skip to primary navigation skip to content

Department of Geography

 

A study of spatial variation in health and human development

Research Team: J. Law with CRISP

Abstract

Findings from the pilot projects of the Canadian Population Health Initiative indicate that there is wide variation among Canadian cities in their mortality rates, and among health regions and neighbourhoods in the number of health problems reported by adults. Thus, living in certain provinces, cities, or regions within Canada is itself a significant risk factor for various health problems. The primary substantive interest of the research is in determining the economic, social and historical factors affecting people’s health and well-being. It pays particular attention to the socioeconomic gradient; that is, the relationship between social outcomes (e.g., children’s health, adult literacy, morbidity and mortality) and socioeconomic status (SES) (i.e., a composite index describing people’s access to and control over wealth, prestige and power). The over-arching goal is to identify factors that improve social outcomes and reduce inequities. A composite index of SES at the Enumeration Area (EA) level was developed. This entails factor analysis of census-derived variables, such as median income, level of education, and unemployment rates. The research developed a computer program in ArcView that incorporates spatial data analysis pertaining to the “smoothing” of data. Further research would look explore the use of multilevel models and spatial regression for studying the factors that may be associated with one’s health outcomes.

Illustrations

Image as described adjacent

Figure 1: Income in New Brunswick: smoothing and interpolation of missing data based on the income values of a census area and its neighbours.

Image as described adjacent

Figure 2: Ordinary Least Squares Estimation – Physical health well-being (PHWB) is positively related to Socio-economic status (SES): The residual standard deviation map (negative residuals, or overprediction in blue; positive residuals, or underprediction in red) would suggest the presence of spatial autocorrelation from visual inspection.