For the Australian Cancer Atlas, we used Bayesian statistical models. This method identifies a range of plausible values to describe the unknown quantities of interest. For the Atlas, these quantities include the cancer diagnosis rates, survival rates within 5 years of diagnosis, screening participation rates, prostate cancer-related hospital treatment rates and proportions of risk factors in each small area. Rather than calculating one specific estimate for each area, the Bayesian statistical models generate a distribution of possible estimates using the observed counts, population data and other (prior) information.
One assumption with these models is that the factors that influence the cancer burden in one geographical area are more similar to those in areas closer to them, than to areas further away. In this way, the analysis enables us to supplement the data observed in each area, which leads to the more stable and robust ‘spatially smoothed estimates’.
The distribution of possible estimates that are obtained for each area reflect the uncertainty of these estimates and enables more appropriate comparisons to be made between areas or with the Australian average.
