Updated method for 2020–2022 Aboriginal and Torres Strait Islander life expectancy estimates

In 2020, the ABS commissioned an independent review to assess the fitness for purpose of the ABS Aboriginal and Torres Strait Islander life expectancy estimation method. Through that review, 31 detailed recommendations relating to methodology, possible additional estimates, planning, communication, and engagement were made. Investigations by the ABS into the recommendations relating to methodology have led to changes in the method for the 2020–2022 estimates. These changes are detailed here and include revised 2015–2017 estimates using the updated methodology. For a summary of the review recommendations see Independent review of the ABS' Aboriginal and Torres Strait Islander life expectancy estimates.

Investigations into the complex methodology and how it behaves under different data conditions are ongoing.  Through the continued focus to improve the estimation method, it is expected that further methodological changes in future iterations of the estimates will occur.

In this paper we provide:

• a summary of the methodology used for producing Aboriginal and Torres Strait Islander life expectancy estimates for 2015–2017 (and 2010–2012).
• a description of the revisions made to the methodology used for producing the 2020–2022 Aboriginal and Torres Strait Islander life expectancy estimates.
• revised life expectancy estimates for 2015–2017 and confidence intervals based on the updated methodology.

For detailed information on the method used to calculate estimates of Aboriginal and Torres Strait Islander life expectancy for 2020–2022 refer to Aboriginal and Torres Strait Islander life expectancy methodology, 2020–2022.

To calculate life expectancies, accurate data for deaths and population are needed. Data for deaths are provided to the ABS by each state and territory Registry of Births, Deaths, and Marriages (RBDM). All states and territories use information acquired from the Death Registration Form (DRF) to identify an Aboriginal and/or Torres Strait Islander death. While it is considered likely that most deaths of Aboriginal and/or Torres Strait Islander people are registered, a proportion of these deaths are not reported as such by the family, health worker or funeral director during the death registration process.

To treat underrepresentation of Aboriginal and Torres Strait Islander deaths in the data, the ABS developed a method to make adjustments in two stages:

1. Death records that were linked to Census records.
2. Census records that were linked to PES data.

Stage 1 used information from Census records that were linked to death registrations in the year following Census night. These linkages provided information on how likely it was that a death was recorded as:

• Aboriginal and/or Torres Strait Islander and linked to a Census record that was also identified as Aboriginal and/or Torres Strait Islander
• Non-Indigenous and linked to a Census record that was identified as Aboriginal and/or Torres Strait Islander
• Not stated and linked to a Census record that was identified as Aboriginal and/or Torres Strait Islander.

The propensities for the records to have the same or different Indigenous status on each collection were used to adjust for underrepresentation in death records.

Stage 2 used propensities of identification from Census and Post Enumeration Survey (PES) linked data. Acknowledging that propensity to identify as Aboriginal and/or Torres Strait Islander can vary with age, three age groups were identified to make age-specific adjustment to the death records:

• 0–14 years
• 15–59 years
• 60 years and over.

From this, a second adjustment was made for underrepresentation in death records.

Using information from the PES to make adjustments to the number of Aboriginal and Torres Strait Islander deaths creates consistency with how population estimates have been adjusted from Census and PES data. Consistency in adjustment for identification across both population data and deaths data is a key strength of the ABS method. For more information on how Aboriginal and Torres Strait Islander populations and deaths have been estimated, see Understanding life expectancy estimates for 2020–2022 and Estimates of Aboriginal and Torres Strait Islander Australians methodology, 30 June 2021.

Key points about the method used:

• The application of age adjustment in the 2015–2017 method was at the national level only. This is because there was insufficient sample from PES in the three age groups at state/territory level to reliably apply age adjustment factors. State and territory estimates were therefore calculated without age adjustment, and as such a non-age adjusted national estimate was also produced for comparison with state/territory estimates.
• Death records that were not able to be linked to Census were not used in adjustments.

For more information, see previous methodology: Life Tables for Aboriginal and Torres Strait Islander Australians methodology, 2015–2017.

In response to recommendations from the 2020 Independent review of the ABS' Aboriginal and Torres Strait Islander life expectancy estimates, the ABS examined the method used to adjust death records for producing life expectancy estimates for 2015–2017. Based on the findings, two methodological enhancements have been made to the 2020–2022 estimates:

1. Adjustment for unlinked deaths records
2. Adaptation of the application of age adjustment.

In the 2015–2017 (and 2010–2012) method to produce life expectancy estimates, adjustment factors were calculated using Census and deaths linked data only. Unlinked death records were excluded from calculating adjustment factors. This method assumed that the rate of unlinked records was not affected by Indigenous status. That is, the relative number of unlinked records was independent of whether it was an Aboriginal and/or Torres Strait Islander record or non-Indigenous record.

However, investigation of the 2022 linkage of deaths to Census data showed that:

• linkage rates varied by Indigenous status and by age group, so that unlinked records were unevenly distributed throughout the deaths records
• the difference between linkage rates of deaths and Census records, between Aboriginal and/or Torres Strait Islander records and non-Indigenous records, has increased since the previous Census cycle
• adjustment factors are sensitive to changes in the relative size of linkage rates by age group for Aboriginal and/or Torres Strait Islander records and non-Indigenous records.

To treat bias introduced by the relative differences in linkage rates between Aboriginal and Torres Strait Islander and non-Indigenous records, both linked and unlinked death records have been used in calculating the adjustment factor. This stabilises the denominator of the equation and removes bias caused by lower linkage rates.

This equation shows the differences in how adjustment factors were calculated for 2015–2017 estimates and 2020–2022 estimates.

When the linkage rate of death records to Census is lower for Aboriginal and Torres Strait Islander records than for non-Indigenous records, both D(ATSI) and E(ATSI) are lower relative to E(NI) and E(NS). This reduces the denominator by more than the numerator and inflates the adjustment factor.

For more information, see Aboriginal and Torres Strait Islander life expectancy methodology, 2020–2022.

For comparison of linkage rates, see Data linkage to derive Aboriginal and Torres Strait Islander Australians deaths identification rates.

For information on previous linkage of Census and death records undertaken as part of the Census Data Enhancement (CDE) study, see Summary results of the CDE study within Life tables for Aboriginal and Torres Strait Islander Australians, 2015–2017.

Adjusting for under-identification of Indigenous status relies on first calculating propensities of identification (comparing where people identify as Aboriginal and/or Torres Strait Islander on the Post Enumeration Survey (PES) relative to how they reported on the Census). The adjustment factors for the 2015–2017 age-adjusted national estimates of Aboriginal and Torres Strait Islander life expectancies were calculated by applying these propensities to the death records that could be linked to the Census, separately for the three age groups:

• 0–14 years
• 15–59 years
• 60 years and over.  

For more information, see previous methodology: Life Tables for Aboriginal and Torres Strait Islander Australians methodology, 2015–2017

For the 2015–2017 estimates, nine propensities were calculated: one for each age group and each Census Indigenous status category (see full age adjustment image below).

Age groups were:

• 0–14 years
• 15–59 years  
• 60 years and over.  

Census Indigenous status categories were:

• Aboriginal and/or Torres Strait Islander
• non-Indigenous
• not stated.

However, making full age adjustments relies on sufficient sample in the PES. Investigations showed that the 2016 PES sample was very small for Census records of non-Indigenous and not stated, particularly for the 60 years and over age group. When propensities are calculated using small numbers, they have a higher degree of uncertainty and the adjustment factors calculated from the propensities are also uncertain.

For the 2020–2022 life expectancy estimates, this uncertainty was reduced by combining all age groups in the Census non-Indigenous and Census not stated categories, while all three age groups remain distinct in the Census Aboriginal and/or Torres Strait Islander category. This is referred to as partial-age adjustment in this paper (see image above). Doing this brings greater certainty to the adjustment factors.

The method of partial age-adjustment was consistently used to derive adjustment factors for Aboriginal and Torres Strait Islander deaths at the national level, for states and territories, for Remoteness Areas and SEIFA Index of Relative Socio-Economic Disadvantage. See Aboriginal and Torres Strait Islander life expectancy methodology, 2020–2022 for details of how partial age-adjustment was defined in adjustment factors.

Using the same method as the 2020–2022 estimates, the 2015–2017 Aboriginal and Torres Strait Islander life expectancy at birth estimates have been recalculated. These revised estimates and their 95% confidence intervals are given in Tables 1, 2 and 3 below. The non-Indigenous life expectancy at birth estimates and differences in life expectancy at birth between Aboriginal and Torres Strait Islander and non-Indigenous Australians were also recalculated and are given in Tables 4, 5 and 6 below.

The revised 2015–2017 life expectancy estimates for Aboriginal and Torres Strait Islander Australians across age groups and sex for selected states and territory and Australia are available as Table 10 under Data downloads. See Aboriginal and Torres Strait Islander life expectancy, 2020 - 2022.