Aim: To determine the relation between the level of initial neonatal care and in-hospital mortality of infants born before 33 wk of gestation in the era of surfactant therapy.
Methods: A 1 y prospective population-based survey was conducted in the north of France, as part of the EPIPAGE (Epidemiologie des Petits Ages Gestationnels) survey. Perinatal data were recorded for 585 very premature newborns transferred to a neonatal intensive care unit in 1997. The relation between the level of the neonatal unit that provided care for the first consecutive 48 h and in-hospital mortality was assessed by multivariate logistic regression, and adjusted for perinatal data and initial disease severity, estimated by the Clinical Risk Index for Babies (CRIB).
Results: The average gestational age (mean +/- SD) was 31.6 +/- 0.62 wk in level I, 30.7 +/- 0.21 in level II, 29.9 +/- 0.13 in non-teaching level III, and 29.0 +/- 0.15 in the level III teaching unit (p < 0.0001). The mean in-hospital mortality rate was 8.4% and did not differ by level of care (p(trend) = 0.17). After adjustment for perinatal data and CRIB, however, with the teaching unit as the reference, the risk of death was significantly higher in level I-II units [adjusted odds ratio (ORa) = 7.9, 95% confidence interval (95% CI) 2.2-29.1], but not in the non-teaching level III units (ORa = 0.8, 95% CI 0.3-2.1).
Conclusion: In-hospital mortality in non-teaching level III units was similar to that in a teaching unit, but significantly higher in level I-level II units. Neonatal care of newborns delivered before 33 wk of gestation should initially occur in level III units.