groupKatrina Sheehan, Lauren Moran, Emma Torrens

menu_bookVolume 11 Number 3

collections_bookmarkReports

lock_openPublic Access

printPrint Article


The National Mortality Dataset (NMD) is an epidemiological dataset that contains annual death registrations for a given reference period. The NMD includes demographic variables including age, sex and country of birth as well as cause of death information coded to the International Classification of Diseases, 10th revision (ICD-10).

Introduction

Cause of death data includes both the underlying cause of death (UCoD) and associated causes of death (ACoD). The UCoD is the disease, condition or external event that initiated the train of morbid events leading to death. The ACoDs refer to all other conditions listed on the medical certificate of cause of death by the certifier. ACoDs can include diseases that are part of the chain of events leading to death, risk factors and co-morbid chronic conditions (Australian Bureau of Statistics 2020a).

Understanding what ACoDs contributed to an individual’s death can provide insight into intervention points to prevent or decrease some causes of death. For example, modifiable risk factors such as smoking or hypertension can be targeted via public health campaigns, changing laws (ie. changing smoking regulations) and treatments (ie. diuretics or ACE inhibitors to manage hypertension) (Pilibosian, Wu, Aldrich and Wheeler 1999). Mental health conditions such as depression or drug and alcohol abuse are known to have a negative impact on health and are focus areas in national suicide prevention strategies (Lee and Jung 2006). Additionally knowing what drugs were present in an overdose can lead to reviews of drug prescription and use patterns (Department of Health 2017).

Health and mortality are influenced by the social and physical environment in conjunction with the personal and family health histories of an individual. The World Health Organization (WHO) (Commission on Social Determinants of Health 2008) cite social determinants of health as key contributors to differences in mortality patterns and life expectancy outcomes amongst community groups, with this being supported by research showing that deaths, especially those due to external causes, occur due to a complex interaction between biological, psychological and psychosocial risk factors (Lee and Jung 2006).

From 2017, the Australian Bureau of Statistics (ABS) have coded a range of psychosocial risk factors into the NMD, not limited to but including history of self-harm, family issues such as domestic violence and divorce, and legal and financial issues. The addition of the psychosocial risk factors were designed to be used in conjunction with other ACoDs including mental health conditions, chronic diseases and injury and poisoning information, extending the utility of the NMD for policy purposes. This paper outlines the history of ACoD capture in the NMD, how the coding of psychosocial risk factors has evolved since the implementation in 2017 and some of the current utility and future opportunities for ACoD coding under ICD frameworks.

Previous processes for coding associated causes of death

The ABS has routinely captured ACoD information as part of the NMD since 1997; being implemented alongside ICD-10 into the Australian mortality coding system. Prior to 1997 only the underlying cause of death was stored for analysis purposes on the NMD. The introduction of ACoD processes meant that all conditions, diseases, injuries, poisonings and external causes were coded and stored for use in statistical outputs. For doctor-certified deaths information on causes of death is taken from the medical certificate of cause of death (MCCD). For coroner-referred deaths the ABS uses the National Coronial Information System (NCIS), an internet based medico-legal database that houses police, toxicology, forensic pathology and coronial findings for each case (Saar, Bugeja and Ranson 2017). The high quality of the NCIS enables for high quality ACoD coding and information to occur. ACoD information is available in each report depending on the context of the death. Historically, ACoD information focussed on four key areas:

Injuries and poisonings

Retaining information on injuries and poisonings for external causes of death was a key enhancement to the NMD with the introduction of ACoD analysis. Under mortality coding rules an injury or poisoning cannot be an underlying cause of death so this information was previously not available. For external causes such as drug overdoses, the poisoning data refers to the drug type. Over the last 20 years this has been used extensively to track changes in profile of those who die from drug overdoses. The data shows that in the late 1990s people who died were more likely to be young and have heroin involved in the death. The current profile of drug overdose deaths shows that prescription opioids are more likely to be involved and a person middle aged.

Natural conditions, including chronic co-occurring conditions

With an ageing population, Australians are more likely to have multiple chronic conditions at death, with an average of 3.3 conditions written on each MCCD. Retaining ACoDs demonstrates that for many people death is not the result of one disease or event, but rather the result of the interaction of multiple disease processes. Mortality coding rules favour the selection of some diseases and conditions over others meaning the true burden of some conditions on mortality such as hypertension and diabetes can be overlooked.

Mental health conditions

People living with mental health conditions have higher rates of mortality. Mental health conditions are known risk factors for avoidable deaths, such as suicide and drug overdose. Depressive disorders, anxiety disorders, patterns of drug and alcohol use and other mental health conditions provide information important for prevention and treatment as well as aid in identification of patterns of these risk factors across preventable deaths. Many mental health conditions are more likely to be an ACoD under mortality coding rules. The addition of ACoD data to the NMD has provided important insights into mental health conditions and how they manifest in different causes of death.

Ill-defined symptoms and acute exacerbation of disease

Other commonly mentioned conditions on a death certificate are ill-defined conditions such as varied organ failure, shock or not formally diagnosed clinical findings such as radiology findings. These conditions are not generally selected as the UCoD under mortality coding rules. While they may not provide detailed information when considered in isolation, when analysed with a UCoD they can provide important information on disease sequence and pathways to death. Additionally, some conditions such as chronic pain or suicide ideation can provide important context for understanding external cause deaths and targeting prevention and intervention policies.  

Coding of psychosocial risk factors

Users of ICD-10 for morbidity would be familiar with Chapter XXI Factors influencing health status and contact with health services and its ability to record diagnoses or other ‘problems’ that are not in themselves a disease or illness that can be assigned to other chapters. The types of factors someone may encounter a health service for or have their health status affected by can be wide ranging, such as environmental factors, social and family factors, and individual factors such as family and personal history of illnesses and conditions. While it may not seem likely that many psychosocial risk factors would be mentioned in mortality reporting, this can depend on the cause of death.

Deaths certified by a doctor tend to only have diagnosable diseases and conditions listed on the MCCD. This is in part a reflection of the structure of the MCCD itself, where a doctor is asked to provide their best medical opinion on the sequence of disease events leading to death and any contributing conditions.

As discussed, for coroner-referred deaths, the ABS access the NCIS and code information from the police, autopsy, toxicology and coroners’ reports. These reports provide a breadth of information relating to these deaths, much of which is highly important from a public health perspective. The investigative nature of coroner-referred deaths increases the likelihood of contextual psychosocial information being mentioned in conjunction with other risk factors such as chronic disease and mental health conditions.

From the 2017 reference period the ABS included psychosocial risk factors in the NMD, with a particular focus on coroner-referred deaths. Psychosocial risk factors were more likely to be reported in deaths that were due to suicide or drug overdose (ABS 2017). This does not mean that other types of external cause deaths or natural causes did not have psychosocial risk factors present, but rather that they were more likely to be collected as part of the investigation in some causes of death.

Table 1 highlights the ten most common psychosocial risk factors mentioned in coronial investigations of people who died by suicide in 2020. Previous history of self-harm was the most common psychosocial risk, followed by family issues such as divorce and issues in spousal relationships. The table also shows that over half of psychosocial risk factors occurred with other psychosocial risks. This supports research showing that multiple inter-related risk factors are known contributors to suicide, with many of these factors modifiable with appropriate intervention and prevention activities (Clapperton, Newstead, Bugeja and Pirkis 2019; Mościcki 1997).

Table 1: Number of co-occurring psychosocial risk factors for suicide deaths, Australia 2020

 Reported AloneReports with other risk factorsTotal occurrences
All psychosocial risk factors1,0321,0742,106
Z915 Personal history of self-harm283431714
Z635 Disruption of family by separation and divorce121283404
Z630 Problems in relationship with spouse or partner133231364
Z653 Problems related to other legal circumstances69184253
Z634 Disappearance and death of family member54196250
Z598 Other problems related to housing and economic circumstances21167188
Z736 Limitation of activities due to disability9376169
Z638 Other specified problems related to primary support group2190111
Z633 Absence of family member1191102
Z562 Threat of job loss108494

The inclusion of psychosocial risk factors (Table 2) has extended the coverage of ACoD information in the NMD. The table below shows that with the addition on psychosocial risk factors, 90% of suicides had a risk factor mentioned as part of the coronial reporting in 2019 and 2020. As cases close and more information is uploaded to the NCIS, that proportion increased to 97% in 2017. While there are limitations to the work, such as availability of information and consistency of recorded information, the data supports the identification of a greater number of prevention and intervention points for deaths due to suicide.

Table 2: Risk factor prevalence, suicides, Australia 2017-2020

 2017 Number 2017 Proportion2018 Number2018 Proportion2019 Number2019 Proportion2020 Number 2020 Proportion
Total suicides3,291100.03,192100.03,318100.03,139100.0
Total suicides reported with mental and behavioural disorder2,36771.92,22369.62,10963.62,05865.6
Total suicides reported with natural disease1,74653.11,83357.41,78253.71,73755.3
Total suicides reported with psychosocial risk factors2,33871.02,31272.42,12464.02,10667.1
Total suicides with any associated cause reported3,19297.02,99493.82,98590.02,84990.8

Updates to the coding and statistical outputs for ACoDs and psychosocial risk factors

The work undertaken to extend the range of ACoDs and psychosocial risk factors captured under the ICD-10 framework has been an enduring part of the NMD for four years. The extension of the work was enabled through a collaboration between the ABS, the Australian Institute of Health and Welfare (AIHW) and the Department of Health as part of the Suicide and Self-Harm Monitoring Project.

Engagement on the concepts underpinning how ACoDs and psychosocial risk factors relate to specific types of death has been ongoing, resulting in small refinements to processes over time. In addition to feedback from stakeholders, coding must reflect and represent societal conditions. Risk factors such as the general socio-economic, cultural and environmental conditions can vary from year to year and be dependent on extrinsic events, for example natural disasters and the current COVID-19 pandemic.

Two examples of updates to coding are presented below.

ACoDs and risk factors in child deaths

Feedback from Child Death Review Teams and other specialists in youth health highlighted the importance of capturing ACoDs and psychosocial risk factors for both natural and external causes of death in children. While there was robust coverage of ACoD and other risk factors for children who died by suicide, other causes of death with known risk factors had low capture of this information in the NMD. Previously, the application of mortality coding rules for deaths that were of unknown causes such as Sudden Infant Death Syndrome (SIDS) (R95) and Sudden Unexpected Death in Infancy (SUDI) (R99) resulted in no ACoDs or risk factors being recorded and only a single UCoD on the certificate. This meant that known risk factors for these deaths, such as hazards in the sleep environment and exposure to drugs and alcohol (Galland et al 2014), were not being recorded.

A review of all coroner-referred child deaths was undertaken for deaths registered in 2020. ACoDs and psychosocial risk factors were coded in 25% of deaths of children (causes included SIDS and SUDI, natural diseases and external causes of death such as assault). Factors relating to the physical environment (e.g. exposure to tobacco smoke) were the most common risk factor for SIDS and SUDI, whereas risk factors relating to the familial context, self-harming behaviour and issues related to education were more commonly recorded for children who died from suicide. These changes to coding processes for child deaths have been implemented as enduring procedures.

COVID-19 as a risk factor for suicide

Tracking the impact of COVID-19 on mortality has been of high importance since the start of the pandemic. This includes deaths due to the virus itself, as well as deaths that may be due to indirect consequences of the pandemic such as changes in access to health care or reduced physical connections to social networks (ABS 2020b). While there was a decrease in deaths due to suicide in 2020, coding of the 2020 coroner-referred data made it evident that the COVID-19 pandemic was an important risk factor for some people. This observation matched reports from Suicide Registers in Victoria and Queensland (Dwyer et al 2021; Leske et al 2021).

When COVID-19 was mentioned as a risk factor it manifested in different ways for individuals. For some people, direct impacts from the pandemic such as job loss, lack of financial security, family and relationship pressures and not feeling comfortable with accessing health care were noted. Others demonstrated a general concern or anxiety about the pandemic and societal changes, or anxiety about contracting the virus itself. Three ICD-10 codes were used to capture COVID-19 as a risk factor in the ACoD and risk factor data. These codes and their descriptions are outlined below.

F41.8 Other specified anxiety disorders: Anxiety and stress related to the COVID-19 pandemic

Z29.0 Isolation: An individual was in isolation in hotel or home for quarantine purposes

Z29.9 Prophylactic measure: The individual was affected by public health measures such as shutdown of business or stay at home orders.

Ensuring that ACoD and risk factor data was expanded to include measures relating to the COVID-19 pandemic has meant that the NMD is able to reflect current social experiences. As the pandemic continues, this information will continue to be coded into the NMD.

Using ACODs and psychosocial risk factor data as part of the NMD

Circumstances relating to a death are complex and multifaceted. For many people who die of a chronic disease, it is often the combined effects of disease and social and demographic circumstances that contribute to a death. On average, people who die in Australia have three to four conditions certified as having contributed to death. Importantly, risk factors should not be considered in isolation.

Similarly, for people who die due to external causes including suicide, it is the combination of multiple factors rather than a single reason that contributes to death. In 2020 people who died by suicide had an average of three to four ACoDs and risk factors reported. These factors can be from one or a combination of mental and behavioural disorders, natural diseases, and psychosocial risk factors and are often interrelated. For example, a risk factor for some people who die by suicide, (especially in older cohorts) is a limitation of activities or loss of ability to do things they enjoy due to disability or illness. This is coded to ICD-10 category Z73.6 Limitation of activities due to disability. While this is important contextual knowledge for that person, the information can be richer when considered alongside other ACoDs. Many people who have Z73.6 coded as a risk factor also have a chronic condition such as cancer which has led to the limitation of activity and other factors such as depression. The three ACoDs and risk factors can be closely related in that the cancer, limitation of activity and depression in combination may have increased a person’s risk of suicide.

When coding ACoDs and risk factors, information available on the MCCD or from the coronial investigation is coded. While some factors may have been present, this does not mean they had a causal effect. Many individuals will experience risk factors, such as those captured in the NMD, but do not experience suicidal behaviours (AIHW 2021). Similarly, an individual may have a chronic disease but it was not linked to the terminal chain of events leading to death. ACoD and risk factors should help inform additional and alternative intervention and prevention points rather than limit existing strategies.

Future opportunities

The ABS will continue to work with data users to further improve and refine the process of capturing risk factors using the ICD-10 framework.

The 11th revision of the ICD was adopted by the World Health Assembly in 2019 and countries are able to implement the classification from 2022 onwards (WHO 2019a). While Australia has no current adoption date for the use of ICD-11 when coding mortality data (due to the need to wait for supporting infrastructure) ICD-11 has extended capabilities in enabling detailed capture of ACoDs.

More detailed code sets will enable for more specified reporting. For example, more precise capture of drug and poisoning agents will be available under new extension codes. Under the current ICD-10 structure, coding of prescription opioids is limited in that natural opioids and synthetic opioids (including semi-synthetic) are unable to be reported in outputs. This means that detail such as identifying fentanyl versus tramadol involvement in a death is not possible. Opioids are the most common drug class present in drug-overdose deaths in Australia and the additional capabilities in reporting will assist in prevention strategies (ABS 2020a).

The ICD-11 also has capability to post-coordinate conditions, adding detail to base codes such as site, pathology and manifestation (WHO 2019b). This will be useful not only for completely capturing a diagnosis in a code string, but also for presenting relevant ACoD information as a succinct piece of information. This will allow relationships between conditions to be more clearly understood and analysed.

The possibilities associated with ICD-11 are extensive but will still be limited by the information available when coding and compiling data.

Conclusion

The inclusion of ACoD information (including psychosocial risk factors) in the NMD has enabled for the development of a richer dataset. Epidemiological studies are able to show the complex nature of mortality both for natural and external causes of death. The recent updates to coding using the ICD-10 framework to capture a broader range of ACoDs including psychosocial risk factors provides further value in aiding to identify prevention and intervention points for key causes of death.

The ACoD information in the NMD is not designed to replace data collected by existing specialist groups such as Child Death Review Teams and Suicide Registers. Instead, it is designed to complement the valuable work of these agencies. The NMD provides a nationally consistent output for all causes of mortality. It can be used as a standalone dataset as well as in data linkage studies where it is linked to datasets including Medicare and the Pharmaceutical Benefits Scheme. Additional information will only support further insights for research and policy.

While there are some limitations, such as being reliant on the information the doctor certifies on the MCCD or availability and consistency of information across jurisdictions for coroner-referred deaths, the data set is able to provide invaluable insight. With continued improvements, innovations and opportunities for the way in which ACoDs are captured, the NMD will continue to provide vital information for public health initiatives in Australia.

Acknowledgements

The ABS would like to acknowledge the jurisdictional Registries of Births, Deaths and Marriages as the data custodians of doctor-certified cause of death data. Additionally the ABS acknowledges the Victorian Department of Justice and Community Safety as the source organisation for coroner-referred cause of death data and the National Coronial Information System as the source database for coroner-referred cause of death cause.


References

  • Australian Bureau of Statistics (2020a) Causes of Death, Australia. Available at: https://www.abs.gov.au/statistics/health/causes-death/causes-death-australia/2020 (accessed 4 November 2021).
  • Australian Bureau of Statistics (2020b) Measuring excess mortality in Australia during the COVID-19 pandemic. Available at: https://www.abs.gov.au/articles/measuring-excess-mortality-australia-during-covid-19-pandemic (accessed 4 November 2021).
  • Australian Bureau of Statistics (2017) Research Paper: Psychosocial risk factors as they relate to coroner-referred deaths in Australia, 2017. Available at: https://www.abs.gov.au/ausstats/abs@.nsf/Latestproducts/1351.0.55.062Main%20Features32017?opendocumentandtabname=Summaryandprodno=1351.0.55.062andissue=2017andnum=andview= (accessed 4 November 2021).
  • Australian Institute of Health and Welfare (2021) Psychosocial risk factors and deaths by suicide. Available at: https://www.aihw.gov.au/suicide-self-harm-monitoring/data/behaviours-risk-factors/psychosocial-risk-factors-suicide (accessed 4 November 2021).
  • Clapperton A, Newstead S, Bugeja L, and Pirkis J (2019) Relative risk of suicide following exposure to recent stressors, Victoria, Australia. Australian and New Zealand Journal of Public Health, 43(3): 254-260.
  • Commission on Social Determinants of Health (2008) Closing the gap in a generation: heath equity through action on the social determinants of health, Final Report of the Commission on Social Determinants of Health, Geneva, World Health Organization. Available at: https://www.who.int/social_determinants/final_report/csdh_finalreport_2008.pdf (accessed 4 November 2021).
  • Department of Health (2017) National Drug Strategy 2017-2026. Available at: https://www.health.gov.au/sites/default/files/national-drug-strategy-2017-2026.pdf (accessed 4 November 2021).
  • Dwyer J, Dwyer J, Hiscock R, O’Callaghan C, Taylor K, Millar C, Lamb A, and Bugeja L (2021) COVID-19 as a context in suicide: early insights from Victoria, Australia.  Australian and New Zealand Journal of Public Health, 45(1): 517-522.
  • Galland BC, Gray A, Sayers RM, Heath AM, Lawrence J, Taylor R, and Taylor BJ (2014) Safe sleep practices in a New Zealand community and development of Sudden Unexpected Death in Infancy (SUDI) risk assessment instrument. BMC Pediatrics, 14(263).
  • Lee SI and Jung H (2006) Psychosocial risk factors for suicide. Psychiatry Investigation, 3(2);15-22.
  • LeskeS, Kolves K, Crompton D, Arensman E, de Leo D (2021) Real-time suicide mortality data from police reports in Queensland, Australia, during the COVID-19 pandemic: An interrupted time-series analysis. Lancet Psychiatry, 8(1): 58-63.
  • Moscicki EK (1997) Identification of suicide risk factors using epidemiologic studies. The Psychiatric Clinics of North America, 20(3): 499-517.
  • Pilibosian ED, Wu M, Aldrich TE, Wheeler FC (1999) The modifiable factors contributing to leading causes of death in South Carolina. Journal of the South Carolina Medical Association, 95(6): 227-230.
  • Saar E, Bugeja L, and Ranson DL (2017) National Coronial Information System: Epidemiology and the Coroner in Australia. Academic forensic pathology, 7: 582-590.
  • World Health Organization (2019a) ICD-11 Implementation or Transition Guide, 2019. Available at: https://icd.who.int/docs/ICD-11%20Implementation%20or%20Transition%20Guide_v105.pdf (accessed 4 November 2021).
  • World Health Organization (2019b) ICD-11 fact sheet. Available at: https://icd.who.int/en/docs/icd11factsheet_en.pdf (accessed 4 November 2021).

Authors

Katrina Sheehan, BHlthSc (Public Health)
Mortality Coding Team Leader
Mortality Data Centre Health and Vital Statistics Section
Australian Bureau of Statistics

Lauren Moran, BBehSc
Acting Director
Health and Vital Statistics Section
Australian Bureau of Statistics

Emma Torrens, BBehSc, MHlthMgt
Mortality Coding Manager
Mortality Data Centre Health and Vital Statistics Section
Australian Bureau of Statistics