Multivariate Bayesian meta‑analysis: joint modelling of multiple cancer types using summary statistics

We know that cancer is a big important problem in Australia, and we also know that certain cancers have markedly different rates between different areas across our country.

However, there currently isn’t an easy way to identify which areas have high (or low) rates of several different cancers together. This is important information, as there can be common risk factors between different types of cancers.

The study identified some differences in relationships among multiple cancers according to the remoteness of a region. Some of the cancer types are related positively. That is, the cancer types have similar risk ( high or low) in major cities of Australia but such patterns are not true in remote and regional areas of Australia. These types of findings create new opportunities for research to find out what is responsible for such discrepancies in different geographic areas and risk of multiple cancers.

This research provides a tool so that people can use readily available data from online cancer atlases, and understand patterns for multiple cancer types together.

The cancer incidence data used in this research can be downloaded from the Australian Cancer Atlas website: One of the co-authors of the publication is Professor Peter Baade, from Cancer Council Queensland.