U.S. Models of Infant Mortality by Race/Ethnicity: One Size Fits All?

Samuel Echevarria-Cruz, Ohio State University

Infant mortality research has historically utilized categorical variations of its two most important proximate determinants: birth weight and gestational age. In addition, these categorical approaches were primarily based on data from the majority population of the U.S. It has also been difficult to ascertain the relative importance of effects between birth weight and gestation age on infant mortality. Utilizing linked birth/infant death vital statistics files for the U.S., we standardize and model continuous birth weight, continuous gestational age and infant mortality by utilizing nonparametric regression strategies (GAM) combined with race/ethnic-specific standards for three groups: Non-Hispanic Whites, Non-Hispanic Blacks and Mexican Americans. We find that 1) All groups exhibit differing standards regarding birth weight effects on infant mortality; 2) Utilizing race/ethnic standards alters the effects of birth weight on infant mortality; and 3) Standardized birth weight and standardized gestational age are relatively equal in their effect on infant mortality.