Causes of Death, Australia methodology

1 This publication contains statistics on causes of death for Australia, together with selected statistics on perinatal deaths.

2 Statistics on perinatal deaths for the 2007-2009 reference years were published separately in Perinatal Deaths, Australia, 2009 (cat. no. 3304.0).

3 In order to complete a death registration, the death must be certified by either a doctor using the Medical Certificate of Cause of Death, or by a coroner. In 2018, 88.1% of deaths were certified by a doctor. The remaining 11.9% were certified by a coroner.

4 In order to complete a perinatal death registration, the death must be certified by either a doctor, using the Medical Certificate of Cause of Perinatal Death, or by a coroner. In 2018, 97.4% of perinatal deaths were certified by a doctor, with the remaining 2.6% certified by a coroner.

5 Although there is variation across jurisdictions in what constitutes a death that is reportable to a coroner, they are generally reported in circumstances such as:

• where the person died unexpectedly and the cause of death is unknown
• where the person died in a violent or unnatural manner
• where the person died during, or as a result of an anaesthetic
• where the person was 'held in care' or in custody immediately before they died
• where the identity of the person who has died is unknown.
   

6 The registration of deaths is the responsibility of the eight individual state and territory Registrars of Births, Deaths and Marriages. As part of the registration process, information about the cause of death is supplied by the medical practitioner certifying the death or by a coroner. Other information about the deceased is supplied by a relative or other person acquainted with the deceased, or by an official of the institution where the death occurred. The information is provided to the Australian Bureau of Statistics (ABS) by individual Registrars for coding and compilation into aggregate statistics. In addition, the ABS supplements this data with information from the National Coronial Information System (NCIS). The following diagram shows the process undertaken in producing cause of death statistics for Australia.

The diagram below outlines the Australian Cause of Death Statistics System. Each death is certified by either a doctor or coroner and the resultant information is provided to the Australian Bureau of Statistics (ABS) through the Registrar of Births, Deaths and Marriages in each state or territory. Information is also provided via the National Coronial Information System for those deaths certified by a coroner. The ABS processes, codes and validates this information, which is then provided in statistical outputs.

7 Ideally, for compiling annual time series, the number of deaths should be recorded and reported as those which occurred within a given reference period, such as a calendar year. However, there can be lags in the registration of deaths with the state or territory registries and so not all deaths are registered in the year that they occur. There may also be further delays to the ABS receiving notification of the death from the registries due to processing or data transfer lags. Therefore, every death record will have:

• a date on which the death occurred (the date of occurrence)
• a date on which the death is registered with the state and territory registry (date of registration); and
• a date on which the registered death is lodged with the ABS and deemed in scope.

8 With exception to the statistics published by Year of Occurrence section (Data Cube 13), all deaths referred to in this publication relate to the number of deaths registered, not those which actually occurred, in the years shown.
 

9 The scope for each reference year of the death registrations includes:

• deaths registered in the reference year and received by the ABS in the reference year;
• deaths registered in the reference year and received by the ABS in the first quarter of the subsequent year; and
• deaths registered in the years prior to the reference year but not received by ABS until the reference year or the first quarter of the subsequent year, provided that these records have not been included in any statistics from earlier periods.

10 From 2007 onwards, data for a particular reference year includes all deaths registered in Australia for the reference year that are received by the ABS by the end of the March quarter of the subsequent year. Death records received by the ABS during the March quarter of 2019 which were initially registered in 2018 (but for which registration was not fully completed until 2019) were assigned to the 2018 reference year. Any registrations relating to 2018 which were received by the ABS from April 2019 will be assigned to the 2019 reference year. Approximately 4% to 7% of deaths occurring in one year are not registered until the following year or later.

11 Prior to 2007, the scope for the reference year of the Death Registrations collection included:

• deaths registered in the reference year and received by the ABS in the reference year;
• deaths registered in the reference year and received by the ABS in the first quarter of the subsequent year; and
• deaths registered during the two years prior to the reference year but not received by the ABS until the reference year.

12 The ABS Causes of Death collection includes all deaths that occurred and were registered in Australia, including deaths of persons whose usual residence is overseas. Deaths of Australian residents that occurred outside Australia may be registered by individual Registrars, but are not included in ABS deaths or causes of death statistics.

13 Deaths registered on Norfolk Island from 1 July 2016 are included in this publication. This is due to the introduction of the Norfolk Island Legislation Amendment Act 2015. Norfolk Island deaths are included in statistics for "Other Territories" as well as totals for all of Australia. Deaths registered on Norfolk Island prior to 1 July 2016 were not in scope for death statistics. Prior to 1 July 2016, deaths of people that occurred in Australia with a usual residence of Norfolk Island were included in Australian totals, but assigned a usual residence of 'overseas'. With the inclusion of Norfolk Island as a territory of Australia in the ASGS 2016, those deaths which occurred in Australia between January and June 2016 with a usual residence of Norfolk Island were allocated to the Norfolk Island SA2 code instead of the 'overseas' category.

14 The current scope of the statistics includes:

• all deaths being registered for the first time;
• deaths in Australia of temporary visitors to Australia;
• deaths occurring within Australian Territorial waters;
• deaths occurring in Australian Antarctic Territories or other external territories (including Norfolk Island);
• deaths occurring in transit (i.e. on ships or planes) if registered in the State of 'next port of call';
• deaths of Australian Nationals overseas who were employed at Australian legations and consular offices (i.e. deaths of Australian diplomats while overseas) where able to be identified; and
• deaths that occurred in earlier reference periods that have not been previously registered (late registrations).

15 The scope of the statistics excludes:

• repatriation of human remains where the death occurred overseas;
• deaths overseas of foreign diplomatic staff (where these are able to be identified);
• stillbirths/fetal deaths (these are included in perinatal counts (see Explanatory Notes 16-20, below). In 2007-2009 these were published separately in Perinatal Deaths, Australia (cat. no. 3304.0), but are now included in this publication.

16 The scope of the perinatal death statistics includes all registered fetal deaths (at least 20 weeks' gestation or at least 400 grams' birth weight) and all registered neonatal deaths (all live born babies who die within 28 completed days of birth, regardless of gestation or birth weight). The ABS scope rules for fetal deaths are consistent with the legislated requirement for all state and territory Registrars of Births, Deaths and Marriages to register all fetal deaths which meet the above-mentioned gestation and birth weight criteria. Based on this legislative requirement, in the case of missing gestation and/or birth weight data, the fetal record is considered in scope and included in the dataset. A record is only considered out of scope if both gestation and birth weight data are present, and both fall outside the scope criteria (i.e. gestation of 19 weeks or less and birth weight of 399 grams or fewer). This scope was adopted for the 2007 Perinatal Deaths collection, and was applied to historical data for 1999-2006. For more information on the changes in scope rules see Perinatal Deaths, Australia, 2007 (cat. no. 3304.0) Explanatory Notes 18-20. These rules have been applied to all perinatal data presented in this publication.

17 The World Health Organization (WHO) definition of a perinatal death differs to that used by the ABS. The WHO definition includes all neonatal deaths, and those fetuses weighing at least 500 grams or having a gestational age of at least 22 weeks, or body length of 25 centimetres from crown to heel. A summary table based on the WHO definition of perinatal deaths is included in the perinatal data cube in this release. See Explanatory Note 81, below, for more details on the interpretation of this table for 2018.

18 Fetal deaths are registered only as a stillbirth, and are not in scope of either the Births, Australia (cat. no. 3301.0) or Deaths, Australia (cat. no. 3302.0) collections. Fetal deaths are part of the Perinatal collection, but not the Causes of Death collection. Neonatal deaths are in scope of the Deaths, Causes of Death and Perinatal collections.

19 This publication only includes information on registered fetal and neonatal deaths. Registered deaths are sourced through jurisdictional Registries of Births Deaths and Marriages (see Explanatory note 6). This scope differs from other Australian data sources on perinatal deaths. For this reason alternative datasets are not directly comparable and caution should be taken when using multiple sources for analysis.

20 Perinatal death data reported by the ABS are not comparable with the National Perinatal Mortality Data Collection (NPMDC) coordinated by the AIHW. As outlined in Explanatory note 19 the ABS data are sourced from state and territory registrars of Births, Deaths and Marriages. This differs from the NPMDC whose data are sourced from health systems, including clinical records. The table below was published in the AIHW Stillbirths and neonatal deaths in Australia 2015 and 2016 report . The table shows that the ABS perinatal dataset is affected by delayed registrations which results in an under count of perinatal deaths, especially those of stillbirths. Caution should be taken when interpreting this data.

21 A range of socio-demographic data are available from the ABS Causes of Death collection. Standard classifications used in the presentation of causes of death statistics include age, sex, and Aboriginal and Torres Strait Islander status. Statistical standards for social and demographic variables have been developed by the ABS. Where these are not released in the Causes of Death published outputs, they can be sourced on request from the ABS.

22 Since the publication of Causes of Death, Australia, 2011, the ABS has released data based on the Australian Statistical Geography Standard (ASGS). The ASGS is a hierarchical classification system that defines more stable, consistent and meaningful areas than those of the Australian Standard Geographical Classification (ASGC), which was used to define geographical areas for output prior to the release of 2011 reference year data. Under the ASGS, causes of death statistics are coded to Statistical Area 2 (SA2) level, and are presented at the state/territory and national level in this publication.

23 The Standard Australian Classification of Countries (SACC) groups neighbouring countries into progressively broader geographic areas on the basis of their similarity in terms of social, cultural, economic and political characteristics. ABS causes of death statistics are coded using the SACC, as the collection includes overseas residents whose death occurred while they were in Australia.

24 For further information, refer to the Australian Statistical Geography Standard (ASGS): Volume 1 - Main Structure and Greater Capital City Statistical Areas, July 2016 (cat. no. 1270.0.55.001) and the Standard Australian Classification of Countries (SACC), 2011 (cat. no. 1269.0).

25 The International Classification of Diseases (ICD) is the international standard classification for epidemiological purposes and is designed to promote international comparability in the collection, processing, classification, and presentation of causes of death statistics. The classification is used to classify diseases and causes of disease or injury as recorded on many types of medical records as well as death records. The ICD has been revised periodically to incorporate changes in the medical field. Currently the ICD 10th revision is used for Australian causes of death statistics.

26 The ICD-10 is a variable-axis classification meaning that the classification does not group diseases only based on anatomical sites, but also on the type of disease. Epidemiological data and statistical data is grouped according to:

• epidemic diseases;
• constitutional or general diseases;
• local diseases arranged by site;
• developmental diseases; and
• injuries.

27 For example, a systemic disease such as sepsis is grouped with infectious diseases; a disease primarily affecting one body system, such as a myocardial infarction, is grouped with circulatory diseases; and a congenital condition, such as spina bifida, is grouped with congenital conditions.

28 For further information about the ICD refer to WHO International Classification of Diseases (ICD).

29 The versions of the ICD 10th Revision are available online.

30 The Update and Revision Committee (URC), a WHO advisory group on updates to ICD-10, maintains the cumulative and annual lists of approved updates to the ICD-10 classification. The updates to ICD-10 are of numerous types including the addition and deletion of codes, changes to coding instructions and modification and clarification of terms.

31 From the 2013 reference year, the ABS implemented a new automated coding system called Iris. The 2013-2017 data coded in the Iris system applied an updated version of the ICD-10 (2013 version for 2013 data, and 2015 version for 2014-2017 data) when coding multiple causes of death, and when selecting the underlying cause of death. For details of further impacts of this change from 2013 data onwards, please see the ABS Implementation of the Iris Software: Understanding Coding and Process Improvements Technical Note, in the Causes of Death, Australia, 2013 (cat. no. 3303.0) publication

32 The 2018 reference year cause of death data presented in this publication was coded using version 5.4.0 of Iris software. This system replaced Iris version 4.4.1 which was used to code the 2013-2017 cause of death data. Version 5.4.0 of the Iris software applied the WHO ICD updates (2016 version) which have resulted in changes to output. For more information see Technical Note Updates to Iris coding software: Implementing WHO updates and improvements in coding processes, in the Causes of Death, Australia, 2018 (cat. no. 3303.0) publication.

33 Prior to the 2013 reference year, the 2006 version of the ICD-10 was the most recent version used for coding deaths, with the exception of two updates that were applied after the 2006 reference year. The first update was implemented in 2007 and related to the use of mental and behavioural disorders due to psychoactive substance use, acute intoxication (F10.0, F11.0...F19.0) as an underlying cause of death. If the acute intoxication initiated the train of morbid events it is now assigned an external accidental poisoning code (X40-X49) corresponding to the type of drug used. For example, if the death had been due to alcohol intoxication, the underlying cause before the update was F10.0, and after the update the underlying cause is X45, with poisoning code T51.9. The second update implemented from the 2009 reference year was the addition of Influenza due to certain identified virus (J09) to the Influenza and Pneumonia block. This addition was implemented to capture deaths due to Swine flu and Avian flu, which were reaching health epidemic status worldwide.

34 The cumulative list of ICD-10 updates can be found online.

35 From the 2013 reference year onwards, the cause of death data presented in this publication was coded using the Iris coding software. This system replaced the Mortality Medical Data System (MMDS), which was used for coding cause of death data for the 1997-2012 reference years. Like MMDS, Iris is an automated coding system. Iris assigns ICD-10 codes to the diseases and conditions listed on the death certificate and then applies decision tables to select the underlying cause of death. Iris version 4.4.1 was used to code 2013-2017 deaths data. Iris version 5.4.0 was used to code 2018 data. For further details on the change to Iris coding software and associated impacts on data, please see the ABS Implementation of the Iris Software: Understanding Coding and Process Improvements Technical Note, in the Causes of Death, Australia, 2013 (cat. no. 3303.0) publication and Technical Note Updates to Iris coding software: Implementing WHO updates and improvements in coding processes, in the Causes of Death, Australia, 2018 (cat. no. 3303.0) publication.

36 All causes of death can be grouped to describe the type of death, whether it be from a disease or condition, or from an injury, or whether the cause is unknown. These are generally described as:

• Natural Causes - deaths due to diseases (for example diabetes, cancer, heart disease etc.) (A00-Q99, R00-R98)
• External Causes - deaths due to causes external to the body (for example intentional self-harm, transport accidents, falls, poisoning etc.) (V01-Y98)
• Unknown Causes - deaths where it is unable to be determined whether the cause was natural or external (R99).

37 Where an accidental or violent death occurs, the underlying cause is classified according to the circumstances of the fatal injury, rather than the nature of the injury, which is coded separately. For example, a motorcyclist may crash into a tree (V27.4) and sustain multiple fractures to the skull and facial bones (S02.7), which leads to death. The underlying cause of death is the crash itself (V27.4), as it is the circumstance which led to the injuries that ultimately caused the death.

38 Ranking causes of death is a useful method of describing patterns of mortality in a population and allows comparison over time and between populations. However, different methods of grouping causes of death can result in a vastly different list of leading causes for any given population. A ranking of leading causes of death based on broad cause groupings such as 'cancers' or 'heart disease' does not identify the leading causes within these groups, which is needed to inform policy on interventions and health advocacy. Similarly, a ranking based on very narrow cause groupings or including diseases that have a low frequency, can be meaningless in informing policy.

39 Tabulations of leading causes presented in this publication are based on research presented in the Bulletin of the World Health Organization, Volume 84, Number 4, April 2006, 297-304. The determination of groupings in this list is primarily driven by data from individual countries representing different regions of the world. Other groupings are based on prevention strategies, or to maintain homogeneity within the groups of cause categories. Since the aforementioned bulletin was published, a decision was made by WHO to include deaths associated with the H1N1 influenza strain (commonly known as swine flu) in the ICD-10 classification as Influenza due to certain identified influenza virus (J09). This code has been included with the Influenza and Pneumonia leading cause grouping in the Causes of Death publication since the 2009 reference year.

40 Since 2015, the ABS includes C26.0 (malignant neoplasm of the intestinal tract, part unspecified) in the WHO leading cause grouping for Malignant neoplasm of colon, sigmoid, rectum and anus (now C18-C21, C26.0). For further details on the reasoning behind the inclusion of C26.0 in this leading cause grouping, see Complexities in the measurement of bowel cancer in Australia, in Causes of Death, Australia, 2015 (cat. no. 3303.0). This change has been applied in this publication to data for all reference years that appear in tables involving leading cause tabulations. This differs to publications prior to 2015, for which C26.0 was not included in this leading cause grouping, and also differs to the suggested WHO tabulation of leading causes for these cancers. Comparisons with data for this leading cause, and associated leading cause rankings, as they appear in previous publications should therefore be made with caution. Time-series data by leading causes has been published in Australia's leading causes of death, 2017 in this publication.

41 The ABS now includes Y87.0 (Sequelae of intentional self-harm), Y87.1 (Sequelae of assault) and Y85 (Sequelae transport accidents) in the WHO leading cause grouping for Intentional self-harm (now X60-X84, Y87.0), Assault (now X85-Y09, Y87.1) and Land transport Accidents (V01-V89, Y85). This change has been applied to harmonise data between the WHO leading cause grouping and subject specific datacubes for intentional self-harm, assault and transport accidents which is published as part of the ABS Causes of Death collection. This change applies to publication data for all reference years that appear in tables involving leading cause tabulations. This differs to previous publications, where Y87.0, Y87.1 and Y85 were not included in these leading cause groupings, and also differs to the suggested WHO tabulation of leading causes. Comparisons with data for these leading causes, and associated leading cause rankings, as they appear in previous publications should therefore be made with caution. Time-series data by leading causes has been published in Australia's leading causes of death, 2017 in this publication.

46 Age-standardised death rates enable the comparison of death rates over time and between populations of different age-structures. Along with adult, infant and child mortality rates, they are used to determine whether the mortality rate of the Aboriginal and Torres Strait Islander population is declining over time, and whether the gap between Aboriginal and Torres Strait Islander and non-Indigenous populations is narrowing. However, there have been inconsistencies in the way different government agencies have calculated age-standardised death rates in the past. The ABS uses the direct method of age-standardisation as it allows for valid comparisons of mortality rates between different study populations and across time. This method was agreed to by the ABS, Australian Institute of Health and Welfare (AIHW) and other stakeholders. For further information see: AIHW (2011) Principles on the use of direct age-standardisation in administrative data collections: for measuring the gap between Indigenous and non-Indigenous Australians. Cat. no. CSI 12. Canberra: AIHW.

47 The direct method has been used throughout the publication and data cubes for age-standardised death rates. Age-standardised death rates for specific causes of death with fewer than a total of 20 deaths have not been published due to issues of robustness.

48 For further information, see Appendix: Principles on the use of direct age-standardisation, from Deaths, Australia, 2010 (cat. no. 3302.0).

49 In this publication, age-standardised and age-specific death rates for Aboriginal and Torres Strait Islander persons for the 2009-2018 reference years have been calculated using 2016-Census-based population estimates (projections and backcasts). Non-Indigenous estimates have been derived by subtracting the 2016-census based Aboriginal and Torres Strait Island population estimates from the total 2016-Census-based estimated resident population (ERP). Rates calculated from population denominators derived from different Censuses may cause artificially large rate differences. Rate comparisons should not be made with previous publications for Aboriginal and Torres Islander data. See Estimates and Projections, Aboriginal and Torres Strait Islander Australians (cat. no. 3238.0) for more information.

50 Causes of death statistics for states and territories in this publication have been compiled based on the state or territory of usual residence of the deceased, regardless of where in Australia the death occurred and was registered. Deaths of persons usually resident overseas which occur in Australia are included in the state/territory in which their death was registered.

51 Statistics compiled on a state or territory of registration basis are available on request.

52 Given the small number of perinatal deaths which occur in some states and territories, some data provided on a state/territory basis in this publication have been aggregated for South Australia, Western Australia, Northern Territory, Australian Capital Territory and Other Territories.

53 In compiling causes of death statistics, the ABS employs a variety of measures to improve quality, which include:

• providing certifiers with certification booklets for guidance in reporting causes of death on medical certificates, see Information Paper: Cause of Death Certification Australia, 2008 (cat. no. 1205.0.55.001);
• seeking detailed information from the National Coronial Information System (NCIS); and
• editing checks at the individual record and aggregate levels.
   

54 The quality of causes of death coding can be affected by changes in the way information is reported by certifiers, by lags in completion of coroner cases and the processing of the findings. While changes in reporting and lags in coronial processes can affect coding of all causes of death, those coded to Chapter XVIII: Symptoms, signs and abnormal clinical and laboratory findings, not elsewhere classified and Chapter XX: External causes of morbidity and mortality are more likely to be affected because the code assigned within the chapter may vary depending on the coroner's findings (in accordance with ICD-10 coding rules).

55 It is the role of the coroner to investigate the circumstances surrounding all reportable deaths and to establish, wherever possible, the circumstances surrounding the death, and the cause(s) of death. Generally most deaths due to external causes will be referred to a coroner for investigation; this includes those deaths which are possible instances of intentional self-harm (suicide).

56 Where a case remains open on the NCIS at the time the ABS ceases processing, and insufficient information is available to code a cause of death (e.g. a coroner certified death was yet to be finalised by the coroner), less specific ICD codes are assigned, as required by the ICD coding rules.

57 The specificity with which open cases are able to be coded is directly related to the amount and type of information available on the NCIS. The amount of information available for open cases varies considerably from no information to detailed police, autopsy and toxicology reports. There may also be interim findings of 'intent'.

58 The manner or intent of an injury which leads to death, is determined by whether the injury was inflicted purposefully or not. When it was inflicted purposefully (intentional), a determination should be made as to whether the injury was self-inflicted (suicide) or inflicted by another person (assault). However, intent cannot be determined in all cases.

59 These published outputs include 2018 and 2017 preliminary data, 2016 revised data and 2015 final data. The standard ABS revisions process has not yet been applied to the 2016 and 2017 reference years that would, in the past, be subject to revisions in this publication. Causes of death revisions data will be released in early 2020.

60 For coroner-certified deaths, the specificity of cause of death coding can be affected by the length of time for the coronial process to be finalised and the coroner case closed. To improve the quality of ICD coding, all coroner-certified deaths registered after 1 January 2006 are subject to a revisions process.

61 Up to and including deaths registered in 2005, ABS Causes of Death processing was finalised at a point in time. At this point, not all coroners' cases had been investigated, the case closed and relevant information loaded into the National Coronial Information System (NCIS). The coronial process can take several years if an inquest is being held or complex investigations are being undertaken. In these instances, the cases remain open on the NCIS and relevant reports may be unavailable. Coroners' cases that have not been closed or had all information made available can impact on data quality as less specific ICD codes often need to be applied.

62 The revisions process to date has focused on cases that remain open on the NCIS database. ABS coders investigate and use additional information from police reports, toxicology reports, autopsy reports and coroners' findings to assign more specific causes of death. The use of this additional information occurs at either 12 or 24 months after initial processing and the specificity of the assigned ICD-10 codes increase over time. As 12 or 24 months pass after initial processing, many coronial cases are closed, with the coroner having dispensed a cause of death and relevant reports become available. This allows ABS coders to assign a more specific cause of death.

63 The Aboriginal and Torres Strait Islander status of a deceased person is captured through the death registration process. It can be noted on the Death Registration Form and/or the Medical Certificate of Cause of Death. However it is recognised that not all such deaths are captured through these processes, leading to under-identification. While data is provided to the ABS for the Aboriginal and Torres Strait Islander status question for around 99% of all deaths, there are concerns regarding the accuracy of the data.

64 The ABS Death Registrations collection identifies a death as being of an Aboriginal and Torres Strait Islander person where the deceased is recorded as Aboriginal, Torres Strait islander, or both on the Death Registration Form (DRF). The Indigenous status is also derived from the Medical Certificate of Cause of Death (MCCD) for South Australia, Western Australia, Tasmania, the Northern Territory and the Australian Capital Territory from 2007. From 2015 data onwards, the Queensland Registry of Births, Deaths and Marriages also used MCCD information to derive Indigenous status. For New South Wales and Victoria, the Indigenous status of the deceased is derived from the DRF only. If the Indigenous status reported in the DRF does not agree with that in the MCCD, an identification from either source that the deceased was an Aboriginal and/or Torres Strait Islander person is given preference over non-Indigenous.

65 There are several data collection forms on which people are asked to state whether they are of Aboriginal and Torres Strait Islander origin. Due to a number of factors, the results are not always consistent. The likelihood that a person will identify, or be identified, as an Aboriginal and Torres Strait Islander person on a specific form is known as their propensity to identify.

66 Propensity to identify as an Aboriginal and Torres Strait Islander person is determined by a range of factors, including:

• how the information is collected (e.g. census, survey, or administrative data);
• who provides the information (e.g. the person in question, a relative, a health professional, or an official);
• the perception of why the information is required, and how it will be used;
• educational programs about identifying as an Aboriginal and Torres Strait Islander person; and
• cultural aspects and feelings associated with identifying as Aboriginal and Torres Strait Islander Australian.
   

67 In addition to those deaths where the deceased is identified as an Aboriginal and Torres Strait Islander person, a number of deaths occur each year for which the Indigenous status is not stated on the death registration form. In 2018, there were 1062 deaths registered in Australia for whom Indigenous status was not stated, representing 0.7% of all deaths registered, representing a slight increase from 2017 (0.6%). This difference is largely driven by a higher number of deaths with a not stated Indigenous status registered in New South Wales (from 600 in 2017 to 731 in 2018), and Victoria (from 144 in 2017 to 173 in 2018). All other states experienced a decrease in deaths where indigenous status was not stated representing an improvement in the dataset. See Explanatory Note 63-66 for further details.

68 Data presented in this publication may therefore underestimate the level of Aboriginal and Torres Strait Islander deaths and mortality in Australia. Caution should be exercised when interpreting data for Aboriginal and Torres Strait Islander Australians presented in this publication, especially with regard to year-to-year changes.

69 Information on causes of death relating to Aboriginal and Torres Strait Islander persons is included in articles throughout this publication. Data cube 12 also provides information on causes of death for Aboriginal and Torres Strait Islander Australians. In data cube 12, numbers and rates of death are reported by jurisdiction of usual residence for New South Wales, Queensland, South Australia, Western Australia and the Northern Territory only. Data for Victoria, Tasmania and the Australian Capital Territory have been excluded in line with national reporting guidelines. For information on issues with Aboriginal and Torres Strait Islander identification, see Explanatory Notes 63-75.

70 Individual state/territory disaggregations of deaths of Aboriginal and Torres Strait Islander Australians by WHO Leading Causes (see Explanatory Notes 38-41) for the 2018 reference year are presented for New South Wales, Queensland, Western Australia and the Northern Territory only. No data are presented for South Australia, due to the small number of deaths by WHO leading causes - most causes have a count of fewer than 20 deaths, which is too small for the production of robust Standardised Death Rates (SDRs). See Explanatory Notes 46-49 for further details.

71 In this publication, age-standardised and age-specific death rates for Aboriginal and Torres Strait Islander persons for the 2009-2018 reference years have been calculated using 2016-Census-based population estimates (projections and backcasts). Non-Indigenous estimates have been derived by subtracting the 2016-census based Aboriginal and Torres Strait Island population estimates from the total 2016-Census-based estimated resident population (ERP). Rates calculated from population denominators derived from different Censuses may cause artificially large rate differences. Rate comparisons should not be made with previous publications for Aboriginal and Torres Islander data. See Estimates and Projections, Aboriginal and Torres Strait Islander Australians (cat. no. 3238.0) for more information.

72 Coronial cases are more likely to be affected by a lag in registration time, especially those which are due to external causes, including suicide, homicide and drug-induced deaths. Due to small numbers these lagged coroner-referred registrations can create large yearly variation in some causes of deaths of Aboriginal and Torres Strait Islander persons. Caution should be taken when making year to year analysis.

73 The ABS undertakes significant work aimed at improving Aboriginal and Torres Strait Islander identification. The ABS is working closely with the state and territory registries of births, deaths and marriages through the National Civil Registration and Statistics Improvement Committee (NCRSIC) to progress towards improved identification in a nationally consistent way.

74 Quality studies conducted as part of the Census Data Enhancement (CDE) project have investigated the levels and consistency of Aboriginal and Torres Strait Islander identification between the 2011 Census and death registrations. See Information Paper: Death registrations to Census linkage project - Methodology and Quality Assessment, 2011-2012 (cat. no. 3302.0.55.004).

75 An assessment of various methods for adjusting incomplete Aboriginal and Torres Strait Islander death registration data for use in compiling Aboriginal and Torres Strait Islander life tables and life expectancy estimates is presented in Discussion Paper: Assessment of Methods for Developing Life Tables for Aboriginal and Torres Strait Islander Australians, 2006 (cat. no. 3302.0.55.002), released on 17 November 2008. Final tables based on feedback received from this discussion paper, using information from the Census Data Enhancement (CDE) study, can be found in Life Tables for Aboriginal and Torres Strait Islander Australians, 2010-2012 (cat. no. 3302.0.55.003).

76 For deaths in the 2018 reference year, 11.9% were certified by a coroner. There are variations between jurisdictions in relation to the proportion of deaths certified by a coroner, ranging from 8.2% of deaths certified by a coroner and registered in Queensland, to 26.1% of deaths certified by a coroner and registered in the Northern Territory. The proportion of deaths certified by a coroner in 2018 is comparable to previous years.

77 The 2018 publication has followed the same process of publication as the 2017 data. The 2017 publication was released six months earlier than usual, allowing for more timely access to mortality data in Australia. For further details on data considerations, see A more timely annual collection: changes to ABS processes (Technical Note), Causes of Death, Australia, 2015 (cat. no. 3303.0).

78 From the 2013 reference year onwards, the cause of death data presented in this publication was coded using the Iris coding software. This system replaced the Mortality Medical Data System (MMDS), which was used for coding cause of death data for the 1997-2012 reference years. Like MMDS, Iris is an automated coding system. Iris assigns ICD-10 codes to the diseases and conditions listed on the death certificate and then applies decision tables to select the underlying cause of death. Iris version 4.4.1 was used to code 2013-2017 deaths data. Iris version 5.4.0 was used to code 2018 data. For further details on the change to Iris coding software and associated impacts on data, please see the ABS Implementation of the Iris Software: Understanding Coding and Process Improvements Technical Note, in the Causes of Death, Australia, 2013 (cat. no. 3303.0) publication and Technical Note Updates to Iris coding software: Implementing WHO updates and improvements in coding processes, in the Causes of Death, Australia, 2018 (cat. no. 3303.0) publication.

79 Users analysing time-series or 2018 cause of death data should take into account a number of issues, as outlined below, which are unrelated to the implementation of Iris.

80 For perinatal data output in the Causes of Death, Australia, 2013 publication, the ABS began a review of its method of coding perinatal deaths, which resulted in an interim change to how this data was output. One significant change was that neonatal deaths were not assigned an underlying cause of death when output in tables of all ages, as had previously occurred. (Details of this change can be found in the Changes to Perinatal Death Coding Technical Note in Causes of Death, Australia, 2013 (cat. no. 3303.0). Further review and consultation has now been undertaken with the national and international coding community, and has resulted in the ABS applying a new method of coding perinatal deaths. The new method creates a sequence of causes on a Medical Certificate of Cause of Perinatal Death which allows for an underlying cause of death to be assigned to a neonatal death. This aligns the output for neonatal deaths to deaths of the general population which are certified using the Medical Certificate of Cause of Death. The change in coding method reinstates the condition arising in the mother being assigned as an underlying cause of death. This method has been applied to the 2014-2018 data, and has also been applied retrospectively to the 2013 neonatal data that is output in tables of all ages in this publication, thus enabling a consistent time-series. Please see the Changes to Perinatal Death Coding Technical Note in Causes of Death, Australia, 2014 (cat. no. 3303.0) for further details.

81 From the 2013 reference year onwards, process changes have led to a reduction in the number of both stillbirths and neonatal deaths where a 'main condition in mother' was recorded, compared to previous years. This has led to a reduction in the number of records assigned within the code block P00-P04: Fetus and newborn affected by maternal factors and by complications of pregnancy, labour and delivery, as main condition in the mother. These changes will affect data output in the Perinatal data cube of this publication only.

82 Doctor-certified neonatal deaths with no cause of death information are coded to Conditions originating in the perinatal period, unspecified (P969). As these deaths have been certified by a doctor, the assumption is made that the neonate died of natural causes. Where a neonatal death is referred to a coroner, but no cause of death information is available, these deaths are coded to Other ill-defined and unspecified causes of mortality (R99). As a reportable death, it cannot be determined whether the neonate died of natural or external causes, in the absence of further information.

83 The count of fetal deaths in scope for the World Health Organization (WHO) definition of perinatal deaths differs to those previously published for 2012 and 2015. This is due to an enhancement to birth weight and gestation information, which resulted in some deaths no longer meeting the World Health Organization definition of a fetal death (that is, a gestational age of at least 22 weeks or weighing at least 500 grams). For 2012, there are two fewer fetal deaths than previously published (1 male and 1 female). For 2015, there are 38 fewer fetal deaths than previously published (18 males, 19 females, 1 death of undetermined sex). Some corresponding death rates have also been affected. Table 14.21 in the perinatal data cube presents fetal and neonatal data according to the WHO scope. No other tables in the perinatal deaths data cube are affected by these changes.

84 In September quarter 2011 the high number of death registrations in New South Wales was queried with the New South Wales Registry of Births, Deaths and Marriages. Information provided by the Registry indicates that these fluctuations may be the result of changes in processing rates. This may have contributed to the increase in the number of deaths registered in New South Wales in 2011. New South Wales deaths in 2011 (50,182) were 5.8% higher than in 2010 (47,453).

85 The number of deaths attributable to Accident to watercraft causing drowning and submersion (V90) increased from 26 in 2010 to 75 in 2011. This increase is primarily due to deaths resulting from an incident in December 2010 when a boat collided with cliffs on Christmas Island. These deaths were registered with the Western Australian Registry of Births, Deaths and Marriages in January 2011, resulting in an increase in the number of deaths coded to V90 in Western Australia.

86 As part of a collection-wide initiative by the ABS to improve specificity of cause of death coding in the 2008 and 2009 reference years, doctor-certified deaths due to Pneumonia, organism unspecified (J18) reduced substantially. This was as a result of the ABS manually interrogating conditions located in Part 2 of the Medical Certificate Cause of Death (MCCD), reallocating them to a more specific cause of death code.

87 In 2010 there was a shift in this pattern. The number of doctor-certified deaths assigned to J18 increased by 690 deaths, or 49.5%. The reason for the 2010 data movement was a more consistent use of coding software decision tables throughout both coding and quality assurance processes. These decision tables provide clear rules for when Pneumonia can be selected as an underlying cause of death, in relation to the information listed in Part 2 of the MCCD.

88 The 2010 increase represented a return to counts observed prior to 2008. In 2007, 2,293 doctor certified deaths were assigned to J18, therefore the 2010 count for this cause of death (2,085) is considered a return to the trend which existed prior to the coding of 2008 and 2009 data. The data from 2011 onwards has been consistent with this trend.

89 There were 1,273 deaths attributed to road crashes (V01-V79, V892, X82, Y32) in 2018. Of these, 37 were of suicidal intent (X82) and there were a further 16 where the intent could not be determined (Y32). When making comparisons between road deaths from the ABS Causes of Death collection and road deaths from other sources, the scope and coverage rules applying to each collection should be considered. It should be noted that the number of road-traffic-related deaths attributed to transport accidents for 2018 is expected to change as data is subject to the revisions process. See Explanatory Notes 59-62 and the Causes of Death Revisions, 2015 Final Data (Technical Note) in Causes of Death, Australia, 2017 for further details.

90 The number of deaths recorded as Assault (X85-Y09, Y87.1) i.e. murder, manslaughter and their sequelae, published in the ABS Causes of Death publication, differ from those published by the ABS in Recorded Crime - Victims, Australia, 2017 (cat. no. 4510.0). Reasons for the different counts include differences in scope and coverage between the two collections, as well as legal proceedings that are pending finalisation. It is important to note that the number of deaths attributed to assault for 2018 is expected to change as data is subject to the revisions process. See Explanatory Notes 59-62 and Causes of Death Revisions, 2015 Final Data (Technical Note) and 2016 Revised Data (Technical Note) in Causes of Death, Australia, 2017 (cat. no. 3303.0).

91 The number of deaths attributed to intentional self-harm for 2018 is expected to increase as data is subject to the revisions process. For further information, see Explanatory Notes 59-62 and the Causes of Death Revisions, 2015 Final Data (Technical Note) and 2016 Revised Data (Technical Note) in Causes of Death, Australia, 2017 (cat. no. 3303.0).

92 From 2006 onwards, the ABS implemented a revisions process for coroner-certified deaths (such as suicides), which has enabled additional suicide deaths to be identified beyond initial processing (see Explanatory Notes 54-62). It is recognised that in the four years prior to the implementation of the revisions process (2002-2005), suicide deaths may have been understated as the ABS began using the National Coronial Information System as the sole source for coding coroner-referred deaths.

93 In addition to the revisions process, new coding guidelines were applied for deaths registered from 1 January 2007. The new guidelines improve data quality by enabling deaths to be coded to suicide if evidence indicates the death was from intentional self-harm. Previously, coding rules required a coroner to determine a death as intentional self-harm for it to be coded to suicide. However, in some instances the coroner does not make a finding on intent. The reasons for this may include legislative or regulatory barriers around the requirement to determine intent, or sensitivity to the feelings, cultural practices and religious beliefs of the family of the deceased. Further, for some mechanisms of death it may be very difficult to determine suicidal intent (e.g. single vehicle incidents, drowning). In these cases the burden of proof required for the coroner to establish that the death was as a result of intentional self-harm may make a finding of suicide less likely.

94 Under the new coding guidelines, in addition to coroner-determined suicides, deaths may also be coded to suicide following further investigation of information on the NCIS. Further investigation of a death would be initiated when the mechanism of death indicates a possible suicide and the coroner does not specifically state the intent as accidental or homicidal. Information that would support a determination of suicide includes indications by the person that they intended to take their own life, the presence of a suicide note, or knowledge of previous suicide attempts. The processes for coding open and closed coroner cases are illustrated in the below diagrams (open/closed case coding decision trees)

95 Over time, the NCIS has worked with jurisdictions to improve the timeliness and completeness of information flowing from the coronial systems to the NCIS database. These improvements lead to changes in the information available to ABS coding staff. It is therefore important that data users are aware of any significant improvements in the management of coronial data to enable better interpretation of data within, and between, reference periods.

96 In 2012, the implementation of JusticeLink in the NSW coronial system significantly changed how information is exchanged between the NSW coroners courts and the NCIS. This system enables nightly uploads of all new information to the NCIS, and as a result information pertaining to NSW coronial cases is available earlier in the investigation process and the information is more complete for the purposes of coding causes of death.

97 There is evidence that the system change in NSW has improved the quality of preliminary coding in relation to deaths due to intentional self-harm. There has been an increase in the number of preliminary intentional self-harm deaths registered in NSW when comparing 2012-2017 counts (708, 694, 798, 819, 799 and 881, respectively) with those of 2011 (568), coupled with fewer cases of deaths of undetermined intent (Y10-Y34).

98 Coronial cases are more likely to be affected by a lag in registration time, especially those which are due to external causes, including suicide, homicide and drug-related deaths. Due to small numbers these lagged coroner-referred registrations can create large yearly variation in some causes of deaths of Aboriginal and Torres Strait Islander persons. Caution should be taken when making year to year analysis.

99 More broadly, change in administrative systems highlights how various factors (including administrative and system changes, certification practices, classification updates or coding rule changes) can impact on the mortality dataset. Data users should note this particular change and be cautious when making comparisons between reference periods. The change does not explain away differences between years, but is a factor to consider. It should also be noted as a factor that may influence the magnitude of any increases in suicide numbers as revisions are applied.

100 The two flow charts below highlight the guidelines used by the ABS when coding a death to intentional self-harm for open and closed coroner cases, where the intent status at the time of coding is neither intentional self-harm nor assault. In these cases, the ABS considers additional information available on NCIS, such as the mechanism and other available data (e.g. the presence of a suicide note or previous suicide attempts) when determining the intent of such deaths for coding purposes.

101 In November 2010, the Queensland Registrar of Births, Deaths and Marriages advised the ABS of an outstanding deaths registration initiative undertaken by the registry. This initiative resulted in the November 2010 registration of 374 previously unregistered deaths which occurred between 1992 and 2006 (including a few for which a date of death was unknown). Of these, around three-quarters (284) were deaths of Aboriginal and Torres Strait Islander Australians. A data adjustment is made for tables which include Aboriginal and Torres Strait Islander data for Queensland for 2010. For further information refer to Technical Notes, Registration of Outstanding Deaths, Queensland, 2010 in Deaths, Australia, 2010 (cat. no. 3302.0) and Retrospective Deaths by Causes of Death, Queensland, 2010, in Causes of Death, Australia, 2010 (cat. no. 3303.0).

102 See the 'Data Used in Calculating Death Rates' Appendix for details of the number of live births registered which have been used to calculate the fetal, neonatal and perinatal death rates shown in this publication. This Appendix also provides data on fetal deaths used in the calculation of fetal and perinatal death rates. These also enable further rates to be calculated.

103 In 2016 the NSW Registry changed the 'proof of identity' requirements for parents registering a new birth. This led to delays in registration of births for 2016 and 2017. The ABS has been working with the NSW Registry to improve the birth registration lag, The recent launch of online birth registration in 2018 by the NSW Registry appears to be improving birth registration time frames for both Aboriginal and Torres Strait Islander and non-Indigenous Australians.

104 In 2016 and 2017 there were lower than expected registration counts for New South Wales. The ABS worked with the NSW Registry of Births, Deaths and Marriages (NSW RBDM) to investigate these counts, highlighting that changes to identity requirements in 2016 had prevented some registrations from being finalised. The NSW RBDM worked with parents to finalise these registrations, enabling many to be included in 2018 counts. Other initiatives also contributed to the higher count of births in NSW in 2018, including the implementation of an online birth registration system and a campaign aimed at increasing registrations among Aboriginal and Torres Strait Islander parents.

105 In 2018, the Northern Territory Registry of Births, Deaths and Marriages identified a processing issue that led to delays in completing the registration of some births that occurred in previous years. These births have since been registered, resulting in 355 additional births being included in 2018 data, the majority of which (339) were of Aboriginal and Torres Strait Islander children. Care should be taken when interpreting changes in birth counts and fertility rates for the Northern Territory in recent years.

106 Multiple causes of death include all causes and conditions reported on the death certificate (i.e. both underlying and associated causes; see the Glossary for further details). As all entries on the death certificate are taken into account, multiple cause of death statistics are valuable in recognising the impact of conditions and diseases which are less likely to be an underlying cause, highlighting relationships between concurrent disease processes, and giving an indication of injuries which occur as a result of specific external events. These features of multiple cause of death data provide a more in depth picture of mortality in Australia.

107 When analysing data on multiple causes of death, data can be presented in two ways: by counts of deaths or by counts of mentions. When analysis is conducted by counts of death, the figures are describing the number of people who have died with a particular disease or disorder. Multiple Cause of Death data derived from counts of mentions is the total number of incidences of a particular disease or disorder on the death certificate. For example, an individual may have had Breast cancer (C50) and then developed Secondary lung cancer (C78.0). This individual would be counted once if counts were by the number of deaths from cancer, but twice if the counts were by the number of mentions of cancer. Care should be taken to differentiate between counts and mentions when analysing multiple cause of death data.

108 Changes in patterns of mortality are studied by policy makers and researchers to improve health outcomes for all Australians. However, changes in patterns of mortality can occur for many reasons. Changes can reflect a real increase or decrease in the prevalence of a disease or disorder, or a change in medical treatment. Mortality data changes can also be a result of administrative processes which can potentially impact on the data, for example, International Classification of Disease (ICD) coding classification changes and updates, and differences in how deaths are certified. Analysis of the multiple causes of death data can give a deeper understanding of how the complete dataset may be affected by both real and administrative changes. For example, in 2009, the World Health Organization (WHO) recommended introducing code J09 (Influenza due to certain identified influenza virus) to the ICD-10 in response to the worldwide epidemics of swine flu and avian flu. There were 98 people who died as a direct consequence of contracting these strains of the flu across 2009 and 2010. In addition there were 51 people who had this flu when they died and for whom this would have been a complicating factor. Additional health risk factors may also be identified. When swine or avian flu was the underlying cause of death, multiple cause data shows obesity and respiratory problems as a common associated cause. In this way, multiple cause data provides policy makers and researchers a greater insight beyond the underlying cause of death.

109 Data cells with small values have been randomly assigned to protect confidentiality. As a result some totals will not equal the sum of their components. Cells with 0 values have not been affected by confidentialisation.

110 Where figures have been rounded, discrepancies may occur between totals and sums of the component items.

111 The Victorian Registry of Births, Deaths and Marriages (RBDM) implemented a new registration system in February 2019. As part of this system implementation, certain policies and procedures have changed within the registry. Of note, the Victorian RBDM has changed their procedures regarding the registration of coroner-referred deaths. Previously coroner-certified deaths were not submitted to the ABS until the case was finalised in the Victorian Coroner Court. From 2019, this has changed and interim registrations (open cases) have been submitted to the ABS. This procedural change has resulted in an additional delay to registrations in 2018 and previous years, but the change to procedure is expected to lead to an increased number of coroner-referred registrations in 2019.

112 ABS published outputs are available free of charge from the ABS website. Click on 'Statistics' to gain access to the full range of ABS statistical and reference information. For details on products scheduled for release in the coming week, click on the Future Releases link on the ABS homepage.

Since 2014 the national Causes of Death dataset has been coded using Iris, a software program which automates the assignment of codes from the International Classification of Diseases 10th Revision (ICD-10) to death records, and assists in the identification of an underlying cause of death. Iris is developed and maintained by the German Institute of Medical Documentation and Information (DIMDI), who produce regular software updates, as well as implement World Health Organization (WHO) updates to the ICD-10 that are embedded within the Iris system.

In order to recognise and statistically represent changing health trends and advances in medical science research, specialist committees within WHO meet annually to review and recommend updates to the ICD-10. Updates can be minor, consisting of spelling updates or small updates to medical terminology within existing codes, or they can be major, consisting of reclassification of medical conditions to different codes, updates to coding rules or the addition or deletion of codes. To maintain consistency in statistical outputs major changes are only implemented on 3 year cycles.

For the coding of 2018 cause of death data the ABS implemented a new version of Iris software (version 5.4.0) which incorporates a new underlying cause of death processing system called the Multicausal and Unicausal Selection Engine (MUSE). Like previous versions of Iris, MUSE assigns codes to medical terms on the Medical Certificate of Cause of Death (MCCD) and applies WHO coding rules to appropriately code and modify multiple causes of death and select an underlying cause of death. This version of Iris also incorporates the most recent major updates to the ICD-10 (2016 coding year updates). The implementation of MUSE, alongside the updates to the ICD-10, align the Australian mortality data up to date with international best practice. The ABS have also implemented extra validation processes with the implementation of MUSE to ensure maximum alignment with WHO guidelines and coding rules.

Key statistical measures need to be considered when interpreting time series data with administrative changes made to processing.

There are generally four ways in which output can change: 

1. A true change in disease or external event.
2. Administrative changes such as changes to certification or events at point of registration.
3. Updates to the WHO ICD-10 classification, Volume 2 coding rules and application of decision tables.
4. Process changes, such as the implementation of new software or changes to local coding practice.
    

Understandably, the focus of health policy is on true changes in patterns of mortality. The ABS uses explanatory notes to highlight administrative and process changes to enable better interpretation of trends in data over time. 

This technical note will provide an overview of software changes, WHO updates and local coding changes to assist users in interpretting changes in the 2018 dataset. The information focusses on factors influencing the selection of underlying causes of death, although the multiple cause dataset is also acknowledged as an integral tool in tracking changes over time.

Updates to WHO guidelines have resulted in changes to causes located in Chapter I, Certain infectious and parasitic diseases:

A40 Streptococcal sepsis - A41 Other sepsis: A common certification issue is the recording of an infection such as sepsis in Part 1 of a death certificate with no preceding cause, with chronic conditions listed on the same certificate in Part 2. Previously, when sepsis was certified in Part 1 of the MCCD and selected chronic conditions were reported in Part 2 of the MCCD, the selection rules did not provide a mechanism for the condition in Part 2 to be selected as the underlying cause. An update to selection rules now allows more chronic conditions, such as neoplasms coded to (C00-C80), in Part 2 to be selected as the underlying cause when sepsis appears in Part 1. In particular, this has resulted in an increase in deaths assigned to Chapter II Neoplasms and a subsequent decrease in the number of deaths assigned to A40-A41 as an underlying cause.

A90 Dengue fever [classical dengue] and A91 Dengue hemorrhagic fever: These codes are no longer valid for causes of death outputs. All deaths previously assigned to A90 or A91 will now be assigned to A97 Dengue

A97 Dengue: This is a new code and includes multiple fourth digit options for coding. A97 replaces deaths previously assigned to A90 and A91.

B94 Sequelae of other and unspecified infectious and parasitic disease: Previously, if an infectious disease was reported with a duration of greater than one year, the sequelae code was assigned (B94). The processing of time intervals as they relate to infectious diseases now means that B94 is only assigned when late or residual effects of the infection are reported. This has resulted in a decrease in deaths assigned to B94.

Updates to WHO guidelines and ABS coding practices have resulted in an increase in deaths assigned to Chapter II Neoplasms.

There are now more causal relationships between acute conditions, including Bacterial sepsis (A40-A41), and Malignant neoplasms (C00-C80). Previously, when sepsis was mentioned in Part 1 of the MCCD and a malignant neoplasm was mentioned in Part 2, the sepsis would commonly be assigned as the underlying cause of death. In accordance with updated decision tables, there are now additional relationships by which neoplasms falling within C00-C80 can be selected as the underlying cause of death when mentioned in Part 2. This has resulted in an increase in deaths assigned to Chapter II Neoplasms and a subsequent decrease in the number of deaths assigned to A40-A41 as an underlying cause.

F05 Delirium, not induced by alcohol and other psychoactive substances: Improvements in coding practices have resulted in a decrease in deaths coded to F05. When F05 is mentioned on the MCCD the decision tables do not provide a mechanism by which the causal condition of the delirium can be chosen as the underlying cause of death. The ABS mortality team have implemented improved coding and validation processes to ensure the causal condition of the delirium is now selected as the underlying cause. There is a notable decrease in deaths coded to delirium as an underlying cause of death as a result of this change.

G23 Other degenerative diseases of basal ganglia: A new fourth-digit code, G233 Multiple system atrophy, cerebellar type [MSA-C], has been added to this category which has resulted in an increase in deaths assigned to G23. Most deaths assigned to G233 were previously coded to G903 Multiple system degeneration.

G83 Other paralytic syndromes: A new fourth-digit code, G835 Locked-in syndrome, has been added to this category. There was 1 death assigned to G835 in 2018.

G90 Disorders of autonomic nervous system: The fourth-digit code, G903 Multiple system degeneration has been removed as a valid code for cause of death coding. This has resulted in a decrease in deaths coded to G90. Causes previously coded to G903 are now coded to G238 Other specified degenerative diseases of basal ganglia.

I67 Other cerebrovascular diseases: Changes in coding processes have been made in relation to I67. Previously, if Chronic cerebrovascular disease was reported with a duration of greater than one year, the sequelae code was assigned (I69). Further, a note under I69 dictates that Chronic cerebrovascular disease is to be coded to I60-I67. Changes have been implemented resulting in the sequelae codes only being used if late or residual effects of the disease are reported. This change has resulted in an increase in deaths assigned to I67 and a subsequent decrease in deaths assigned to I69.

I69 Sequelae of cerebrovascular disease: Changes in coding processes have been made in relation to I69. Previously, if Chronic cerebrovascular disease was reported with a duration of greater than one year, the sequelae code was assigned (I69). Further, a note under I69 dictates that Chronic cerebrovascular disease is to be coded to I60-I67. Changes have been implemented resulting in the sequelae codes only being used if late or residual effects of the disease are reported. This change has resulted in a decrease in deaths assigned to I69 and a subsequent increase in deaths assigned to I67.

L98 Other disorders of skin and subcutaneous tissue, not elsewhere classified: A new fourth-digit code, L987 Excessive and redundant skin and subcutaneous tissue, has been added to this category. No conditions were assigned to this code in 2018.

M19 Other arthrosis: A relationship between J18 Pneumonia, organism unspecified and M19 Other arthrosis has been removed in 2018. When J18 was reported in Part 1 of the MCCD and M19 in Part 2, previous relationships allowed M19 to be chosen as the underlying cause of death. The removal of this relationship no longer assigns M19 as the underlying cause in these cases. This has resulted in a decrease in deaths assigned to M19.

O94 Sequelae of complication of pregnancy, childbirth and the puerperium: This is a new code in 2016. O94 is used for morbidity coding only and therefore has not affected output in this publication.

P91 Other disturbances of cerebral status of newborn: This category has a new fourth-digit code, P917 Acquired hydrocephalus of newborn. No conditions were assigned to this code in 2018.

W26 Contact with other sharp object(s): There are now multiple fourth-digit options for W26. Previously, when a death occurred as a result of W26, there was no option to further specify the type of sharp object involved. ABS mortality coders are now required to choose from multiple fourth-digit options to further specify the death:

• W260 Contact with knife, sword or dagger
• W268 Contact with other sharp object(s), not elsewhere classified
• W269 Contact with unspecified sharp object(s)

As a result, deaths previously assigned to W26 will now be assigned a fourth-digit during coding. No conditions were assigned to W260-W269 in 2018.

1  Deaths that are referred to a coroner can take time to be fully investigated. To account for this, the ABS has implemented a revisions process for those deaths where coronial investigations remained open at the time a preliminary cause of death was assigned. Data are deemed preliminary when first published, revised when published the following year and final when published after a second year. This technical note focusses specifically on final data for 2016 coroner-certified deaths.

2  The revisions process has been applied to all reference periods from 2006 onwards. Revisions are one of two measures implemented to enable timely data to be released on coroner-certified deaths (see Explanatory Notes 54-62 for further information). The second measure, referred to as 'open coding', ensures that all available documentation is taken into account when assigning a cause of death to coronial cases that are yet to be finalised. The combination of these two measures, along with ongoing enhancements in the timeliness and completeness of documentation on the National Coronial Information System (NCIS), have resulted in significant improvements to the quality of preliminary Causes of Death data. 

3  There are three main improvements to the Causes of Death data which are gained through the revisions process. Firstly, for deaths from natural causes a more specified condition may be identified. For example, a death may be coded to a condition such as cardiac arrest at preliminary coding, but with the later addition of an autopsy report, an underlying ischaemic heart condition could be identified. Secondly, for deaths from external causes (accidents, assaults and suicides) more information might be provided on mechanism. For example, a death coded to an unspecified accident with a fracture of hip, may later be found to have been caused by a fall down steps. Lastly, external causes may also have the intent of death updated through revisions. For example, a drug overdose where the intent of death was not determined at preliminary coding, may be updated to an intentional drug overdose when a coronial finding has been made.

4 Until the 2014 reference period, the ABS released the annual Causes of Death dataset 15 months after the end of each reference period (i.e. data for the 2014 reference period was published in March 2016). The 2015 release of Causes of Death, Australia was released 6 months earlier, representing a significant change in processing of the national mortality dataset. 

5 Bringing forward the release of Causes of Death data meant that preliminary coding of coroner-certified deaths occurred approximately 6 months earlier than in previous years. Given that the timeliness of report availability on the NCIS is critical to the ABS's ability to assign specific cause of death codes, considerable analysis was undertaken to ensure the preliminary dataset would be of sufficient quality to be fit for purpose. See Technical Note 1 A More Timely Annual Collection: Changes to ABS Processes in the 2015 publication.

6  With earlier release of preliminary data, there is now a period of 30 months between the release of preliminary and final data. The table below shows the impact of this changed revisions process at the International Classification of Diseases, 10th revision (ICD-10) chapter level. The continued earlier release of data resulted in more deaths assigned at preliminary coding to the Symptoms, signs and abnormal clinical and laboratory findings, not elsewhere classified (Symptoms and signs) (R00-R99) chapter for the 2016 reference period. Consequently, a larger number of deaths have been reassigned from R00-R99 to other chapters over the 2016 revisions period. The redistribution of deaths to more specified ICD-10 codes is discussed in more detail below.

7 The table below provides the counts of deaths by ICD-10 chapter for the 2016 reference period across the revisions process. Revisions are most likely to result in decreases in the number of deaths assigned to Symptoms and signs (R00-R99) with corresponding increases in other chapters.

8 Deaths which are originally coded to the Symptoms and signs (R00-R99) chapter can be reassigned to specific natural or external causes of death. The majority of those reassigned are subsequently found to be deaths from natural causes (76.7%), with Diseases of the circulatory system (I00-I99) being the most common natural cause. Of those reassigned to external causes of death, 18 were deaths due to intentional self-harm (suicide) and 39 were due to Accidental drug poisoning (X40-X44).

9 The expected outcome of the revisions process is to improve data quality. Enhancements to underlying cause of death data quality may include updates to either mechanism or intent or identifying an underlying cause where not previously possible. While the revisions process has a minimal impact on statistical output at the chapter level of the ICD-10 (with the exception of R00-R99), data improvements become more apparent when considering movements within individual chapters.

10 The table below shows selected data for coroner-certified deaths only at the sub-chapter level. There were key data improvements for specification of mechanism for external causes of deaths over the 2016 revisions period. There were 263 deaths where intent was coded but mechanism was unspecified at preliminary coding. Through the revisions process a mechanism was identified for 163 (62.0%) of these deaths. The majority of these records had no change in intent, but were assigned a more specific mechanism. For example, a suicide death where the mechanism was unspecified at preliminary coding (Intentional self-harm by unspecified means (X84) may be reassigned to a suicidal drowning (Intentional self-harm by drowning (X71)) during the revisions process when an autopsy becomes available for analysis.

11 The table below demonstrates that for deaths which are certified by a coroner, the reduction in the number of cases assigned to Other ill-defined and unspecified causes of mortality (R99) decreased by 39.8% over the full revisions process.

12 The table below shows changes at the sub-chapter level for the 2016 reference period, with a focus on the External causes of morbidity and mortality (V01-Y98) chapter.

Notable increases in deaths due to external causes over the full revisions process include:

• Falls (W00-W19) increased by 63 deaths. For many of the deaths reassigned to a fall, the type of injury was known at preliminary coding (e.g. neck of femur fracture), yet the mechanism was unknown (e.g. the broken hip was caused by an unspecified accident). Over the full revisions process, additional information about the nature of the mechanism became available allowing these records to be reassigned to a fall (e.g. the broken hip was identified to be due to a fall down stairs)
• Accidental drug poisoning (X40-X44) increased by 37 deaths. Many of the deaths reassigned to an Accidental drug poisoning (X40-X44) were originally assigned to Ill-defined causes of mortality (R99). Drug-induced deaths require intensive investigations to accurately determine not only the cause and manner in which the death occurred, but also the attribution of a drug(s) to the death. Over time, as investigations are finalised, more information on the NCIS becomes available allowing these deaths to be reassigned to Accidental drug poisonings (X40-X44)
• Intentional self-harm (X60-X84, Y870) increased by 43 deaths. Of these 43 deaths, the majority were reassigned from Accidental drug poisonings (X40-X44), Undetermined intent (Y10-Y34) and Ill-defined causes of mortality (R99)

• Intentional drug poisoning (X60-X64) increased by 22 deaths. Of these 22 deaths, the majority were reassigned due to updated intent information becoming available (especially the final coronial finding). The highest number of Intentional drug poisonings (X60-X64) were reassigned from Accidental drug poisoning (X40-X44)

• Car occupant injured in transport accident (V40-V49) increased by 20 deaths. The majority of these deaths were reassigned as more specificity surrounding the circumstances of the death became available. Deaths were predominantly reassigned from Unspecified motor vehicle and transport accidents (V89, V98-V99) and Exposure to unspecified factor (X59)

• Assault (X85-Y09) increased by 18 deaths. Of these 18 deaths, the majority were reassigned due to updated intent information becoming available. Deaths were predominantly reassigned from Ill-defined causes of mortality (R99) and Undetermined intent (Y10-Y34)
   

13 Various improvements to the availability and timeliness of national mortality information have been undertaken over several years. One major improvement undertaken by the NCIS is the more timely upload of reports and information for open coroner cases. This information can then be used at an earlier point by the ABS to improve open coding data quality. Earlier availability of reports can reduce the number of deaths from Ill-defined and unspecified causes of mortality (R99) present in the dataset at preliminary coding. The improved timeliness in report attachment on the NCIS was a key factor in enabling the ABS to bring forward the publication of annual causes of death data. A comparison of 2014, 2015 and 2016 final Ill-defined and unspecified causes of mortality (R99) counts indicate a substantial reduction, from 956 in 2014 to 722 in 2015, then to 606 in 2016.

14 There are some specific causes of death that may be more impacted by the changed revisions process. These include Accidental drug poisoning (X40-X44), Intentional drug poisoning (X60-X64) and Sudden Infant Death Syndrome (SIDS) (R95). Deaths from these causes require intensive investigations to accurately determine the cause and manner in which the death occurred. Therefore some key reports may not be available on the NCIS when preliminary coding of these deaths occur. These deaths are particularly sensitive to the revisions process in that more detailed information regarding the context of the death is often gained over time as information from investigations becomes available on the NCIS. 

15 The number of deaths assigned to SIDS (R95) increased by 10 deaths between preliminary and final coding. Of these 10 deaths, 9 were previously assigned to Ill-defined causes of mortality (R99). While revised data captures a significant proportion of SIDS deaths, the rules for classifying these deaths are influenced by specific terminology used in coronial findings. Data users should consider combining deaths coded to SIDS in conjunction with infant deaths coded to Ill-defined and unspecified causes of mortality (R99) when seeking to understand how many sudden unexplained deaths in infants occur in total. 

16 Over the revisions process there was an increase of 62 drug-induced deaths (includes all intents: Accidental (X40-X44), Intentional (X60-X64) and Undetermined (Y10-Y14). Accidental drug poisoning (X40-X44) contributed the largest increase across intent types for drug poisonings over the 2016 revisions process, accounting for 59.7% of the increase. 

17 The process for determining that a death was caused by Accidental drug poisoning (X40-X44) is complex as multiple factors such as drug type, intent and presence of pre-existing natural disease need to be considered. Just under half (47.6%) of the deaths reassigned to an Accidental drug poisoning (X40-X44) were initially coded to Ill-defined and unspecified causes of mortality (R99). These deaths typically did not have toxicology and/or pathology reports available on the NCIS at the time of preliminary coding. A further 19.5% of those reassigned to this category were initially coded to Intentional drug poisoning (X60-X64) followed by Undetermined drug poisoning (Y10-Y14) (12.8%). 

18 Determining deaths from Intentional drug poisoning (X60-X64) is similarly complex. Over one-third (37.9%) of deaths reassigned to an Intentional drug poisoning (X60-X64) were coded at preliminary as Accidental drug poisoning (X40-X44) deaths. These deaths typically had only an initial police report available at preliminary coding, where details on the intent of death can be unclear. A further 20.7% of reassigned Intentional drug poisoning (X60-X64) deaths were initially coded to Ill-defined causes of mortality (R99). These deaths typically did not have police, toxicology and/or pathology reports available on NCIS at the time of preliminary coding.

19 The revisions process has traditionally focused on improving specificity of the underlying cause of death. More recently, there has been growing interest in associated cause statistics which can provide a more complete picture of the diseases and/or circumstances that contributed to a death. Associated causes include the type of injuries sustained by a deceased person, drug type in a drug-induced death (e.g. heroin, cannabis), chronic disease (e.g. cancer) and mental and behavioural disorders (e.g. depression, anxiety). The ABS has maximised the use of improved report attachment on the NCIS to enhance associated cause statistics through the revisions process. Analysis of associated causes of death can better enable targeted policy and prevention initiatives, especially for those deaths which are deemed preventable. For this reason, the revisions process typically focusses on associated cause of death enhancements for two key areas - drug specification in drug-induced deaths and mental and behavioural disorders implicated in deaths from external causes.

20 There are multiple complex factors which need to be considered when a death is certified as due to a drug poisoning. The timing between the death and toxicology testing can influence the levels and types of drugs detected, making it difficult to determine the true level of a drug at the time of death. Individual tolerance levels may also vary considerably depending on multiple factors, including sex, body mass and a person’s previous exposure to a drug. Consideration of contextual factors around the death must also be considered such as pre-existing natural disease and reports from friends and families regarding the circumstances surrounding death. For these reasons, the certification of a death as a drug poisoning can take significant time to complete, making these deaths particularly sensitive to the revisions process.

21 Policies directed at reducing deaths due to drug poisoning employ a variety of strategies depending on drug type. Information regarding the type of drug(s) in a drug poisoning can often depend on the availability of an autopsy, toxicology or coronial finding report. When these reports are not available, the drug type is unknown and coded to Other and unspecified drugs, medicaments and biological substances (Unspecified drug) (T509). Importantly, deaths coded with an Unspecified drug (T509) are still counted as a drug poisonings at preliminary output, but they may be enhanced with more specific information about drug type via the revisions process. 

22 From preliminary to final, the number of drug-induced deaths in 2016 where drug type was not specified (T509) decreased from 110 to 4. As a result there was an increase in the number of specified drug types (see table below) with Benzodiazepines (T424) recording the largest increase (134 mentions) when analysed by single drug type. This was followed by Other and unspecified antipsychotics and neuroleptics (T435) (81 mentions) and Other and unspecified antidepressants (T432) (69 mentions).

23 Associated causes of death may provide important contextual information for deaths due to Intentional self-harm (X60-X84, Y870). At preliminary coding, 79.7% of suicides in 2016 had associated causes mentioned as contributory factors to death. Through the revisions process, this proportion increased to 84.0%. The table below shows the top 5 increases for associated causes of death as they relate to Intentional self-harm (X60-X84, Y870). The number of deaths with Mental and behavioural disorders due to psychoactive substance use (F10-F19) mentioned as an associated cause increased by the most over the revisions period, followed by Mood disorders (F30-F39), including depression, and Suicide ideation (R458).

24 Associated causes of death may also provide critical insights into deaths due to Accidental drug poisoning (X40-X44). The table below shows the top 5 increases for associated causes of death as they relate to Accidental drug poisoning (X40-X44). As additional evidence and documentation was added to the NCIS there were 130 accidental drug overdoses where a Mental and behavioural disorders due to psychoactive substance use (F10-F19), such as addiction or chronic substance misuse, was identified. Mood disorders (F30-F39) were identified as being a factor in 69 accidental drug-induced deaths via the revisions process and anxiety and stress-related disorders (F40-F49) were identified as a factor in 49 deaths.

1  Deaths that are referred to a coroner can take time to be fully investigated. To account for this, the ABS has implemented a revisions process for those deaths where coronial investigations remained open at the time a preliminary cause of death was assigned. Data are deemed preliminary when first published, revised when published the following year and final when published after a second year. This technical note focusses specifically on revised data for 2017 coroner-certified deaths.

 2  The revisions process has been applied to all reference periods from 2006 onwards. Revisions are one of two measures implemented to enable timely data to be released on coroner-certified deaths (see Explanatory Notes 54-62 for further information). The second measure, referred to as 'open coding', ensures that all available documentation is taken into account when assigning a cause of death to coronial cases that are yet to be finalised. The combination of these two measures, along with ongoing enhancements in the timeliness and completeness of documentation on the National Coronial Information System (NCIS), have resulted in significant improvements to the quality of preliminary Causes of Death data. 

 3  There are three main improvements to the Causes of Death data which are gained through the revisions process. Firstly, for deaths from natural causes a more specified condition may be identified. For example, a death may be coded to a condition such as cardiac arrest at preliminary coding, but with the later addition of an autopsy report, an underlying ischaemic heart condition could be identified. Secondly, for deaths from external causes (accidents, assaults and suicides) more information might be provided on mechanism. For example, a death coded to an unspecified accident with a fracture of hip, may later be found to have been caused by a fall down steps. Lastly, external causes may also have the intent of death updated through revisions. For example, a drug overdose where the intent of death was not determined at preliminary coding, may be updated to an intentional drug overdose when a coronial finding has been made.

4 Until the 2014 reference period, the ABS released the annual Causes of Death dataset 15 months after the end of each reference period (i.e. data for the 2014 reference period was published in March 2016). The 2015 release of Causes of Death, Australia was released 6 months earlier, representing a significant change in processing of the national mortality dataset. 

5 Bringing forward the release of Causes of Death data meant that preliminary coding of coroner-certified deaths occurred approximately 6 months earlier than in previous years. Given that the timeliness of report availability on the NCIS is critical to the ABS's ability to assign specific cause of death codes, considerable analysis was undertaken to ensure the preliminary dataset would be of sufficient quality to be fit for purpose. See Technical Note 1 A More Timely Annual Collection: Changes to ABS Processes in the 2015 publication.

6  With earlier release of preliminary data, there is now a period of 18 months between the release of preliminary and revised data. The table below shows the impact of this changed revisions process at the International Classification of Diseases, 10th revisions (ICD-10) chapter level. As anticipated, the magnitude of changes is the largest for deaths assigned to the Symptoms, signs and abnormal clinical and laboratory findings, not elsewhere classified (Symptoms and signs) (R00-R99) chapter, decreasing by 26.7% for the 2017 reference year. This is comparable to the decrease in 2015 (26.9%), the first year the publication was released 6 months earlier. The redistribution of deaths to more specified ICD-10 codes is discussed in detail below.

7 The table below provides the counts of deaths by ICD-10 chapter for the 2017 reference period from preliminary to revised. Revisions are most likely to result in decreases in the number of deaths assigned to the Symptoms and signs (R00-R99) chapter with corresponding increases in other chapters.  

 8 Deaths which are originally coded to the Symptoms and signs (R00-R99) chapter can be reassigned to specific natural or external causes of death. The majority of those reassigned are subsequently found to be deaths from natural causes (70.7%), with Diseases of the circulatory system (I00-I99) being the most common natural cause. Of those reassigned to external causes of death, 20 were found to be suicides.

9 The expected outcome of the revisions process is to improve data quality. Enhancements to underlying cause data quality may include improved understanding of either mechanism or intent or identifying an underlying cause where not previously possible. While the revisions process has a minimal impact on statistical output at the chapter level of the ICD-10 (with the exception of R00-R99), data improvements become more apparent when considering movements within individual chapters.

10 The table below shows data for coroner-certified deaths only at the sub-chapter level. There were key data improvements for specification of mechanism for external causes of deaths over the 2017 revisions period. There were 148 deaths where intent was coded but mechanism was unspecified at preliminary coding. Through the revisions process a mechanism was identified for 66 (44.6%) of these deaths. The majority of these records had no change in intent, but were assigned a more specific mechanism. For example, a suicide death where the mechanism was unspecified at preliminary coding (Intentional self-harm by unspecified means (X84)) may be reassigned to a suicidal drowning (Intentional self-harm by drowning (X71)) during the revisions process when an autopsy becomes available for analysis. Intentional self-harm by drowning (X71) during the revisions process when an autopsy becomes available for analysis. 

11 The table below further demonstrates that the number of coroner-certified deaths assigned to Other ill-defined and unspecified causes of mortality (R99) decreased by 43.6% from preliminary to revised.

12 The table below provides information on changes at the sub-chapter level for the 2017 reference period, with a focus on the External causes of morbidity and mortality (V01-Y98) chapter.

Notable increases in deaths due to external causes include:

• Accidental drug poisoning (X40-X44) increased by 98 deaths. Many of the deaths reassigned to an accidental drug poisoning death were originally assigned to Ill-defined causes of mortality (R99). Drug-induced deaths require intensive investigations to accurately determine not only the cause and manner in which the death occurred, but also the attribution of a drug(s) to the death. Over time, as investigations are finalised, more information on the NCIS becomes available allowing these deaths to be reassigned to an accidental drug poisoning.

• Intentional self-harm (X60-X84, Y870) increased by 69 deaths. Of these 69 deaths the majority were reassigned due to updated intent information becoming available (especially the final coronial finding). Deaths were predominantly reassigned from Undetermined intent (Y10-Y34), Accidental drug poisoning (X40-X44) and Other ill-defined and unspecified causes of mortality (R99).

• Intentional drug poisoning (X60-X64) increased by 36 deaths. Of these 36 deaths the majority were reassigned due to updated intent information becoming available (especially the final coronial finding). Deaths were predominantly reassigned from Accidental drug poisoning (X40-X44).
   

13 Various improvements to the national mortality system have been undertaken over several years. One major improvement undertaken by the NCIS is the more timely upload of reports and information for open coroner cases. This information can then be used at an earlier point by the ABS to improve open coding data quality. Specifically, earlier availability of reports can reduce the number Ill-defined and unspecified causes of mortality (R99) present in the dataset at preliminary coding. These improvements are now being reflected in the mortality dataset. A comparison of 2015 and 2017 preliminary R99 counts of coroner-certified deaths indicate a substantial reduction, from 1,427 in 2015 to 1,199 in 2017. 

14 There are some specific causes of death that may be more impacted by the changed revisions process. These include Accidental drug poisoning (X40-X44), Intentional drug poisoning (X60-X64) and Sudden Infant Death Syndrome (SIDS) (R95). Deaths from these causes require intensive investigations to accurately determine the cause and manner in which the death occurred. Therefore some key reports may not be available on the NCIS when preliminary coding of these deaths occurs. These deaths are particularly sensitive to the revisions process, in that more detailed information regarding the context of the death is often gained through revisions. 

15 The number of deaths assigned to SIDS (R95) increased by 6 deaths between preliminary and revised coding. All 6 deaths were initially coded to Ill-defined causes of mortality (R99). While revised data captures a significant proportion of SIDS deaths, the rules for classifying these deaths are heavily influenced by specific terminology used in coronial findings. Data users should consider combining deaths coded to SIDS (R95) in conjunction with infant deaths coded to Ill-defined and unspecified causes of mortality (R99) when seeking to understand how many sudden unexplained deaths in infants occur in total. 

16 Over the revisions process there was an increase of 100 drug-induced deaths (includes all intents: Accidental (X40-X44), Intentional (X60-X64) and Undetermined (Y10-Y14)). Accidental drug poisonings (X40-X44) contributed the largest increase across intent types for drug poisonings over the 2017 revisions process. 

17 The process for determining that a death was caused by an Accidental drug poisoning (X40-X44) is complex, as multiple factors such as drug type, intent and presence of pre-existing natural disease need to be considered. Of the deaths reassigned to Accidental drug poisoning (X40-X44), approximately 62.3% were initially coded to Other ill-defined and unspecified causes of mortality (R99). A further 23.1% of those reassigned to this category were initially coded to Undetermined drug death (Y10-Y14). These deaths typically had only an initial police report available at preliminary coding, where circumstances surrounding death can be unclear and often appear similar to deaths from natural causes.

18 Determining deaths from Intentional drug poisoning (X60-X64) is similarly complex. Around 37.3% of deaths reassigned to an Intentional drug poisoning (X60-X64) were coded at preliminary as Accidental drug poisoning (X40-X44). These deaths often had only an initial police report available at preliminary coding, where details on the intent of death can be unclear. A further 23.5% of those reassigned to this category were initially coded to Ill-defined and unspecified causes of mortality (R99). These deaths typically did not have toxicology and/or pathology reports available on NCIS at the time of preliminary coding.

19 The revisions process has traditionally focussed on improving specificity of the underlying cause of death. More recently, there has been growing interest in associated cause statistics which can provide a more complete picture of the diseases and/or circumstances that contributed to a death. Associated causes include the type of injuries sustained by a deceased person, drug type in a drug-induced death (e.g. heroin, cannabis), chronic disease (e.g. cancer) and mental and behavioural disorders (e.g. depression, anxiety). The ABS has maximised the use of improved report attachment on the NCIS to enhance associated cause statistics through the revisions process. Analysis of associated causes of death can better enable targeted policy and prevention initiatives, especially for those deaths which are deemed preventable. For this reason, the revisions process typically focusses on associated cause of death enhancements for two key areas - drug specification in drug-induced deaths and mental and behavioural disorders implicated in deaths from external causes.

20 There are multiple complex factors which need to be considered when a death is certified as drug-induced. The timing between the death and toxicology testing can influence the levels and types of drugs detected, making it difficult to determine the true level of a drug at the time of death. Individual tolerance levels may also vary considerably depending on multiple factors, including sex, body mass and a person’s previous exposure to a drug. Consideration of contextual factors around the death must also be considered such as pre-existing natural disease and reports from friends and families regarding the circumstances surrounding death. For these reasons, the certification of a death as being drug-induced can take significant time to complete, making these deaths particularly sensitive to the revisions process.

21 Policies directed at reducing deaths due to drug poisoning employ a variety of strategies depending on drug type. Information regarding the type of drug(s) in a drug poisoning can often depend on the availability of an autopsy, toxicology or coronial finding report. When these reports are not available, the drug type is unknown and coded to Other and unspecified drugs, medicaments and biological substances (Unspecified drug) (T509). Importantly, deaths coded with an Unspecified drug (T509) are still counted as a drug-induced death at preliminary output, but they may be enhanced with more specific information about drug type via the revisions process. 

22 From preliminary to revised, the number of drug-induced deaths in 2017 where drug type was not specified (Unspecified drug (T509)) decreased from 100 to 22. As a result, there was an increase in the number of specified drug types (see table below) with Benzodiazepines (T424) recording the largest increase (142 additional mentions) when analysed by single drug type. This was followed by Other opioids (T402) (85 additional mentions) and Other and unspecified antipsychotics and neuroleptics (T435) (64 additional mentions).

23 Associated causes of death may provide important contextual information for deaths due to Intentional self-harm (X60-X84, Y870). At preliminary coding, approximately 78.7% of suicides in 2017 had associated causes mentioned as contributory factors to death. Through revisions, this proportion increased to 82.3%. The table below shows the top 5 increases for associated causes of death as they relate to Intentional self-harm (X60-X84, Y870). Mood disorders (F30-F39), which include depression and bipolar affective disorder, were the most common associated causes of death identified during the revisions process, followed by Mental and behavioural disorders due to psychoactive substance use (F10-F19) and Suicide ideation (R458).

24 Associated causes may also provide critical insight into deaths due to Accidental drug poisoning (X40-X44). The table below shows the top 5 largest increases in associated causes for Accidental drug poisonings (X40-X44). As additional evidence and documentation was added to the NCIS there were 101 accidental drug overdoses where a Mental and behavioural disorders due to psychoactive substance use (F10-F19) such as addiction or chronic substance misuse was identified. Mood disorders (F30-F39) were identified as being a factor in 66 accidental drug-induced deaths via the revisions process and anxiety and stress-related disorders (F40-F49) were identified as a factor in 55 deaths.

1 As part of the ABS's revisions process for Causes of Death, the ABS updates causes for coroner-certified deaths at 12 and 24 months after initial processing, to reflect the latest available information. Revisions have now been applied to 2016 and 2017 data. As coronial investigations regarding deaths due to suspected suicide can be extensive, it is a cause of death which may be more heavily impacted by revisions. It is important from a public health perspective to have accurate counts of suicides. As such, this technical note focusses on how the revisions process has changed suicide counts in 2016 and 2017. 

2 Over time there has been a reduction in the number of deaths that are reassigned to suicide through the revisions process. In 2006 and 2007, the first years for which revisions were applied, the number of suicide deaths increased by 17.7% and 18.5%, respectively. In 2016, the final suicide count was 1.5% higher than the preliminary count. Several factors have impacted on the increased quality of preliminary data, including enhanced coding practices, enabling greater use of documents available on the National Coronial Information System (NCIS) and more timely report attachment.

3 The final number of deaths due to suicide recorded for 2016 is 2,909, a net increase of 43 deaths (1.5%) from the preliminary count of 2,866. There was an increase of 45 suicides over the first revision period and a reduction of 2 suicides in the second revision period. 

4 Deaths which have been reassigned to suicide through the revisions process were most likely to be coded to an Accidental poisoning (X40-X44) (22 deaths) or Event of undetermined intent (Y10-Y34) (19 deaths) when initially coded. There were 18 deaths that were initially coded to Other ill-defined and unspecified causes of mortality (R99) and later identified as suicide deaths through the revisions process. There were also some minor changes in the recorded mechanism of death associated with additional information becoming available, especially toxicology and pathology reports. 

5 The revised number of suicides in 2017 is 3,197, a net increase of 69 suicide deaths (2.2%) over the first year of the revisions process. Most deaths reassigned to suicide were initially coded to Event of undetermined intent (Y10-Y34) (32 deaths). There were 20 deaths that were reassigned from both Accidental poisoning (X40-X44) and from Other ill-defined and unspecified causes of mortality (R99). The table below shows the total suicide counts for Australia at each stage of the revisions process for 2016 and 2017. 

6 New South Wales (NSW) recorded the largest increase in deaths due to suicide for the 2017 revisions period. The revised number of suicides for NSW in 2017 is 929, a net increase of 49 (5.6%). Updated suicide data for jurisdictions is provided at the end of this technical note. 

7 The number and age-standardised death rate of deaths due to intentional self-harm by state and territory from 2009 to 2018 are shown in the tables below. These tables provide an updated time series that includes the revisions for 2016 and 2017 and should now be used in preference to those published in September 2018. A more detailed table which includes revised suicide counts by mechanism (ICD-10 codes X60-X84 and Y87.0) are provided in the Revisions data cube in the Data downloads section of this publication. Further tabulations are available on request. Please contact the National Information and Referral Service on 1300 135 070.