Catch up on the Actuaries Institute’s COVID-19 Mortality Working Group’s latest analysis of excess deaths.
The COVID-19 Mortality Working Group has examined the latest provisional mortality statistics up to 30 April 2022, released by the Australian Bureau of Statistics (ABS) on 29 July.
We have compared the observed deaths to our ‘baseline’ predicted number of deaths for doctor-certified deaths (by cause) and coroner-referred deaths (for all causes combined).
We have also provided an indicative estimate of excess mortality due to COVID-19 only, for the three months subsequent to the ABS data (May to July 2022).
Our previous Actuaries Digital article discussed in some detail how we have arrived at our baseline predicted deaths. In short, our baselines are set by extrapolating linear regression models fitted to standardised death rates (SDRs), which are then re-expressed as numbers of deaths. For the pandemic years, we have included different years to train the regression models:
- For 2020 and 2021, we have used the 2015-19 experience.
- For 2022, we have:
- used 2015-19 experience to set our baseline for deaths from respiratory disease and dementia, on the basis that 2020 and 2021 experience for these causes was materially affected by the pandemic;
- also shown the average experience in 2020-21, for reference, when analysing mortality from these causes; and
- used 2015-21 experience to set our baseline for deaths from all other causes and for coroner-referred deaths, on the basis that it is likely that 2020 and 2021 experience more closely reflects a slow-down in underlying mortality improvement than the impacts of the pandemic.
- used 2015-19 experience to set our baseline for deaths from respiratory disease and dementia, on the basis that 2020 and 2021 experience for these causes was materially affected by the pandemic;
Our estimates of excess deaths remain ‘in the absence of the pandemic’ for each of the three years 2020 to 2022. We have not included any COVID-19 deaths in the baseline, as these would not exist in the absence of the pandemic.
As always, it is important to note that predicted death numbers are increasing faster from demographic changes (aging and population size) than they are reducing due to mortality improvement. Therefore, our model predicts higher numbers of deaths in each successive year.
Comparison to ABS reporting
The results we present here differ from those quoted by the ABS in its commentary surrounding the release of the Provisional Mortality Statistics. For 2022, the most significant difference is in the determination of the baseline, where the ABS uses a simple average of the number of deaths from 2017, 2018, 2019 and 2021, with no allowance for mortality trends or demographic changes. In our view, this understates the baseline and therefore overstates the measure of excess deaths.
For example, the ABS has reported about 8,500 (17%) excess deaths in the first four months of 2022, in comparison with our estimate of 6,800 (13%).
Excess deaths to 30 April 2022
Figure 1 and Table 1 below summarise the results of our analysis. This article concentrates on the experience in 2022, but we have also shown the 2020 and 2021 results for context.
Deaths were above the upper end of the prediction interval (i.e. above the 97.5th percentile) for all but one week of 2022 so far.
Table 1 – Excess deaths in Australia
For the month of April 2022, total deaths were 9% (or +1,200) higher than predicted. More than half of the excess deaths were doctor-certified deaths due to COVID-19. Both ischaemic heart disease and other unspecified causes also made a significant contribution to excess deaths, as did coroner-referred deaths (which will include some COVID-19 deaths).
For the first four months of 2022, total deaths were 13% (or +6,800) higher than predicted. This compares with excess deaths of -4,500 for 2020 and +3,400 for 2021.
By cause, in aggregate for these four months:
- there were 3,630 doctor-certified deaths from COVID-19, representing just over half of the excess deaths;
- doctor-certified deaths from respiratory disease continued to be lower than expected (down 6% or 250), noting that there are normally relatively few such deaths before May;
- doctor-certified deaths from cancer were close to expected, as has been the case so far throughout the pandemic;
- doctor-certified deaths from heart disease, cerebrovascular disease, diabetes and dementia were all higher than predicted (by between 5% and 11%, or a total of 1,200);
- doctor-certified deaths from other unspecified diseases were higher than predicted (by 11% or 1,400), continuing a trend observed since April 2021. Note this is a large ‘catch-all’ category and it is difficult to infer the reason for this large increase; and
- coroner-referred deaths were 10% (+680) higher than expected, noting that some COVID-19 deaths (such as when people die at home) will have been referred to the coroner.
Data included in the ABS article COVID-19 Mortality in Australia released on 29 July 2022 shows that, in the first four months of 2022, there were 714 deaths in people who were COVID-19 positive at death but where COVID-19 was not the primary cause of death. This means that COVID-19 was a potential contributory factor in around one quarter of the excess mortality from non-COVID-19 causes.
COVID-19-related deaths in May to July 2022
While the ABS provisional mortality statistics data is only available until the end of April 2022, surveillance COVID-19 deaths are available daily. Figure 2 shows the number of such deaths in each month of 2020, 2021 and first seven months of 2022.
Figure 2 – COVID-19 deaths in Australia, reported from surveillance systems*
*On 31 March 2022, NSW Health released a report that detailed the outcomes of a reconciliation between their daily death counts and data held by Births, Deaths and Marriages (which is the basis of the ABS data). This reconciliation identified 331 deaths that had not been captured in the daily counts. We have shown these deaths separately in the chart, allocated to the months when the deaths occurred.
There were 912 surveillance COVID-19 deaths in 2020 and a further 1,402 deaths in 2021. In the seven months to 31 July 2022, there have been 9,550 COVID-19 deaths. Of these, 4,633 occurred in the three months to 31 July 2022 (where we have no provisional mortality statistics for all causes). Surveillance deaths in the month of July 2022 are the highest yet recorded.
The latest COVID-19 Mortality in Australia article by ABS shows that, since the Omicron wave, COVID-19 was the underlying cause of death on a reducing proportion of death certificates mentioning COVID-19. Figure 3 shows a comparison of deaths ‘from’ COVID-19 versus those ‘with’ COVID-19, noting that, for the most recent months, a large proportion of deaths have not yet been registered (the estimated numbers of unregistered deaths are also shown). Figure 3 also contains a line showing the proportion of registered COVID-19-related deaths that were ‘from’ rather than ‘with’ COVID-19.
Figure 3 – A comparison of deaths ‘from’ COVID-19 to those ‘with’ COVID-19 (source: ABS)
The proportion of registered COVID-19 deaths from’COVID-19 has reduced in 2022, from 88% in January to 76% in May and June. A large number of deaths are still to be registered in June, so this percentage could change – hence, this point is shown as a preliminary estimate.
As such, we consider it reasonable to assume that in May to July 2022, around 76% of COVID-19 deaths may have been from COVID-19 rather than with COVID-19. Therefore, we estimate that, of the 4,633 deaths reported in May to July 2022, around 3,500 may be due to COVID-19.
Our prediction model suggests that, without a pandemic, there would have been a total of around 45,600 deaths in the three months from May to July 2022.
Thus, COVID-19 deaths represent around 8% extra mortality from May to July 2022 (around 7% in both May and June and 9% in July). There may have been fewer deaths from respiratory diseases in this period than our pre-pandemic predictions. However, we consider it likely that mortality from non-COVID-19, non-respiratory causes will be higher than our pre-pandemic predictions, given the higher-than-expected mortality in 2021 and early 2022, and that this will outweigh the benefit from respiratory disease. Therefore, we expect that total excess mortality in May to July 2022 will have been higher than the 8% explained by COVID-19.
Excess mortality across the pandemic
Figure 4 combines our preliminary estimate of excess deaths due to COVID-19 only for May to July 2022 with our detailed excess death estimates shown earlier, to reveal cumulative excess mortality since the start of 2020.
Figure 4 – Estimated cumulative excess deaths in Australia since 1 January 2020
Deaths from the Delta wave in the latter part of 2021 and the Omicron wave in early 2022 had fully eroded the negative excess deaths experienced earlier in the pandemic by early January 2022. Cumulatively across the pandemic, our conservative estimate is that Australia had experienced around 9,200 excess deaths by the end of July 2022. This represents an average excess mortality rate of about 2% over that time.
These impacts by year can be seen separated into excess deaths from COVID-19 and other causes in the table below.
Table 2 – Excess deaths separated into COVID-19 and other causes
We estimate that there have been 9,200 deaths from COVID-19 in Australia to the end of July 2022.
For non-COVID-19 causes, lower than predicted deaths in 2020 have been fully offset by higher than predicted deaths in 2021 and the four months to 30 April 2022.
What could be causing the non-COVID-19 excess deaths?
Higher numbers of deaths than predicted does not tell us why this is occurring. There are a number of reasons hypothesised around the world (where this effect is occurring to a greater or lesser extent):
- Undiagnosed COVID-19: Some of the excess deaths could actually be from non-identified COVID-19. This effect happened early in the pandemic, but seems less likely in 2022, as testing is much more available, particularly for those who are seriously ill. Also, for any deaths where COVID-19 may be suspected, post-mortem testing is occurring in Australia.
- Post-acute COVID-19 complications or interactions with other causes of death: An earlier COVID-19 illness could be causing later illness and death, and/or COVID-19 could have worsened other diseases which ultimately caused death. To some extent, this shows on death certificates in the 714 deaths in the first four months of 2022 where COVID-19 is listed as a contributory cause, and a further 60 deaths were identified as from Long COVID. However, it is possible that there could be more of these deaths than identified.
- Delayed deaths from other causes: Some of the reduction in deaths in 2020 and 2021 that resulted from the absence of many respiratory diseases may be reversing. People who otherwise may have died of flu or other respiratory diseases in those years had their systems been stressed may now be succumbing to their underlying illnesses.
- Delay in emergency care: Pressure on the health, hospital and aged care systems could lead to people not getting the care they require, either as they avoid seeking help, or their care is not as timely as it might have been in pre-pandemic times.
- Delay in routine care: Lack of earlier diagnostic testing for non-COVID-19 causes could lead to later mortality. While this does not yet appear to be occurring for cancer deaths, it may be a factor in higher deaths from other causes, such as ischaemic heart disease, diabetes, and the large ‘other’ category.
- Pandemic-influenced lifestyle changes: There is evidence from the UK that a higher proportion of people made less healthy lifestyle choices during lockdowns (e.g. drinking more alcohol, exercising less, higher rates of childhood obesity), and that these less healthy practices have continued. It is unclear to what extent similar factors may be affecting mortality in Australia in 2022.
It isn’t possible to identify from death counts alone whether any or all of these issues are causing the non-COVID-19 excess deaths, but we think that they are the most likely explanations.
Leading causes of death
In this section, we assess where COVID-19 sits in terms of leading causes of death in Australia.
By way of background, each death in Australia is coded with a primary cause of death based on the International Classification of Diseases, version 10 (ICD-10). The ICD-10 classification system is hierarchical, so depending on how you add up the classifications, the leading causes of death may differ. Cancer, for example, is the cause of death in about 50,000 deaths per annum putting ‘all cancers’ clearly as the leading cause of death. However, the individual cancer that causes the highest number of deaths is lung cancer at about 8,500 per annum, making it the fourth leading cause excluding COVID-19.
The ABS reports on the top 20 leading causes of death by grouping individual ICD-10 codes. Cancers are grouped based on the region of the body. In this analysis, we have followed the same classification system.
So where does COVID-19 fit? At best, we can only make an estimate as:
- the detailed cause of death data is not yet available for deaths in 2020, 2021 or 2022 (data is available for deaths registered in 2020, but data for deaths occurring in 2020 is incomplete); and
- the deaths by cause shown in Table 1 above are for doctor-certified deaths only, and cancer deaths are not available broken down by type of cancer.
We have estimated deaths for the leading causes for the first seven months of 2022. To do this, we have:
- taken doctor-certified deaths by cause to 30 April as shown in Table 1;
- added our predicted doctor-certified deaths for the three months of May to July 2022; and
- included allowance for coroner-referred deaths (using the historical ratio of doctor-certified to coroner-referred deaths).
For the leading cancer causes, we have adopted a top-down approach. We have followed the same approach as above for cancer deaths as a whole. We have then estimated the leading cancers by using the same breakdown of cancer deaths by body part as the average for 2015-2020. Lung cancers make up around 18% of all cancer deaths while colon cancers make up around 12% of all cancer deaths. These proportions have been stable over the recent time period examined.
Table 3 – Excess deaths separated into COVID-19 and other causes
Ischaemic heart disease has been the leading cause of death in Australia for many years. However, the mortality rate for this cause has been declining at the same time as the population has been aging, resulting in an increasing number of dementia deaths. Our estimates of the number of deaths from ischaemic heart disease and dementia in the first seven months of 2022 are therefore the same at around 10,000 deaths from each cause.
Turning to COVID-19 deaths, we estimate deaths from COVID-19 in the first seven months of 2022 are about 7,100:
- 3,600 doctor-certified deaths in four months to April (noting there will also be a small number of coroner-referred deaths that we have ignored in this analysis); plus
- 3,500 deaths in the three months from May to July.
This would put COVID-19 as the third leading cause of death for the first seven months of 2022, well behind ischaemic heart diseases and dementia with around 10,000 deaths each, but materially higher than cerebrovascular diseases (largely stroke) at around 5,500 and cancers of the lung at 5,400.
The remainder of this article shows actual versus predicted deaths to 30 April 2022. Analysis and discussion of individual causes of death refer to doctor-certified deaths, while coroner-referred deaths are for all causes combined.
Following this is a discussion of how COVID-19 deaths are reported and recorded in Australia.
Figure 5 – Weekly actual and predicted doctor-certified deaths in Australia – COVID-19
There were 654 doctor-certified deaths from COVID-19 in April 2022, compared with 890 surveillance deaths (236 fewer). We would not expect the numbers to be identical, because:
- there are delays between time of death and lodgement of the doctor’s certificate;
- the reporting criteria are different, with the main difference being that the surveillance reporting includes all deaths in people who have died while COVID-19 positive and without another clearly obvious unrelated cause (e.g. trauma), whereas the ABS deaths included in this chart only include deaths where COVID-19 was the primary cause of death; and
- some COVID-19 deaths will be referred to the coroner (e.g. deaths occurring at home).
While COVID-19 deaths in 2022 have far exceeded deaths from this cause earlier in the pandemic, COVID-19 deaths in April 2022 were lower than in January and February.
Deaths from respiratory disease
Figure 6 – Weekly actual and predicted doctor-certified deaths in Australia – all respiratory diseases
In the four weeks of April 2022, deaths from respiratory disease were close to or lower than predicted, following the trend throughout most of the pandemic.
The following figures present a breakdown of respiratory disease into influenza, pneumonia, lower respiratory disease, and other respiratory disease.
Figure 7 – Weekly actual and predicted doctor-certified deaths in Australia – influenza
There was one influenza death in the first four months of 2022, occurring in the last week of April.
Figure 8 – Weekly actual and predicted doctor-certified deaths in Australia – pneumonia
Deaths from pneumonia in April 2022 have been fewer than predicted, and at the bottom end of the 95% prediction interval. Deaths so far in 2022 are similar to the levels seen in 2020 and 2021.
Figure 9 – Weekly actual and predicted doctor-certified deaths in Australia – lower respiratory diseases
Deaths from lower respiratory disease in April 2022 have broadly been close to predicted.
Figure 10 – Weekly actual and predicted doctor-certified deaths in Australia – other respiratory diseases
Deaths from other respiratory diseases continue to be close to our predictions in April 2022.
Non-COVID-19 and non-respiratory deaths
Figure 11 – Weekly actual and predicted doctor-certified deaths in Australia – all causes other than respiratory diseases and COVID-19
Excluding deaths from COVID-19 and respiratory diseases, deaths were higher than expected for all weeks in April 2022, and above the prediction interval for the first two of those weeks. This is mainly driven by deaths from ‘other’ causes.
The following figures show a breakdown of non-respiratory/non-COVID-19 deaths into cancer, heart disease, cerebrovascular disease, diabetes, dementia, and all other causes.
Figure 12 – Weekly actual and predicted doctor-certified deaths in Australia – cancer
Cancer deaths continue to be close to predicted numbers for most weeks, albeit more commonly above than below. With diagnostic testing down in 2020, there were concerns that there would be a spike in cancer deaths in 2021 and beyond. While it is still early days, we are not yet seeing any clear evidence of this effect.
Figure 13 – Weekly actual and predicted doctor-certified deaths in Australia – ischaemic heart disease
Deaths from ischaemic heart disease were significantly higher than predicted in the second week of April, but within the prediction interval for the other three weeks. Very few weeks have been below the predicted line since March 2021.
Figure 14 – Weekly actual and predicted doctor-certified deaths in Australia – cerebrovascular disease
For cerebrovascular disease, deaths were close to predicted in April 2022.
Figure 15 – Weekly actual and predicted doctor-certified deaths in Australia – diabetes
Deaths from diabetes were also close to predicted in April 2022.
Figure 16 – Weekly actual and predicted doctor-certified deaths in Australia – dementia
Deaths from dementia were higher than predicted in April 2022 but were within the prediction interval.
Figure 17 – Weekly actual and predicted doctor-certified deaths in Australia – other unspecified diseases
Deaths from other causes i.e. those not explicitly reported on by the ABS, were again much higher than predicted in April 2022. All weeks were at or above the upper limit of the prediction interval.
Figure 18 – Weekly actual and predicted coroner-referred deaths in Australia – all causes
Coroner-referred deaths continued to be higher than predicted every week in April 2022 and were above the prediction interval for the first and last week of that month.
How are COVID-19 deaths reported in Australia?
This appendix sets out our understanding of how COVID-19 deaths are reported in Australia. This is based on our understanding built up over many years of dealing with death statistics in different circumstances. However, we are not medical professionals, so don’t have first-hand experience of how deaths are certified.
First off, we have the daily surveillance reporting. The purpose of this data is to provide an indication of how the pandemic is trending. Note that data has been collected for years for other communicable diseases such as flu, measles, etc, and is used by epidemiologists and health professionals to manage the health response. It was not designed for other users, including actuaries!
By its nature, the daily surveillance reporting is intended to be quick, to give the best picture possible of what is happening NOW. To facilitate this, the decision has been taken to report deaths both ‘from’ COVID-19 and ‘with’ COVID-19. Anyone with a positive COVID-19 diagnoses who dies will be included in the daily count, excluding anyone with an obvious non-COVID cause such as trauma.
The data is not perfect, but it is timely. We have seen some of the limitations, including:
- reporting of 105 deaths by Victorian DHHS on 29 July due to a data feed issue;
- in the second wave in Victoria, deaths not being reported by some aged care homes to the state health authorities in a timely manner;
- in NSW, some deaths not being reported to NSW Health at all and only being captured after comparing to the Registry of Birth, Deaths, Marriages (BDM); and
- deaths that are referred to the coroner and are therefore not reported in the daily surveillance until a finding has been made.
These issues mean that the daily reporting does not always reflect deaths that occurred in the last day or two.
It’s important to note that the daily surveillance data is NOT the official record of COVID-19 deaths in Australia.
The official record of COVID-19 deaths comes from each state/territory’s BDM. The ABS bases its reporting on the BDM and is the ‘gold standard’.
Deaths are included in the BDM once they have been registered. There are two ways this can happen:
- The treating doctor/regular GP registers the death when the circumstances of death are quite straightforward and the cause of death known.
- The death is referred to the coroner, who will undertake investigations such as autopsy, toxicology, etc, and will establish the cause of death.
The officially recorded cause of death in the public records is the cause of death as determined by the treating doctor who knows the patient, or by a coroner after extensive investigation. It is not an arbitrary decision made by a bureaucrat.
The cause of death determined by the doctor/coroner will show whether the death was ‘from’ COVID-19 or ‘with’ COVID-19. It captures separately deaths occurring during the acute phase of illness and deaths from long COVID.
Some COVID-19 deaths that may not necessarily be captured are those where the person may have apparently recovered from the acute phase, but then died from heart failure (for example) caused by COVID-19. Depending on the circumstances of death, some of these deaths will be referred to the coroner and COVID-19 may be found to be the underlying cause. But there will be other deaths where this does not happen, making analysis of excess mortality important.
Doctor-certified deaths are typically registered much more quickly than deaths referred to a coroner, but there can still be quite lengthy delays between when a death occurs and when a death is registered.
As an example, the ABS released its latest provisional mortality statistics on 29 July 2022. This covered all deaths that occurred up to 30 April 2022 and were registered by 30 June 2022. So, it does not include any deaths that occurred in the latest three months. And around 2% of doctor-certified deaths and 10% of coroner-referred deaths that occurred in April won’t have been registered by 30 June.
So, there is a trade-off: quick information that is a bit inaccurate from the daily surveillance reporting, versus the best possible information that comes out quite slowly.
Members of the COVID-19 Mortality Working Group:
CPD: Actuaries Institute Members can claim two CPD points for every hour of reading articles on Actuaries Digital.