Statistical analyses
To explore the variables that predicted MIR per health region, we first conducted univariate analyses with distance to radiotherapy center and each sociodemographic variable as the independent variables, and MIR as the dependent variable. All significant (p ≤ 0.05) independent variables in the univariate analyses were included in a step-wise backwards elimination method of multiple regression analysis (ordinary least squares) to determine the best predictors of MIR per health region. However, the Variance Inflation Score (VIF) for the proportion of Aboriginals variable was >10 (12.0) at the onset of the regression analysis, indicating high multi-collinearity with the other sociodemographic variables in the model. Given that this was one of our variables of specific interest, we could not continue the regression analysis. 
For an alternative analysis to multiple regression, we used recursive partitioning (RPA) as an exploratory analysis to dichotomize our two variables of interest (proportion of Aboriginals and distance to radiotherapy centre) and to determine their impact on MIR. RPA is a method used to classify subjects and variables, and can be useful in identifying synergistic interactions among factors \cite{Cook1984}. In the medical context, it has been useful in determining prognostic and risk groups in cancer patients \cite{Chang_2015}, and in creating clinical algorithms for patient treatment \cite{Fonarow2005}. Although it is a robust method in analyzing multiple variables, we included only the two variables of interest as multi-collinearity would continue to be an issue \cite{Dormann_2012}. Once the groups were created, we conducted the Student's t statistic (assuming unequal variances due to unequal sample sizes) and one-way analyses of variance (ANOVA) to determine the impact of the proportion of Aboriginals and distance to radiotherapy on MIR individually and together, using non-parametric Wilcoxon Method comparisons due to a low sample size in some groups. 
All statistical tests were conducted in JMP version 12 (XX). Choropleth maps were generated using Tableau version 10.4 (XX). 
Saskatchewan – rates same in each region that were initially combined
Nova Scotia manipulated data for CCHS health regions (2013 vs 2015)
 
Role of the funding source
The funder had no role in the study design, collection, analysis or interpretation of the data, or writing of the report. JC had full access to all data used in the study, and the final responsibility to submit for publication.
 
RESULTS
All-cancer MIRs for the 2010-2012 period were calculated for the 112 Canadian health regions. MIRs ranged from .35 in Ontario's York Regional Health Unit, to .88 in Nunavut (Figure 1). Ontario's York Regional Health Unit also had the lowest proportion of self-identified Aboriginals in 2011 at .4%, and the highest proportion was in Quebec's Région des Terres-Cries-de-la-Baie-James at 96%. Overlaying the Canadian radiotherapy centers showed that the vast majority of radiotherapy centers in 2012 were located along the country's southern border, with a similar visual geographic pattern of health regions with higher MIR also having a higher proportion of self-identified Aboriginals, both being further away from radiotherapy facilities.