Modelling Chronic Heart Disease (CHD) mortality in Australia by geography, Indigenousness and SES groups

Primary Supervisor: Dr Haider Mannan (opens in a new window)

CHD is the leading single cause of premature death in Australia. In 2016, it was the underlying cause of death for males, accounting for 19,800 or 12% of all deaths (AIHW, 2014). People who live in regional and remote areas and in lower SES groups generally experience poorer health than their major cities and this reflects by the measures of mortality (Stevenson et al., 2012; AIHW web report, 2018]. This is supported by the findings that in 2015 CVD mortality in remote and very remote areas were 30% higher than those living in major cities, 40% higher in the lowest SES group compared to the highest SES group and 90% higher among Indigenous Australians than non-Indigenous Australians (AIHW web report, 2018). Although the CHD death rates vary across the states and territories to some extent (AIHW, 2008; Wilkinson et al., 2000] being particularly high at Northern Territory (Wilkinson et al., 2000), these could be due to a large number of drivers including lower socioeconomic status, poorer risk factor profiles, differential access to health services and the high proportion of indigenous people in areas with higher CHD death rates (AIHW, 2008). However, there are greater variations in CHD mortality within different states and territories particularly up to 2-fold among different statistical divisions suggesting an inequitable distribution of the determinants of CHD mortality (Wilkinson et al., 2000). Also, there are wide variations in CVD mortality by remoteness and to a lesser extent by SES groups (Jacobs et al., 2018). Thus, small-area surveillance of CHD or CVD mortality is important because it can reveal patterns that are masked at the population level.

The ultimate objective of this study is to apply repeated measures multilevel modelling under a Bayesian (Khana et al., 2018) framework (because of small area surveillance of CHD mortality), to study regional and between-group (eg., SES groups and Indigenous/non-Indigenous groups) variations in the temporal association between selected social and economic determinants of health, classical CHD risk factors, sedentary behaviour or physical activity and/or non-traditional CHD risk factors, and CHD mortality. To our knowledge there has been no research on this topic in Australia and hence it deserves attention.
This study will address specific epidemiological and methodological questions that could be explored, such as:

  • Clustering of individuals by statistical divisions, remoteness, Indigenous status, SES groups, based on CHD mortality
  • The relative importance of social and economic determinants compared to CHD risk factors
  • Whether sedentary behaviour and physical activity follow different causal/biological pathways in influencing CHD mortality
  • The role of non-traditional health risk factors in predicting CHD mortality
  • The above three objectives are in context of spatio-temporal modelling to determine the predictors of CHD mortality
  • Methodological issues pertaining to predicting CHD mortality in the presence of missing data in the health risk factors and/or social determinants of health

The proposed dataset for the study is the Australian Diabetes, Obesity and Lifestyle study (AusDiab). AusDiab is the first longitudinal Australian population-based study established to examine the prevalence and incidence of diabetes and its complications, as well as heart disease and kidney disease. The baseline AusDiab study (1999-2000) surveyed the ge~~neral Australian population aged 25 years and over residing in 42 randomly selected urban and rural areas (census collector districts) in six states and the Northern Territory. In 2004-05, an additional survey site in Canberra was added to the original 42 sites used in 1999-2000. In 2011-12, further sites were added such as Busselton from Western Australia, Townsville and Bundaberg from Queensland, and Byron Bay from New South Wales.

Advanced knowledge of biostatistics and good basic knowledge of epidemiology is required by the PHD candidate to conduct this research.