The changing epidemiology of coronary heart disease in people with diabetes in Western Australia

INTRODUCTION

Diabetes and coronary heart disease (CHD) are two common chronic conditions with contrasting epidemiological trends. CHD incidence and mortality rates have been declining for some decades but the prevalence of diabetes has been increasing. The association between CHD and diabetes is complex and multifactorial. The pathological processes which underlie CHD are exacerbated by the presence of diabetes. Historical data show that once CHD is manifest, people with diabetes have worse short-term outcomes, receive less evidence-based treatment, and have poorer long-term outcomes compared with their non-diabetic counterparts. The ability to determine whether changing approaches to acute CHD treatment and management, and primary and secondary prevention are having a positive impact requires long-term population level data. However, there are limited international and no Australian population level data detailing trends in the incidence of CHD in people with diabetes. There are also few settings where the full spectrum of incidence, short-term case fatality and longer-term outcomes following CHD have been measured in the same population, both overall and specifically for people with diabetes.The primary aim of this thesis was to investigate the changing epidemiology of CHD in people with diabetes in the context of the changing epidemiology of CHD and cardiovascular disease more broadly in the population of Western Australia. The specific objectives were:

1. To establish the reliability of the coding of diabetes in WA hospital morbidity data for CHD patients

2. To estimate the impact of the increasing prevalence of diabetes on the population-level incidence of CHD; and

3. To measure the risk of short and long-term outcomes following myocardial infarction (MI) in diabetic people.

METHODS

The study objectives were investigated using high-quality linked health data extracted from the Western Australian Data Linkage System for the period 1985 to 2011. The main study dataset contained linked hospitalisation and mortality data for the period 1985 to 2011 enabling a variety of analyses of trends in incidence, recurrence and outcomes to be conducted using logistic, Poisson and Cox regression modelling.

RESULTS

This thesis is presented as a series of papers. Chapters 4–9 contain the six papers undertaken for the thesis. Chapters 4–8 are final manuscript versions of papers that have already been published and Chapter 9 is the version of the manuscript that has been submitted to a peer-reviewed journal for publication.

Chapter 4 describes a study undertaken to determine the accuracy of the recording of diabetes status in hospitalisation data for CHD patients. Comparing the recording of diabetes in hospital data and data collected from hospital records, concordance was high with the use of an extended lookback period (sensitivity >90%, positive predictive value 92%).

Chapter 5 analyses demonstrate that the incidence and recurrence of hospitalised CHD declined between 2000 and 2007. This was supported by reductions in CHD mortality rates over the same period, and concurrent falls in cerebrovascular disease (CeVD) and peripheral arterial disease (PAD) incidence and recurrence rates. Rates of polyvascular disease declined more rapidly than single territory disease. Age-standardised prevalence of diabetes in incident cases increased from 21.5% to 26.0% in men and from 26.3% to 29.0% in women over the same period.

Chapter 6 analyses with stratification of MI cases by diabetes status showed that MI incidence rates in people with diabetes fell substantially between 1998 and 2010 whereas concurrent reductions in rates in non-diabetic patients were marginal. The trends in diabetic people occurred despite increasing levels of chronic kidney disease, hypertension and prior CHD hospitalisations in these patients.

Chapter 7 focuses on short-term outcomes following incident MI in people with and without diabetes. The results show that age- and sex-adjusted 30-day case fatality declined by 10.6%/year in people with diabetes and 6.9%/year in people without diabetes. The trends meant that by the end of the study period there was no longer any difference in the adjusted risk of 30-day death following incident MI according to diabetes status.

Chapter 8 shows that long-term MI recurrence and mortality remain high following incident MI in the WA population, particularly in 70–84-year-olds. Men and women aged 35–54 years and 55–69 years have similar event probability through eight years of follow-up. When trends in mortality outcomes were analysed by diabetes status (Chapter 9), unadjusted 5-year all-cause mortality following incident MI was about two-fold higher in people with diabetes and there was no significant change in the multivariable-adjusted hazard ratio comparing people with diabetes to those without from 1998 to 2009.

CONCLUSION

In conclusion, trends in a whole-population setting show that there is decreasing incidence, recurrence and mortality rates of CHD and vascular disease more broadly, with concurrent increasing diabetes prevalence. In this context, there are mixed results for CHD outcomes in people with diabetes. Trends in incidence rates of MI have fallen substantially in people with diabetes, as have short-term deaths following incident MI over an extended contemporary period. In contrast, longer-term all-cause mortality has not improved in diabetic people. These data are extremely important for the implications of the likely effect of changing primary prevention approaches and acute care following MI for people with diabetes, whereas secondary prevention, particularly in people with diabetes, appears to have had little impact on the survival differential by diabetes status.