Psychosocial risk factors as they relate to coroner-referred deaths in Australia

The research paper: Psychosocial risk factors as they relate to coroner referred deaths in Australia, describes the methods and initial results of a pilot study conducted on deaths that occurred in 2017, focussing particularly on suicide accidental and drug-induced deaths.

The circumstances relating to suicide and drug deaths are complex. Existing mortality data captures information on the intent (accident, assault or self-harm) and mechanism of deaths, as well as information on associated causes such as drug or alcohol addictions, mental health conditions, chronic and terminal diseases. The risk factor pilot study has enabled additional information, including a past history of self-harm, relationship issues, legal issues, bereavement, unemployment, homelessness and disability, to be added to what's known about the circumstances of these deaths.

The ABS will work with stakeholders to refine study methods and further analyse results. It is hoped that insights provided by this study will assist those working in suicide research and prevention and can be embedded as part of future national mortality datasets.

• Care should be taken when reporting suicide deaths. Please refer to the Mindframe website for further guidance.
• When reporting ABS data, the Australian Bureau of Statistics (or ABS) must be attributed as the source.
• For media requests and interviews, contact the ABS Communications Section on 1300 175 070 (8.30am - 5pm Mon-Fri).
• Subscribe to our email notification service and get media releases or products sent to you on release.
   

Lifeline: 13 11 14

Suicide Call Back Service - 1300 659 467

Kids Helpline (for young people aged 5 to 25 years): 1800 55 1800 

The national Causes of Death dataset is used extensively as a foundation for policy, research and planning. The purpose of the dataset is to provide information on the sequence of diseases, injuries and external causes leading to a person’s death, as well as associated causes which may have contributed to mortality.

It is widely recognised that to enhance the value of mortality data for public health purposes, cause of death information cannot be considered in isolation. For example, when causal information is combined with demographic variables such as age, sex and socio-economic status, important insights are made into health profiles and potential areas of health intervention.

Health and mortality are also influenced by the social environment of an individual. The social environment may include employment, housing, social support and family factors, which may serve as both risk and protective factors. The WHO (CSDH, 2008) cite social determinants of health as key contributors to differences in mortality patterns and life expectancy outcomes amongst community groups. This idea is supported by research showing that external cause deaths, and those due to suicide in particular, occur due to a complex interaction between biological, psychological and psychosocial risk factors (Lee & Jung, 2006).

Psychosocial factors are not collected systematically on death certificates in Australia. The legal requirement for a doctor or coroner is to provide disease or injury information related to the death, alongside key demographic variables including age, sex and ethnicity. Subsequent coding and statistical output by the Australian Bureau of Statistics (ABS) is designed to capture these disease and injury processes and provide an overview of causes of death in Australia. Hence, data gaps exist in the national mortality dataset as they relate to social determinants such as psychosocial risk factors.

The ABS accesses the National Coronial Information System (NCIS) to code all coroner-certified deaths in Australia. The NCIS is a medico-legal database which holds a range of information pertaining to each death, including reports from police, toxicologists and pathologists as well as the coroner. These reports contain information on not only diseases, mental health conditions and external events and injury, but also a wealth of information on psychosocial factors as they relate to the deceased individual.

The ABS has undertaken a pilot study to enhance the national Causes of Death dataset by capturing information on psychosocial risk factors for deaths referred to a coroner in the 2017 reference period. Psychosocial risk factors were identified in coronial reports including police, pathology and coronial findings and codes from the International Classification of Diseases 10th revision (ICD-10) were assigned, in order to summarise the information in a standard way.

This paper provides information on how this project was conceptualised and how data was coded and analysed. Initial results are presented with an aim to assess the quality of outputs and the fitness for purpose of the dataset into the future. A set of supplementary data cubes demonstrating further use of this dataset are available in the Data downloads section of this publication.

Enhancement of causes of death datasets is not a new phenomenon. Over the last fifty years, there have been many changes to the causes of death dataset, always with the aim of enhancing data for public health purposes. The changes span many areas and include updates to the coding classification (the International Classification of Diseases) to incorporate more detailed categories of disease, injury and poisoning. A brief overview of changes to the ABS data set is discussed below.

The most widely used output of mortality is that tabulated by the underlying cause of death. The underlying cause of death is the disease or external event which initiated the train of morbid events leading to death. It is widely recognised that in the context of an ageing population where multiple comorbidities are present, and for external causes where injury outcomes and contextual events are extremely important for intervention and prevention, that one cause of death is not sufficient for detailed analysis.

When the ABS commenced using ICD-10 in 1997, major changes were made to the national Causes of Death dataset. For deaths coded to ICD-8 and ICD-9, deaths were stored with one ICD code only (the underlying cause of death). From ICD-10 onwards, the ABS began capturing and storing all information on diseases and conditions mentioned on the Medical Certificate of Cause of Death (MCCD). This practice is known as multiple cause coding. Multiple cause data adds an enhanced dimension to mortality statistics by providing more accurate representations of the prevalence of a disease at death, highlighting common associations between diseases and giving a more complete view of causes of death (Moriyama, Loy & Robb-Smith, 2011). On average each death in Australia has between 3-4 diseases certified by the practitioner.

The NCIS holds a wealth of information regarding contextual factors as they relate to death in the police, toxicology, pathologist and coroner finding reports. Contextual information may include existing mental health conditions such as depression or addiction. In recent years, concerted effort has been made by the ABS mortality coding team to identify and include information on mental health conditions, diseases and drug and alcohol usage on the national mortality dataset. As deaths referred to a coroner can be subject to long investigation times, the ABS now updates associated cause information as part of its annual revisions process to ensure as much detail as possible is recorded against a case (ABS, 2017). This change has been particularly beneficial for suicides and drug-induced deaths, where it is known that understanding risk factors can lead to improved policy, interventions and prevention activities (Ayuso-Mateos, 2019).

The ICD is subject to updates to reflect changes in medical understanding and incorporate more detailed categories to improve data outputs. The ABS works closely with the World Health Organization (WHO) on this update process. ABS data has been coded to many versions of the ICD over time, with each revision demonstrating the above mentioned improvements.
 

A revision of the ICD has been conducted by the WHO in recent years and has now been approved by the World Health Assembly. ICD-11 introduces the concept of ‘extension codes’ which allow for additional information to be appended to a cause of death code. Specifically, one set of extension codes titled ‘Proximal risk factors for intentional self-harm’ closely aligns with the work set out in this pilot study. Another set of extension codes ‘Context of assault’ has also been included in ICD-11, and provides categories for contextual factors to death such as bullying and intimidation. The ABS will work with the WHO and other international partners to ensure these code sets align with requirements for this type of study in the future.

It is widely acknowledged that mortality is affected by a complex interaction of factors including biological (such as disease), psychological (such as mental illness, drug addiction) and psychosocial factors (such as family and education). While there is uncertainty about the extent of causation between psychosocial factors and mortality, it is generally accepted that having this information allows for a range of public health responses to disease and injury (Singh-Manoux, Macleod & Smith, 2003). In addition, there are some preventable causes of death, including suicide, where psychosocial factors are known to be associated with higher mortality.

The ABS currently produces a national mortality dataset which provides in-depth information on biological factors contributing to death and also psychological factors in the form of mental and behavioural disorders (including those related to drug and alcohol misuse). For external causes, injury, mechanism and intent of death are tabulated.

The principles behind ICD-10 coding of mortality data do not cover the capture of information on psychosocial factors, yet this information is clearly required by data users. This effectively creates a key data gap for mortality. This is largely due to the construct of death certification where only diagnosable conditions, diseases and external events causing injury are required to be listed on the MCCD. However, as coroner referred deaths are coded in a way that provides a wealth of information via the NCIS, the ABS was presented with a unique opportunity to pilot the coding of psychosocial factors to create an enriched, nationally consistent mortality dataset. Specific aims of the pilot study were to:

• develop methods for capturing more of the extensive information available on the NCIS to expand the utility of the national mortality dataset;
• develop and test a framework for capturing this information and to inform development of embedded constructs in the revision of the ICD;
• demonstrate the utility of this additional information to policy makers and those working in various areas of public health; and
• embed psychosocial factor coded outputs into the national mortality dataset on a permanent basis.

For the purposes of this pilot study a psychosocial factor was defined to be social processes and social structures which can have an interaction with individual thought or behaviour and health outcomes. Although many psychosocial factors play a protective role in an individual’s life, this pilot study captured only those factors which were deemed to have a negative effect on the death. This is partly due to investigative process relating to death, where a police officer, coroner or pathologist are more likely to report factors thought to contribute to risk of death than those that might reduce it.

Examples of psychosocial factors include relationship status, employment status, bereavement, contact with the legal system and educational outcomes. A complete list of psychosocial factors captured are included in the annex of this publication. It is important to utilise the annex whilst interpreting results of the pilot.

The pilot study was conducted on all coroner-referred deaths registered in the 2017 reference period. In total, 19,130 deaths were in scope for analysis. Table 1 provides an overview of certifier type and mechanism of deaths registered in 2017.

Psychosocial risk factors were coded to entries in ‘Chapter 21 – Factors influencing health status and contact with health system’ of the ICD-10. This chapter is not currently used for mortality coding by the ABS, with its purpose directed at morbidity coding to capture circumstances where people may encounter the health services with diagnoses or problems not necessarily classified as an illness. Chapter 21 contains many entries where code descriptions align closely with the risk factors examined in the study, particularly in block ‘Z55-Z65 Persons with potential health hazards relating to socioeconomic and psychosocial circumstances’. A broad overview of categories in chapter 21 is provided below:

Z00-Z13: Persons encountering health services for examination and investigation
Z20-Z29: Persons with potential health hazards related to communicable diseases
Z30-Z39: Persons encountering health services in circumstances related to reproduction
Z40-Z54: Persons encountering health services for specific procedures and health care
Z55-Z65: Persons with potential health hazards related to socioeconomic and psychosocial circumstances
Z70-Z76: Persons encountering health services in other circumstances
Z80-Z99: Persons with potential health hazards related to family and personal history and certain conditions influencing health status

There are some limitations to using Chapter 21 as the framework. There are psychosocial risk factors which do not align with the provided codes or are part of a non-specific code which cannot be easily analysed once tabulated. One example of this is bullying. This is known to be an important risk factor and one of high public policy interest, especially in relation to suicide deaths, but there is no separate code for bullying in the ICD-10. Where important risk factors could not be captured with existing codes, decisions were made to alter code inclusions to better enable capture. All cases of bullying were coded to Z608 Other problems related to social environment. When Z608 is seen in the dataset it can be considered as a death where bullying was mentioned as a factor within the investigation. Code descriptions have been updated accordingly within tables (see annex for further information).

Although there are limitations to the inclusions in Chapter 21 Factors influencing health status and contact with health system, there are also a number of benefits to utilising this framework:

• ABS metadata remains consistent with ICD-10 principles for all causes of death;
• Frameworks for psychosocial factors align broadly with pertinent risk factors identified in literature; and
• Codes can be mapped to ICD-11 when it is implemented in Australia and this will enable better analysis of time series data.

The range of psychosocial risk factors included and examined in the ABS pilot study was identified through an iterative process. This process included:

• examination of existing literature relating to psychosocial risk factor studies undertaken previously and the types of risk factors that might relate to deaths from external causes;
• examination of policy, including highlighted areas of data gaps, especially those pertaining to suicide;
• mapping of potential risk factors to entries in the Chapter 21 Factors influencing health status and contact with health system. When a suitable category was not available an assessment was made on how to include the psychosocial factor. Some categories are quite broad with inclusions terms (e.g. problems related to legal circumstances, see annex) whereas other categories were altered to suit data needs (e.g. bullying, as discussed above);
• utilising the combined knowledge of the ABS mortality coding team and their understanding of the content of key reports on the NCIS including police and pathology reports as well as coronial findings;
• conducting initial trials in coding risk factors in accordance with categories identified;
• reviewing the suitability of risk factor categories/contents in accordance with actual documentation on the NCIS; and
• refinement and finalisation of categories and content for the completion of the pilot study.

ABS mortality coders accessed police, toxicology, pathology and coronial finding reports via the NCIS. All documents were read and factors which were considered pertinent to an individual’s death were assigned a relevant psychosocial factor code.

Coronial findings and police reports were the most commonly used reports for this pilot study. The investigative nature of these reports increase the likelihood of relevant contextual information being included, and it is from these summaries that most information was extracted. Autopsy reports were used less commonly, although in some jurisdictions summaries of lifestyle circumstances are provided by the pathologist. This information was valuable to the pilot. Toxicology reports, though providing a wealth of information on drug type and usage, do not identify contextual lifestyle factors and hence were not used as a key component of this study.

Each State and Territory has its own legislation and processes relating to coroner-certified deaths meaning that the type of information collected and stored in the NCIS database differs slightly by jurisdiction. As such, the most common report type used for extraction of psychosocial factors differed by State and Territory.

There were 2,474 coroner-referred deaths where a psychosocial factor was identified, accounting for approximately 13% of deaths in scope for the pilot study. On average, when a risk factor was identified there were 1.6 per death, highlighting the likelihood of co-occurring risks. The most common identified risk factor was a personal history of self-harm, followed by the disruption of a family unit due to separation or divorce. The top 20 psychosocial factors identified in the pilot study are listed below.

Suicide was the most common intent of death to have a psychosocial factor listed in investigative reports at death, with 1,966 intentional self-harm deaths in 2017 recording at least one factor. This was followed by accidents, with 344 deaths recording a psychosocial factor. The most common type of accidental death to have a psychosocial factor mentioned were drug overdoses (X40-X44).

Across all mechanisms and intents of death, males had a higher number of deaths with a psychosocial factor recorded. The table below outlines the number of deaths by intent and sex for coroner-referred deaths in 2017 with a contributing psychosocial factor.

Due to the large majority of psychosocial factors being recorded against suicide deaths and drug overdose deaths, the remainder of this section will focus on these two causes.

Suicide is the leading cause of death for people under 45 years of age in Australia. Statistical outputs on suicide most commonly focus on the rate of death per 100,000 population. Although this provides insights into changes over time and mortality profiles of the current population, it does not provide insight into intervention and prevention needs. Multiple inter-related risk factors are a known contributor to suicide, with many of these factors known to be modifiable with appropriate intervention and prevention activities (Clapperton, Newstead, Bugeja & Pirkis, 2019; Mościcki 1997). The introduction of psychosocial risk factors to the national suicide dataset will serve to expand the breadth of information and potential response mechanisms surrounding these deaths.

There were a total of 1,966 suicide deaths where one or more psychosocial risk factors were identified (see Table 4 below), accounting for 62.9% of all suicide deaths. The proportion of deaths where a risk factor was identified was similar for males and females (62.4% and 64.2% respectively).

Although the coverage of psychosocial risk factors was consistent for deaths due to suicide for all ages, results show a difference in the type of risk factors across the lifespan. While a history of self-harm is the most common identified risk factor for those under 65 years of age, it becomes less prominent in older age groups. Difficulties with life management due to chronic disease was the most common risk factor in those aged over 65. Some factors were more prevalent among particular age groups. One example of this is bullying which was more commonly identified as a risk factor in suicides for those aged under 25 than any other cohort.

Psychosocial factors were identified in 61.2% of suicide deaths of Aboriginal and Torres Strait Islander people. This is similar to the proportion of psychosocial factors identified in suicides of the non-Indigenous population (60.3%). The table below shows that while the top 5 most common psychosocial factors were the same for both Aboriginal and Torres Strait Islander and non-Indigenous people, the ranking of these factors differed. The most common psychosocial factor identified in suicide deaths for Aboriginal and Torres Strait Islander people was problems in relationship with spouse or partner. This factor was the third most common psychosocial factor identified in non-Indigenous suicide deaths.

In this section of the paper, the term 'associated cause' is used to cover the factors relating to a death, whether they be traditional mortality associated causes such as mental health conditions, drug or alcohol abuse, chronic or terminal diseases or the psychosocial risk factors identified in this study. All of these associated causes are noted in police, coroner or pathology reports as being important factors relating to the death. However, it is important to note that these factors generally do not occur in isolation and therefore should not be considered in isolation. They instead should be considered as part of a complex interaction between biological, psychological and psychosocial factors (Lee & Jung, 2006).

There is benefit to analysing psychosocial risk factors in combination with other known risks for health outcomes. For suicide deaths in particular, separating mental health conditions from chronic and other natural diseases provides important insights into circumstances for an individual. The following sections will provide analysis on how psychosocial risk factors interplay with mental health conditions and natural diseases. For tabulation purposes the factors are presented in broad groups: psychosocial factors encapsulate all codes in scope for this study (see annex for a full list of codes), mental health conditions include those coded to F00-F99 Mental and Behavioural Disorders of the ICD-10 and chronic and other natural diseases includes chronic health conditions (e.g. cancer), acute conditions which may have been exacerbated by injury and ill-defined conditions and symptoms which do not fit definitively into other categories. More detailed analysis of specified mental health conditions and other natural diseases are contained within the supplementary data cubes.

Psychosocial factors were associated with a similar proportion of suicide deaths to that of mental health conditions, and were more likely to be mentioned in a police, pathology or coronial finding than a contributing natural disease. The table below shows that, when all types of associated factors are considered, there are close to 90% of suicides where a relevant contributing factor to death is known.

Psychosocial risk factors have a consistent presence across all ages as they relate to individuals who die due to suicide. The interaction between psychosocial risk factors with biological and psychological factors is clearly demonstrated by the table below with mental health being a far more prominent factor in younger age cohorts and natural disease more prevalent in older age groups (see Table 8).

Coverage of associated causes of death and in particular psychosocial factors, was not consistent across Australian jurisdictions for deaths due to suicide. The Australian Capital Territory (ACT) had the highest proportion of psychosocial factors identified for suicide deaths, with close to 80% coverage. The ACT also had the highest number of associated causes in general with nearly 100% of suicides having a known factor contributing to death identified. South Australia had the lowest coverage of associated causes identified with just over three quarters (76.7%) of suicide deaths having a factor identified. Coverage of psychosocial causes for South Australia was particularly low compared with other states and territories, with only 36.5% of suicides having a factor identified in the reports accessed via the NCIS.

There are particular combinations of codes which are best considered in combination to enhance understanding of the circumstances surrounding a persons’ decision to end their life. Three pertinent examples are included below. The annex also contains detail on how to interpret certain combinations of codes.

Suicide and severe psychiatric illness of a family member, friend or other key primary support is well documented as a risk factor for suicide (Qin, Agerbo & Mortensen 2002; Rudd & Roberts, 2019). Similarly, bereavement can be a risk factor for suicidal behaviour (Latham & Prigerson, 2011). Bereavement when the death of a family member or friend is due to suicide is known to increase suicide risk when compared to deaths due to natural causes or accidents (Rostila, Saarela & Kawachi, 2013).

Bereavement and history of suicide in the primary support group can therefore co-occur, or they can exist as separate risk factors for an individual. When investigation of outcomes from a suicide death indicated that an individual was feeling grief from bereavement and had experienced a recent suicide in the primary support group, ABS mortality coders assigned two codes to these deaths: Z634 Disappearance and death of family member (or primary support group) and Z818 Family history of suicide. The table below shows that while there were 258 suicide deaths where either bereavement or history of suicide in the primary support group were mentioned as risk factors, there were 18 suicides where bereavement was specifically due to the suicide of a person in the primary support group. When these two codes appear together in the dataset, this unique relationship between known risk factors should be considered.

Limitation of activities due to a chronic disease, multiple debilitating co-morbidities or injury is known to be a risk factor for increasing suicidal ideation (Lim, 2010). This risk factor was captured under code Z736 Limitation of activities due to disability or chronic health condition in the ABS pilot study. More context can be provided to the source of disability by considering co-morbidities appearing with this risk factor. The table below outlines chronic conditions most commonly appearing with this psychosocial factor among suicide deaths.

Data in the results have focussed on fine level outputs by specific life events. Although it is valuable to understand the depth of detail available in the dataset, it is also important that the outputs are able to be tabulated to broad level psychosocial events.

Social determinants is a widely utilised health concept which describes the way people are born, grow, live, work and age (CSDH, 2008). These determinants have strong links to health outcomes, including mortality. There are many models which represent determinants of health. The model applied in this example is based on the framework developed by Dahlgren and Whitehead (1991) which separates out interpersonal factors (e.g. actions taken by individuals on health behaviour), social and community network factors (e.g. support shown by family), living and working conditions (e..g housing) and economic, environmental and cultural factors. For this purposes of this paper living and working conditions have been combined with economic, environmental and cultural factors. The psychosocial factors applied for the pilot study has strong alignment with the social determinant model. The framework allows data to be grouped into categories relating to individual factors (e.g. history of self-harm), social and community factors (e.g. family discord) and socio-economic and environmental factors (e.g. financial issues).

The graph below provides an example of this, grouping psychosocial factors as they relate to suicide deaths. Although males experience a higher number of all risk factors and have a higher rate of suicide, proportionally female suicide deaths have a slightly higher mention of factors relating to interpersonal and social and community issues.

Psychosocial risk factors were much less likely to be identified for accidental deaths, with only 344 or 6.7% of coroner-certified accidental deaths having one or more risk factors identified. It was most commonly accidental drug induced deaths where risk factors were identified, accounting for 266 of the 344. There were a total of 1,189 accidental drug induced deaths in 2017, so risk factors were identified among 22.4% of those deaths.

Personal history of self-harm was identified as a risk factor in 62 accidental drug-induced deaths, being the most common risk factor overall. Recent release from prison was the second most identified psychosocial factor for accidental drug-induced deaths, and the most frequent for men. The table below outlines the ten most common psychosocial factors by sex.

In addition to psychosocial risk factors, associated causes of death including mental health conditions, diseases and ill-defined conditions are also important factors in accidental drug-induced deaths. The table below shows that, similar to suicide deaths, there is close to 90% of accidental drug overdoses where associated contextual information as it relates to the death is known. However, there is a clear differentiation between mental health conditions, natural disease and psychosocial factors. Mental health conditions, which include drug addiction and misuse are mentioned as a contributing factor in over 70% of cases. This highlights the importance of considering the interaction of risk factors.

The type of drug or drugs involved in deaths due to acute toxicity have implications for treatment and intervention, as well as prevention activities. The pattern of drug types can be considered for various sub-groups of cases, such as those with a code for a particular psychosocial factor.

The graph below shows, for the accidental drug overdose cases involving each of the top five identified psychosocial factors, the proportion recorded as involving the known illicit substance heroin, amphetamine type substances and cocaine compared to all remaining substances. The proportion was highest for the cases with mention of unemployment or recent release from prison.

This pilot study undertaken by the ABS mortality team aimed to develop and refine methods for capturing psychosocial factors and to assess the value of the outputs for informing public health initiatives.

The new method for capturing psychosocial factors proved feasible, though subject to some limitations. The psychosocial factors recorded align very closely with those discussed in the literature. For suicide deaths in particular the psychosocial factors align well with those collected and disseminated by suicide registers. In a recent report utilising the Victorian Suicide Register, exposure to stressors including mental illness, drug and alcohol abuse, divorce and separation and contact with police were highlighted as priorities for suicide prevention (Clapperton, Newstead, Bugeja & Pirkis, 2019).

The most commonly identified psychosocial factor was personal history of self-harm. Previous studies have found that a suicide attempt is one of the strongest risk factors for suicide (Department of Health and Ageing, 2007). A history of self-harm has also been associated with premature mortality of any kind, including accidental drug overdose (Carr et al, 2017)

The method improves information on two types of death in particular, suicide and drug overdose deaths. The relatively frequent presence of psychosocial information on these types of death is likely due to the intensive investigation processes put in place by jurisdictional authorities. Police forms generally incorporate a section to flag potential risk factors relating to suicides, and the coronial finding outlines potentially modifiable risk factors.

The breadth of data requirements relating to suicide deaths was investigated several years ago through a consultation process led by Suicide Prevention Australia. This process highlighted many psychosocial factors which would be of particular value to those working in suicide prevention, and while it was clear that some of that information was contained in NCIS reports, the complex and often narrative nature of those reports limited its accessibility to many potential users. Systematic coding and capture of information on these factors should enhance both utility and accessibility.

The framework developed for this study was applied consistently between sexes and across age groups, with particular psychosocial factors found to be prominent at particular stages of the life cycle. This aligns with existing literature which highlights different life events affecting mental and physical health over the life span, but also birth cohort effects (Blumenthal & Kupfer, 1990).

Coverage of psychosocial factors was consistent for suicide deaths of Aboriginal and Torres Strait Islander and non-Indigenous people. Although the top five most commonly identified psychosocial factors were the same for the Indigenous and non-Indigenous populations, the ranking of risk factors was different, with relationship issues being most common for Aboriginal and Torres Strait Islander people. Research shows that risk factors relating to social connectedness and grief and loss and how these interplay with culture require concerted focus for Aboriginal and Torres Strait Islander people (Dudgeon & Holland, 2017). The ABS will work in collaboration with Aboriginal and Torres Strait Islander stakeholders to assess the suitability of the dataset when taking culture into consideration.

The capture of information on psychosocial factors alongside existing mortality data constructs of underlying and associated causes of death is a clear strength of this study, providing insights into the interaction between psychosocial factors, mental health conditions and natural diseases. Results demonstrated that whilst the cause of death data already provide useful information on associated risk factors for death, the psychosocial factors add a new dimension, with much relevance to public policy activity. When combined with demographic indicators including sex and age, this framework provides a vehicle to better capture the complex interactions between factors relating to an individual’s death.

Limitations in regards to the data must be taken into account when utilising outputs for research purposes. Although all coroner-referred deaths registered in 2017 were in scope for the pilot, psychosocial factors were only recorded for 13% of these deaths. This is in part due to the different investigative approaches used by authorities for different types of deaths (i.e. natural cause deaths versus external cause deaths) and the type of information captured through those investigations. Relevant information could be captured for suicide and drug overdose deaths, but was not reported in police and autopsy reports for natural disease deaths. It is also important to note that coronial investigations are an iterative process undertaken by a multitude of agencies and that coding of risk factors was undertaken at a point in time.

The ICD-10 framework used in this study was designed to capture risk factors which influence health. Motor vehicle accidents are subject to many environmental factors including wet weather, speed and mobile phone use. Although these are important factors for these deaths, they did not fit the study definition of a psychosocial risk factor and were therefore not in scope for this pilot. The future implementation of ICD-11 will provide a more detailed framework for capture of psychosocial risk factors and may also enable capture of other types of risk factors.

Psychosocial risk factors were present in close to 63% of suicides and 22% of accidental drug overdoses. With the general nature of some of the psychosocial factor categories there is a possibility that a greater coverage should have been expected. The 2017 coroner-referred death dataset is preliminary and open cases will undergo two rounds of updates as cases close and more information becomes available for analysis. It is likely this will enable additional risk factors to be added to the pilot dataset over time, thereby increasing coverage. However, it is also important to note that existing information on associated causes of death (i.e. mental health conditions, drug and alcohol issues and other natural diseases) can also be relevant risk factors. When associated causes and psychosocial risk factors are considered together, at least one risk factor was identified for close to 90% of all suicide and drug overdose deaths.

Protective factors were not included in the pilot study. Although the role they play in an individual’s life is extremely pertinent, it did not appear that information regarding protective factors was collected systemically as part of the coronial investigation.

Some of the categories used to capture psychosocial risk factors are broad and consist of many inclusions terms. An example of this is Z653 Problems related to legal circumstances. This category was used to code deaths where upcoming court cases, domestic violence orders and general illegal activities were documented. There were also psychosocial factors, for example bullying, where no suitable category was available. Given the breadth of factors captured in individual categories and the need to amend some categories to enable capture of particular factors, it is important that the attached annex is used for interpretation of data. The lessons learned from undertaking this study in terms of the suitability of particular codes will provide useful insights for future use of ICD-11, and the ABS will work with WHO to maximise it's suitability for this type of data capture.

There are clear differences in quality and number of psychosocial factors listed in NCIS reports across jurisdictions. There is a possibility that jurisdictions with lower numbers of risk factors mentioned have a higher number of open cases and numbers will change as cases are closed and data is revised. The ABS will also engage with key stakeholders, including the NCIS in this space to discuss potential areas of improvement.

The overall aim of this study was to create a method and framework for coding of psychosocial factors for coroner-referred deaths on the national mortality dataset. Results showed consistent coverage across demographic cohorts, with factors most commonly identified across suicide and drug-induced deaths. Outputs align with existing literature. The ABS will continue to work with stakeholders to address limitations, and will use information gathered through the study and through feedback to provide feedback to WHO on ICD-11 constructs.

The resources required to undertake this work are significant and the work is therefore not sustainable without funding. The ABS will work with interested parties in research, policy and planning to identify ways of making this work sustainable into the future.

This dataset offers a nationally consistent approach to capturing information on psychosocial risk factors which is currently unavailable in the Australian setting. Initial outputs indicate that it will be a valuable resource for policy makers and importantly, align with data needs in current suicide prevention strategies.

The principle authors of this report were Lauren Moran, James Eynstone-Hinkins, Fiona Khoo and Katrina Sheehan. They would like to gratefully acknowledge the contributions and support received from Professor James Harrison of the National Injury and Surveillance Unit and team members from the Mental Health and Suicide Prevention Branch at the Department of Health. Significant contributions were also made by the ABS mortality coding team.

For further information about these and related statistics, contact the National Information and Referral Service on 1300 135 070.

This release previously used catalogue number 1351.0.55.062