PHSW02 - Where you live matters: improving geographic estimates of cancer risk across Australia

Overview

Cancer registries and other health datasets in Australia are often recorded using postcode as the geographic unit. However, postcodes are not well-suited to spatial analysis because they vary greatly in size and can group together people from very different communities. The Australian Bureau of Statistics (ABS) has designed a hierarchy of Statistical Areas, of which Statistical Area Level 2 (SA2) is a common small area unit for health research in Australia. SA3 and SA4 represent progressively coarser levels of the same hierarchy. This project addresses the practical problem of estimating cancer risk at these preferred geographic units when the underlying data are recorded at the postcode level. The ABS provides correspondence files describing the proportion of each postcode's population falling within each SA2, providing a basis for reassigning cases across spatial units. The project is connected to ongoing collaborative work with Cancer Council Queensland and is aligned with the Australian Cancer Atlas (https://atlas.cancer.org.au/), a national resource mapping cancer risk across Australia. The project sits at the intersection of biostatistics, spatial epidemiology, and public health.

Research engagement

The student will review relevant literature on spatial misalignment in health geography, examine how ABS Statistical Area hierarchies and correspondence files work in practice, and explore how uncertainty arises when reassigning counts across geographic boundaries. They will be introduced to Bayesian statistical concepts at an accessible level with supervisor support.

Research activities

The student will work with academic supervisors in the School of Public Health and Social Work at QUT and will have access to expertise from Cancer Council Queensland. Activities include summarising relevant methodological literature, implementing weighted aggregation methods to estimate cancer risk at SA3 and SA4 levels from SA2-level data, applying bootstrapping to quantify uncertainty in these estimates, and comparing results across aggregation strategies. Where time permits, the student may explore a model-based approach to the postcode-to-SA2 misalignment problem. All coding will be done using R software (https://www.r-project.org/).

Research skills

The student will develop practical skills in spatial data handling in R, working with administrative geographic data and ABS correspondence files, uncertainty quantification using resampling methods, and scientific writing. They will be introduced to core concepts in Bayesian statistics and spatial epidemiology and will gain experience working in a collaborative research environment with academic and non-academic partners.

Outcomes

The primary aim is to produce a well-documented reproducible R workflow that takes postcode-level cancer data and produces estimates of cancer risk at SA2, SA3, and SA4 levels with associated uncertainty measures. A secondary aim is to clearly document the assumptions underlying each step so that end users understand the limitations of the estimates. Outputs will contribute to cancer surveillance work with Cancer Council Queensland and may form the basis of a technical report or journal paper.

Skills and experience

Students should have completed intermediate statistics coursework including regression modelling and some exposure to statistical inference. Experience with R is desirable but not essential. No prior knowledge of Bayesian statistics or spatial analysis is required. An ideal candidate will have an interest in public health or population health research and be comfortable working with data. Students from statistics, mathematics, public health, or related quantitative disciplines are encouraged to apply.

Start date

2 November, 2026

End date

19 February, 2027

Location

QUT, Kelvin Grove Campus

Additional information

The student will have access to academic supervisors with expertise in biostatistics and spatial epidemiology, R software and relevant packages, ABS correspondence files and geographic data, and guidance from Cancer Council Queensland collaborators. Regular supervision meetings will be scheduled throughout the project.

Keywords

• spatial epidemiology
• public health
• cancer surveillance
• quantitative analysis
• statistics
• bayesian statistics
• cancer risk

Contact

Associate Professor Darren Wraith

0731380863

d.wraith@qut.edu.au