Differences Between Infant Mortality Data from the Linked Birth Infant Death File

The linked files include information from the birth certificate such as age, race, and Hispanic origin of the parents, birthweight, period of gestation, plurality, prenatal care usage, maternal education, live birth order, marital status, and maternal smoking, in addition to information from the death certificate such as age at death and underlying and multiple cause of death. The linked birth and infant death data set is a valuable tool for monitoring and exploring the complex interrelationships between infant death and risk factors during pregnancy or presented at birth.

Although the time periods are the same, numbers of infant deaths and infant mortality rates in the period linked files are not identical to the numbers in the 1996 vital statistics mortality file. Although the overall infant mortality rate of 7.3 for 1996 is the same between the two data sources, infant mortality rates by characteristics, such as race, Hispanic origin and cause of death are not identical. Differences in numbers of infant deaths between the two files can be traced to three different causes:
1. geographic coverage differences,
2. additional quality control, and
3. weighting.

Differences in geographic coverage are due to the fact that for the vital statistics mortality file, all deaths occurring in the 50 States and the District of Columbia are included regardless of the place of birth of the infant. In contrast, to be included in the linked file, both the birth and death must occur in the 50 States and the District of Columbia. Also, the linkage process subjects infant death records to an additional round of quality control review. Every year a few records are voided from the file at this stage because they are found to be fetal deaths, deaths at ages over 1 year, or duplicate death certificates. Finally, although every effort has been made to design weights that will accurately reflect the distribution of deaths by characteristics, weighting may contribute to small differences in numbers and rates by specific variables between these two data sets.

This page last reviewed January 11, 2007