To promote internal consistency and objectivity, the following principles were applied:
- Attribute the burden to the condition where the health loss was experienced (‘prevalence principle’). This principle was used mostly when mapping diseases or conditions that can be a long-term result of an earlier condition; diseases that are risk factors or sequelae for other diseases; or diseases that can be counted in more than one disease group. Examples include:
-
the burden from liver cancer or chronic liver disease due to hepatitis was counted where the condition manifested or was experienced (that is, in cancer or gastrointestinal conditions), not as a long-term sequelae of hepatitis. This is consistent with global studies and with the mapping practice for other conditions that are now known to be the result of previous infectious diseases.
-
the overlap in cardiovascular disease, chronic kidney disease and diabetes was dealt with by attributing the health loss to the condition experienced, rather than the underlying cause (for example, renal complications due to diabetes mellitus was counted under chronic kidney disease). The AIHW explored the overlap between these diseases to quantify their indirect impacts and collective burden. Results from these studies were published in the report Diabetes and chronic kidney disease as risks for other diseases (AIHW 2016c).
-
Classify diseases according to Australian disease monitoring activities. Australian disease monitoring classifications were given priority over the GBD to provide better information for Australian health priority setting. For example, the GBD classified all neoplasms together, regardless of malignancy. In Australia, monitoring of neoplasms is restricted to malignant neoplasms, so they were classified separately to other neoplasms.
The proposed mappings of ICD-10 codes to diseases in the ABDS disease list were reviewed by disease specific expert groups before being finalised.
Assigning diseases to disease groups
Under the ABDS disease hierarchy, each disease is allocated to a single disease group. The allocation of particular diseases to a disease group affects the estimates of burden and ranking by disease group that are reported in the published analyses. Alternative disease group presentations of the ABDS 2018 results can be readily developed from the existing disease list. For example, gastrointestinal disorders do not include gastrointestinal infections, or gastrointestinal cancers, but the estimates for these diseases could be added to the gastrointestinal disorders group to obtain a broader picture of the burden for this area of interest.
For the most part, assigning diseases to disease groups relied heavily on the chapter structure of ICD-10. However, for a small number of diseases it was less straightforward, as they appeared potentially to bear some characteristics of more than one group. These diseases were allocated after discussion with experts from both potential disease groups, and, as with the prevalence principle, assigned according to where the health loss is actually experienced.
Major decisions referred to experts for advice included:
-
suicide and self-inflicted injuries – the burden was included under injuries, consistent with ICD-10 coding and previous national and GBD studies.
-
accidental poisonings involving drugs and alcohol (ICD-10 codes X41, 42 and 45) – the burden was included under injuries rather than substance use disorders, consistent with coronial assessment, on the basis that where the coroner found evidence of an underlying dependence, the cause of death would reflect this and be assigned to substance use disorders. The drug and alcohol experts expressed concerns about the reliability of distinctions between opioid overdose fatalities that are due to accidental overdose or those due to opioid dependence. There is evidence in Australian studies that most overdose deaths occur among people with a history of dependence, and very few deaths are deliberate. However, as the coding for X42 (Accidental poisoning by and exposure to narcotics and psychodysleptics [hallucinogens], not elsewhere classified) includes several drugs, not just opioids, this assumption would have to be made for those other drugs as well.
-
gestational diabetes – the burden was counted in the reproductive & maternal disease group, rather than endocrine disorders, due to this condition only arising during pregnancy, and is consistent with previous national and GBD studies.
-
cerebral palsy – the burden was allocated to the infant & congenital conditions disease group, rather than neurological conditions, as, in most cases, cerebral palsy is acquired in the prenatal and perinatal period and emerges as a leading cause of death for children aged under 5. As a sequela, cerebral palsy is acquired through several other infant & congenital conditions, such as birth trauma and birth asphyxia
-
fetal alcohol spectrum disorders (FASD) – although counted under mental health and substance use disorders in the GBD 2010, the burden was assigned to infant & congenital conditions in the ABDS as the main sequelae are learning difficulties and disfigurement, and the burden is experienced by the child (not the mother).
-
postnatal depression – the burden was not included as a separate disease in the ABDS due to data limitations. As available data did not distinguish whether the depressive disorder was associated with childbirth, postnatal depression was included in estimates for depressive disorders, within the mental & substance use disorders disease group. This is consistent with previous national and GBD studies.
Selection and assessment of data sources
All potential data sources to estimate disease burden (whether published or unpublished) were assessed for comparability, relevance, representativeness, currency, accuracy, validation, credibility and accessibility/timeliness (see Additional material for the guidelines used to direct data selection). Only data sources that met the guidelines were included in the study.
Potential data sources were required to: have case definitions appropriate to the disease or risk factor being analysed; be relevant to the Australian population; and be timely, accurate, reliable and credible. Where possible, national data sources, rather than sources relating to particular regions or subpopulations, were used.
Administrative data sources (for example, disease registers, hospitalisations) were evaluated for their level of ascertainment (how well the data correspond to the disease or sequela in question) and coverage (the proportion of the population included in the data).
Surveys were evaluated for their representativeness, potential selection bias, and measurement bias (validity and reliability of measurement).
Epidemiological studies were evaluated for the quality of their study design, their timeliness, credibility, representativeness, and sources of bias or error.
There are new data sources for many diseases in the ABDS 2018, notably greater use of linked hospital/deaths data.
The key data source used in estimating mortality is described in Estimating fatal burden, and key data sources used in estimating morbidity are listed in Estimating non-fatal burden.
Methodological choices specific to Indigenous estimates
Additional factors needed to be considered when calculating burden of disease estimates for Aboriginal and Torres Strait Islander people. As a general principle in the ABDS, the methods used to produce Indigenous burden of disease estimates were consistent with those used to produce national estimates. For example, the same reference life table, disability weights and disease list were used. However, it was not always possible to adopt completely consistent methods due to differences in data availability, data quality and population size and characteristics.
Indigenous under-identification
While in recent decades major improvements have been made to the quality and availability of information about Indigenous Australians, existing data are subject to several limitations regarding data quality and availability. These include under-identification of Indigenous Australians in administrative data sets (and changes in people’s inclination to identify as Indigenous over time), and lack of available data on the prevalence of certain diseases in the Indigenous population. Methods employed to address these issues in the ABDS are discussed in the relevant sections of this report on fatal and non-fatal burden.
Dealing with small numbers
An important consideration for Indigenous burden of disease is the robustness and reliability of estimates produced, and the level of disaggregation supported by the data, given the small size of the Indigenous population compared with the much larger non-Indigenous population.
To ensure validity of the results, the AIHW combined several years of data and/or age groups as necessary to produce Indigenous estimates. Additionally, the level of disaggregation used to report Indigenous estimates was broader than that reported for the total Australian population. This included collapsed age groups for those aged 0–4 and 85 and over.
Measuring the gap between Indigenous and non-Indigenous Australians
Direct age-standardisation was used to compare rates between Indigenous and non-Indigenous Australians, and to measure the gap in burden between the 2 populations. The direct method was chosen, following a series of sensitivity analyses undertaken by the AIHW, which looked at the impact and robustness of using the direct method compared with the indirect method on resulting Indigenous YLL estimates (see AIHW 2015 for more information). The direct method enables multiple comparisons (for example, disease by sex) and can be used for comparisons over time. A limitation of the direct method is that less reliable estimates can be produced when it is applied to a small number of deaths and prevalent cases; this should be kept in mind when interpreting gap results for less common diseases and conditions.
Age-standardised rate differences and rate ratios were reported as measures of the gap. Rate differences provide a measure of the absolute gap between 2 populations, while rate ratios are a measure of the relative gap between 2 populations.
For the most accurate estimate of the gap in disease burden between Indigenous and non-Indigenous Australians, comparisons have been made to estimates calculated for the non-Indigenous population. Estimates for the total Australian population should not be compared with those for Indigenous population.
Choice of population denominator for Indigenous estimates
In estimating the Aboriginal and Torres Strait Islander population for the years prior to each Census, the Australian Bureau of Statistics makes a number of assumptions regarding past mortality rates, migration and improvements in life expectancy. As such, several population backcast and projection series are produced in addition to the Estimated Resident Population for each Census year.
Following sensitivity analyses by the AIHW to look at the impact of using different Indigenous population denominators in burden of disease rate calculations, it was agreed to use the backcast population series based on the 2016 Census, which applies the Indigenous identification level in 2016 to earlier years. Using this backcast population for all reference years provides consistency between the denominators used for the Indigenous burden of disease estimates in the ABDS 2018.
For more information on these choices, see Impact and causes of illness and death in Aboriginal and Torres Strait Islander people 2018 (AIHW forthcoming 2022).
Methodological choices specific to sub-national estimates
Sub-national estimates include state/territory, remoteness categories and socioeconomic groups. These are defined as:
- state and territory classifications – the 8 Australian jurisdictions: New South Wales, Victoria, Queensland, Western Australia, South Australia, Tasmania, Australian Capital Territory and Northern Territory. Disaggregation by state/territory is well supported by the data, with the majority of data sources (except for epidemiological studies and small surveys) defining and reporting state or territory in a standard way.
- remoteness categories – based on the 2016 Australian Statistical Geographic Standard (ASGS) for 2018 and 2015 estimates, or the 2011 ASGS for 2011 estimates. The ASGS is divided into 5 remoteness areas: Major cities, Inner regional, Outer regional, Remote and Very remote. Remoteness areas aggregate to states and territories and cover the whole of Australia. Most major data sources, except for epidemiological studies and small surveys, were able to be broken down by remoteness area. This study reported estimates for 4 remotes areas: Remote and Very remote were combined.
- socioeconomic groups – presented as quintiles of lowest to highest socioeconomic position. Ideally, it would be better if detailed individual-level measures of socioeconomic characteristics were available in key data sources. But the most consistently available approach across the national data sources was the geographically-based proxy of socioeconomic group based on the relative socioeconomic characteristics of the area of residence, known as SEIFA (Socio-Economic Indexes for Areas). SEIFA is a measure of socioeconomic disadvantage developed by the ABS that ranks geographic areas in Australia according to relative socioeconomic advantage and disadvantage. The ABS broadly defines relative socioeconomic advantage and disadvantage in terms of ‘people’s access to material and social resources and their ability to participate in society’. The AIHW generally reports analyses of socioeconomic differences using SEIFA divided into population-based quintiles. It is also the standard for the majority of national agreement indicators. This approach ensures that, regardless of the underlying geographical unit, about 20% of the population is allocated to each quintile. SEIFA contains 4 indexes, with the Index of Relative Socioeconomic Disadvantage (IRSD) historically being the most commonly used at the AIHW for health-related analyses. For more information on SEIFA. SEIFA was only used for disaggregation of national estimates.
Sub-national methodology
Sub-national estimates were based on breaking down national estimates at a level of disaggregation (disease, sex and broad age group) supported by the underlying data, rather than being derived using separate data sources. This ensured that comparisons across each disaggregation were based on common data definitions, which is often not the case when sub‑national data sources are combined.
The preferred approach for sub-national estimates was to derive sub-national disaggregation directly from the primary data source using geographical identifiers. When this was not available, secondary data sources were used to identify health loss gradients between the sub-national regions that could then be applied to the national data. Lastly, when neither of these approaches were possible, the national sex/age prevalence rates were applied to the population structure of the sub-national unit. This assumed no difference in disease prevalence rates between sub-national and national populations.
Specific details on the methods used for sub-national estimates for mortality and morbidity are included in Disease specific methods – mortality and Disease specific methods – morbidity.
Key considerations
The validity of sub-national results is influenced by the availability and quality of data at the level of disaggregation, and by the population size in the various groups.
For state and territory estimates, analyses used the same age groups as the national analysis. For remoteness and socioeconomic group analyses, age groups were restricted to 5-year age groups 0, 1–4, 5–9, …, 85+ to overcome limitations with data.
Indigenous sub-national estimates
Indigenous sub-national estimates were considered reliable to calculate and report at the disease group level, but not at the specific disease level. This was due to:
- limited availability of Indigenous data for individual diseases at the geographical levels of interest
- limited availability of Indigenous identification adjustment factors at sub-national levels for relevant administrative data collections
- small numbers if Indigenous estimates were broken down at sub-national levels.
Indigenous sub-national estimates were considered adequate to report for 4 states and territories (New South Wales, Queensland, Western Australia, and the Northern Territory). Estimates were not calculated for Victoria, South Australia, Tasmania or the Australian Capital Territory due to small numbers of Indigenous deaths in these jurisdictions, and lack of suitable mortality adjustment factors. However, these jurisdictions account for the majority of burden in most cases (see Table 2.3).
Estimates for all 5 categories of remoteness were reported (Major cities, Inner regional, Outer regional, Remote and Very remote).
For Indigenous burden estimates by level of socioeconomic disadvantage, an Indigenous-specific index (the Indigenous Relative Socioeconomic Outcomes Index) (Biddle & Markham 2017) was used. This was considered to more accurately reflect levels of disadvantage in the Indigenous population than the SEIFA index used for the national component. As such, the Indigenous estimates by socioeconomic disadvantage were not compared with national estimates by socioeconomic disadvantage.
Indigenous sub-national estimates of YLL were calculated directly from mortality data (adjusted for Indigenous under-identification) using state/territory and remoteness specific adjustment factors.
Hospitalisation data (adjusted for under-identification), ABS health survey data (2018–19 Australian Aboriginal and Torres Strait Islander Health Survey; 2012–2013 AATSIHS biomedical data), or population proportions (depending on the disease group) were used to break down the national-level Indigenous YLD into subnational categories. Hospitalisation data were used for 10 disease groups, and health survey data were used for 5 disease groups for state/territory and remoteness estimates. A combination of Indigenous health survey data and data from the Australia and New Zealand Dialysis and Transplant Registry (ANZDATA) was used for 1 disease group (kidney & urinary diseases), and the subnational Indigenous population structure distribution data was used for the final disease group (skin disorders). For estimates by socioeconomic group, hospitalisation data were used for all disease groups, as Statistical Area Level 2 data (required to calculate the Indigenous Relative Socioeconomic Outcomes Index) were available from this data collection.
The data sources used to break down Indigenous YLD into subnational categories can be found in Table 2.2. The proportions used to break down Indigenous YLD estimates for each disease group can be found in tables 2.3 to 2.5.
State-level data were not generally used to build the national burden of disease estimates for the Indigenous population (that is, fatal burden estimates were calculated using national mortality adjustment factors, and non-fatal burden estimates were largely calculated using national prevalence estimates sourced from national data collections). As a result, Indigenous estimates reported at the national level are not subject to the same data quality issues as the state and territory estimates.
For more information on the methods used for Indigenous subnational estimates see Australian Burden of Disease Study: impact and causes of illness and death in Aboriginal and Torres Strait Islander people 2018 (AIHW forthcoming 2022).
Table 2.2: Data source used for subnational distribution of 2018 Indigenous non–fatal burden estimates
|
State/territory |
Remoteness |
Socioeconomic group |
Blood/metabolic |
Adjusted hospitalisations |
Adjusted hospitalisations |
Adjusted hospitalisations |
Cancer |
Adjusted hospitalisations |
Adjusted hospitalisations |
Adjusted hospitalisations |
Cardiovascular |
Adjusted hospitalisations |
Adjusted hospitalisations |
Adjusted hospitalisations |
Endocrine |
2012–13 AATSIHS |
2012–13 AATSIHS |
Adjusted hospitalisations |
Gastrointestinal |
Adjusted hospitalisations |
Adjusted hospitalisations |
Adjusted hospitalisations |
Hearing/vision |
2018–19 NATSIHS |
2018–19 NATSIHS |
Adjusted hospitalisations |
Infant/congenital |
Adjusted hospitalisations |
Adjusted hospitalisations |
Adjusted hospitalisations |
Infections |
Adjusted hospitalisations |
Adjusted hospitalisations |
Adjusted hospitalisations |
Injuries |
Adjusted hospitalisations |
Adjusted hospitalisations |
Adjusted hospitalisations |
Kidney/urinary |
2012–13 AATSIHS and ANZDATA |
2012–13 AATSIHS and ANZDATA |
Adjusted hospitalisations |
Mental & substance use |
2018–19 NATSIHS |
2018–19 NATSIHS |
Adjusted hospitalisations |
Musculoskeletal |
2018–19 NATSIHS |
2018–19 NATSIHS |
Adjusted hospitalisations |
Neurological |
Adjusted hospitalisations |
Adjusted hospitalisations |
Adjusted hospitalisations |
Oral |
Adjusted hospitalisations |
Adjusted hospitalisations |
Adjusted hospitalisations |
Reproductive/maternal |
Adjusted hospitalisations |
Adjusted hospitalisations |
Adjusted hospitalisations |
Respiratory |
2018–19 NATSIHS |
2018–19 NATSIHS |
Adjusted hospitalisations |
Skin |
Population distribution |
Population distribution |
Adjusted hospitalisations |
Note: AATSIHS = Australian Aboriginal and Torres Strait Islander Health Survey, ANZDATA = Australia and New Zealand Dialysis and Transplant Registry, NATSIHS = National Aboriginal and Torres Strait Islander Health Survey.
Table 2.3: Sub-national proportions used for distribution of 2018 non–fatal burden estimates by state/territory, by Indigenous status
|
Indigenous
NSW |
Indigenous
Qld |
Indigenous
WA |
Indigenous
NT |
Indigenous
Remainder |
Non-Indigenous
NSW |
Non-Indigenous
Qld |
Non-Indigenous
WA |
Non-Indigenous
NT |
Non-Indigenous
Remainder |
Blood/metabolic |
20.8 |
25.5 |
14.4 |
22.5 |
16.8 |
23.8 |
21.0 |
11.9 |
0.4 |
42.9 |
Cancer |
30.5 |
27.6 |
11.7 |
7.5 |
22.7 |
27.5 |
23.1 |
10.6 |
0.4 |
38.4 |
Cardiovascular |
26.8 |
27.6 |
15.0 |
12.1 |
18.5 |
30.5 |
20.3 |
9.6 |
0.6 |
39.0 |
Endocrine |
32.8 |
25.0 |
17.2 |
14.8 |
10.2 |
35.6 |
21.7 |
9.6 |
0.9 |
32.2 |
Gastrointestinal |
30.0 |
24.8 |
14.5 |
10.1 |
20.6 |
30.9 |
20.5 |
9.7 |
0.6 |
38.3 |
Hearing/ vision |
33.4 |
27.9 |
10.8 |
6.8 |
21.1 |
31.3 |
20.2 |
10.0 |
0.6 |
37.9 |
Infant/ congenital |
31.4 |
27.2 |
11.6 |
10.0 |
19.8 |
32.6 |
18.5 |
10.8 |
0.6 |
37.5 |
Infections |
23.2 |
25.3 |
17.0 |
21.4 |
13.1 |
29.7 |
21.7 |
9.7 |
0.6 |
38.3 |
Injuries |
24.1 |
24.7 |
19.4 |
16.7 |
15.1 |
30.1 |
20.4 |
10.5 |
0.8 |
38.2 |
Kidney/ urinary |
16.1 |
24.6 |
17.2 |
27.4 |
14.7 |
56.7 |
3.4 |
1.5 |
1.5 |
36.9 |
Mental & substance use |
32.7 |
26.9 |
12.7 |
8.3 |
19.4 |
31.2 |
19.6 |
9.6 |
0.6 |
39.0 |
Musculoskeletal |
37.6 |
25.1 |
10.5 |
5.0 |
21.8 |
31.8 |
19.4 |
9.9 |
0.5 |
38.4 |
Neurological |
29.9 |
27.6 |
13.8 |
8.7 |
20.0 |
25.0 |
22.4 |
10.6 |
0.4 |
41.6 |
Oral |
22.4 |
27.4 |
14.0 |
13.1 |
23.1 |
25.3 |
18.8 |
13.8 |
0.4 |
41.7 |
Reproductive/ maternal |
27.8 |
29.7 |
13.5 |
12.4 |
16.6 |
30.3 |
21.3 |
10.6 |
1.0 |
36.8 |
Respiratory |
38.0 |
26.2 |
10.9 |
3.8 |
21.1 |
29.3 |
19.7 |
9.9 |
0.5 |
40.6 |
Skin |
33.2 |
27.8 |
12.6 |
9.2 |
17.2 |
31.9 |
19.8 |
10.3 |
0.7 |
37.3 |
Note: See Table 2.2 for data sources used for proportional splits.
Table 2.4a: Sub-national proportions used for distribution of 2018 non–fatal burden by remoteness, by Indigenous status
|
Indigenous
Major cities |
Indigenous
Inner regional |
Indigenous
Outer regional |
Indigenous
Remote |
Indigenous
Very remote |
Blood/metabolic
|
28.0
|
19.7
|
20.2
|
11.2
|
20.9
|
Cancer
|
37.5
|
26.4
|
20.4
|
6.7
|
9.0
|
Cardiovascular
|
29.0
|
20.0
|
23.0
|
10.9
|
17.1
|
Endocrine
|
27.3
|
16.4
|
17.2
|
11.7
|
28.1
|
Gastrointestinal
|
32.6
|
24.5
|
23.5
|
8.8
|
10.6
|
Hearing/ vision
|
40.9
|
25.2
|
18.8
|
6.1
|
9.0
|
Infant/ congenital
|
37.5
|
26.1
|
19.9
|
6.5
|
10.0
|
Infectious diseases
|
26.4
|
17.4
|
20.3
|
14.0
|
21.9
|
Injuries
|
30.7
|
18.9
|
20.4
|
11.4
|
18.6
|
Kidney/ urinary
|
9.2
|
11.2
|
35.6
|
22.3
|
21.7
|
Mental
|
35.7
|
27.0
|
19.6
|
6.3
|
11.2
|
Musculoskeletal
|
41.7
|
24.8
|
20.6
|
6.0
|
7.1
|
Neurological
|
38.4
|
26.1
|
18.5
|
8.3
|
8.7
|
Oral
|
34.4
|
22.7
|
19.9
|
8.7
|
14.4
|
Reproductive/ maternal
|
35.0
|
22.6
|
23.2
|
7.5
|
11.8
|
Respiratory
|
44.6
|
27.3
|
18.5
|
4.5
|
5.1
|
Skin
|
37.7
|
23.9
|
20.2
|
6.6
|
11.6
|
Note: See Table 2.2 for data sources used for proportional splits.
Table 2.4b: Sub-national proportions used for distribution of 2018 non–fatal burden by remoteness, by Indigenous status
|
Non-Indigenous
Major cities |
Non-Indigenous
Inner regional |
Non-Indigenous
Outer regional |
Non-Indigenous
Remote |
Non-Indigenous
Very remote |
Blood/ metabolic
|
70.4
|
19.5
|
8.8
|
1.0
|
0.3
|
Cancer
|
68.3
|
21.6
|
8.9
|
0.9
|
0.3
|
Cardiovascular
|
67.2
|
22.1
|
9.3
|
1.0
|
0.4
|
Endocrine
|
64.8
|
22.2
|
11.1
|
1.4
|
0.5
|
Gastrointestinal
|
69.6
|
20.6
|
8.5
|
0.9
|
0.4
|
Hearing/ vision
|
70.8
|
19.2
|
8.2
|
1.3
|
0.5
|
Infant/ congenital
|
74.1
|
17.6
|
7.1
|
1.0
|
0.3
|
Infectious diseases
|
69.7
|
19.9
|
8.9
|
1.1
|
0.4
|
Injuries
|
69.6
|
19.9
|
8.9
|
1.1
|
0.5
|
Kidney/ urinary
|
77.0
|
12.1
|
10.7
|
0.1
|
0.0
|
Mental
|
71.6
|
18.0
|
9.0
|
0.9
|
0.3
|
Musculoskeletal
|
68.4
|
20.9
|
8.9
|
1.4
|
0.5
|
Neurological
|
71.6
|
20.0
|
7.4
|
0.7
|
0.3
|
Oral
|
72.2
|
18.6
|
8.0
|
0.9
|
0.3
|
Reproductive/ maternal
|
74.6
|
15.8
|
8.2
|
1.1
|
0.4
|
Respiratory
|
72.2
|
18.2
|
7.8
|
1.2
|
0.5
|
Skin
|
73.2
|
17.6
|
7.8
|
1.0
|
0.4
|
Table 2.5: Subnational proportions used for distribution of 2018 non–fatal burden by socioeconomic group (IRSEO Index), Indigenous Australians
|
1 (most disadvantaged) |
2 |
3 |
4 |
5 (least disadvantaged) |
Blood/ metabolic |
19.2 |
20.0 |
24.9 |
16.9 |
19.0 |
Cancer |
25.5 |
27.6 |
25.0 |
14.5 |
7.4 |
Cardiovascular |
18.6 |
21.9 |
25.3 |
18.9 |
15.4 |
Endocrine |
17.3 |
22.8 |
23.9 |
19.2 |
16.8 |
Gastrointestinal |
22.4 |
25.3 |
26.0 |
17.5 |
8.9 |
Hearing/ vision |
21.0 |
24.5 |
22.6 |
17.4 |
14.5 |
Infant/ congenital |
21.9 |
27.3 |
27.7 |
14.9 |
8.3 |
Infections |
15.9 |
20.5 |
24.1 |
19.0 |
20.5 |
Injuries |
20.4 |
22.5 |
22.8 |
17.4 |
16.9 |
Kidney/ urinary |
22.6 |
24.4 |
25.0 |
16.3 |
11.8 |
Mental & substance use |
27.7 |
24.0 |
23.8 |
15.1 |
9.4 |
Musculoskeletal |
27.2 |
26.5 |
23.1 |
14.1 |
9.2 |
Neurological |
26.8 |
25.6 |
25.3 |
14.9 |
7.5 |
Oral |
23.1 |
25.8 |
23.9 |
16.5 |
10.6 |
Reproductive/ maternal |
20.6 |
26.4 |
27.2 |
15.9 |
10.0 |
Skin |
15.6 |
21.2 |
22.9 |
20.5 |
19.8 |
Note: All proportions calculated from the NHMD.
Methodological choices specific to 2003, 2011, 2015 and 2018 estimates
Comparable YLL, YLD, DALY and attributable burden estimates were produced for each disease for the national population. Sub-national estimates for 2003 were not within the scope of this study.
As the 2003, 2011, 2015 and 2018 estimates are point-in-time estimates, their comparison with each other does not constitute a time-series analysis. Several issues must be considered before analysing and interpreting time trend data. A key issue is that 4 points in time can provide misleading information about changes over time – assuming that there is a straight-line trend between these 4 points might mask variation that exists but is not measured in this analysis, and results must be interpreted with this in mind. In addition, interpretation of changes over time also needs to take into account other aspects, such as the impact of confounders over time related to the estimates, and changes in metadata between reference periods. Any major changes between the previous studies and 2018 data that have an impact on the interpretation are highlighted in the relevant chapters in this report.
2015, 2011 and 2003 estimates
Where there were no changes in methods or data sources, the 2015 estimates from the ABDS 2015 were adjusted because the underlying population estimates were updated following the release of the 2016 Census. In contrast, the 2011 and 2003 estimates were kept the same because the underlying populations are based on 2011 census. If there were changes in methods or data source for 2018, the estimates for previous years were re-estimated using the new methods to keep comparability across all four years.
Specific details on methods for previous years estimates for mortality, morbidity and risk factors are included in disease specific methods – mortality, disease specific methods – morbidity and risk factor specific methods.
Indigenous 2011 and 2003 estimates
Issues relating to changing Indigenous identification over time and potential inconsistencies in identification in numerator data and population denominators have an impact on the comparability of Indigenous burden of disease rates over time. These issues also have implications on the choice of population denominator used for 2003 and 2011 Indigenous burden of disease estimates.
Where possible, adjustments have been made to account for changes in Indigenous identification over time in the numerator data used for rate calculations of disease burden. For example, Indigenous deaths and hospitalisations for 2003 and 2011 estimates were adjusted using factors based on identification levels relevant to these reference years.
The population denominator used for 2003 and 2011 Indigenous burden of disease estimates were consistent in terms of Indigenous identification with that used for 2018 estimates, which is important for assessing rate changes over time. Indigenous population estimates based on the 2016 Census were used, which applies the Indigenous identification level in 2016 to earlier years in the series, including for 2011 and 2003.
Additional material