Health & GIS Article Review:
Urban housing and its relationship to health
Purpose
Badland et al. (2017) used spatial analysis to further explore how adequate and affordable housing can be a major determinant of health. This study attempted to conceptually map and spatially test area-level measures of housing with selected health and wellbeing outcomes. The authors wanted to demonstrate how “a suite of housing ‘liveability’ indicators could be developed, applied, and monitored in the future to create communities that support health and well-being” (p. 18 Badland et al. 2017). Their study site was in Melbourne, Australia and their study sample was composed of 7,753 adults who resided in the urban area (those living in non-urban and non-metropolitan areas were excluded).
Methods/Spatial Analysis
The authors addressed three questions in this study; “1) conceptualise the range of pathways through which housing in urban settings impact health and wellbeing behaviours and outcomes; 2) spatially operationalize measures that map to these pathways; and 3) test associations for how selected housing measures to relate to health and well-being in an urban context” (p. 18).
To conceptualize pathways on how housing impacts health, Badland et al. (2017) developed a housing conceptual framework that considered upstream neighbourhood attributes and downstream (perceptions, behaviours, outcomes) determinants of health. They hypothesized three models of causal and response variables: housing density reflects fear of crime, community satisfaction and mental and physical health; housing tenure reflects community satisfaction and mental and physical health, and affordable housing reflects mental and physical health. Simply put, Badland et al. (2017) predicted that housing density, housing tenure, and affordable housing influenced health and wellbeing.
They collected the neighbourhood spatial data from the Australian Bureau of Statistics (ABS) using the SA1 administrative unit, which group populations into 200-800 people, similar to Canadian dissemination areas. Badland et al. (2017) divided up each measure into quartiles in order to determine if “there were ‘thresholds’ in the associations between the housing measures and the outcome variables” (p. 19) also to comply with their regression models reliance on similar distributions amongst the groups.
Demographic, perception, behaviour, and outcome data were acquired through a survey that included residential addresses allowing the authors to geocode the data. Badland et al. (2017) used ArcGIS to spatially join the survey respondents answers to the SA1’s for those living in urban areas.
The authors then used a multivariate multilevel regression models conducted in Stata IC v.13.1. to test the associations between the spatially derived housing measures and outcomes. They used this type of regression model because they wanted to carry out a single test of the joint effect of an explanatory variable on several dependent variables. This model also allows conclusions to be drawn on the extent to which the correlations depend on the individual and on the group level.
Results
By looking at the significant associations between variables, Badland et al. (2017) found that overall, “compared with the reference categories, as dwelling, proportion of rental properties, and housing unaffordability increased, the odds of reporting poorer self-related health also increased” (p. 21). I rate this paper a 7/10 due to the lack of visual maps. I also believe this approach was far too simplistic to gain a clear understanding of health in relation to neighbourhood spatial measures. Although the authors did state that they were aiming for simplicity so that urban planners could apply this reasoning to planning a more liveable city.
Reference
Badland, H., Foster, S., Bentley, R., Higgs, C., Roberts, R., Pettit, C., & Giles-Corti, B. (2017). Examining associations between area-level spatial measures of housing with selected health and wellbeing behaviours and outcomes in an urban context. Health & Place, 43, 17-24. doi:10.1016/j.healthplace.2016.11.003