Measuring Australia's excess mortality during the COVID-19 pandemic

Provisional deaths data for measuring changes in patterns of mortality during the COVID-19 pandemic and recovery period.

Cite
Print
Released
30/03/2022
Source
Provisional Mortality Statistics, Jan 2020 - Dec 2021

Key Statistics

  • In 2021 Australia had higher than expected mortality, but only a small number of weeks recorded deaths that reached statistical significance. 
  • Victoria had the highest number of statistically significant excess deaths. 
  • Deaths due to respiratory diseases remained lower than expected in 2021. 

Introduction

COVID-19 has continued to impact the lives of people in Australia. Analysing all-cause mortality can provide a more complete assessment of the overall impact of the pandemic (rather than a focus on COVID-19 mortality alone). This measure is referred to as excess mortality, 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. Excess mortality measures can account for deaths due to COVID-19, potentially misclassified or undiagnosed COVID-19 deaths, and other mortality that may be indirectly related to the pandemic (e.g. relating to social isolation or changed access to health care).

The ABS has previously published excess mortality estimates on the first two waves of COVID-19 in Australia with two articles:

This article provides an update on excess mortality in Australia until the end of 2021. This covers the time period when infections and deaths from the Delta variant were most prevalent. While the Omicron variant was first detected in Australia in November 2021, deaths from the virus itself have been highest in early 2022. A further article on excess mortality will be published in May 2022 providing updated estimates which will cover the Omicron wave. 

Excess mortality is one of many possible methods to measure the impact of the COVID-19 pandemic. Age-standardised death rates also provide an indication of mortality through the pandemic period. Refer to Provisional Mortality Statistics reports for monthly provisional counts of deaths and age-standardised death rates compared against 5-year historic averages, and the Causes of Death articles for more detailed updates focusing on COVID-19 deaths.

COVID-19 pandemic in Australia

Australia has had multiple waves of COVID-19 infections since the start of the pandemic in March 2020.

The first wave was recorded from mid-March to mid-April, with most states in Australia recording active infections and associated mortality.

The second wave started in June 2020, with the largest number of active infections and deaths occurring in Victoria. ‘Stay at Home’ orders were implemented in Victoria from July 2020 until the end of October 2020.

Between the second and third wave, many jurisdictions managed smaller outbreaks with localised lockdowns.

A third wave began in June 2021 with the spread of the Delta variant. Most jurisdictions recorded COVID-19 infections during this time period. New South Wales and Victoria recorded the highest number of infections and mortality due to COVID-19 during the Delta variant wave and both states implemented a number of public health measures to manage this, including vaccinations, social distancing and lockdowns. 

The Omicron variant was first identified in Australia in November 2021 and began to spread throughout the country at the end of 2021 leading to a fourth wave which has continued into the first quarter of 2022. 

Key events during the COVID-19 pandemic in Australia are noted in graphs where relevant.

Measuring excess mortality

Across the world, health and statistical authorities have sought to measure excess mortality during the COVID-19 pandemic. Different methodologies can be applied, with the goal to predict an expected number of deaths for a given year. Estimating the expected future seasonality for deaths can be a a challenge in many models. The suitability of a model can be dependent on factors such as country context, data quality and collection methods, or outcomes sought from analysis.

The ABS has adopted aspects of a methodology used by New South Wales (NSW) Health, applying a cyclical linear regression with a robust estimation procedure to produce both an expected number of deaths and a range of expected deaths for 2021. The model has been applied to all-cause and cause-specific mortality to identify significant changes in patterns of mortality over time. See the methodology section at the end of this article for detailed information about the model applied.

Interpreting results

Outputs from excess mortality estimates will differ depending on the calculation applied and the scope of the input data. When interpreting the results in this report the following factors must be taken into consideration: 

  • Only deaths certified by a doctor are included in the analysis. Approximately 87-89% of deaths annually are certified by a doctor, with some variability across jurisdictions. 
  • The expected number of deaths in 2021 has been projected using 2016 as the beginning of the baseline period. 
  • This report places emphasis on observed numbers of excess deaths that are statistically significant. In any given time period, even if no temporary health hazards influence the number of deaths (such as community transmission of influenza or COVID-19) there is some natural variation in patterns of mortality. While the actual number of deaths may be different from the expected number of deaths, it should fall within an expected range (i.e. there is a 95% chance that the expected number of deaths would lie between the upper and lower bounds of the confidence intervals). When actual observations (counts of death) exceed the upper threshold or drop below the lower threshold this indicates a statistically significant change in the pattern of mortality. 
  • A single week above threshold does not necessarily suggest statistically significant excess mortality. Prolonged periods (2 or more weeks) where counts exceed thresholds suggest more strongly that the numbers of deaths are above or below normal. 

Each section below presents two graphs –

The first graph is a time series of doctor certified deaths from January 2016 to December 2021 that has the number of actual observations plotted against the expected number of deaths estimated from the regression. The upper and lower thresholds (1.96 standard errors) of the regression are also plotted.

The second graph focuses on 2020 and 2021, allowing closer inspection of patterns of death during the COVID-19 pandemic. 

Annual all-cause mortality: Australia

  • There were over 5,000 deaths more than expected in Australia during 2021. This is the result of sustained periods during the year which had higher than expected deaths. Only a small number of time periods reached statistical significance.
  • There were 1,734 fewer deaths than expected in 2020 with sustained weeks across the winter months reaching statistical significance. 
  • There were 3,630 more deaths than expected in 2017 driven by a severe flu season, but higher numbers of these excess deaths were statistically significant when compared with 2021. 
Annual excess mortality estimates for doctor certified deaths, Australia (a)(b)(c)(d)
Expected no. of deathsObserved No. of deathsDifferenceExcess (%)Statistically significant number above the limit of usual variation
2016138,742138,725-170Expected range
2017140,108143,7383,6302.61,501
2018141,329139,326-2,003-1.4-19
2019142,395143,6721,2770.9Expected range
2020145,855144,121-1,734-1.2-209
2021144,108149,1985,0903.5108
  1. Data is provisional and will change as additional registrations are received. 
  2. 2016 is the starting point for the regression to model the expected number of deaths.
  3. Data includes all doctor certified deaths occurring until 31 December and registered by 28 February 2022 and received by the ABS. 
  4. See methodology section for more information. 

Weekly all-cause mortality: Australia

  • There were about 5,000 more deaths than expected in Australia in 2021. For most weeks in 2021 the number of deaths were higher than the projection but within the expected range. 
  • There were five weeks in 2021 where the actual number of deaths exceeded the upper threshold reaching statistical significance. These are the weeks beginning 18 January 2021, 24 May 2021, 7 June 2021, 26 July 2021 and 20 December 2021. Collectively, a total of 108 deaths exceeded the upper threshold (i.e. statistically significant excess mortality) during these five weeks. While these are single points in time, they all fall within extended periods of higher than expected mortality.
  • Deaths due to COVID-19 make an impact on all-cause mortality from the week beginning 2 August 2021 (see graph 2 below). 
  • There were 1,734 fewer deaths than expected in 2020 with periods during the winter months being statistically significant. The lower than expected number of deaths provide insight into how COVID-19 public health measures may have impacted on mortality. For example measures may have reduced infectious diseases transmitting from person to person and may have contributed to a decrease in deaths due to respiratory diseases (excluding COVID-19). 
  • Excess mortality at the all-cause level was statistically significant in the week starting 30 March 2020 (48 deaths above the upper threshold). This time period coincided with higher rates of COVID-19 infections and the implementation of stricter lockdown measures in Australia during the first wave. 
  • A severe influenza season in 2017 contributed to 3,630 more deaths than expected in 2017. Between mid-July to mid-September the number of deaths reached statistical significance with over 1,500 deaths exceeding the upper threshold. 
  1. Dates for key events are indicative only and may differ to other sources. 
  2. Data is provisional and will change as additional death registrations are received.
  3. 2016 is the starting point for the regression when modelling the projected number of deaths. 
  4. Data includes all doctor certified deaths occurring by the end of December 2021 and registered and received by the ABS until 28 February 2022. 
  5. See the methodology section at the end of this article for more information.
  1. Dates for key events are indicative only and may differ to other sources. 
  2. Data is provisional and will change as additional death registrations are received.
  3. 2016 is the starting point for the regression when modelling the projected number of deaths. 
  4. Data includes all doctor certified deaths occurring by the end of December 2021 and registered and received by the ABS until 28 February 2022. 
  5. See the methodology section at the end of this article for more information.

Annual all-cause mortality: State and territory

Public health measures put in place to manage the spread of COVID-19 were implemented in Australia at national, state and local levels. States and territories were impacted differently by the pandemic, with those differences including infection rates, COVID-19 mortality and subsequent public health measures. 

  • All jurisdictions except the Northern Territory recorded a higher than expected number of deaths in 2021. 
  • Victoria and South Australia both had periods of sustained statistically significant excess mortality (2 or more weeks) during 2021. 
  • In 2021, although New South Wales, Queensland, Western Australia, Tasmania and Australian Capital Territory recorded a higher than expected number of deaths, they were generally within the expected range.
  • Victoria has recorded the highest number of deaths above expectation during the COVID-19 pandemic, including the highest number of statistically significant excess deaths. 
  • New South Wales, Queensland, Western Australia and Tasmania recorded lower than expected deaths in 2020, with some periods reaching statistical significance in these states (except Western Australia). 
  • A severe influenza season in 2017 contributed to some periods of statistically significant excess mortality in New South Wales, Victoria, Queensland, South Australia and Tasmania. 

Weekly analysis of excess mortality in Victoria, New South Wales and Queensland has been included in graphs below. Weekly data for all other states and territories is included in the data downloads. Data for small jurisdictions should be interpreted with caution.

All-cause doctor certified mortality by state and territory (a)(b)(c)(d)(e)
 New South WalesVictoriaQueenslandSouth AustraliaWestern AustraliaTasmaniaNorthern Territory (e)Australian Capital Territory (e)
Expected        
  201647,57433,58427,42210,92512,4614,1098012,012
  201747,60433,99928,32811,18912,4293,9968042,042
  201847,89434,19229,08511,29412,3963,9008082,048
  201948,43634,15229,69511,23812,3643,8218222,043
  202050,09634,48530,69511,21512,5583,8298732,053
  202150,31233,37730,46010,64012,3103,7208751,970
Observed        
  201647,67033,57127,12610,92012,5824,0368042,013
  201749,48034,61528,93911,41212,1804,2487912,073
  201847,06333,59228,62911,05112,4013,6848552,053
  201948,96334,22929,89611,26212,5853,9487961,994
  202048,80434,86729,90911,31712,4953,7618912,077
  202150,32536,15931,31311,53113,1144,0867792,072
Difference        
  201696*-13-296-5121-7331
  20171,876*616*611*223-249*252*-1331
  2018-831-600-456-2435-216475
  20195277720124221*127-26-49
  2020-1,292*382*-786*102-63-681824
  2021132,782*853891*804366-96102

*notes years that have periods of statistically significant excess mortality or lower than expected mortality (2 weeks or more above or below the thresholds). Other years fall within the expected range.

  1. Data is provisional and will change as additional death registrations are received.
  2. 2016 is the starting point for the regression when modelling the projected number of deaths. 
  3. Data includes all doctor certified deaths occurring by the end of December 2021 and registered and received by the ABS until 28 February 2022. 
  4. See the methodology section at the end of this article for more information.
  5. Data for small jurisdictions should be interpreted with caution. Small numbers of weekly deaths can cause volatility with the modelled number of expected deaths. 

Weekly all-cause mortality: New South Wales

  • In 2021 there were 13 more deaths than expected in New South Wales. This annual pattern falls within the expected range of doctor certified mortality and the difference is not considered statistically significant. 
  • In the weeks beginning 7 June 2021 and 19 July 2021 the number of excess deaths reached statistical significance exceeding the upper threshold by 20 deaths collectively. These are both single data points in time and the significance of these events should be interpreted with caution. 
  • Between the week beginning 4 October 2021 and the end of November 2021 mortality was generally lower than expected but still within the expected range. 
  • COVID-19 mortality is notable in all-cause mortality figures from the week beginning 2 August 2021 (see graph 2 below). 
  • In 2020 there were 1,292 less deaths than expected occurring in New South Wales. Between the week beginning 11 May 2020 and the end of November 2020 mortality was lower than expected. There were four weeks where the reported number of deaths reached statistical significance (128 less than the lower threshold collectively). 
  • In 2017 there were 1,876 more deaths than expected with a virulent influenza season occurring. Across the winter months there was sustained statistically significant excess mortality with 967 deaths exceeding the upper threshold during this period. 
  1. Dates for key events are indicative only and may differ to other sources. 
  2. Data is provisional and will change as additional death registrations are received.
  3. 2016 is the starting point for the regression when modelling the projected number of deaths. 
  4. Data includes all doctor certified deaths occurring by the end of December 2021 and registered and received by the ABS until 28 February 2022. 
  5. See the methodology section at the end of this article for more information.
  1. Dates for key events are indicative only and may differ to other sources. 
  2. Data is provisional and will change as additional death registrations are received.
  3. 2016 is the starting point for the regression when modelling the projected number of deaths. 
  4. Data includes all doctor certified deaths occurring by the end of December 2021 and registered and received by the ABS until 28 February 2022. 
  5. See the methodology section at the end of this article for more information.

Weekly all-cause mortality: Victoria

  • In 2021 there were 2,782 deaths in Victoria above the expected number. For every week except two the number of deaths was above expectation.
  • There were two periods of statistically significant excess mortality in 2021. The first began in the week beginning 12 April and remained until early May (61 observations above the upper threshold). The second began at the start of October and remained mostly above the upper threshold until the end of the year (303 observations above the upper threshold). 
  • Deaths due to COVID-19 are notable from the week beginning 13 September 2021 (see graph 2). 
  • There were periods of statistically significant excess mortality in 2020 coinciding with the first and second waves of the pandemic. During this period there were 110 observations that exceeded the upper bound of expectation. 
  • During the second wave in 2020 statistically significant excess mortality could be largely attributed to deaths due to COVID-19 itself (see graph 2). 
  • During the entire pandemic period (start of March 2020), Victoria has recorded 3,152 deaths above expectation. There are a number of periods that are outside the expected range indicating statistical significance. There were 504 observations that exceeded the upper bound of expectation during this period. 
  1. Dates for key events are indicative only and may differ to other sources. 
  2. Data is provisional and will change as additional death registrations are received.
  3. 2016 is the starting point for the regression when modelling the projected number of deaths. 
  4. Data includes all doctor certified deaths occurring by the end of December 2021 and registered and received by the ABS until 28 February 2022. 
  5. See the methodology section at the end of this article for more information.
  1. Dates for key events are indicative only and may differ to other sources. 
  2. Data is provisional and will change as additional death registrations are received.
  3. 2016 is the starting point for the regression when modelling the projected number of deaths. 
  4. Data includes all doctor certified deaths occurring by the end of December 2021 and registered and received by the ABS until 28 February 2022. 
  5. See the methodology section at the end of this article for more information.

Weekly all-cause mortality: Queensland

  • In 2021 there were 853 deaths higher than the expected number in Queensland. While there were single weeks of statistically significant excess mortality, the number of deaths primarily remained within the expected range. 
  • In the weeks beginning 1 February, 22 February, 24 May, 27 September and 20 December 2021, statistically significant excess mortality was recorded (52 deaths exceeded the upper threshold over these weeks). These are single data points in time and the significance of these events should be interpreted with caution. 
  • From the end of November 2021 mortality has been sustained at higher than expected levels, although only one week (20 December) reached statistical significance. 
  • The number of deaths due to COVID-19 have minimal impact on all-cause mortality for Queensland as most deaths from the virus have occurred in 2022. 
  • There were 786 less deaths than expected in Queensland in 2020. From mid-April until mid-December 2020, the number of deaths primarily remained lower than expected with three weeks recording statistically significant lower mortality (weeks beginning 15 June, 29 June and 7 September 2020). 
  • From the week beginning 20 January 2020 mortality was higher than expected until the end of March 2020. Of this period, two weeks reached statistical significance, with 26 observations exceeding the upper bound. This was prior to the first COVID-19 death being reported in Australia. There were no other weeks in 2020 when excess deaths were observed in Queensland.
  • There were some periods of statistically significant excess mortality during 2017 and 2019 during more severe influenza seasons. 
  1. Dates for key events are indicative only and may differ to other sources. 
  2. Data is provisional and will change as additional death registrations are received.
  3. 2016 is the starting point for the regression when modelling the projected number of deaths. 
  4. Data includes all doctor certified deaths occurring by the end of December 2021 and registered and received by the ABS until 28 February 2022. 
  5. See the methodology section at the end of this article for more information.
  1. Dates for key events are indicative only and may differ to other sources. 
  2. Data is provisional and will change as additional death registrations are received.
  3. 2016 is the starting point for the regression when modelling the projected number of deaths. 
  4. Data includes all doctor certified deaths occurring by the end of December 2021 and registered and received by the ABS until 28 February 2022. 
  5. See the methodology section at the end of this article for more information.

Cause-specific mortality

While all-cause mortality provides a more accurate picture of patterns of mortality during the pandemic it is important to look at cause-specific mortality to provide insights into how individual causes have changed. Specific causes may experience significant changes which can be masked at the all-cause level. The following section provides excess mortality analysis for selected causes of death certified by a doctor in 2021. Deaths are analysed by underlying cause of death only.

Respiratory diseases (J00-J99)(excluding COVID-19)

  • There have been 5,966 lower than expected deaths due to respiratory diseases (including diseases such as influenza, pneumonia and chronic lower respiratory conditions) between 2020 and 2021. There have been prolonged periods where the difference in expected and observed deaths was statistically significant (1,646 observations were beyond the lower bound). 
  • Since the week beginning 6 April 2020 deaths due to respiratory diseases have been lower than expected. 
  • From the week beginning 25 May 2020 until the week beginning 23 November 2020 the difference to expectations for respiratory disease deaths was statistically significant. This pattern was replicated from late June 2021 until the end of November 2021. 
  • Sustained statistically significant mortality due to respiratory diseases was last recorded in 2017. 
  1. Dates for key events are indicative only and may differ to other sources. 
  2. Data is provisional and will change as additional death registrations are received.
  3. 2016 is the starting point for the regression when modelling the projected number of deaths. 
  4. Data includes all doctor certified deaths occurring by the end of December 2021 and registered and received by the ABS until 28 February 2022. 
  5. See the methodology section at the end of this article for more information.
  1. Dates for key events are indicative only and may differ to other sources. 
  2. Data is provisional and will change as additional death registrations are received.
  3. 2016 is the starting point for the regression when modelling the projected number of deaths. 
  4. Data includes all doctor certified deaths occurring by the end of December 2021 and registered and received by the ABS until 28 February 2022. 
  5. See the methodology section at the end of this article for more information.

Influenza and pneumonia (J09-J18)

  • There have been 4,235 lower than expected deaths due to influenza and pneumonia through the COVID-19 pandemic. Since April 2020 the number of deaths has not once exceeded the expected number of deaths in a weekly period. 
  • For the majority of the COVID-19 pandemic the difference in observed and expected deaths due to influenza and pneumonia has been statistically significant. 
  • There have been two deaths certified by a doctor due to influenza in 2021 and 43 in 2020. 
  • Public health measures put in place to prevent the spread of COVID-19 infections are also effective in limiting the spread of other infectious agents including influenza.
  • The influenza season resulted in excess mortality in 2016, 2017 and 2019.
  • In 2017 there were 1,061 deaths due to influenza and pneumonia that were higher than expected. There were sustained periods of statistically significant excess mortality with 706 deaths exceeding the upper threshold. 
  1. Dates for key events are indicative only and may differ to other sources. 
  2. Data is provisional and will change as additional death registrations are received.
  3. 2016 is the starting point for the regression when modelling the projected number of deaths. 
  4. Data includes all doctor certified deaths occurring by the end of December 2021 and registered and received by the ABS until 28 February 2022. 
  5. See the methodology section at the end of this article for more information.
  1. Dates for key events are indicative only and may differ to other sources. 
  2. Data is provisional and will change as additional death registrations are received.
  3. 2016 is the starting point for the regression when modelling the projected number of deaths. 
  4. Data includes all doctor certified deaths occurring by the end of December 2021 and registered and received by the ABS until 28 February 2022. 
  5. See the methodology section at the end of this article for more information.

Chronic lower respiratory conditions (J40-J47)

  • There have been 188 higher than expected deaths due to chronic lower respiratory conditions in 2021. There were two weeks where deaths exceeded the upper threshold (weeks beginning 29 March and 12 April) reaching statistical significance. These are single data points above the upper threshold and the significance of the increase should be interpreted with caution.  
  • Between the end of November 2020 and May 2021 deaths were sustained above expectation but were primarily within expected range. 
  • Between June and November 2021 deaths due to chronic lower respiratory conditions were generally lower than expected for a sustained period but still within the range of expectation. 
  • The lower than expected number of deaths due to chronic lower respiratory conditions was statistically significant for the majority of the winter months in 2020. 
  • Significant excess mortality was recorded in the winter months of 2017. 
  1. Dates for key events are indicative only and may differ to other sources. 
  2. Data is provisional and will change as additional death registrations are received.
  3. 2016 is the starting point for the regression when modelling the projected number of deaths. 
  4. Data includes all doctor certified deaths occurring by the end of December 2021 and registered and received by the ABS until 28 February 2022. 
  5. See the methodology section at the end of this article for more information.
  1. Dates for key events are indicative only and may differ to other sources. 
  2. Data is provisional and will change as additional death registrations are received.
  3. 2016 is the starting point for the regression when modelling the projected number of deaths. 
  4. Data includes all doctor certified deaths occurring by the end of December 2021 and registered and received by the ABS until 28 February 2022. 
  5. See the methodology section at the end of this article for more information.

Cancer (C00-C97, D45, D46, D47.1, D47.3-D47.5)

  • There have been 724 deaths due to cancer that were higher than expected in 2021. Deaths due to cancer have remained primarily in the expected range during this period. 
  • Over the two week period beginning 22 November 2021 deaths due to cancer were significantly in excess. This two week period followed a week of statistically lower than expected deaths and the significance of these events should be interpreted with caution. 
  • Between 2016-2019 cancer mortality has generally fallen between expected ranges. Intermittent points where the number of deaths has exceeded the upper threshold or dropped below the lower threshold of expectation have not been prolonged. This same pattern of death is seen for cancer mortality in 2020.
  • This analysis has covered all cancers grouped together. Results may differ for specific cancer types.
  1. Dates for key events are indicative only and may differ to other sources. 
  2. Data is provisional and will change as additional death registrations are received.
  3. 2016 is the starting point for the regression when modelling the projected number of deaths. 
  4. Data includes all doctor certified deaths occurring by the end of December 2021 and registered and received by the ABS until 28 February 2022. 
  5. See the methodology section at the end of this article for more information.
  6. This analysis has covered all cancers grouped together. Results may differ for specific cancer types.
  1. Dates for key events are indicative only and may differ to other sources. 
  2. Data is provisional and will change as additional death registrations are received.
  3. 2016 is the starting point for the regression when modelling the projected number of deaths. 
  4. Data includes all doctor certified deaths occurring by the end of December 2021 and registered and received by the ABS until 28 February 2022. 
  5. See the methodology section at the end of this article for more information.
  6. This analysis has covered all cancers grouped together. Results may differ for specific cancer types.

Diabetes (E10-E14)

  • In 2021, there were 75 less than expected deaths from diabetes. There was one week in 2021 where diabetes deaths exceeded the upper threshold of expectation (week beginning 8 March 2021) and two weeks where deaths were below expectation (weeks beginning 25 January and 22 February 2021). This comes after a sustained period of higher than expected deaths due to diabetes in 2020. 
  • There were 165 higher than expected deaths due to diabetes observed in 2020. A period of two consecutive weeks where excess diabetes deaths were statistically significant was observed during the first wave of the COVID-19 pandemic from the week beginning 30 March 2020.
  • People with diabetes can be susceptible to infections due to a compromised immune system. Mortality can be influenced by infectious disease activity in the community. 
  1. Dates for key events are indicative only and may differ to other sources. 
  2. Data is provisional and will change as additional death registrations are received.
  3. 2016 is the starting point for the regression when modelling the projected number of deaths. 
  4. Data includes all doctor certified deaths occurring by the end of December 2021 and registered and received by the ABS until 28 February 2022. 
  5. See the methodology section at the end of this article for more information.
  1. Dates for key events are indicative only and may differ to other sources. 
  2. Data is provisional and will change as additional death registrations are received.
  3. 2016 is the starting point for the regression when modelling the projected number of deaths. 
  4. Data includes all doctor certified deaths occurring by the end of December 2021 and registered and received by the ABS until 28 February 2022. 
  5. See the methodology section at the end of this article for more information.

Dementia (F01, F03, G30)

  • Deaths from dementia during the COVID-19 pandemic were generally within expected ranges. The majority of observed deaths were below expected deaths in 2020 and above expected deaths in 2021. 
  • In 2021 there were 523 deaths due to dementia that were higher than expected. In various weeks over the year, 54 observations collectively were above the upper threshold reaching statistical significance. Many of these are single weeks and should be interpreted with caution. 
  • From May 2020 to October 2020 dementia mortality was generally below expected projections. In the weeks beginning 29 June 2020 and 13 July 2020 the number of deaths due to dementia dropped below the lower threshold. 
  • People who have dementia can be at higher risk of dying from acute respiratory infections including influenza and pneumonia. The level of activity of acute respiratory disease can affect the death rate of dementia.
  • There were excess deaths recorded due to dementia in 2017 during the winter months. This was likely related to the severe influenza season recorded in 2017. 

 

  1. Dates for key events are indicative only and may differ to other sources. 
  2. Data is provisional and will change as additional death registrations are received.
  3. 2016 is the starting point for the regression when modelling the projected number of deaths. 
  4. Data includes all doctor certified deaths occurring by the end of December 2021 and registered and received by the ABS until 28 February 2022. 
  5. See the methodology section at the end of this article for more information.
  1. Dates for key events are indicative only and may differ to other sources. 
  2. Data is provisional and will change as additional death registrations are received.
  3. 2016 is the starting point for the regression when modelling the projected number of deaths. 
  4. Data includes all doctor certified deaths occurring by the end of December 2021 and registered and received by the ABS until 28 February 2022. 
  5. See the methodology section at the end of this article for more information.

Data downloads

Download all (614.48 KB)

Excess mortality in Australia, Jan 2020 - Dec 2021

Data files

Methodology

The analysis of 2020 and 2021 mortality data undertaken by the ABS is based on a model developed by Serfling¹⁰ and later adapted by the US Center for Disease Control (CDC) and New South Wales Health (NSW Health). This section provides an overview of how the model has been developed over time, key aspects of the model, and how the model has been adapted and applied by the ABS in this analysis.

Historical development of model

Historically this type of model has been used to estimate excess mortality caused by influenza.

  • As early as 1932, Collins determined that excess mortality during winter months in the United States of America (USA) was a consequence of epidemic influenza and therefore could be used as an indicator for the recognition of influenza outbreaks.²
  • In 1963, Serfling described a cyclic regression model, based on the seasonal pattern of pneumonia and influenza (P&I) deaths, to infer excess deaths due to influenza.¹⁰
  • Since then, models based on Serfling's have been applied in a number of temperate countries (USA,³,⁴ France,⁴ Australia,⁴ and Italy⁵) to demonstrate that excess mortality occurring during winter months is associated with pandemic and seasonal epidemics of influenza.
  • The CDC took the approach further and established the 122 Cities Surveillance System to provide timely, prospective information on excess mortality due to influenza. ⁶,⁷
  • The New South Wales mortality surveillance was included as part of the routine Influenza Surveillance Program and uses an adaptation of the Serfling model to monitor for influenza epidemics. This system primarily monitors the proportion of influenza positive specimens among all respiratory specimens from the major public health laboratories, the proportion of emergency department visits diagnosed with influenza, and outbreaks of influenza reported to Public Health Units by residential care facilities.⁹ 

Aspects of the Serfling model and its implementations

The Serfling model uses historical data to predict current patterns of mortality. This then enables the identification of any deviation from that prediction that may signal an influenza epidemic. In developing this model, Serfling recognised that past influenza epidemics could overly influence the regression, limiting the ability of the model to identify any new epidemic. For this reason, Serfling manually excluded past epidemics when fitting the model. 

An important feature of this methodology is the identification of a threshold that can be used as the upper and/or lower limit for the data. Observations that lie between these upper and lower limits are considered to be normal. The chosen threshold will vary depending on the desired specificity of the analysis and the methods used for calculation. In statistics there are default or familiar thresholds used in measuring these boundaries such as 1.96 standard errors either side of the mid-point.

An upper threshold was chosen to define the upper limit of the expected number of deaths in the absence of an influenza epidemic. This upper threshold was calculated as the predicted number of deaths in a given week plus a constant multiple of the standard error of the time series (the differences between each value predicted by the model and the actual observed values, or the 'model residuals'). Serfling chose the constant multiple to be 1.64 standard errors, and considered 2 consecutive weeks above the threshold to indicate epidemic behaviour.¹⁰

The NSW Health model used for the Influenza Surveillance Program is based on the Serfling model. This variant of the model fits a cyclic linear regression to data for the previous 5 calendar years and then forecasts the current year's time series from data up to the end of the previous year. Receiver operating characteristic curves are used to identify the most useful balance between sensitivity (the true positive rate) and specificity (=1-false positive rate) when determining the threshold that helps identify the start of an influenza epidemic.¹⁰

Data source

Refer to the methodology section of Provisional Mortality Statistics for a description of the scope of data used in this model.

ABS adaptation of the model

The ABS have applied key characteristics of the model used in the New South Wales Influenza Surveillance Program. The ABS model is also based on the application of a cyclic linear regression model to the time series of weekly number of deaths. The model has been applied to doctor certified deaths from all causes and to deaths from specific causes. Most diseases covered in both the Provisional Mortality Statistics reports and in this analysis show some cyclical pattern in numbers of deaths, indicating that the model should prove suitable for this broader application. 

The approach used by the ABS involves use of a 'robust' estimation procedure for fitting the model. Epidemics can cause increased mortality which leads to outliers in time series data. By down-weighting outliers it aims to fit the model to patterns of mortality that are expected to occur in the absence of an epidemic. This robust regression down-weights the influence of extreme observations (outliers) and is applied to a five-year baseline time series (weeks commencing on a Monday). Depending on the mortality activity in 2020 some adjustments had to be made to selected states and territories and selected causes of death to avoid over-fitting of the model. The regression is then used to forecast expected numbers of deaths for the current year.

The cyclical regression model includes: a linear time term, t, with values 1, 2, 3, ... for each week of the time series, and the square of the time term, t2, to accommodate long-term linear and curvilinear changes in the background proportion of the cause of death arising from factors such as population growth or improved disease prevention or treatment.

Also included are annual seasonal harmonic variables to describe the cyclical seasonal background pattern. The harmonic variables are functions of the week number, t, and the periodicity in the same units – in this case, yearly (52.18 weeks). The 2 harmonic variables in this case are: sine(2π t/52.18) and cosine(2π t/52.18).

The final model was:

\(Expected(proportion) = A + Bt + Ct2 +D sine(2πt/52.18) + E cosine(2πt/52.18)\)

where A, B, C, D, and E are the coefficients calculated from the regression.

To evaluate the approach, the model was fitted using PROC ROBUSTREG in SAS Software with the simplest, default 'M estimation' method.¹² The PROC SCORE procedure was then used to forecast values for 2020 and 2021. The standard error and threshold is derived from the stdi option in PROC ROBUSTREG which is run a second time with the 2020 and 2021 results attached.

Threshold identification

Identifying the threshold at which to signal excess mortality will vary depending on the goals of the research, application of the method, and seasonality of diseases of interest. Several national statistical organisations are using different models including Farrington surveillance algorithms, z -scores and prediction intervals to inform such thresholds.

A method for calculating the threshold for this analysis was selected which had a broader application and was more easily interpreted across a range of conditions, namely 1.96 standard errors or a 95% confidence interval. Use of the 95% confidence interval does introduce the expectation that random weekly variation could lead to counts that exceed thresholds 5% of the time. As such, individual weeks that exceed thresholds should be interpreted with caution. However, any prolonged period of weeks, 2 or more, exceeding the expected range can suggest excess mortality has been signalled.

The SAS function PROC ROBUSTREG offers the STDI parameter to calculate the 'standard error of prediction' of the model.¹² Which is used to define 1 standard error. The standard error of prediction is a more logical choice for assigning the threshold of excess mortality than the root mean square error (standard error of the model residuals) because it incorporates not only the variance of the residuals but also the variance of the model parameter estimates. This provides an estimate of the expected variability of the observed values in the absence of an epidemic.

References

1. CDC. Excess Deaths Associated with COVID-19 https://www.cdc.gov/nchs/nvss/vsrr/covid19/excess_deaths.htm#techNotes

2. Collins SD. Excess mortality from causes other than influenza and pneumonia during influenza epidemics. Public Health Rep 1932;47:2159–2180.

3. Reichert T, Simonsen L, Sharma A, Pardo S, Fedson D, Miller M. Influenza and the winter increase in mortality in the United States, 1959–1999. Am J Epidemiol 2004;160:492–502.

4. Viboud C, Boelle P, Pakdaman K, Carrat F, Valleron A, Flahault A. Influenza epidemics in the United States, France, and Australia, 1972–1997. Emerg Infect Dis 2004;10:32–39.

5. Rizzo C, Bella A, Viboud C, Simonsen L, Miller M, Rota M, et al. Trends for influenza-related deaths during pandemic and epidemic seasons, Italy, 1969–2001. Emerg Infect Dis 2007;5:694–699.

6. Centers for Disease Control and Prevention. 122 Cities Mortality Reporting System, Manual of Procedures. Atlanta (Georgia, United States): US Department of Health and Human Services; 2004.

7. Simonsen L, Clarke MJ, Stroup DF, Williamson GD, Arden NH, Cox NJ. A method for timely assessment of influenza-associated mortality in the United States. Epidemiology 1997;8:90–395.

8. O'Brien K, Barr IG. Annual Report of the National Influenza Surveillance Scheme, 2006. Commun Dis Intell 2007;31:167–179.

9. NSW Health, Communicable Diseases Branch. New South Wales Influenza Surveillance Report. NSW Health.

10. Serfling RE. Methods for current statistical analysis of excess pneumonia-influenza deaths. Public Health Rep 1963;78:494–506.

11. The ROBUSTREG procedure. SAS/STAT(R) 12.3 User's Guide. Cary (USA): SAS Institute, 2013. Available from: https://support.sas.com/documentation/onlinedoc/stat/123/rreg.pdf

12. Model Fit and Diagnostic Statistics. SAS/STAT(R) 9.2 User's Guide. Cary (USA): SAS Institute, 2008.