Demonstrating the utility of the COVID-19 Register

The Australian Institute of Health and Welfare (AIHW) has developed a linked data set using COVID‑19 case information which includes data on health service use and health outcomes to provide greater insights into the impact of COVID‑19 on the health of the Australian population and the health system.

The COVID‑19 Register is constantly being expanded and updated, and it is expected to contain linked data for around 6 million COVID‑19 cases from 7 jurisdictions as at August 2023. This report demonstrates some of the types of analysis that will be possible with a linked data set that allows person-based, patient pathway analysis, which will expand our understanding of the impact, including the long-term impact, of COVID‑19 in Australia.

What linked data is available for researchers?

Version 1

In December 2022, the first version (version 1) of the COVID‑19 Register was made available for approved researchers to use. Version 1 includes COVID‑19 case information from New South Wales, South Australia, Tasmania, the Australian Capital Territory, and the Northern Territory linked to a range of health administrative datasets, including the Medicare Benefits Schedule (MBS), the Pharmaceutical Benefits Scheme (PBS) and the National Death Index (NDI). There are just under 250,000 linked individuals with COVID‑19 in version 1, with varying notification dates (ranging from the start of the pandemic in early 2020 to March 2022 for some jurisdictions, and to November 2021 for others). The analysis examples presented in this report use version 1 of the data to demonstrate how this data could be used.

Version 2

The next version (version 2) includes an expanded scope of cases (for some states notifications are supplied up to September 2022), with around 6 million COVID‑19 cases (including those notified in Queensland and Victoria) and includes data on cases treated in intensive care units from the Australian and New Zealand Intensive Care Society (ANZICS). Version 2 is a whole-of-population data set, meaning that it includes health service information on both those diagnosed with COVID‑19 and the remaining population without a diagnosis at the time of linkage, which can be used as a comparator population. AIHW aims to make version 2 available to both government and non-government researchers before the end of 2023.

A detailed description of the data included and the linkage rates for each version are described in the report COVID‑19 linked data set: Linkage results.

Which types of analyses are demonstrated in this report?

To demonstrate the utility of the linked data for researchers, this report presents 3 analysis examples. These demonstration examples focus on individuals whose first COVID‑19 notification has a diagnosis between 16 June and 14 December 2021 (analysis cohort), at the time when the Delta variant was dominant in Australia. This time period was chosen as there was the most overlap of cases in the participating jurisdictions and allowed health service use to be analysed 6 months before and after diagnosis. Health service use was investigated in this report using MBS and PBS data, to show how the data could be used to describe health service use patterns before and after a COVID‑19 diagnosis. It is important to note that the findings simply demonstrate the utility of the data, rather than making any significant conclusions about the impact of COVID‑19 on Australian’s health and service usage. The data used in this report only partially captures COVID‑19 activity at the time so broader conclusions cannot be drawn from the findings. Researchers can use these examples to inform their own analysis planning for future versions of the data.

The COVID‑19 Register can be used for person-based rather than case-based analysis

Existing reports on the impact of COVID‑19 in Australia have focused on how many ‘cases’ there are in the community, rather than how many individual ‘people’ have been diagnosed with COVID‑19 (Department of Health and Aged Care 2023). The COVID‑19 Register allows for each person in Australia to have a unique record in the data and distinguishes if that person has had multiple infections over time. In the first demonstration analysis, there were 248,572 people in the entire version 1 cohort, and 75,024 people in the analysis cohort, that is, those who had a COVID‑19 diagnosis in the defined analysis period and met the other cohort criteria described above. Around 91% of the analysis cohort using version 1 of the COVID‑19 Register had a usual residence within NSW. This can be explained by the differential timing of data availability from the jurisdictions in the first version of the COVID‑19 Register data. This example also highlighted the value of linking COVID‑19 cases nationally. Around 1,410 people in the analysis cohort had a positive COVID‑19 test recorded in a state or territory outside of where they usually lived. Around 192 people (or <1%) had another COVID‑19 diagnosis in the analysis cohort. Reinfections will be able to be examined in depth once the data is updated to include more people diagnosed when Omicron variants were dominant.

Health outcomes following a COVID‑19 diagnosis can be monitored using the COVID‑19 Register

During the pandemic, existing surveillance systems were used to try to monitor how many people were dying from the acute effects of COVID‑19, however to date there has been little information on all cause and longer-term mortality and morbidity patterns related to the COVID‑19 pandemic. The COVID‑19 Register captures all deaths that have occurred in Australia over many years, regardless of whether this death was caused by COVID‑19. The second demonstration analysis explored deaths (as identified as individuals with a linked record on the National Death Index) following a COVID‑19 diagnosis. This is the first time nationally (as opposed to individual state-based data) COVID‑19 notifications have been linked to official death records. It differs from existing publications on COVID‑19 deaths and excess mortality which only use death records and not COVID‑19 case notifications. The example analysis was able to demonstrate that as at 30 June 2022, around 1% of the analysis cohort died (842 people), and the rate was higher for males than females in all age groups. Of these deaths, 86% died within 6 months of diagnosis. It is important to note this represents deaths from all causes, and not necessarily from COVID‑19 and that deaths recorded in the COVID‑19 Register may differ from those reported by state and territory health departments due to differences in definitions and the nature of probabilistic linkage. Future versions of the data will have more recent and detailed information on underlying and associated causes of death.

Researchers can use the data to explore changes in health service use following a COVID‑19 diagnosis

There has been a concern that COVID‑19 has longer-term impacts on Australian’s health service needs, but to date, there has been minimal data to investigate this impact. The COVID‑19 Register could be used to explore health service patterns, such as GP visits, before and after a COVID‑19 diagnosis. Given that the COVID‑19 Register allows for following individual patients over time, the demonstration analysis explored non-referred attendances (GP visits) at 1, 3 and 6 months following a COVID‑19 diagnosis when compared to the same period before their diagnosis. These time periods were chosen due to the availability of the MBS and PBS data in version 1 and could be broadened with future versions of the data to look at longer term health service use.

The demonstrations also show the ability to look at health usage by different MBS service categories such as specialist attendances.

The report also presents an example of how the data could be used to look at prescription patterns following a COVID‑19 diagnosis. Future analysis could use the prescriptions data to look at the effectiveness of antivirals or the impact of COVID‑19 on diabetes prescriptions.

A rich resource for monitoring COVID‑19 in Australia

This report is the first example of how the data in the AIHW’s COVID‑19 Register can be used. Future reports aim to include analyses by priority populations and explore the effects of vaccines, different variants, use of antivirals, and impact of multiple COVID‑19 notifications. The data included in the COVID‑19 Register will be updated regularly, to allow researchers to follow people who have had COVID‑19 over time through the health system.

Reference

Department of Health and Aged Care (2023) Coronavirus disease 2019 (COVID-19) epidemiology reports, Australia, 2020–2023, Department of Health and Aged Care, Australian Government, accessed 14 April 2023.