Background: Heart disease is a leading cause of the health gap between Aboriginal and non-Aboriginal people in Australia. Higher incidence of acute myocardial infarction (AMI) and higher mortality from AMI are major contributors to the greater burden of disease in Aboriginal people. Much of the research on reducing rates of AMI focuses on individual risk factors, such as smoking, physical activity, cholesterol level and diabetes. However, broader contextual and structural factors, including features of the geographic areas where individuals live, and the hospitals they attend, can have an important impact on health outcomes. Identifying and quantifying contextual and individual factors that influence the higher rates of AMI events and mortality in Aboriginal people will assist in better development and targeting of interventions to tackle these disparities. In this thesis, I develop methods for classifying Aboriginal people in routinely collected hospital data, and use these data to investigate the influence of individual, area of residence, and hospital factors on rates of AMI, mortality from AMI, and procedures after AMI, in Aboriginal people in New South Wales (NSW), Australia. Methods: Routinely collected hospital data for the entire NSW population for the period July 2000 to December 2008 were linked to mortality data from July 2000 to December 2009 using probabilistic methods. Firstly, I investigated the recording of Aboriginal status in the hospital and deaths data, and used linked data to develop and test algorithms to enhance the reporting of Aboriginal status. Then I used (i) multilevel Poisson regression models to estimate the relative rates of first AMI events, accounting for area of residence; (ii) multilevel logistic regression models to estimate the relative mortality after AMI admission, accounting for hospital and admission; and (iii) multilevel Cox proportional hazards models to estimate the relative procedure rates after AMI admission, accounting for hospital of admission. I also sequentially accounted for other individual risk factors, such as the presence of comorbid conditions, to determine their influence on the disparities in outcomes for Aboriginal people. Results: Sixty per cent of the variation in recording of Aboriginal status in routinely collected hospital data was due to the hospital of admission, and status recording was worse in major city compared with more regional and remote hospitals, and in private compared with public hospitals. The number of people reported as Aboriginal, and estimated admission rates and mortality ratios, varied according to the algorithm used to enhance the reporting of Aboriginal status. After accounting for age, sex, and year of admission, rates of AMI in Aboriginal people were more than two times those in non- Aboriginal people, even when comparing within areas of residence. The disparities were particularly large for women and those in younger age groups. There was significant variation in AMI rates by geographic area, with higher rates outside of major city areas and in areas of lower socioeconomic status. The relative Aboriginal to non-Aboriginal disparity in rates was also particularly large in these areas. Aboriginal patients had a similar 30-day mortality risk to non-Aboriginal patients, after adjusting for age, sex, year and hospital, but a higher risk of dying within one year. The latter difference became non-significant after adjustment for comorbid conditions. There was a higher 30-day mortality risk for patients admitted to smaller, more remote hospitals without on-site angiography facilities compared with larger hospitals and those with on-site angiography, respectively. Aboriginal patients had a revascularisation rate 37% lower than non-Aboriginal patients of the same age, sex, year of admission, and AMI type, but a rate 18% lower within the same hospital. Adjustment for comorbid conditions, such as diabetes and renal disease and other individual factors, explained the remaining disparity. Hospitals varied markedly in procedure rates, and this variation was associated with hospital size, remoteness, and facilities. Conclusions: Hospital-level interventions, such as better training of staff, are required to improve the recording of Aboriginal status, particularly in major city and private hospitals. Data linkage of routine administrative data can improve reporting of Aboriginal status, although the impact of the algorithm used to enhance reporting should be explored using sensitivity analysis. My research identified the importance of contextual influences when examining disparities in rates of AMI, and in mortality and procedures after admission for AMI. There was significant variation in overall AMI rates by area, which was partly explained by area-level disadvantage. Even when comparing within areas, Aboriginal people had higher rates of AMI than their non-Aboriginal counterparts. Priority areas for area-level interventions were those with a higher than average disparity and a higher than average rate of AMI for Aboriginal people. While disparities in longer-term mortality and procedure rates within hospitals did not persist after fully adjusting for individual risk factors such as comorbidities, these disparities will remain as long as Aboriginal people have higher rates of comorbid conditions (e.g. diabetes and renal disease) that complicate treatment and survival. For residents of rural and regional areas, both Aboriginal and non- Aboriginal, improving access to larger hospitals or those with specialist treatment facilities could improve surgical rates and outcomes after AMI. However, the main priority must be reducing the early onset of AMI and comorbid chronic conditions, such as diabetes and renal disease, and the subsequent early mortality among Aboriginal Australians. This will require major efforts in primordial, primary and secondary prevention. Priorities include targeting individual risk behaviours, such as smoking, improving the management of early symptoms of cardiac disease, reducing barriers to accessing primary care and cardiac rehabilitation services, and changing community norms about smoking and health behaviours. Interventions must acknowledge the wider historical and contextual causes of the current Aboriginal health disadvantage, and must deal with macro, contextual and individual levels of influence in order to have a significant impact.
| Date of Award | 2015 |
|---|
| Original language | English |
|---|