Understanding life expectancy estimates for 2020–2022

The production of life expectancy estimates is a very complex and technical exercise. Understanding how these estimates are produced and what they represent is important to interpreting them. This article explains what life expectancy estimates are, some of the challenges in producing them, and what this means for understanding and using them.

Life expectancy estimates provide a measure of how long, on average, people of a particular age and sex are expected to live. They are not a measure of how long an individual is expected to live.

Life expectancy estimates are often used to assess a population’s health. Aboriginal and Torres Strait Islander life expectancy estimates are used to measure progress toward the national ‘Closing the Gap’ target 1: Close the Gap in life expectancy within a generation, by 2031. They also contribute to the production of Aboriginal and Torres Strait Islander population estimates and projections, providing annual estimates of the population, which are used for other national reporting indicators.

Life expectancy estimates are derived using the number of deaths in a period and the estimate of the population at the midpoint of that time. Understanding the quality and completeness of these data sources is important for understanding how life expectancy estimates can be used and whether they can be compared over time.

To calculate Aboriginal and Torres Strait Islander life expectancy estimates, we need two key pieces of information:

1. The number of Aboriginal and Torres Strait Islander deaths that occurred (for each age group and sex)
2. The size of the Aboriginal and Torres Strait Islander population (for each age group and sex).

Life expectancy estimates are calculated for a point in time. The estimates in this release use:

• the average number of Aboriginal and Torres Strait Islander deaths for 2020, 2021 and 2022
• the estimate of the Aboriginal and Torres Strait Islander population at 30 June 2021.

The average number of deaths across three years is used as this smooths out yearly fluctuations in the number of deaths and produces more representative life expectancy estimates. The population estimate at 30 June 2021 is used as this is the mid-point of the three years of deaths data.

To create life expectancy estimates and life tables, the age-specific death rates are calculated by dividing the three-year average of deaths by the population. This is done separately for males and females and for different age groups (as, for example, the age-specific death rate for females aged 85 years and over will be different to the age-specific death rate for males aged 20–24 years). These age-specific death rates are the basis for calculating life expectancy. A detailed description of the method is provided in Aboriginal and Torres Strait Islander life expectancy methodology, 2020–2022.

Generally, Aboriginal and Torres Strait Islander life expectancy estimates are not comparable over time. This is because the number of deaths and population estimates recorded at each point reflect things such as changing identification and undercount in the source data used to produce them. To compare life expectancy estimates over time, the deaths and the population need to be based on the same (or largely the same) group of people. 

Increases or decreases in either population size or numbers of deaths can be due to factors other than natural (or demographic) change, such as how the data are collected and recorded. It is these changes in the data used to derive life expectancy that disrupts the comparison of them over time. This is because the characteristics and risk factors for each of these distinct datasets may be different, and therefore the resulting life expectancies produced may simply be a reflection of differences between datasets rather than an improvement in health outcomes.

Understanding the components of change over time within both population and deaths data is therefore important for making a judgement about whether the resulting life expectancy estimates are comparable between time periods.

The next sections look at some of the factors that affect the numbers of people counted and numbers of deaths recorded, for reasons other than natural increase/decrease.

Population estimates are based primarily on Census counts, with adjustments made for undercount as measured through the Post Enumeration Survey (PES). Factors that can increase or decrease the number of Aboriginal and/or Torres Strait Islander people recorded in each Census for reasons other than natural/demographic (otherwise termed ‘non-demographic’) include:

• Changes in whether a person identified (or was identified) as Aboriginal and/or Torres Strait Islander in a particular Census.
• Changes in Census coverage and response, for example people who did not complete a Census form in 2016 but did complete a Census form in 2021.

The figure above shows the increase in the count of people who identified as Aboriginal and/or Torres Strait Islander in the Census from 2011 to 2016 and from 2016 to 2021. There was an increase of 25.2% or 163,557 people between 2016 and 2021. This was higher than the increase between 2011 and 2016 (18.4%), and over half (56.5%) of the increase was due to non-demographic change.

The non-demographic component of change leads to a shift in the Aboriginal and Torres Strait Islander population structure. An example is given in Understanding change in counts of Aboriginal and Torres Strait Islander Australians: Census, 2021 and included in the following box.

Compositional change like this in the count of people who identified as being Aboriginal and/or Torres Strait Islander in the Census flows through to Aboriginal and Torres Strait Islander population estimates. As population estimates are used to calculate age-specific death rates and life expectancy, these changes will also flow through to them. When there are compositional population changes reflected in life expectancy estimates, it affects how accurately changes in life expectancy over time. This is because changes may be a result of how people report their Indigenous status, rather than changes in population health.

Changes in the count of people who reported being Aboriginal and/or Torres Strait Islander in the Census by state/territory are reported in Understanding change in counts of Aboriginal and Torres Strait Islander Australians: Census, 2021. This shows that the largest non-demographic changes occurred in New South Wales, Victoria and Tasmania, while Northern Territory had the smallest non-demographic change. Changes in life expectancy estimates for areas with smaller non-demographic changes are more likely to result from changes in population health.

Data on 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 identified as being of an Aboriginal and/or Torres Strait Islander person by the family, health worker or funeral director during the death registration process. Over time, jurisdictions have supplemented information from the Medical Certificate of Cause of Death (MCCD) to improve the identification of Aboriginal and/or Torres Strait Islander deaths in the data. Even with these improvements there remain a number of death records for which the Indigenous status is unknown; in 2022 this was 975 records (or 0.5% of all death records).

In 2022, the New South Wales Registry of Births, Deaths and Marriages starting using information on the MCCD to improve identification of Aboriginal and/or Torres Strait Islander deaths. A deceased person was identified as Aboriginal and/or Torres Strait Islander if they were recorded as such on either the DRF or the MCCD. This change is consistent with how deaths data is processed in most other jurisdictions (with the exception of Victoria), where both the DRF and the MCCD have been used for several years. This change means that more Aboriginal and/or Torres Strait Islander deaths have been identified, and life expectancy estimates for 2020-2022 cannot be compared with previous life expectancy estimates for New South Wales. This also affects the comparability of national estimates.

The population and deaths data used to calculate life expectancy are both estimates. This means that multiple sources of information have been used to produce an estimate that is as close as possible to what the true figure is.

For population estimates: Census + PES

For estimates of deaths: death records + Census and PES

Using the PES to derive estimates for both population and deaths enables mortality rates and life expectancy to be calculated using a consistent application of adjustments for underreporting. This helps to ensure that the population and deaths estimates are about the same (or a similar) group of people.

However, making these adjustments also introduces uncertainty (in statistical terms, error), because the PES is based on a sample of the population. Confidence intervals give a measure of this error.

Population estimates are based on Census counts, with adjustments made using the PES. The PES measures how many people were missed or counted more than once in the Census, as well as how many people were not identified as Aboriginal and/or Torres Strait Islander when they should have been and vice versa. Together this information is used to estimate the undercount of the Aboriginal and Torres Strait Islander population in the Census and to make adjustments to account for this in the estimates of the population.

Deaths are registered with state and territory RBDMs. All state and territory RBDMs use information from the Death Registration Form (DRF) and most also use information from the Medical Certificate of Cause of Death (MCCD) to identify an Aboriginal and/or Torres Strait Islander death. Even using these two pieces of information, we know that the number of Aboriginal and Torres Strait Islander deaths is underreported (refer to Deaths, Australia methodology, 2022).

By linking death records to their Census record we can see where the Aboriginal and/or Torres Strait Islander status differs in each set of data for the same person and where it is the same. This enables us to make an initial adjustment to death records to lift the representation of Aboriginal and Torres Strait Islander deaths based on their status in the Census.

We also know that Aboriginal and/or Torres Strait Islander status in Census records is underreported, so we use information from the PES to make a further adjustment and bring the number of Aboriginal and Torres Strait Islander deaths to a level consistent with how population has been adjusted.

A confidence interval gives a range of values within which we are 95% confident that the true value lies.

Confidence intervals are published alongside the life expectancy estimates and can be use used to help interpret them.

The table above shows life expectancy estimates for males and females for different states/territory along with the error measurement, the 95% confidence interval.

Here is an example of how to use the confidence interval when comparing across groups. For males in New South Wales, life expectancy is 73.8 years compared with 72.9 years in Queensland. It seems that life expectancy is almost a year longer in New South Wales. But when we look at the confidence intervals, we can see that they have a large overlap.

When confidence intervals overlap, it means that the life expectancy estimates are not different according to a statistical test. When the confidence intervals do not overlap, then the life expectancy estimates are statistically different. This means that life expectancy estimates for males in New South Wales and Queensland are not statistically different, because the confidence intervals overlap. Looking at other comparisons for males, we can see that life expectancy is lower in Northern Territory compared with Queensland and New South Wales, but not compared with Western Australia.

If we compare life expectancy between males and females, it is higher for females in New South Wales, Queensland and Northern Territory (the confidence intervals for males do not overlap with those for females) but is not statistically different in Western Australia (where the confidence intervals for males and females overlap).