In univariate analyses using these new categories, health regions inhabited by < 23% versus ≥ 23% of Aboriginals had significantly lower MIR (0.42 versus 0.53, respectively; t(14) = -3.64, p = .003) and shorter Euclidean distance to nearest radiotherapy centre (121km vs. 799km, respectively; t(14) = -4.63, p = .0004). The one-way ANOVA including both proportion of Aboriginals and distance to nearest radiotherapy centre was significant (F(1,111) = 33.07, p < .0001). MIRs in health regions inhabited by ≥ 23% Aboriginals were not significantly different based on distance (≥ 922 versus < 922 kilometres: Z = 1.41, p = .158). However, both groups had significantly higher MIRs compared to health regions with < 23% Aboriginals, regardless of whether they were greater or less than 37 kilometres away (≥ 23% Aboriginals and ≥ 922 kilometres versus < 23% Aboriginals and < 37 kilometres: Z = 3.48, p = .0005 ; ≥ 23% Aboriginals and ≥ 922 kilometres versus < 23% Aboriginals and ≥ 37 kilometres: Z = -3.52, p = .0004; ≥ 23% Aboriginals and < 922 kilometres versus < 23% Aboriginals and < 37 kilometres: Z = 3.88, p = .0001; ≥ 23% Aboriginals and < 922 kilometres versus < 23% Aboriginals and ≥ 37 kilometres: Z = 3.27, p = .001). Health regions with < 23% Aboriginals located ≥ 37 kilometres away from nearest radiotherapy centre also had significantly higher MIRs compared to those < 37 kilometres away (Z = 3.15, p = .002).
DISCUSSION
MIR is high among health regions with more Aboriginal population. Reasons for this: sociodemographic factors. Collinearity = all variables play a role in predicting outcome.
Areas of highest population density are where RT centers are located, furthest from health regions with highest indigenous population. However, distance to RT was not significant in predicting MIR, meaning while for other residents if they live furt
RPA -- hard to split further for >20% group because few regions where highly populated with Aboriginals and live closer to radiotherapy centre..
Talk about other studies with their cut-off points. 34km is small but other studies have shown same thing. Liu lung cancer shorter times - this looked at all cancer, so hard to say.
Limitations: ecological fallacy. 2012 data - New centers built, would be interesting to see if things have changed since then. CCHS indicators did not include two of the populous indigenous health regions in Quebec, and excluded on-reserve Aboriginal populations, but this still includes the off-reserve populations and CCHS variables not used anyway due to high multicollinearity, speaking to the point that its the common sociodemographic factors that affect the region that are more important...? n = 5. health region centroid estimate.