To analyze the recent accelerated decline in infant mortality for three race/ethnicity designations in New York City and to investigate whether shifts in birth weight distribution or changes in birth weight-specific death rates were more important in determining these declines between 1988 to 1989 and 1992 to 1993.Methods
Two complete cohorts of linked birth-death certificate files consisting of all live births in New York City in 1988 to 1989 and 1992 to 1993 were examined. For each cohort, separate multinomial logistic regressions were estimated by race/ethnicity to analyze the probability of a neonatal or postneonatal death relative to survival as a function of a spectrum of covariates. The coefficients from these regressions were used to construct direct and indirect standardization exercises to predict changes in infant mortality holding characteristics of the cohort, including birth weight distribution, constant over time, or holding the influence of determinants, including birth weight-specific death rates, constant over time.Results
For whites, Hispanics, and blacks, infant mortality rates declined by 27.4%, 24.8%, and 22.7%, respectively, between 1988 to 1989 and 1992 to 1993. For whites and blacks, the largest decreases occurred for neonatal mortality rates, whereas for Hispanics, postneonatal rates fell the greatest. Although infant mortality rates among very low birth weight infants (<1500 g) fell by 27.8%, 19.3%, and 16.6% for whites, Hispanics, and blacks, the greatest decreases in rates were seen among normal birth weight infants (>2500 g). Infant mortality rate declines for this category of infants reached 31%, 31.7%, and 31.3%, respectively, for whites, Hispanics, and blacks. Direct and indirect standardization exercises indicated that the most important factor in determining these declines were decreases in birth weight-specific death rates, not improvements in the birth weight distribution over time.Conclusions
We conclude that the large decreases in infant mortality rates witnessed in New York City between 1988 to 1989 and 1992 to 1993 were attributable not to improvements in birth weight distribution of the population but to declines in birth weight-specific death rates and that normal birth weight infants showed the greatest improvement. Pediatrics 1998;101:682-688.