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 then included in a step-wise backwards elimination method of multiple regression analysis (ordinary least squares, OLS) to determine the best predictors of MIR per health region. Variables were dropped in order of least significance and if found to have high collinearity within the model, based on a Variance Inflation Score (VIF) of >10, until all remaining variables were significantly associated at p ≤ 0.05 .
The results of the OLS model determined which variables were included in a subsequent geographically-weighted regression (GWR) analysis. GWR is a regression modeling method that accounts for spatial structure, and can therefore produce a more robust model than OLS when spatial dependency is suspected (\cite{Ford_2016},\cite{Nakaya_2005}). The global R2 of the OLS and GWR models were compared to assess the model of fit, with a higher R2 indicating that a greater proportion of the variance in MIR was explained by the model. The Akaike Information Criterion (AIC) was also used to compare the two models, with a lower value indicating higher accuracy (\cite{me1996},\cite{a2009}).
All statistical tests were conducted in JMP version XX (XX). Calculation of Moran's I statistics, in addition to the GWR analyses were conducted using GeoDa software version XX (XX). Choropleth maps were generated using Tableau version XX (XX).
Saskatchewan – rates same in each region that were initially combined
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 were calculated for 111 health regions in Canada, for the 2010-2012 period, which revealed the lowest MIR in Ontario's York Regional Health Unit at .35, and the highest MIR in Nunavut at .88 (Figure 1). Overlaying the radiotherapy centers in Canada with MIRs per health region showed that the vast majority of radiotherapy centers in 2012 were located along the country's southern border.
Figure 1.
The global Moran's I for MIR was .35 (p = .001), indicating statistically significant spatial autocorrelation with a tendency towards clustering of health regions. Mapping of the clusters and further analysis using the LISA test revealed areas of significantly higher MIRs in all territories (Yukon = .67, Northwest Territories = .55, Nunavut = .88), the most northern health regions in Saskatchewan (Athabasca Health Authority = .51), Manitoba (Northern Regional Health Authority = .56), and Quebec (Région du Nord-du-Québec = .52, Région du Nunavik = .49) (Figure 2). The most northern regions in Ontario also showed significant clustering of high MIR (Porcupine Health Unit = .47, Thunder Bay District Health Unit = .45), with the exception of the Northwestern Health Unit. Health regions with significantly lower MIRs were seen in
Figure 2.
All-cancer MIRs by health region across Canada exhibited positive statistically significant global Moran’s I index values, with a tendency towards clustering (Moran’s I = .346, p = .001). Mapping of clusters showed areas of higher MIRs (range .45–.88) in all of Canada’s northern regions (Nunavut, Northwest Territories and Yukon), and in the north of certain provinces (Saskatchewan, Manitoba, Ontario and Quebec). Clusters with lower MIRs (range .35–.43) were observed in southern regions of British Columbia, Alberta, Saskatchewan, Ontario and in New Brunswick.
Discussion
Areas of highest population density are where RT centers are located.
--
Research in context panel
Evidence before this study: This section should include a description of all the evidence that the authors considered before undertaking this study. Authors should briefly state: the sources (databases, journal or book reference lists, etc) searched; the criteria used to include or exclude studies (including the exact start and end dates of the search), which should not be limited to English language publications; the search terms used; the quality (risk of bias) of that evidence; and the pooled estimate derived from metaanalysis of the evidence, if appropriate.
Added value of this study: Authors should describe here how their findings add value to the existing evidence.
Implications of all the available evidence: Authors should state the implications for practice or policy and future research of their study combined with existing evidence. Research in context panels should not contain references; key studies mentioned here should be referenced in the main text.