An ecological analysis of neighborhood effects on health outcomes: The case of excess mortality in an urban area

Census and vital records data for 328 urban census tracts are used to understand the relationship between ecological characteristics and disparities in survival. The central thesis of this study, drawing upon ecological theory, is that mortality differentials among area neighborhoods is subject to the influence of four factors: ecological impoverishment, low occupational structure, population instability, and lack of success. Despite obvious similarities, neighborhoods are perceived to have unique structural characteristics that result in differential responses to threats and challenges. The test of this model is based on census and vital records data for Cuyahoga County for ages under 65 years for the period 1989-1991.;Three questions are posed for this analysis: (a) To what extent do variations in excess mortality correspond to differences in ecological conditions of neighborhoods? (b) What is the relative importance of dimensions of ecological conditions in explaining spatial variations in excess mortality? and (c) Does an ecological model fit equally well for all the subareas in the analysis?;Descriptive analyses reveal that excess mortality is a widely dispersed phenomenon, and more than 60 percent of the tracts experienced higher mortality than was expected. Tracts with extreme poverty (over 40%) and with longer periods of concentrated poverty exhibited higher level of excess mortality. Pockets of excess mortality are almost exclusively concentrated within the confines of the city of Cleveland.;Multiple regression analysis explained 64% of the variance in excess mortality. Ecological impoverishment is shown to be the single most important variable followed by low occupational structure. The findings also showed significant interaction effects between impoverishment and access, and instability and access on excess mortality. The slope of excess mortality on access at the highest level of impoverishment and instability did not attain statistical significance. The implications of this finding are discussed.;The results of this study suggest that the main targets of policy and program intervention should be the structural correlates that are most predictive of poor health outcomes in neighborhoods.