Statistical Methods in Populated-Based Cancer Survival Studies

The increasing quantity of detailed cancer related data with access to large, linked electronic databases has stimulated novel applied research. Advances in statistical methodology and software are not so rapid.

With more detailed data combined with greater understanding of when effects can be considered causal, there needs to be reconsideration of the most appropriate ways to analyse and report comparisons from population-based cancer studies.

This project will concentrate on survival analysis methods using large registry-based data.AIM 1: Rethinking how survival is estimated within population-based cancer studies through appropriate integration of causal inference methods.AIM 2: Improving the understanding and communication of risk in population-based cancer studies through further development of reference-adjusted measuresAIM 3: Transferal of methods into applied work through collaboration, software development, and education material.Aims 1 and 2 concentrate on methodological development.

Aim1 will develop causal inference methods within the relative survival framework for use in population-based cancer survival studies.

Aim 2 will further develop reference adjusted measures, an alternative approach to quantifying survival differences making all-cause and crude probabilities comparable between population groups.

Aim 3 will ensure that the novel methods are used in practice through leading and collaborating on applied studies, software development, and education material.