Demonstrating the utility of the COVID-19 Register

This section demonstrates how the COVID‑19 Register can be used to explore demographic patterns in a cohort of COVID‑19 cases. Using linked data allows for missing data in one data set (such as age) to be supplemented by complete fields in other linked data sets and makes this a much richer analysis. This analysis also demonstrates how different definitions of location (state of notification versus state of usual residence) can tell a different story. The results shown in this report do not inform about the demographic characteristics of COVID‑19 cases in Australia, since it relates to a specific, non-representative sample. The aim is to give an indication of the richness of the demographic details that can be used in analyses as the data becomes more representative.

Age and sex

In this example, a hierarchical approach was used to select the variables for age and sex, depending on the completeness of the variables across the data sets once they were joined to the cohort. While the variables used to derive age have no missing values in the Medicare Consumer Directory, individuals in the cohort may not have a record in these datasets if they have never been enrolled in Medicare at the time of linkage.

For example, we found that once linked, around 5% of the analysis cohort did not have a value for age in the Medicare Consumer Directory. To address this, where the age was missing in one dataset, it was supplemented with an age variable from another dataset if present, using a hierarchical approach. This initial exploration of the completeness of variables in a study population once linked to administrative records is something for future researchers to consider and build into their own analysis projects.

Once missing age data had been supplemented so that every record had an age, in the analysis cohort, the median age was 29 years (interquartile range 15–44). Around 52% of the cohort were male and there were more males than females for all age groups between 25–94 (Figure 2). The highest proportion of people diagnosed with COVID‑19 was people aged 15–19 years (963 persons per 100,000 population), and the lowest was people aged 70–74 (205 per 100,000 population). These results were consistent with previously reported surveillance reports. The COVID‑19 Weekly Surveillance in NSW report for the week ending 4 December 2021 reported a median age of 28 years for cases, interquartile range 15–44 (NSW Health, 2021).

It is important to note that COVID‑19 testing rates were high during the analysis cohort time period as the Delta wave was occurring and this may influence the age and sex patterns outlined in this report.

Figure 2: Linked analysis cohort by age group and sex

Notes:

1. Age group is based on age as at 31 December 2021.
2. For details on how COVID‑19 proportions (people diagnosed with COVID‑19 per 100,000 population) were calculated and for how age and sex was derived, refer to Technical notes.

Source: AIHW analysis of COVID‑19 Register (version 1).

State/territory of usual residence

To explore geographical information in this report, the geographical region in which a person usually resides has been used, as represented by a ‘state of usual residence’. Linked data allows researchers to look at cross border effects and identify people who may have been diagnosed in a jurisdiction that is not where they usually live, which can be useful for health service planning purposes.

Around 91% of the analysis cohort using version 1 of the COVID‑19 Register had a usual residence within NSW (Table 1). This can be explained by the varying dates in which data has been supplied for this project, as detailed in Table A1 (see Technical notes). While Victoria, Queensland and WA were not included in the first version of the data used for this report, there are some people (1,410) included in the cohort who usually reside in these areas and who had a positive COVID‑19 test notified in one of the other participating jurisdictions. This demonstrates the utility of linking COVID‑19 cases on a national level, given that people move locations over time and may seek health care in a jurisdiction that differs from where they have a diagnostic test. Around 4% of people had a missing SA2, future reports will explore supplementing this missing field with alternative sources of geography, for example postcode from the notifications data.

1. People who usually reside in WA, Qld, Vic, and Other Territories (Jervis Bay, Cocos (Keeling) Islands, Christmas Island and Norfolk Island) who had a positive COVID‑19 test notified in one of the other participating jurisdictions (NSW, SA, Tas, ACT, NT).

Remoteness area

This section demonstrates how information from the Medicare Consumer Directory in the COVID‑19 Register could be used to explore remoteness areas of the residence of COVID‑19 cases. About 28% of the Australian population live in regional and remote areas (ABS 2022). Data show that people living in rural and remote areas have higher rates of hospitalisations, deaths, injury and also have poorer access to, and use of, primary health care services, compared with people living in Major cities (AIHW 2022b).

The majority (90%) of people in the analysis cohort usually resided in Major cities (Figure 3). This pattern was similar when looking at the COVID‑19 proportions (people diagnosed per 100,000 population) in the Australian population for this cohort. After adjusting for age, the highest number of people diagnosed per 100,000 population was in Major cities (841 per 100,000 population). There were slightly more males than females affected per 100,000 population in Major cities (847 versus 778 per 100,000 population, respectively).

Figure 3: Linked analysis cohort by remoteness area and sex

Notes:

1. Remoteness area is based on usual place of residence.
2. Excludes records with missing information on remoteness area.
3. For more details on how the COVID‑19 proportions (people diagnosed with COVID‑19 per 100,000 population) were computed and how sex was derived, refer to Technical notes.

Source: AIHW analysis of COVID‑19 Register (version 1)

Socioeconomic groups

This section demonstrates how information from the Medicare Consumer Directory in the COVID‑19 Register can be used to explore socioeconomic groups, as having access to material and social resources (such as secure housing and a stable income) are key determinants of health. There is little known about the impacts of COVID‑19 in different population groups or by factors such as education level or income, and this example demonstrates the rich source of information researchers can use in the COVID‑19 Register to explore these questions. A composite measure of socioeconomic position known as the Index of Relative Socio-economic Disadvantage (IRSD) (ABS 2018) was used based on the usual area of residence.

Nearly half (45%) of all people in the analysis cohort were in the lowest socioeconomic group as shown in Figure 4. This pattern was consistent after adjusting for age and population size in each group, with the highest proportion occurring in the lowest socioeconomic group (1,463 per 100,000 population). There was a slightly higher proportion among males than females in all socioeconomic groups.

The higher proportion in the lowest socioeconomic group may be explained by more risk factors for exposure to the virus in this group, for instance they may be more likely to be casual frontline workers unable to work from home (Roder et al 2020).

Figure 4: Linked analysis cohort by socioeconomic group and sex

Notes:

1. Socioeconomic groups are classified according to population-based quintiles using the Index of Relative Socio-economic Disadvantage (IRSD) based on Statistical Area level 2 (SA2) of current residence.
2. Excludes records with missing information on IRSD ranking.
3. For more details on how COVID‑19 proportions (people diagnosed with COVID‑19 per 100,000 population) were computed and how sex was derived, refer to Technical notes.

Source: AIHW analysis of COVID Register (version 1)