Objective: To examine birth-weight-specific and age-specific mortality among US infants to determine if the large infant mortality decrease in 1990 was due to surfactant use.
Design: Population-based analysis of data from the 1983-1991 National Linked Birth and Infant Death files. Mortality trends from 1983 to 1989 were used to calculate expected infant mortality rates for 1990 to 1991.
Setting: United States. PARTICIPANTS AND STUDY POPULATION: All singleton infants with known birth weight born in the United States from 1983 to 1991.
Interventions: None.
Main outcome measures: Mortality at less than 1 day of life, 1 to 6 days, 7 to 27 days, or 28 to 364 days. Observed mortality rates were divided by the expected rates in 250-g birth-weight categories to create mortality ratios.
Results: The observed infant mortality rate in 1990 was 8.05, significantly lower than the expected rate of 8.36. Infants weighing 750 to 1749 g had mortality ratios of approximately 0.8 for 1- to 6-day mortality, with ratios significantly less than 1.0 for mortality in all age groups except less than 1 day. Observed mortality among infants weighing less than 750 g or from 1750 to 2499 g was not significantly lower than expected at any age. Postneonatal mortality among infants weighing 2500 g or more was significantly lower than expected. Infants weighing less than 1500 g accounted for almost 700 fewer infant deaths than predicted in 1990. Infants weighing 2500 g or more accounted for approximately 550 fewer deaths than expected.
Conclusions: The hypothesis that surfactant was partially responsible for the overall infant mortality drop in 1990 is supported by the lower than expected mortality among infants weighing 750 to 1749 g. However, the unexpected improvement in postneonatal mortality among infants weighing 2500 g or more was responsible for a substantial portion of the overall decline and suggests that other factors also acted to decrease US infant mortality in 1990.