Conclusions: Gestational age and normative maturation explains most variation in LOS in NICU survivors. Hence, future NICU LOS studies should also examine PCAD. Our analysis indicates that controlling for gestational age, SGA, SNAP scores, and BPD all independently predict PCAD, with excess LOS due to hospital variation also noted independent of these risks.