A multivariate assessment of the effects of residence on infant mortality

This research examines the relationship between residence and infant mortality. The purpose of the study was to identify the effects of maternal residence on infant mortality, using a multivariate model which included both individual and county-level variables known to be associated with suboptimal birth outcome. Data on all births in Florida during 1987 were drawn from birth and infant death certificates. In addition, information concerning county sociodemographic structure and medical resources were gathered and linked to the individual records. After examining the distributions of selected risk variables across a five-category measure of residence (from most urban to most rural), a logit model was estimated to predict the odds of an infant death associated with maternal residence. At the bivariate level, rural residents were found to have increased odds of an infant death compared to residents of all other residence categories. Second, a logit model was estimated that controlled for the influence of important maternal, infant, and county risk characteristics. The results of this second, more fully specified model indicate that residence did not have an independent direct effect on infant mortality when the influence of the other risk factors was controlled. We conclude that although residence does not influence infant mortality directly, it does influence mortality indirectly through its association with key risk factors. In particular, because population characteristics and medical resources are differentially distributed across rural and urban areas, residence remains an important factor to be considered when predicting health outcomes. The implications of these findings for policy-makers and health planners, as well as for health services researchers, are also discussed.