Survival and place of delivery following preterm birth: 1994-96.

AIM To compare the survival of premature infants, adjusted for disease severity, in different types of neonatal intensive care setting. METHODS A prospective observational study in the Trent Health Region was carried out of all infants born to resident mothers at or before 32 weeks of gestation between 1 January 1994 to 31 December 1996 inclusive. The 16 neonatal units in Trent were subdivided into five relatively large units which regularly took outside referrals and 11 smaller units which provided intensive care for a variable proportion (sometimes nearly 100%) of their local population. Data regarding obstetric management, neonatal care, and outcome were collected by independent neonatal nurses who visited the units on a regular basis. Survival rates were compared with an expected rate calculated using the Clinical Risk Index for Babies (CRIB). For either setting to be abnormally good or bad actual deaths had to exceed the 95% confidence interval of the CRIB estimate. RESULTS Actual survival rates for infants < or = 32 weeks gestation and for the group of babies < or = 28 weeks gestation fell within the 95% confidence interval of the rate predicted by CRIB for both the larger referral units and the smaller district units. Similarly, compared with the CRIB prediction, infants transferred in utero or postnatally were not adversely affected in terms of the number who died. CONCLUSION Previous results from this geographical population, showing that survival of babies < or = 28 weeks gestation was better when their care was provided by referral units, are no longer sustained. Significant changes to the neonatal services over time make the current results plausible. However, the new structure poses potential threats to the teaching, training, and research base of the neonatal service as a whole.