Bird, S. T.; and Bauman, K. E. Am J Public Health
, vol. 85, pgs. 26-9 (1995)OBJECTIVES. This study compared the relative strength of the associations of a set of structural (social, economic, and political) variables and a set of health services variables with state-level infant, neonatal, and postneonatal mortality. It also examined whether health services mediate the relationships between structural variables and state-level infant, neonatal, and postneonatal mortality. METHODS. With the state as the unit of analysis, data for all 50 states were analyzed by means of multiple regression. RESULTS. Structural variables accounted for substantially more variance in infant, neonatal, and postneonatal mortality than health services variables, and health services variables were more strongly related to infant mortality than to neonatal or postneonatal mortality. When health services variables were controlled, the strengths of the associations between the structural variables and infant, neonatal, and postneonatal mortality were reduced but remained statistically significant. CONCLUSIONS. A substantial portion of the variance in state-level infant mortality is accounted for by states' structural characteristics, which are partially mediated by health services.