Technical notes
Data sources
Hospitalisations data are sourced from the Australian Institute of Health and Welfare’s (AIHW) National Hospital Morbidity Database (NHMD). The NHMD is a compilation of episode-level records from admitted patient morbidity data collection systems (APC NMDS) in Australian public and private hospitals. It includes episodes of care for admitted patients in all public and private acute and psychiatric hospitals, free standing day hospital facilities and alcohol and drug treatment centres in Australia. Hospitals operated by the Australian Defence Force, corrections authorities and in Australia's offshore territories may also be included. Hospitals specialising in dental, ophthalmic aids and other specialised acute medical or surgical care are included. Data quality statements for the NHMD are available on the AIHW MyHospitals website. For more information about data contained in the NHMD refer to the AIHW MyHospitals technical notes.
Emergency department data are sourced from the National Non-admitted Patient Emergency Department Care Database (NNAPEDCD). Data quality statements for this dataset are available on the AIHW MyHospitals website. For the 2021–22 NAPEDC NMDS/NBEDS, diagnosis information was reported using the ED ICD-10-AM version 11 shortlist that can be found on the website of the Independent Hospital Pricing Authority.
For more information about data contained in the NNAPEDCD refer to the MyHospitals technical notes for recent years.
Deaths data are sourced from the AIHW National Mortality Database (NMD). When a person is declared dead, information about their death is recorded on a death certificate by either a medical practitioner or coroner. Registration of all deaths is compulsory in Australia and is the responsibility of the Registrar of Births, Deaths and Marriages of the relevant state or territory, under jurisdiction-specific legislation.
Deaths data are assembled, coded, and published on behalf of the Registrars by statistical agencies. These agencies have varied since 1900 and have included state-based offices and what is now the Australian Bureau of Statistics (ABS). Information is also provided to the ABS via the National Coronial Information System (NCIS) for those deaths certified by a coroner.
The ABS codes causes of death according to the International statistical classification of diseases and related health problems, 10th revision (ICD-10) (WHO 2019) and, after checks and de-identification, creates the Cause of Death Unit Record File (CODURF).
The CODURF contains characteristics of the person who died (for example, age, sex, and Indigenous status), and characteristics of their death (for example, causes of death, date, and place where the person usually lived). The AIHW maintains these data in the NMD.
The data quality statements underpinning the AIHW NMD can be found on the following ABS internet pages:
- Australian Bureau of Statistics (2021), Deaths, Australia methodology
- Australian Bureau of Statistics (2021), Causes of Death, Australia methodology
For more information on mortality coding refer to Causes of Death, Australia Methodology, 2021, Australian Bureau of Statistics, ABS 2022).
Population data are used for demographic analyses and as the denominator in calculating rates. All population level calculations are based on the estimated resident population (ERP) calculated as at the midpoint of each financial year. For example, for the reporting period 2021–22, the denominator population is the June 2021 ERP + the June 2022 ERP, divided by 2. This is used as the denominator for age specific/crude and age standardised rates.
The ERP as at 30 June 2001 is used as the standardising population throughout the report (ABS 2003).
All population data are sourced from the Australian Bureau of Statistics (ABS) as follows:
- General populations are from National, state and territory population
- First Nations populations are from Estimates and Projections, Aboriginal and Torres Strait Islander Australians (ABS 2019)
- Remoteness populations (available on request from ABS)
Since estimates of Aboriginal and Torres Strait Islander (First Nations) populations are only provided for 30 June, estimates for 31 December are calculated by adding 2 consecutive 30 June estimates and dividing by 2 (for example, the estimate for 31 December 2021 is calculated by adding estimates for 30 June 2021 and 30 June 2022 then dividing by 2).
For non-Indigenous people, population denominators were derived by subtracting the estimated First Nations population from the Australian national estimated resident population, as at 31 December of the relevant year.
Injury hospitalisations
Injury case identification
A diagnosis of injury is defined as ICD-10-AM codes in the range S00–T75 or T79, using ‘Chapter 19 Injury, poisoning and certain other consequences of external causes’. A primary diagnosis of injury is when one of the specified codes is the first diagnosis code reported, while an additional diagnosis of injury is when one of the specified codes is reported but not as the first diagnosis.
A person may have more than one incident of injury resulting in hospitalisation in a financial year and each case of hospitalisation will be counted separately in this report. This is because we are counting incidents of injury resulting in hospitalisation, rather than the number of people who were hospitalised, in a given financial year. If a single incident led to an admission in more than one hospital, the incident has only been counted once. Therefore, counts of injury cases will be lower than the count of hospital records indicating injuries.
- Records with the maximal snapshot id in any database where the date of separation falls within the timeframe defined in the report.
- NHMD records with a principal diagnosis in the ICD 10 AM range S00–T75 or T79, using ‘Chapter 19 Injury, poisoning and certain other consequences of external causes’.
- NHMD records with a separation date between 1 July 2021 to 30 June 2022
- Records were excluded where the AIHW ‘standard analysis’ flag was absent, i.e. care type was newborn with unqualified days only (7.3), organ procurement - posthumous (9), or hospital boarder (10).
- Injuries due to Complications of surgical and medical care (T80 – T88) and Sequelae of injuries, of poisoning and of other consequences of external causes (T90 – T98) are excluded.
Each record in the NHMD refers to a single episode of care in a hospital. Some injury incidents result in more than one episode of care and, therefore, more than one record.
To minimise the impact of overcounting where a person experienced multiple episodes of care relating to the same condition, the following criteria are applied to estimate incidents:
- Excludes records where admission mode is transfer from another hospital (1)
- Excludes records where admission mode is statistical admission (2) and care type is not acute (1, 7.1, 7.2)
- Excluding records where care involving use of rehabilitation procedures (Z50) appears as an additional diagnosis and care type is not acute (1, 7.1, 7.2)
Diagnosis, intervention, activity, place of occurrence and external cause data for 2021–22 were reported to the NHMD using classifications from the 11th edition of the International statistical classification of diseases and related health problems, 10th revision, Australian modification (ICD-10-AM) (ACCD 2019a), incorporating the Australian classification of health interventions (ACHI).
In tables and figures, information on diagnoses, external causes, activity, place of occurrence and interventions are presented using the codes and abbreviated descriptions of the ICD-10-AM/ACHI. Full descriptions of the categories are available in ICD-10-AM/ACHI publications on the Independent Health and Aged Care Pricing Authority (IHACPA) website (ACCD 2019a, ACCD 2019b, ACCD 2019c).
Where data are presented in a time series incorporating previous reporting periods, these have been coded according to the following editions of ICD‑10‑AM:
- 7th edition for 2011–12 and 2012–13 hospital data
- 8th edition for 2013–14 and 2014–15 hospital data
- 9th edition for 2015–16 and 2016–17 hospital data
- 10th edition for 2017–18 and 2018–19 hospital data
- 11th edition for 2019–20, 2020-21 and 2021–22 hospital data
The NHMD is structured so that the first listed external cause for a record relates to the first listed injury diagnosis (principal diagnosis). While multiple external causes may be recorded for a separation, we report only one cause for each injury, referred to as ‘nominal external cause’ in these notes. The following steps are followed to determine the nominal external cause for each injury hospitalisation:
- The first reported external cause is taken to be the nominal external cause
- If the nominal external cause, as determined by step 1, is U90.0 (Staphylococcus aureus) or a supplementary factor (Y90–Y98), then the second reported code is taken to be the nominal external cause
- If the nominal external cause, after steps 1 and 2, relates to complications of medical and surgical care (Y40–Y84), sequelae of external causes of morbidity and mortality (Y85–Y89), or a supplementary factor code (Y90–Y98), then the record is excluded.
The categorisation of external causes using ICD-10-AM codes are detailed in Appendix tables to technical notes for Injury in Australia.
Type of injury and site of injury on the body is based on the patient’s principal diagnosis. Principal diagnosis is the diagnosis chiefly responsible for occasioning the episode of care for the patient as defined by ICD-10-AM codes. The principal diagnosis details the type of injury sustained such as fractures, dislocations, nerve injuries and burns, and the body part injured such as head, neck, ankle and foot.
To categorise injuries by type and body part injured, Injury in Australia’s principal diagnosis matrix has been applied (as outlined in the Appendix tables to technical notes for Injury in Australia).
The sum of injuries by body part may not equal the total number of hospitalised injury cases because some injuries are not described in terms of body region.
The Australian ERP as at 30 June 2001 is used as the standardising population throughout the report. Age‑standardisation of rates enables valid comparison across years and/or jurisdictions without being affected by differences in age distributions.
Population‑based rates of injury tend to have similar values from one year to the next. Exceptions to this can occur (for example, due to a mass‑casualty disaster), but are unusual in Australian injury data. Some year‑on‑year variation and short‑run fluctuations are to be expected, so small changes in a rate over a short period do not provide a firm basis for asserting that a trend is present.
All rate calculations utilise a denominator based on the estimated resident population (ERP) calculated as at the midpoint of each financial year. For example, for the reporting period 2021–22, the denominator population is the June 2021 ERP + the June 2022 ERP, divided by 2. This is used as the denominator for age specific/crude and age standardised rates. Rates are calculated for each financial year unless otherwise noted.
Measure | Numerator | Denominator | Calculation |
---|---|---|---|
Population (used for rates) | June 2021 population + June 2022 population | 2 | Numerator ÷ Denominator |
Crude or age-specific rate of hospitalisation | Number of cases of injury hospitalisation per defined category (e.g. age group) | Estimated Australian population as at mid-point of financial year | (Numerator ÷ Denominator) x 100,000 |
Age-standardised rate (ASR). | Expected events per age group in standard population= crude rate of hospitalisation x standard population (for each corresponding age group) | Nil | The direct method of standardisation is used. (Sum of numerators across all age groups ÷ total standard population) x 100,000 |
Change in rates | Nil | Nil | Estimated trends in age-standardised rates were reported as average annual percentage changes. |
Remoteness
Remoteness areas are based off the patient’s usual place of residence and are defined using the ABS’ Australian Statistical Geography Standard (ASGS) Remoteness Structure 2016 (ABS 2016). The ASGS Remoteness Structure 2016 categorises geographical areas in Australia into remoteness areas, described in detail on the ABS website which also includes detail of the nature of changes between the ASGS 2011 and ASGS 2016.
The remoteness classification is as follows:
- Major cities – for example, Sydney, Melbourne, Brisbane, Adelaide, Perth, Canberra and Newcastle
- Inner regional – for example, Hobart, Launceston, Wagga Wagga, Bendigo and Murray Bridge
- Outer regional – for example, Darwin, Moree, Mildura, Cairns, Charters Towers, Whyalla and Albany
- Remote – for example, Port Lincoln, Esperance, Queenstown and Alice Springs
- Very remote – for example, Mount Isa, Cobar, Coober Pedy, Port Hedland, Tennant Creek and Norfolk Island.
This report defines men as adult males over the age of 19. Therefore, only records where sex is specified as Male, and age is 19 or over, are included. Injury cases with missing age and/or sex information are not included in this analysis.
Percentages and rates (crude/age-specific and age-standardised) are rounded to 1 decimal place. Percentages may not add up to 100.0 because of rounding. Both crude/age-specific rates and age-standardised rates are calculated per 100,000 population.
Data may be suppressed to maintain the privacy or confidentiality of a person, or because a proportion or other measure is related to a small number of events and may therefore not be reliable. Data may also be suppressed to avoid attribute disclosure. The abbreviation ‘n.p.’ (not published) has been used in tables to denote these suppressions. The suppressed information remains in the totals.
The AIHW operates under a strict privacy regime which has its basis in Section 29 of the Australian Institute of Health and Welfare Act 1987 (AIHW Act). Section 29 requires that confidentiality of data relating to persons (living and deceased) and organisations be maintained. The Privacy Act governs confidentiality of information about living individuals. The AIHW is committed to reporting that maximises the value of information released for users while being statistically reliable and meeting legislative requirements described in the AIHW Act and the Privacy Act. Aggregated injury hospitalisations data are usually presented in tables, graphs, or maps. To maintain attribute disclosure and minimise risk of potentially re-identifying a person, data suppression rules have been applied.
Consequential suppression may also be applied to prevent a suppressed cell from being calculated. This is often done by suppressing table cells in the same row or column or suppressing the table totals.
Calculated data may also be suppressed due to quality and reliability reasons.
Counts
- Counts less than 5 are suppressed and consequential suppression is applied.
- When data are disaggregated by geography location, counts for areas where the population is less than 1,000 are suppressed.
Crude rates
- Crude rates with counts (numerator for calculation) less than 10 are suppressed.
- If the corresponding counts measure is suppressed, the crude rate has been suppressed.
- When data are disaggregated by geography location, counts for areas where the population is less than 100 are suppressed.
Age-standardised rates
- Age-standardised rates with counts (numerator for calculation) less than 20 are suppressed.
- If the corresponding counts measure is suppressed, the age-standardised rate has been suppressed.
- When data are disaggregated by geography location, counts for areas where the population is less than 30 are suppressed.
Z-score
No suppression applied.
A summary of data notes and data quality issues for the NHMD can be found in the Admitted Patient Care technical notes and appendices on MyHospitals.
The key issues are:
First Nations status
The AIHW report Indigenous identification in hospital separations data: quality report (AIHW 2013) presents the latest findings on the quality of First Nations identification in hospital separations data in Australia, based on studies conducted in public hospitals during 2011. Private hospitals were not included in the assessment. The findings indicate that, overall, the quality of First Nations identification in hospital separations data was similar to that achieved in a previous study (AIHW 2010). However, the survey for the 2013 report was performed on larger samples for each jurisdiction/region and is therefore considered more robust than the previous study. An estimated 88% of First Nations patients were correctly identified in Australian public hospital admission records in 2011-12 (AIHW 2013). This under counting of First Nations patients is a known issue across states and territories with proportions ranging from 58% (confidence interval, 46-69%) in the Australian Capital Territory and 98% (96-99%) in the Northern Territory over the same time period.
Variation in state and territory coding practices
Changes in New South Wales admission practice
The emergency department admission policy was changed for New South Wales (NSW) hospitals in 2017–18. Episodes of care delivered entirely within a designated emergency department or urgent care centre are no longer categorised as an admission regardless of the amount of time spent in the hospital. This narrowing of the categorisation has had the effect of reducing the number of admissions recorded in NSW from the 2017–18 financial year. For NSW, the effect was a significant decrease (3.7%) in all public hospital admissions in 2017–18 compared to 2016–17. The impact of the change was felt disproportionately among hospitalisations for injury and poisoning. According to NSW Health, the number of hospitalisations for injury and poisoning in NSW decreased by 7.6% between 2016–17 and 2017–18, compared to a usual yearly increase of 2.8% (Centre for Epidemiology and Evidence 2019).
The change in NSW’s emergency department admission policy may have had different effects on case numbers within different external cause categories. This is because different types of injury have a different likelihood of requiring prolonged care in an emergency department, but without an admission to a hospital ward.
Due to the size of the contribution of NSW data to the national total, Australian data from 2017–18 should therefore not be compared with data from previous years.
Cross-border flow of patients
Data on state or territory of hospitalisation should be interpreted with caution because of cross-border flows of patients. This is particularly the case for the Australian Capital Territory. In 2021–22, 19% of separations for Australian Capital Territory hospitals were for patients who lived in New South Wales.
Injury deaths
The underlying cause of death (UCoD) code represents the disease or injury that initiated the train of morbid events leading to a person’s death, according to information available to the coder. If a death was due to an injury, the ICD-10 requires that the external cause be entered as the UcoD.
Multiple causes of death (McoD) codes represent all the morbid conditions, diseases and injuries which are listed on the death certificate. They include all the factors in the morbid train of events leading to death: the underlying cause, the immediate cause, any intervening causes, and any conditions that contributed. This is especially helpful for chronic conditions, which often involve more than one illness.
Coding is according to the ICD-10 (WHO 2019), which includes a chapter for injuries and another for external causes.
A death due to injury is defined by 2 criteria:
- The UcoD was an external cause code in the range V01-Y36
- At least one McoD was an external cause in the range V01-X59 or Y10-Y34, and at least one other was a code for an injury (S00-T75 or T79). McoD is not considered for records where the UcoD indicates self-harm or assault.
The sum of the counts of death by cause may be greater than the total number of injury deaths because some deaths are due to multiple causes.
Deaths data are commonly recorded according to the calendar year in which the death was registered. However, in this report data are presented according to the financial year in which each death occurred.
The code range V01–Y36 includes all unintentional (accidental) deaths, intentional self-harm (suicide), homicides, and deaths where intent remained undetermined. These codes provide information around the circumstances of the death, such as details of a transport accident, drowning, asphyxiation, effects of radiation, heat, pressure, deprivation, and maltreatment.
The code range S00–T75 and T79 includes traumatic injuries (such as fractures and lacerations), burns, poisoning and toxic effects of substances. The codes also provide information about the single, multiple, or unspecified body regions affected: such as head, shoulder, knee and foot.
External cause | Cause-specific criteria |
---|---|
Transport | Records that included the following ICD‑10 codes were included:
Suicide and homicide deaths (UcoD X60–Y09) were excluded. |
Drowning | Records that included the following ICD 10 codes were included:
Suicide and homicide deaths (UCoD X60–Y09) were excluded. |
Choking and suffocation | Records that included the following ICD 10 codes were included:
Suicide and homicide deaths (UCoD X60–Y09) were excluded. |
Accidental poisoning | Records that included the following ICD 10 codes were included:
Suicide and homicide deaths (UCoD X60–Y09) were excluded. |
Falls | Records that included the following ICD 10 codes were included:
The codes for fractures are S02, S12, S22, S32, S42, S52, S62, S72, S82, S92, T02, T08, T10, T12, and T14.2. |
Thermal causes | Records that included the following ICD 10 codes were included:
Suicide and homicide deaths (UCoD X60–Y09) were excluded. |
Contact with objects | Records that included the following ICD 10 codes were included:
Suicide and homicide deaths (UCoD X60–Y09) were excluded. |
Electricity and air pressure | Records that included the following ICD 10 codes were included:
Suicide and homicide deaths (UCoD X60–Y09) were excluded. |
Contact with living things | Records that included the following ICD 10 codes were included:
Suicide and homicide deaths (UCoD X60–Y09) were excluded. |
Exposure to forces of nature | Records that included the following ICD 10 codes were included:
Suicide and homicide deaths (UCoD X60–Y09) were excluded. |
Overexertion, travel and privation | Records that included the following ICD 10 codes were included:
Suicide and homicide deaths (UCoD X60–Y09) were excluded. |
Suicide | Records that included the following ICD 10 codes were included: the UCoD was Intentional self-harm (X60–X84). |
Homicide | Records that included the following ICD 10 codes were included: the UCoD was Assault (X85–Y09) or Legal intervention and operations of war (Y35–Y36).vx |
Event of undetermined intent | Records that included the following ICD 10 codes were included:
|
Measure | Numerator | Denominator | Calculation |
---|---|---|---|
Population (used for rates) | June 2021 population + June 2022 population | 2 | Numerator ÷ Denominator |
Crude/age-specific rate of death | Number of injury deaths per defined category (e.g. age group) | Estimated Australian population as at mid-point of financial year | (Numerator ÷ Denominator) x 100,000 |
Age-standardised rate (ASR) | Expected events per age group in standard population= crude rate of hospitalisation x standard population (for each corresponding age group) | Nil | The direct method of standardisation is used. (Sum of numerators across all age groups ÷ total standard population) x 100,000 |
Changes in rates | Nil | Nil | Estimated trends in age-standardised rates were reported as average annual percentage changes. |
The sum of the counts of death by cause may be greater than the total number of injury deaths because some cases of death have multiple causes which contributed to the death. Cases with multiple causes are counted once per cause group.
This report defines men as adult males over the age of 19. Therefore, only records where sex is specified as Male, and age is 19 or over, are included. Injury cases with missing age and/or sex information are not included in this analysis.
Crude/age-specific rates and age-standardised rates are calculated per 100,000 population and are rounded to 1 decimal place (e.g. 3.4 per 100,000).
Data may be suppressed to maintain the privacy or confidentiality of a person, or because a proportion or other measure is related to a small number of events and may therefore not be reliable. Data may also be suppressed to avoid attribute disclosure. The abbreviation ‘n.p.’ (not published) has been used in tables to denote these suppressions. The suppressed information remains in the totals.
Counts
Counts less than 3 are suppressed and consequential suppression is applied.
Crude rates
- Crude rates with counts (numerator for calculation) less than 10 are suppressed.
- If the corresponding counts measure is suppressed, the crude rate has been suppressed.
Age-standardised rates
- Age-standardised rates with counts (numerator for calculation) less than 20 are suppressed.
- If the corresponding counts measure is suppressed, the age-standardised rate has been suppressed.
Z-score
No suppression applied.
Data quality statements
The data quality statements underpinning the AIHW National Mortality Database can be found in the following Australian Bureau of Statistics (ABS) publications:
- ABS quality declaration summary for Deaths, Australia methodology
- ABS quality declaration summary for Causes of Death, Australia methodology
- For more information on the AIHW National Mortality Database see Deaths data at AIHW.
Errors in deaths data
The data presented in this report are subject to 2 types of statistical error—non-random and random (a third type of statistical error, sampling error, does not apply in this report, because none of the data sources used involved probability sampling).
Non-random error
Some level of non-random error is to be expected in administrative data collections, such as the NMD on which this report relies. For example, non-random error could occur if the approach to assigning cause codes to deaths were to differ systematically between jurisdictions, or over time. While systems are in place to encourage uniform data collection, and coding and scrutiny of data during analysis include checking for patterns that might reflect non-random error, some error remains.
Random error
The values presented in the report are subject to random error, or variation. Variation is relatively large when the case count is small (especially if less than about 10). Variation is small enough to be mostly unimportant when the case count is larger (that is, more than a few tens of cases).
Some of the topics for which results are reported compare groups that vary widely in case count, largely due to differences in population size (for example, the population of New South Wales is more than 30 times as large as the Northern Territory population, and the population of Major cities is nearly 90 times that of Very remote areas). In this situation, year to year changes in counts or rates for the smaller-population groups might be subject to large random variation. Such fluctuations could potentially be misinterpreted as meaningful rises or falls.
ABS (Australian Bureau of Statistics) (2003) Population by age and sex, Australian states and territories, 2001: Census edition final. ABS cat. no. 3201.0. Canberra: ABS.
ABS (2016) Australian Statistical Geography Standard (ASGS): Volume 5—Remoteness structure, July 2016. ABS cat. no. 1270.0.55.005. Canberra: ABS.
ABS (2019) Estimates and projections, Aboriginal and Torres Strait Islander Australians, 2006 to 2031. ABS cat. no. 3238.0. Canberra: ABS.
ACCD (Australian Consortium for Classification Development) (2019a). The international statistical classification of diseases and related health problems, 10th revision, Australian modification (ICD-10-AM), 11th edn. Tabular list of diseases and alphabetic index of diseases. Adelaide: Independent Hospital Pricing Authority (IHPA), Lane Publishing.
ACCD (2019) The Australian classification of health interventions (ACHI), 11th edn. Tabular list of interventions and alphabetic index of interventions. Adelaide: IHPA, Lane Publishing.
ACCD (2019) The international statistical classification of diseases and related health problems, 10th revision, Australian modification (ICD-10-AM), 11th edn. Australian coding standards for ICD-10-AM and ACHI. Adelaide: IHPA, Lane Publishing.
AIHW (2013) Indigenous identification in hospital separations data: quality report. Cat. no. IHW 90. Canberra: AIHW.
WHO (World Health Organization) (2019.) The international statistical classification of diseases and related health problems, 10th revision (ICD-10) Geneva: WHO. Accessed 7 May 2024.