Infant mortality: A multi-level analysis of individual and community risk factors

This paper reports the results of an analysis of infant mortality based on a conceptual model that combines micro-level and macro-level variables taken from demographic, sociological and epidemiological research traditions. Using generalized hierarchical linear modeling techniques, we analyze 1988 and 1989 linked birth and death records for Upstate New York matched with county-level data from government and private sources. Net of health and sociodemographic risk factors, our results show that the number of per capita primary care physicians and local government expenditures on health care services and hospitals are positively linked to an increase in the probability of infant death and that our indicator of hospital facilities is negatively related to risk of death. We also find that some negative health behaviors and health resources of mothers are mediated by the local health care environment. Our results demonstrate the utility of combining perspectives from several disciplines when evaluating infant death, especially the impact of policy-related issues concerning health care service infrastructure.