Spatial Non-Stationarity Effects of Unhealthy Food Environments and Green Spaces for Type-2 Diabetes in Toronto
My Session Status
The rising number of diabetes cases worldwide is emerging as one of the fastest-growing global healthcare emergencies of the 21st century [1]. Using Toronto, Ontario, Canada, as an example, more than one in ten adults was diagnosed with Type 2 Diabetes mellitus (T2DM) [2]. Physical inactivity and unhealthy food environments with energy-dense food sources were commonly found to be the significant predictors of diabetes prevalence [3]. People are more likely to exercise in a well-designed environment, reducing the risk of diabetes [4]. On the other hand, having supermarkets with fresh food sources around neighborhoods promote a balanced diet, yet unhealthy food services increase the risk of diabetes [5]. However, environmental factors will affect dissimilarly at different geographic locations when relationships are measured across space, so-called spatial non-stationarity [6,7]. Previous research often assumed that the relationships between environmental factors and health outcomes were stable or stationary over space. Such presumptions may result in biased and incorrect conclusions about environmental impacts on health outcomes [7]. Recognizing non-stationarity effects helps to investigate ignored factors, improve our understanding of the spatial phenomena of T2DM, and provide suggestions for allocations of management and prevention resources [8]. As a result, it is crucial to control and reduce the risk of T2DM prevalence by understanding the variations of spatial associations between T2DM and the environment to design targeted prevention programs.
This study investigated the spatial non-stationarity effects of unhealthy food environments and green spaces on the T2DM prevalence rate at the neighborhood level in Toronto. This study also compared how the results vary between different age groups: adults (20 and above), young (from 20 to 44), middle (from 45 to 64), and seniors (65 and above). To fill the research gaps, this study employed the Geographically Weighted Regression (GWR) model to examine the spatial non-stationarity associations between contextual variables and the T2DM prevalence rate for all neighborhoods in Toronto. Unhealthy food outlets and green space densities were included in the GWR as independent variables. In addition to environmental characteristics, socioeconomic status also significantly affects the T2DM prevalence rate. Various studies showed that education and income were negatively correlated with T2DM [9–12]. Specific immigrant groups, such as South Asians and black immigrants, were reported to have a higher chance of developing T2DM earlier in their lifetime in Canada [13,14]. As a result, this study included socioeconomic status in the GWR analysis as independent control variables, including medium annual income, unemployment rate, low education rate, and immigration rate.
The results from this study reveal that environmental variable dissimilarly affected T2DM prevalence rates among different age groups and neighborhoods in Toronto after controlling for socioeconomic status. For example, green space and unhealthy food outlet density yielded positive associations with diabetes prevalence rates for elder generations but had negative relationships for younger age groups around Toronto East. The observed associations will provide beneficial suggestions to support government and public health authorities in designing education, prevention, and intervention programs targeting different neighborhoods to control the burden of diabetes.