Using data linkage to estimate the burden of diabetes in Tasmania

Background
Diabetes is a metabolic disease occurring when there is not enough insulin secreted from the pancreas, or insulin secreted could not regulate blood glucose levels due to impaired action. Diabetes is considered to be an emerging public health challenge with increasing prevalence, devastating and costly complications, and high mortality rates. To comprehensively estimate the burden of diabetes, there is a need for datasets that contain a variety of information reflecting different aspects of the burden that diabetes imposes. Data linkage, also known as record linkage, is a method used to match records belonging to the same person from various sources. This method has been developed to generate new and comprehensive datasets for various purposes, especially research. Although promising, the potential of the data linkage approach in estimating the burden of disease has not been well investigated in Tasmania, a rural island state of Australia.
Objectives
Using a linked dataset that comprised approximately 87% of the adult population in Tasmania where few studies investigating diabetes have been published, this thesis aimed to estimate the burden of diabetes, with a focus on the economic burden. To be more specific, the primary aims of this thesis were to:
1. Identify the common themes, and knowledge gaps of data linkage research on diabetes in Australia.
2. Quantify the direct medical costs of people with diabetes from the healthcare system perspective, compare these costs to their non-diabetes counterparts, and identify trends in these costs over time.
3. Estimate the risk of emergency department/inpatient visit due to complications in people with diabetes, and compare them to their non-diabetes counterparts.
4. Estimate healthcare costs of diabetes complications in the year of first occurrence and the second year, and quantify the incremental costs for people with diabetes versus no diabetes related to each complication.
5. Visualise the geographic variations of diabetes burden and identify areas where targeted interventions are needed.
Thesis outline, methods, and summary of key results
To comprehensively address these aims, five studies were conducted, each study specifically addressing a corresponding aim. The thesis includes the following chapters:
Chapter 1 introduces a general overview of diabetes and data linkage, as well as some important health economic concepts that have been applied in the thesis.
Chapter 2 provides a summary of the methods used in this thesis, particularly the methods for identifying participants and the costing methods used.
Chapter 3 presents a systematic review of the application of the data linkage approach in diabetes research across Australia. In this study, we identified six common themes after reviewing 118 Australian data linkage studies. We also identified gaps related to health economic research using linked data that guided the remaining studies in this thesis. More specifically, there were few data linkage studies estimating healthcare costs. Those studies were either outdated or focused on certain groups which would not provide a comprehensive picture of the total costs associated with diabetes management.
Chapter 4 presents an estimate of the direct medical costs attributable to diabetes, as well as the incremental costs of diabetes, defined as the difference in costs between people with diabetes and those without. Time trends in costs were also analysed. From the linked dataset comprising information related to hospital episodes, emergency visits, pathology tests, and cause of death in Tasmania during 2014-2017, 51,324 people with diabetes were identified using either diabetes diagnostic criteria or hospital diagnostic codes. Based on propensity score matching, each was matched with two people without diabetes of similar sex, age (decile), and geographical area, sourced from the same dataset. The direct medical costs and incremental costs due to hospital admissions, emergency department visits and pathology tests in people with diabetes were calculated and adjusted to Australian dollars 2020-2021. On average, the annual costs of people with diabetes ($5,209 [95% confidence interval 5,112-5,317]) were almost double those of their non-diabetes counterparts ($2,782 [2,738- 2,826]; incremental costs: $2,427 [$2,322-2,543]). Furthermore, an upward trend in the incremental costs over time was observed. In general, this study provided an overall picture of the burden of diabetes in Tasmania.
In Chapters 5 and 6 the burden due to diabetes complications was estimated. More specifically, chapter 5 presents the estimates of the risk of emergency department/inpatient visit related to complications in people with diabetes, compared to people without diabetes. This study demonstrated that diabetes complications (especially macrovascular complications) are responsible for a significant burden on hospital services, and emphasised the importance of preventing and appropriately managing microvascular complications.
Chapter 6 presents the estimates of healthcare costs related to diabetes complications in the first year the complications occur and the second year, and the incremental costs of diabetes versus non-diabetes due to each complication. This study revealed that diabetes complications are costly, and the costs are higher in people with diabetes, compared to those without diabetes. Furthermore, this study also identified complications that significantly contribute to the economic burden that diabetes imposes. The results can be used as input information for diabetes simulation models and will support future policy analyses.
Chapter 7 presents the visualisation of the burden of diabetes in Tasmania using an interactive map. This study showed the geographic variations in diabetes burden across Tasmania, as well as identified areas with high diabetes burden to inform policy making to reduce the burden of diabetes. The map is a cutting-edge tool which was developed using the Shiny package in R and is available online to allow easy communication of the burden of diabetes to important stakeholders.
Finally, chapter 8 presents the conclusions of this thesis and suggests promising directions for future studies.