473 PRENATAL SELECTION BIAS IN THE EVALUATION OF NEONATAL INTENSIVE CARE (NIC)

Evaluation of NIC of very low birth weight infants (VLBW) has been based largely on infants treated in referral centers. The characteristics of the population of infants from which the treated infants come are unknown. To address the issue of maternal residence, all VLBW (500-1499g) livebirths in Hamilton-Wentworth County (1974 pop. 406,900) from 1964-69 and 1973-77 (before and after the introduction of the McMaster Regional Perinatal Program) were reviewed. The newborn mortality at final discharge by maternal residence was as follows:Mortality rates decreased between the 2 time periods from .582 to .397, a difference of .185. Adjusting for maternal residence by applying 1964-69 mortality rates to the number of births in 1973-77 results in an expected mortality rate of .531. 28% (.051/.185) of the decrease in mortality is “explained” by maternal residence. Maternal selection may introduce a bias in the evaluation of NIC of VLBW infants.