Predictors of infant mortality among college-educated black and white women, Davidson County, Tennessee, 1990-1994.

Strategies to reduce US infant mortality rates often focus on the black-white disparity in rates. Linked Infant Birth and Death Files for Davidson County, Tennessee, from 1990 through 1994 were used to determine infant outcomes for infants born to college-educated white and black women. Risks for adverse outcomes were identified by comparing infant deaths to live births using logistic regression analyses. The following variables entered the logistic model process: maternal and paternal age; race and education; nativity status; maternal risk factors; interpregnancy interval; parity; infant gender; tobacco or alcohol use; number of prenatal visits; trimester in which prenatal care began; marital status; gestational age; and birthweight. After adjustment for the effects of the other variables, a gestational age < 28 completed weeks of gestation was the most significant independent predictor of infant death. Black race was not identified as a significant predictor of infant mortality. Regardless of race, a decrease in infant mortality rates among college-educated women in this country depends on the prevention of preterm births. Strategies to diagnose early preterm labor must proceed from a comprehensive maternal care program for all women. Open channels of communication between patient and provider will form the cornerstone for preterm prevention-intervention programs. Analysis of state and local infant mortality data may identify regional differences in infant mortality rates and differences in risk factors associated with adverse infant outcomes.