Main data sources
The data in this report are mainly drawn from national data collections and surveys. These include the following:
Australian Aboriginal and Torres Strait Islander Health Survey
The 2012–13 Australian Aboriginal and Torres Strait Islander Health Survey AATSIHS included a sample of 13,000 Aboriginal and Torres Strait Islander people for the core sample and sub-samples for various components of the survey such as voluntary biomedical data for adults. The AATSIHS sample was specifically designed to select a representative sample of Aboriginal and Torres Strait Islander peoples and therefore overcome the problem inherent in most national surveys (i.e. small and unrepresentative Indigenous samples). Survey results are subject to sampling errors as only a small proportion of the population is used to produce estimates that represent the whole population—see RSEs above. Non-sampling errors may occur where there is non-response to the survey or questions in the survey, misunderstanding of questions or errors in recording, coding or processing the survey. Information recorded in this survey is ‘as reported’ by respondents. Responses may be affected by imperfect recall or individual interpretation of survey questions. Any data that are self-reported are also likely to underestimate circumstances about which the respondent is unaware, or may be reluctant to report (e.g. certain health conditions, weight, drug use). Selected non-Indigenous comparisons are available from the 2011–12 Australian Health Survey (AHS). The AHS was conducted in major cities, regional and remote areas, but very remote areas were excluded from the sample. Further information on AATSIHS data quality issues can be found in the User Guide for the survey ABS cat. no. 4727.0 (ABS, 2013c). Time series comparisons for some indicators are available from the 2004–05 NATSIHS, 2008 NATISS, 2002 NATSISS, 2001 NHS and the 1994 National Aboriginal and Torres Strait Islander Survey (NATSIS).
The National Aboriginal and Torres Strait Islander Social Survey
The 2014–15 National Aboriginal and Torres Strait Islander Social Survey (NATSISS) was conducted from September 2014 to June 2015. Information was collected by personal interview from approximately 11,000 Aboriginal and Torres Strait Islander people living in private dwellings in both non-remote and remote parts of Australia. The NATSISS sample was specifically designed to select a representative sample of Aboriginal and Torres Strait Islander peoples. The NATSISS uses the standard Indigenous status question to identify Aboriginal and Torres Strait Islander households from which the sampling process is then undertaken. Information recorded in this survey is ‘as reported’ by respondents, or from child proxies (usually parents), on behalf of selected children aged 0–14 years. Data may differ from those that might be obtained from other sources or by using other collection methodologies. Responses may also be affected by imperfect recall or individual interpretation of survey questions. Selected non‑Indigenous comparisons are available from the 2014 General Social Survey and the 2014–15 National Health Survey. Further details can be obtained from ABS cat. no. 4714.0 (ABS, 2016). Time series comparisons for some indicators are available from the 2002 and 2008 NATSISS and the 1994 National Aboriginal and Torres Strait Islander Survey (NATSIS).
The Census uses the standard Indigenous status question and it is asked for each household member.
There are 4 principal sources of error in Census data: respondent error, processing error, partial response and undercount. Quality management of the Census programme aims to reduce error as much as possible, and to provide a measure of the remaining error to data users, to allow them to use the data in an informed way.
The Census form may be completed by one household member on behalf of others. Incorrect answers can be introduced to the Census form if the respondent does not understand the question or does not know the correct information about other household members. Many of these errors remain in the final data.
The processing of information from Census forms is now mostly automated. Quality assurance procedures are used during Census processing to ensure processing errors are minimised. Sample checking is undertaken during coding operations, and corrections are made where necessary.
When completing their Census form, some people do not answer all the questions that apply to them. In these instances, a ‘not stated’ code is allocated during processing, with the exception of non-response to age, sex, marital status and place of usual residence. These variables are needed for population estimates, so they are imputed using other information on the Census form, as well as information from the previous Census.
Other Census data issues relate to the accuracy of the Census count itself, e.g. whether people are counted more than once, or not at all.
The significant volatility in Aboriginal and Torres Strait Islander Census counts and the variable quality of data on births, deaths and migration of Aboriginal and Torres Strait Islander peoples do not support the use of the standard approach to population estimation. Due to the inherent uncertainties in estimating the Aboriginal and Torres Strait Islander population, data which use these estimates should be interpreted with caution (e.g. life expectancy estimates, mortality rates). Given these uncertainties, changes in health outcomes such as mortality rates are difficult to assess.
Following each Census, assumptions are made about past levels of mortality to produce back cast population estimates.
In the 2011 Census, the ABS implemented improvements to the 2011 Census Indigenous enumeration procedures and enhanced the Census Post Enumeration Survey. There was a 30% increase in the estimate of the Indigenous population between the 2006 and 2011 Censuses. The trends involving population rates have needed to be revised for this edition of the HPF based on these updated population estimates (ABS, 2014).
National Perinatal Data Collection
Birth notification forms are completed for all births of 20 weeks or more gestation, or a birthweight of 400 grams or more. The Perinatal National Minimum Data Set includes all births in Australia in hospitals, birth centres and the community. Jurisdictional level data in the HPF are based on place of mother’s usual residence rather than place where birth occurred. A standard data item for Indigenous status is specified in the Perinatal National Minimum Data Set for the mother and now also for the baby. Data on Indigenous status are not reported for Tasmania prior to 2005, as the ‘not stated’ category for Indigenous status was included with the non-Indigenous category. Numbers are small in jurisdictions such as ACT and Tasmania and therefore need to be interpreted with caution.
Studies in Australia linking perinatal data with birth registration data and hospital admissions show that Aboriginal and Torres Strait Islander data are under-reported (Taylor, 2000; Comino, 2007; Kennedy, 2009). In 2007, the AIHW completed an assessment of the quality of Indigenous status information in perinatal data in each state and territory. This involved a survey that was sent to the midwifery managers across Australia to determine how many hospitals in each jurisdiction obtain Indigenous status information of mothers giving birth from admission records and how many collect this information independently. The assessment also involved analysis of the variability in the number and proportion of mothers recorded as Aboriginal and/or Torres Strait Islander in the perinatal data collection over time and across jurisdictions for the period 1991–2004. The outcomes of this assessment showed that Indigenous status data from NSW, Vic, Qld, WA, SA and the NT are suitable for trends analysis and national reporting. Perinatal data from Tasmania, although improving, were deemed to be of insufficient quality. Although the most recent data in the ACT were of publishable quality, the data were not yet of sufficient stability to support trends analysis (Leeds, 2007).
All jurisdictions are working towards improving the quality of Indigenous status in perinatal data collections. States and territories have agreed to improve Aboriginal and Torres Strait Islander data collection procedures in key data collections including implementation of the Best Practice Guidelines for the collection of Indigenous status in health data sets.
AIHW has worked with the states and territories in the development of an enhanced Perinatal National Minimum Data Set (NMDS) to include nationally consistent data items on antenatal care, smoking and alcohol use during pregnancy, and Indigenous status of the baby. Nationally consistent data items on smoking during pregnancy, gestational age at first antenatal visit, and Indigenous status of the baby have been added to the Perinatal NMDS (from 1 July 2009, 1 July 2010, and 1 July 2012 respectively). A data item on number of antenatal visits was included in the Perinatal NMDS from 1 July 2013, although Victoria was not able to commence collection until 1 January 2015.
Analysis of perinatal data between jurisdictions and over time requires some caution due to definitional changes, changing models of care (e.g. antenatal care being increasingly provided outside of hospitals) and changes to reporting systems.
National Hospital Morbidity Data
This data collection includes all completed admitted patient episodes in public and private hospitals across Australia. For 2014–15, almost all public hospitals provided data for the NHMD. The exception was a mothercraft hospital in the ACT. The great majority of private hospitals also provided data, the exceptions being the private free-standing day hospital facilities in the ACT, the single private free-standing day hospital in the NT, and a private free-standing day hospital in Victoria.
The number and pattern of hospitalisations can be affected by differing admission practices among the jurisdictions and from year to year, and differing levels and patterns of service delivery. Some jurisdictions have slightly different approaches to the collection and storage of the standard Indigenous status question and categories in their hospital collections. The incompleteness of Indigenous identification means the number of hospital separations recorded as Aboriginal and Torres Strait Islander is an underestimate of hospitalisations involving Aboriginal and Torres Strait Islander people.
Between 2006 and 2008, the AIHW completed an assessment of the level of Indigenous under-identification in selected public hospitals in all states and territories. Results from this assessment indicated that NSW, Vic, Qld, WA, SA and the NT have adequate Indigenous identification (20% or less overall under-identification of Aboriginal and Torres Strait Islander patients) in their hospital separations data. Historically, Indigenous hospitalisation data from 1998 was reported only for the four jurisdictions with the best quality identification (NT, WA, SA and Qld). In 2011–12, the AIHW completed another study to re‑assess the level of under-identification in public hospitals data. The study found that all jurisdictions now have sufficient quality Indigenous identification for reporting and a correction factor of 1.09 is used at the national level to adjust for under‑identification in hospital data.
Interpretation of results should take into account the relative quality of the data from the jurisdictions. Time series analyses may be affected by changes in the quality of Indigenous identification over time. An AIHW study estimated that 89% of Indigenous patients were correctly identified in Australian public hospital admission records in 2007-08 (AIHW, 2010). A more recent study of public hospital data from 2011–12 found no statistically significant change in identification rates at the national level, although there were some changes at the jurisdictional level (AIHW, 2013). Time series presentations in this report include data from both public and private hospitals across several jurisdictions, so the overall effect of changes in Indigenous identification over time is unclear. This should be taken into account when interpreting time series data.
Between 2009–10 and 2010–11, there were significant changes in the coding of diagnoses for diabetes and obstetrics and for imaging procedures. There were also significant changes made to coding practices for diabetes and related conditions for the 2012–13 year, resulting in increased counts for these conditions. There have also been changes in the hepatitis coding. Caution should be used in comparing 2011–12 and 2012–13 data with data from previous HPF reports. For more information about these issues, please refer to Database quality statement summaries in the ‘Australian Hospital Statistic 2012–13’ report, available at www.aihw.gov.au.
For more information on the quality of emergency department care data, please refer to the data quality statement in the ‘Australian hospital statistics 2012–13: emergency department care’ report, available at www.aihw.gov.au.
Hospitalisations for which the Indigenous status of the patient was not reported are excluded from hospitalisations data for non-Indigenous Australians.
National Mortality Database
The count of deaths for Aboriginal and Torres Strait Islander peoples can be influenced by the accuracy of identification of Aboriginal and Torres Strait Islander deaths, late registration of deaths, as a result of coronial inquiry, and changes to death forms and/or processing systems. Due to the small size of the Aboriginal and Torres Strait Islander population, these factors can significantly impact on trends over time and between jurisdictions.
All jurisdictions broadly comply with the standard wording for the Indigenous status question for their death registration forms. However, the response categories tend to differ between a number of jurisdictions, most notably WA provides an additional 'Unknown' response category and the NT does not provide clear instructions on how to select both Aboriginal and Torres Strait Islander origin (e.g. by ticking both boxes) (ABS & AIHW, 2005).
Almost all deaths in Australia are registered. However, the Indigenous status of the deceased is not always recorded/reported and/or recorded correctly. The incompleteness of Indigenous identification means the number of deaths registered as Aboriginal and Torres Strait Islander is an underestimate of deaths occurring in the Aboriginal and Torres Strait Islander population (ABS, 2013b). As a result, the observed differences between Aboriginal and Torres Strait Islander and non‑Indigenous mortality are underestimates of the true differences. While the identification of Aboriginal and Torres Strait Islander peoples in deaths data is incomplete in all state and territory registration systems, 5 jurisdictions (NSW, Qld, WA, SA and the NT) have been assessed by the ABS and the AIHW as having adequate identification. Longer term mortality trend data are presented for these jurisdictions from 1998 onwards. The ABS calculated the identification rate of Aboriginal and Torres Strait Islander deaths for the period 2011–12 through data linkage between the Census and death registration records to examine differences in reporting of Indigenous status across the two datasets (ABS, 2013b).
Aboriginal and Torres Strait Islander deaths identification rate, state/territory and Australia, 2011–12
|Vic, SA, Tas, ACT, Overseas territories||0.40|
|Australia (not age-adjusted)||0.72|
Source: (ABS, 2013b).
Mortality rates use the count of deaths in the numerator and population estimates in the denominator. Analysis of trends in mortality rates is therefore also influenced by the assumptions made about mortality in the population backcast and projections which is somewhat circular. The 2011 Census based backcast mortality assumptions were that Aboriginal and Torres Strait Islander life expectancy at birth at the Australia level increased by 0.2 years per year for males and 0.15 years per year for females for the period 30 June 1996 to 30 June 2006 and then increased by 0.3 years per year for males and 0.15 years per year for females for the 2006–2011 period. Under this assumption, life expectancy at birth in 1996 would be 65.5 years and 71.4 years for Aboriginal and Torres Strait Islander males and females respectively. Whether Aboriginal and Torres Strait Islander life expectancy at birth has changed at a faster or slower rate is unknown (ABS, 2014).
The AIHW has also developed an Enhanced Mortality Database by linking death registration records to several additional data sources that contain information on Aboriginal and Torres Strait Islander deaths (see details in Data Development section).
BEACH (GP Survey data)
The Bettering the Evaluation And Care of Health (BEACH) survey (now ceased) collected information about consultations with GPs, including GP and patient characteristics, patient reasons for the visit, problems managed and treatments provided. The survey was conducted annually between 1998 and 2016. Information was collected from a random sample of approximately 1,000 GPs from across Australia each year. Each participating GP provided details of 100 consecutive consultations.
Although the questionnaire contains an Indigenous identifier, it is unknown whether all GPs ask their patients this question. The reliability of the results of the BEACH survey was tested in a sub-study of about 9,000 patients encountered during the survey in 2003. The sub-study found that when the question on Indigenous status is asked of the patient within the context of a series of questions about origin and cultural background, 2.2% will self-report as Aboriginal and/or Torres Strait Islander, which is twice the rate routinely recorded in BEACH. However, this difference was not considered to be statistically significant (AIHW 2010) .
Under the National Indigenous Reform Agreement (NIRA), governments have agreed to the implementation of the Best Practice Guidelines for the collection of Indigenous status in health data collection, which include recommended strategies for local service providers such as GPs and practice nurses to improve Indigenous identification data.General practices and Aboriginal and Torres Strait Islander health services that meet accreditation requirements are eligible for financial incentives when they sign on to the Practice Incentives Program—Indigenous Health Incentive (PIP IHI), obtain consent from their Aboriginal and Torres Strait Islander clients to be registered for chronic disease management through the IHI, and provide a targeted level of care for each registered patient. Participation in the PIP IHI also provides eligibility for other initiatives under Medicare.
A Voluntary Indigenous Identifier (VII) was introduced into the Medicare database from November 2002. By July 2016 approximately 484,000 people had identified as Aboriginal, Torres Strait Islander, or both (around 66% of the estimated Indigenous population). Medicare data presented by Indigenous status have been adjusted for the under-identification in the Medicare VII database. Indigenous rates are therefore modelled and should be interpreted with caution. Estimates generated by the adjustment methodology for a given period will vary according to the point in time at which they are calculated, as the adjustment factors will be updated regularly to account for the ongoing change in the population coverage of the VII sample.
MBS data do not give a comprehensive picture of actual service delivery for Indigenous Australians, as some service delivery (particularly in Remote areas) is not part of these systems. For example, MBS data does not include services provided in the public health system or under other arrangements that do not attract an MBS claim (for example, some AMS and state/territory health services).