Descriptive statistics for each independent variable considered in the regression analysis are presented in Table 1. In univariate analyses, distance to radiotherapy and each sociodemographic variable were significantly associated with MIR (p < .05), except for physical activity (% moderately active or active) (Table 1). In the OLS regression model, greater distance to radiotherapy, poorer food security and higher smoking rates were still found to be statistically significant predictors of increased MIR (R2 =.70, F = 81.50, p < .01, Akaike Information Criterion = -415.37), and all exhibited a VIF of less than 2. These variables remained significant in the GWR model and resulted in an improved fit when accounting for spatial heterogeneity (R2 =.74, Akaike Information Criterion = -426.47) (Table 2). Note that for both regression analyses, the number of health regions included was 110 due to the exclusion of two health regions in Quebec (Région du Nunavik and Région des Terres-Cries-de-la-Baie-James ) from the CCHS.