Owing to the nature and definition of the data used, there were several study limitations, one of which was the use of mortality rather than incidence. The lack of information on non-lethal cancer cases may have served to bias the analysis. In the absence of a population-based incidence registry covering the entire country, we used mortality data.
