Indirect standardization of neonatal mortality for birth weight

Recent advances in perinatal medicine, together with the proliferation of regionalized perinatal networks, have increased the need for monitoring of trends and variations in birth weight-specific perinatal mortality rates. Since the United states has no national system of linked birth and death records, the only available method for comparing weight-specific mortality among geographic areas or medical care facilities in through indirect standardization of mortality for birth weight. This paper investigates the characteristics of the Standardized Mortality Ratio (SMR) obtained by indirect standardization using 1974-77 neonatal mortality and birth weight data for 202 geographic areas across the United States. The results show that the SMR is not sensitive to the choice of standard rates even when the area's birth weight-specific mortality rates are not a constant multiple of standard rates. A method for estimating the standard error of the SMR without knowledge of the weight-specific rates is presented and shown to be nearly unbiased. There is also a substantial amount of variation among geographic areas in both neonatal mortality rates, SMRs, and birth weight distributions. Thus, despite certain limitations, the SMR provides a useful summary measure for monitoring trends and variations among geographic areas or medical care facilities in birth outcome.