Spatial Analysis of Geographic Variation in Mental Health Visits and Its Association with Social and Built Environment in Toronto Neighbourhoods
MetadataShow full item record
Introduction: Mental health is a growing concern in Canada. Existing studies that examine mental health related factors generally focus on individual-level characteristics, which often neglect contextual and spatial effects. This study explores the geographic variation in mental health visits (MHV) in Toronto and identifies the social and built environment factors associated with MHV at the neighbourhood level adopting spatial analytical methods. Methods: MHV are defined as individuals aged 20+ having had a mental health and addictions related primary care visit according to physicians’ billing claims during the 2011 and 2012 fiscal years. MHV data were retrieved from the Toronto Community Health Profiles; social and built environment factors derived from various original data sources were obtained from the Toronto Community Health Profiles and Toronto Open Data. The Global Moran’s I Statistic and Kulldorff’s Spatial Scan Statistic were applied to evaluate the overall geographic variation in MHV and detect the locations of high and low risk clusters for MHV, respectively. This study quantified the effects of social and built environment on MHV fitting two spatial regression models, the spatial error model and the spatial lag model. All-subset selection using BIC as the selection criterion was employed as an ancillary tool to help determine which factors are most important to the relationships between social and built environment and MHV. Results: Overall, the geographic distribution of MHV exhibited a clustering pattern, and the locations of hot and cold spots for MHV were further identified and visualized in Toronto neighbourhoods. Two social factors and two built environment factors were identified as the most salient factors affecting MHV. Income inequality and the proportion of households in need of major repairs were associated with increased MHV, while the proportion of East Asian residents and the number of health providers per 10,000 residents were negatively correlated with MHV. The spatial regression models showed superior performance compared to the non-spatial OLS model, and the spatial lag model provided the best model fit as indicated by BIC. Conclusions: This study indicates that both social and built environment factors can contribute to variation of population mental health. The results can provide useful strategy basis for both locally tailored and general population mental health promotion programs. The cluster maps that visualized specific areas of high mental health concern can be utilized to target neighbourhoods in need of more focused investigations and mental health initiatives. Stakeholders may develop appropriate campaigns that serve to improve mental health in neighbourhoods with high levels of income inequality and deliver culturally tailored mental health services in East Asian communities. The findings also point to the need to improve housing quality and supply of general healthcare providers for addressing population mental health problems. Limitations related to data, the modifiable areal unit problem, and ecological fallacy are also discussed. Future studies can conduct attitude surveys among Toronto residents to gain better understandings of neighbourhood mental health.
Cite this version of the work
Yujie Yang (2019). Spatial Analysis of Geographic Variation in Mental Health Visits and Its Association with Social and Built Environment in Toronto Neighbourhoods. UWSpace. http://hdl.handle.net/10012/14848