And for England, by chance, I had data from OHID (not OWID :)) and the estimates from IHME are bonkers ... so basically any analysis on IHME is building a castle on moving sands
(unfortunately)
In fact IHME wants to work with "cause underlying and contributing" / for the US, you can get the data from CDC wonder but for all other countries, you need to either dig deep into supplementary excels or make wet finger in the air estimations
Ahhhh!!!! That could explain part of it!
But there is still the issue that total deaths don't match with the national figures (e.g. France)
ADDENDUM: I didn't mention these other (obvious) design limitations in the paper
- it uses cases not infx so relies on the (shaky) assumption of equal testing propensity between vaxed and unvaxed
- the calculated impact of vaccines on covid mortality is "crude" and not yet corrected for HVE
I had asked AGES at the time (2022) to confirm their definition (as I did with every health agency in countries I followed) as I report it here
Of course, I may have missed an update
That's why I asked a clarification question to the paper's authors over on twitter
END
So, despite the totals looking "roughly equal" in 2020-2021 between / "deaths with covid as underlying and contributing cause"
/ AGES' dataset (ie "any death <28d after + Test") used in the paper
the profile of the deaths is likely to be quite off
5/
But also before 2022, those countries who tested much less than Austria had "deaths with covid" LOWER than deaths involving covid per death register (e.g. UK)
Typically, covid deaths of non-testers are being missed this way (eg at home)
4/
The reason is very simple: Nobody tested as much as Austria, and this by a mile
The more you test, the more you pick up less severe / symptomatic cases and the more you pick up incidental deaths and this was particularly massive in 2022 in Austria
3/
Why do I believe the authors use "deaths with covid"? Because
/ they mention explicitly AGES which only produced "deaths with covid" datasets AFAIK
/ their total up to July 14, 2022 matches perfectly the dataset from AGES I have
The issue: In 2022, these deaths were way off
2/
Hmm: AFAIU, this study published yesterday on determinants of covid mortality in Austria did so using deaths "with covid" ie any death <28d after a + test, not causal deaths
The issue? Deaths with covid massively over-estimate covid causal deaths in AT because AT was testing world champion
1/
PS: I think there are preliminary results from this notified trial
clinicaltrials.gov/study/NCT073...
This may be well-known in the world of Long Covid patients but it's now to me so I share it
Teams in the Netherlands found that auto-antibodies could be implicated in developing LC
It would be great to get to the root of this mysterious illness...
www.eurekalert.org/news-release...
I remain surprised that, despite this bias being well-documented in the literature, not a single study so far that looks at vax induced illness or mortality signal has actively sought to correct for the bias
And so we have another merry-go-round of unbaked claims ... Sigh
END
If you want an example, I applied this "least bad approach so far" on the Netherlands a few years back
cf here
x.com/Jean__Fisch/...
This is a real issue and one which I only see one way to address:
- look at mortality rates LONG after the vaccination
- Assume these are "the baseline"
- Calculate excess vs these baselines separately for vaxed and unvaxed
- Derive a relative risk ratio
It's not perfect but it's the best we can do
We know since the seminal study from Denmark in 2024 that, even after correcting for socio economical, age, and reported health factors, vaccinated have
- higher illness incidences
- lower mortality incidences
x.com/Jean__Fisch/...
This pop level study in Canada found a lower sudden death rate among young healthy vaxed than unvaxed
Its conclusion "this does not support an increase of sudden death risk post vax" does not hold
Why? It did not correct for the known under-diagnostic of illnesses among unvaxed
Of course, because if a cancer patient dies of pneumonia he caught somewhere, clearly the last straw was respiratory but if the medic believes that the person would have survived had he not had cancer, the underlying cause is cancer
But it's hugely judgemental (no critique, just a fact)
That the vaccines cannot be a major factor was already clear when looking at the picture across ages, as I did last summer
bsky.app/profile/jean...
Deaths involving cancer (either as underlying or contributing cause) excluding covid are now more or less complete for 2025 in the US
Here the view with population effects taken out of the picture
But not knowing if the issue is concerning 1,000 or 1,000,000, everyone can come with his priorities about what is the adequate response, from "We should never have locked down" to "we should have locked down much earlier"
That's why I wrote, the conclusions are useless ... alas
Just for clarity: Obviously, delayed care and restricted access to NHS has led to deaths or will lead to deaths (from missed screening)
The point here is that without an idea of the order of magnitude (1k? 100k?), you cannot draw any recommend any adequate changes
Not only that
So the UK covid inquiry's only data point on the impact of delayed care / reduced access to NHS is absolutely not supporting the claim
So.... The UK inquiry drew conclusions without an inkling of the size of the problem
Not good (because essentially useless)
END
But could it be that there is a signal possibly among the young people? Here I can do even better because not only do I have the official data of deaths but I also mapped it against vax and covid waves
As you see, the suspected link to covid is even reinforced
And here is the data ...
a) As you see, the excess is driven by the covid (and later flu) waves
b) As you don't see (and need to trust me😀), the excess for Mar to Dec 2022 is ... 200 and not statistically significant
So how big was the excess? Is there some truth to it after all?
We are in luck: ONS did share a special file (a user request) by illness, by occurrence date, by week!
I used this and mapped the actuals against the pre-pandemic trend of mortality (as one should do)
First, the source of the excess (ONS report) and they are two problems
a) it's not Mar to Dec 2022 but Mar 2020 to Dec 2022 so the whole pandemic
b) ONS measures excess against 15-19 average deaths in 2020 and 2021 and 16-19+21 average for 2022: That overestimates expected deaths
The latest UK inquiry report flags issues from the limited access to healthcare
To underline this, the report cites an excess mortality in colorectal cancer from March to December 2022 per ONS
The issue? Nothing is correct and certainly not a link with delayed care ... ⤵️
