Geographical variation in the prevalence of problematic substance use in Canada.

OBJECTIVE: The prevalence of substance-related problems has been shown to vary between Canadian provinces, but little else is known about the pattern of geographical differences. In this study, we modelled these differences, using methods of spatial analysis, and attempted to determine whether they are explained by known risk factors.
METHODS: We used data from Cycle 1.2 of the Canadian Community Health Survey. We tested interprovincial differences, before and after adjustment for covariates, and also examined differences between urban areas. We then used interpolation techniques to model variation in prevalence without reference to administrative boundaries. Finally, we performed a spatial cluster scan for areas of heightened prevalence.
RESULTS: The prevalence of problematic substance use is lower in Ontario and Quebec than in the rest of the country. This pattern is due principally to low prevalence in Toronto, Montreal, and surrounding areas. Prevalence is higher in mid-sized cities than in larger ones or in rural areas. Problematic substance use shows a fairly high degree of spatial clustering, especially within major cities. Interprovincial differences and clustering are generally not explained by known risk factors.
CONCLUSIONS: The pattern of large-scale differences is consistent with existing research and is probably part of a larger disparity among regions of Canada. The persistence of variation after adjustment for covariates suggests the influence of unmeasured, geographically varying factors, of which there are several candidates, including latitude and immigrant settlement patterns.