301 Approaches to ethnic group classification based on Census ancestry data for examination of health inequalities in Australia

OP 13: Research Methods, B304 (FCSH), September 3, 2025, 17:00 - 18:00 Aims Ethnicity is not collected routinely in Australian health data despite increasing ethnic diversity of the Australian-born population. Ancestry data is collected in the Census, but this data has two responses and enormous granularity and is difficult for researchers to use. We aimed to work with an expert and community panel to develop an approach to using ancestry data to examine health inequalities. Methods An expert panel was formed including representatives from the Federation of Ethnic Communities’ Councils (FECCA), Australian Bureau of Statistics (ABS), and researchers. The work was also informed by a community panel and yarning circles led by Aboriginal and Torres Strait Islander academics. We analysed 2016 Census data linked to death registrations for 2016-2021 in 20.9 million people and compared how different approaches to categorising ethnicity impacted on detection of health inequalities between groups. Results We developed the following approach: (1) Recategorise those reporting Australian only or New Zealand only ancestry from the Oceania to the European category; (2) For those reporting two ancestries: prioritise ethnic minority over national identities (e.g. Australian) and for the remainder create multiethnic groups based on the combined ancestries; (3) First analyse data at the maximum level of granularity (most detailed ethnicity categories) possible, but then collapse up granularity where no health inequalities are identified to simplify results and messaging. Analysis of mortality data demonstrated that these approaches improved identification of mortality inequalities by ethnicity. Conclusions We used an iterative consensus approach to develop a set of principles for how Census Ancestry data can be used to identify ethnic inequalities in health. The ability to link health datasets to the Census through the Person Linked Integrated Data Asset (PLIDA), mean that such recommendations are important to be able to fully harness the value of Census data in identifying inequalities.