One of the common questions people have asked me about our excess deaths work is "have you looked at it by age?"
Here's a bit of insight into that question, noting it comes with a lot of caveats... #covid19Aus#excessdeaths
For anyone who missed it, here is my thread of excess deaths in Australia to the end of May 2022
First off, the ABS provisional mortality statistics show a breakdown by age/gender. But the age groups are very broad (0-44, 45-64, 65-74, 75-84, 85+). So there could be trends occurring within the age bands that are hidden when grouped together.
Eg, its not unreasonable to expect that mortality for 45-50yrs was less impacted by the pandemic than 60-64yrs, but they are in the one age band.
Even combining 65-69 with 70-74 might mask a bigger change in the older half cf the younger half.
And the 0-44 age band is very big!
There is also no breakdown by age/gender x cause. Cause is really helpful as it can help explain why we might be seeing a certain trend. We know mortality in 2020 and 2021 was impacted by lower respiratory deaths, but can only guess the impacts of this by age.
This graph shows the Aust population from 2015-2022 in the 5 age bands.
[Aside: the lack of migration in 2020 and 2021 shows clearly here in the under 65s]
Contrast the population graph with this one showing deaths in the first five months of each year from 2015-2022. The 0-44s make up 60% of pop but only 5% of deaths, while 85+ make up only 2% of pop but 40% of deaths.
The graph clearly shows a big increase in numbers of deaths in the 75-84 and 85+ age groups in 2022, more so that population increases would suggest.
The 45-64 and 64-75 age groups are less clear. While deaths are up a little, numbers are small and may be driven by pop changes.
A VERY rough back of the envelope calculation suggests excess mortality in 2022 by age group of:
- around 15% for both 75-84 and 85+
- ballpark of around 5% for 65-74
- less than 5% for under 64
And the excess for males maybe a few % points worse than for females.
Does this help with identifying what might be causing excess deaths?
For me, it is yet more evidence that vaccines are not a driver. Given that all age groups other than children are very highly vaccinated, we would have seen more deaths in younger cohorts.
It also supports that drugs/alcohol/diet are unlikely to be a big contributor as they are more likely to show up in younger age cohorts.
But I think everything else is still on the table in terms of our four most likely reasons.
The end. Any questions?
[And before you ask, yes, I am planning to do a look by state/territory.]
• • •
Missing some Tweet in this thread? You can try to
force a refresh
Excellent summary, noting the baseline is no longer “assuming no pandemic”.
Actual weekly deaths are mostly within the 95% confidence interval. However, most weeks in June, July and August were above the prediction and this is significant in aggregate for the eight-month period.
The Mortality Working Group has today published our Research Paper covering excess mortality in 2020-2023. actuaries.asn.au/public-policy-…
It is a massive paper (100+ pages), so I'm not going to cover off all of it!
This thread gives a taster of what is included in the paper, and is aimed at those who may want to delve further into the nitty-gritty.
There are four main sections of our paper (plus an into in section 1).
Section 2 cover excess mortality for Australia. This formally documents the excess mortality blogs we published throughout 2023. For those who have been following me, the numbers will look familiar.
The Mortality Working Group of the Actuaries Institute has released their latest estimate of excess mortality for Q1 2024.
TLDR: excess mortality is 1%, noting this is measured against the new baseline. actuaries.digital/2024/07/10/exc…
Our new baseline measures 2024 mortality against 2023 (after allowing for some mortality improvement), and includes and allowance for COVID-19 deaths.
Deaths from all causes have been within the 95% confidence interval in each of the first thirteen weeks of 2024, although they have been towards the top of that range for three of the last four weeks of March.
"Australia appears to have had at least a week where no COVID-associated deaths were recorded for the first time in more than two years."
This is simply not true. The statement is based on this Fed Health graph.
Yes, it does show a 7-day average of zero deaths for the most recent data points. But this ignores the fact that the graph is compiled using date of death, and it is almost impossible for someone to die, have their death registered, and included in the Fed data within a week!
The Mortality Working Group of the Actuaries Institute has released our estimate of excess mortality for the full year 2023.
TLDR: Excess mortality for 2023 is 5% (95% CI: 3%-7%) or +8,400 deaths.
A thread/ actuaries.digital/2024/04/05/exc…
Note that our expected number of deaths (baseline) is our expectation had the pandemic not happened.
The baseline allows for changes in the age and size composition of the population over time, plus allows for pre-pandemic mortality trend to continue.
We finished the year with deaths being much higher than expected for each of the four weeks in December (as they were in November).
Australian deaths counts: surveillance vs death certificates.
For some time now I have been quizzically looking at the death surveillance reporting, thinking the numbers looked too low.
A thread/
This is because of the relationship between hospitalisations and deaths.
Normally deaths track hospital admissions quite closely, but with a little bit of a lag. But for Dec23 and Jan24, this relationship changed - deaths are lower than expected based on hospitalisations.
A few theories were floating around about why (eg. JN.1 less severe), but it also didnt make sense to me from looking at the Vic death reporting, and knowing the Vic wave was a bit earlier than the rest of Aust. Things just werent stacking up nicely.