groupEmma Torrens, Lauren Moran, James Enyestone-Hinkins

menu_bookVolume 11 Number 1

collections_bookmarkReports

lockMember Access

printPrint Article


The COVID-19 pandemic has driven increased demands for timely, high quality statistics that can help governments to respond to the social, health and economic impacts of the disease. Data has also been critically important for informing the public during this time of high uncertainty.

During the early stages of the pandemic, Australia relied on health surveillance systems to provide critical information on numbers of COVID-19 infections and deaths, with these systems continuing to play an important role in managing subsequent outbreaks. As the pandemic progressed, the need to understand the broader impacts on mortality, both direct and indirect, also became apparent. These impacts could not be measured through disease surveillance systems but could be captured through civil registration based mortality data. The Australian Bureau of Statistics (ABS) recognised both the challenge and opportunity in seeking to deliver rapid registration based mortality data, and after consultation with the Australian Health Protection Principal Committee, agreed to find a way to deliver the required data.

Several innovations have since changed the traditional methods for processing and disseminating mortality data in Australia, and enabled the production of new monthly Provisional Mortality Statistics reports. These new reports provide timely insights into changes in patterns of mortality in Australia, with further analytical articles focussing both on COVID-19 deaths and measures of excess mortality.

Existing processes for producing the National Mortality Dataset

As Australia’s national statistical agency, the ABS has undertaken the role of producing the National Mortality Dataset (NMD) for many years. The NMD is an epidemiological dataset containing all deaths registered in Australia for a given reference period. It is used as a foundation for population estimates, health policy, health service planning and evaluation.

The civil registration system is the source of official mortality statistics produced by the ABS. When an individual dies, a medical certificate of cause of death is completed by either a doctor or a coroner and lodged, along with a death registration statement, with the relevant jurisdictional Registry of Births, Deaths and Marriages (RBDM). When both of these forms are received, the death registration can be completed by the RBDM and subsequently notified to the ABS.

Cause of death data is compiled and published on an annual basis, with annual reports released approximately 9 months after the end of each reference period. For example, the 2018 cause of death report was released in September 2019. Annual reports contain statistical information on all causes of death occurring in Australia, as well as specialised articles and information on key indicators of mortality such as leading causes of death and preventable causes such as suicide and drug overdoses.

Building a framework for rapid reporting

Shortly after the COVID-19 pandemic began, countries around the world recognised the importance of monitoring changes in patterns of mortality beyond those directly attributable to COVID-19. This additional information would help identify any potential misclassified or undiagnosed COVID-19 deaths, and would also enable monitoring of changes in mortality that might be indirectly related to the pandemic (i.e. relating to issues of social isolation or changes in access to health care).

Civil Registration and Vital Statistics (CRVS) systems are able to deliver the breadth of data required to monitor changes in patterns of mortality, with the timeliness and completeness of registration, legislative frameworks for registration and established data flows between agencies as the key factors in determining how and when those insights could be produced.

In Australia, a death registration cannot be completed until after a burial has occurred. Depending on the timing of the burial it can range between approximately two weeks and a number of months before the RBDM is able to complete a death registration and lodge the information with the ABS. This is especially the case with coroner referred deaths where coronial investigations may necessitate delays to burial and other administrative processes relating to death. Managing this time lag between death, registration and reporting was especially important in shaping the Australian method for rapid reporting.

Not all CRVS systems are the same. For instance, in the United Kingdom (UK) there is a legislated requirement for registration to occur prior to burial or cremation (Births, deaths, marriages and care 2021). This results in registration occurring within days of a death and supports very rapid reporting of registration based mortality data. Weekly reporting of death registrations already occurred in the UK prior to the COVID-19 pandemic and rapid reporting of cause specific mortality was able to be established by the UK’s Office for National Statistics soon after the pandemic began (Office for National Statistics 2021).

Rapid reporting in Australia needed to be able to account for the variable delays between death and registration so that the outputs were sufficiently robust to measure change over time. Investigations showed that historically, around 95% of deaths that had been certified by a doctor were reported to the ABS two months after the date of occurrence of the death. This provided a reasonable level of completeness to enable initial comparison with historic counts.

Coroner referred deaths take longer on average to be registered and reported (approximately three to four months after the occurrence of the death) due to the time required to investigate those deaths. Even after registration, information on causes of those deaths may not be available for many months, making rapid reporting on Coroner referred deaths especially challenging. For those reasons a decision was made to exclude them from rapid reporting.

Considering the strengths and limitations of the Australian CRVS system, it was possible to establish a method for releasing monthly reports on doctor certified deaths which could provide a robust comparison of numbers of deaths in 2020 by specific causes against those from the past five years. Details of those initial reports are provided in the section below.

Around the world, many countries focussed on measuring excess deaths, typically defined as the difference between the observed number of deaths in a specified time period and the expected numbers of deaths in that same time period. Measures of excess mortality can account for deaths due to COVID-19 and as such can provide a more complete measure of the overall impact of the pandemic. While initial ABS reports could highlight abnormalities in counts of deaths by cause that might indicate possible excess mortality, they were not population weighted and did not provide indications of whether changes exceeded the natural variability in deaths on a weekly or monthly basis.

Strong interest in measuring excess deaths in Australia led to further analysis of provisional outputs using more robust statistical techniques that measure the statistical significance of changes in mortality over time. Data analysed and reported using these methods provide additional guidance to those using the data to interpret changes in patterns of mortality over time and respond accordingly. 

Provisional Mortality Statistics reports

In June 2020 the ABS (2020) released its first Provisional Mortality Statistics report. Reports have subsequently been released every 4-5 weeks during 2020 and will continue to be published in 2021. These reports are focussed on measuring change over time, taking account of the requirements and constraints described previously. As such, the scope and content of the reports differs substantially from traditional annual Cause of Death reports. Differences are outlined below.

Data is published for doctor certified deaths only

To enable more timely reporting, provisional data is limited to doctor certified deaths only. Over 85% of deaths in Australia are certified by a doctor. Data received each month from the RBDMs for doctor certified deaths include complete cause of death information, so detailed disease specific information can be compiled and released in each Provisional Mortality Statistics report.

Data is published by date of occurrence of death

Data is compiled based on the date the death occurred and reported on a weekly basis. Reporting by date of occurrence minimises the impact of administrative factors (i.e. delays between death and registration) that could otherwise hinder comparisons over time. When each report is published, deaths are included for the most recent month, and any deaths that occurred in previously reported periods are added to counts for those weeks. Annual Cause of Death reports are compiled according to the year in which the death was registered, so data in the provisional and annual reports are not directly comparable.

Data is provisional

Statistics provided in the monthly cause of death release are considered provisional and may change between the monthly and annual publication as further quality assurance processes are applied.

2020 data is measured against a baseline

Data for 2020 in the provisional release is compared to an average baseline over the previous five years of deaths based on year and week of occurrence. As mortality rates generally remain steady over time, this average acts as a proxy for the expected number of deaths in 2020.

Provisional data can be mapped against health surveillance data

As the provisional monthly reports are compiled on the date on which the deaths occurred, they can be directly mapped against other data compiled by occurrence.  Tracking deaths against surveillance counts of COVID-19 infections has proven helpful in relating any changes in mortality to changes in infection rates over time. This is demonstrated in Figure 1.

Figure 1: Doctor certified deaths, COVID-19 infections, Australia, 1 January – 27 October 2020 vs 2015-19 benchmarks

Source: Australian Bureau of Statistics 2020

COVID-19 mortality

CRVS data has been able to provide additional insights into COVID-19 mortality, complementing the rapid reporting provided by health surveillance systems. Data from Medical Certificates of Cause of Death (MCCDs) provide comprehensive information on the symptoms of COVID-19 and how it leads to death as well as pre-existing chronic conditions a person may have been suffering. Examining COVID-19 deaths in the context of disease outcomes and comorbid conditions assists in planning for health service requirements to treat COVID-19 patients, as well as understanding who is at higher risk of death from COVID-19. This information from the MCCD provides a more comprehensive picture of COVID-19 mortality in Australia.

Future opportunities

The Provisional Mortality Statistics reports have been well received and have provided important insights into mortality during the COVID-19 pandemic period. Australia experienced statistically lower than expected rates of mortality during 2020, with the provisional data providing the foundations for identifying and understanding this change. However further opportunities exist to strengthen the efficacy of these data in Australia.

Due to the exclusion of coroner data, the report is unable to answer important questions related to mortality of young people during 2020, mortality due to external causes of death such as accidents and suicides and whether there were any changes in coronial certification rates. Finding ways to deliver those insights would certainly enhance the utility of these data.

The ABS is also working to understand data supply pathways as they relate to the registration process to address the timeliness of data availability. While the Provisional Mortality Statistics reports are much timelier than the annual publication, information can still only be released two or three months after deaths occur. This constitutes a barrier to rapid identification of issues and could delay any subsequent response. Opportunities to overcome existing limitations of the system are being explored, with key changes such as the implementation of electronic death certificates by RBDMs likely to provide more timely source data.

Conclusion

The COVID-19 pandemic has been a catalyst for changing the way mortality data is output and made available throughout the world. The benefits of making CRVS data available more quickly are widely recognised and the utility of these data extend beyond informing changes during the COVID-19 pandemic period. For example, CRVS data could provide more complete mortality data to complement disease surveillance data during the influenza season and could provide insights during natural disasters such as the bushfires. Provisional Monthly Statistics have already proven helpful as governments have responded during the pandemic, and further enhancements to timeliness and the breadth of data provided would expand the utility of these data in guiding future health interventions.


References


Authors

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

Lauren Moran, BBehSc
Assistant Director
Mortality Data Centre Health and Vital Statistics Section Australian Bureau of Statistics     

James Eynstone-Hinkins, BSc
Director
Mortality Data Centre
Health and Vital Statistics Section Australian Bureau of Statistics