5 Filters

Latest ONS data - can anyone suggest an explanation

#1

ONS has released data of deaths in the vaccinated by time since vaccination.

I’m just posting the two graphs from this promising new source Excess Burden, who took the trouble to just download the ONS data and graph the data by age groups, that were already there.
One is Covid deaths and the other Non-covid.
As you’ll see, the issue isn’t the ages but the time of the deaths.

Covid deaths in the vaccinated, by time elapsed since jab.

image
image533×739 30.7 KB

Non-covid deaths in the vaccinated, by time elapsed since jab.

image
image534×740 24.8 KB

In the 1st graph (covid deaths) the issue is obvious. In the three oldest age ranges, the deaths peak 3 weeks after the jab. Even in the less spectacular 60-69 and 50-59 (green and yellow lines) they peak at week 3 after the jab also. In the two youngest age groups; the curiously grouped 10-39 (why this grouping?) and 40-49, the peak is at two weeks post jab.

In the second graph (Non-covid deaths) the peak for the highest 5 age groups is at 9-10 weeks after the jab, and the weely death rate rises steadily towards this peak.

Surely in both case (especially the non-covid deaths) the deaths post-jab should be at a steady rate?

One thing that occurred to be was that the counting of covid deaths within the first 2 weeks of a jab as unvaccinated (and I’m not sure all the sources do this) might have led to biased counting or deliberate miscounting in the first 3 weeks. Imagine some poor sap with the job of moving the data away from the first 2 weeks.
But even if this was happening there would still be a big peak in covid deaths in the first 3 weeks since the jab, however they were spread.

You’ll find the whole thing here https://excessburden.substack.com/

Excess Burden didn’t comment, maybe wondering if there were subtleties in the data.
So not worth a huge amount of time perhaps, hopefully someone (Norman Fenton’s team?) will pick it up soon.

Cheers

1 Like
#2

Interesting and weird…

You can get the data from the link he provides here

And, sure enough, table 9 is the table that these graphs are from, however something is not adding up (literally!)

Here’s an example showing my confusion. If I take a cohort (ages 70-79 for example) and add up all the people who died from covid x number of weeks after vaccinaton (i.e. add up all the numbers in your first chart for the age group 70-79, which is column 3 in table 9) I get:

  • Total number of vaccinated people (70-79) who died of covid: 7616 (table 9)

Ok. I now look at table 2 which lists total deaths from covid by month for the same period. I would expect that if I add up all the vaccinated people who died of covid over the year (column 6 in table 2), I should get the same number as the one I just calculated - 7616. But I don’t.

  • Total number of vaccinated people (70-79) who died of covid: 6780 (table 2)

So the total covid deaths from one table to another differ? The total number of covid deaths over the year should be the same… (or at least very close as I made some very small adjustments).

Clearly I just don’t understand exactly what table 9 is actually showing and how it relates to the other data in that spreadsheet. I feel like the ONS could have been a lot clearer here.

If you can make any sense of what the data is actually showing then I’ll see if I can add any potentially interesting thoughts.

Cheers
PP

1 Like
#3

Hi PP.
I agree with your calculations in Table 2 and 9 (I took “<3” as meaning 2, as you seem to have done as well :slightly_smiling_face :).

Funny you should alight on this possible discrepancy. Suspicious as I am, I was already thinking about that little ruse. Norman Fenton said he thought it was being done, and I’d read it somewhere.

The vaccinations in Table 9 are defined by when the vaccination was, as that table has the purpose to show by time elapsed. If the ‘vaccinations’ in Table 2 were subject to the interpretation I referred to, this would account for the difference. I just don’t remember whether it was ONS or PHE that were doing this*.

I’d probably still assume the data in Table 9 were correct as the definition of ‘days since vaccination’ is precise.
Cheers

Edit: It was the ONS that Fenton was referring to (during an interview with Del Bigtree) about that trick with the vaccinated who die within 2 weeks counting as unvaccinated. He wasn’t sure if they did that. The CDC certainly did, it was on their website (but isn’t now).

They had this discussion:

DB: "You have to be 2 weeks after your second dose in a two-dose series (Pfizer and Moderna) so that month after your first shot you are considered unvaccinated if you get sick and die you’re going to land, categorised as unvaccinated. And for the JJ vaccine you have to wait two weeks after that single dose.
That there could create this exact same anomaly too

NF: It could, it would - if you just did it by one week you would get (speaks assertively) identical figures to what I’ve shown, and you can prove it’s identical, my paper - (indistinct, mentions blog) …
To be fair, I have to say, the ONS is telling us, that we don’t do that 14 days, that even if it’s less than 14 days, if they’ve died they’re supposed to be recorded as vaccinated, but the way things work in hospitals and surgeries, if they die shortly afterwards - the paperwork isn’t going to be submitted. It seems to me unlikely that the information on when they got vaccinated is going to be there.
DB Especially in a situation where a person goes to a clinic, or their personal doctor, to get vaccinated. If they end up in the ER, or dying, they’re in the nearest hospital, they could be in a different system. Many times what we’re seeing is that you’re only listed as having been vaccinated if you’re vaccinated in the same hospital you end up in for your critical care, so right there you can see problems and anomalies.
NF Exactly…"

Not sure if I buy all of that either, but I’d still expect the data with the precise definition to be correct.
Maybe your query shows that they are doing a trick in Table 2 !?

1 Like
#4

Hiya

Yes, to my mind 2 < 3 so happy days!

I think you’re on to something in the discrepancy of the data. It might well be along the lines that you’re suggesting. I checked and table 3 is consistent with table 2, so at least some of the tables make sense! I notice that table 9 has a different source than the others, so that’s likely part of the problem.

It seems that the definition of who is vaccinated is a lot broader in table 9 than the rest of the spreadsheet (lots more people in that category). That’s surprising as there are so many vaccination categories in the other tables… Somehow adding up all those categories still doesn’t reach the definition/number used in Table 9.

I did look at a few charts of the data in table 2 to see if the patterns were at least similar but I got nothing on my initial, cursory look.

Without knowing how they came to get that number of people in table 9, I don’t really know how they classified if someone was vaxxed or not and thus I don’t really know how to interpret the graphs. I have some simple thoughts but don’t really feel like they’re worth sharing given the uncertainties…

Very annoyingly obfuscatory.

Anyway, I’m interested if you get any more clarity

Cheers

#5

Well I’d say that there is something altogether VERY odd about these charts, showing that a high proportion of people catching COVID die around three weeks later if they’ve been vaccinated. This seems to go with the observation that vaxed people are getting COVID worse than unvaxed people - perhaps suffering a hyper immune reaction to infection, as predicted. But the extraordinary similarity between graphs of different ages suggests some specific cause. They certainly need explanation!

#6

Yes, there’s a fair amount that needs explaining in those numbers…

#7

Post wasn’t aimed at PP, rather the world - or interested 5f readers, whichever group is smaller :slightly_smiling_face:

One thing I think is encouraging is it’s clear the data isn’t simply made up.
I feel sure of this, because they wouldn’t have made THIS up (from Table 2 that we were discussing):

image
image1116×465 154 KB

This is the best type of data there is. Age-standardised (great because vaccine rollout was determined by age, and age also greatly determines fatality rate from covid, so two great confounders are dealt with). It shows death rate per head of the relevant population, so removes the problem of a numerically dominant group.

For cause of death, the table has covid, non-covid and all-cause.
The snippet shown is just of non-covid, and just one age group.

I just want to point out one thing in the non-covid deaths - the startling rate of deaths in the category ‘First dose at least 21 days ago’ (5883) versus unvaccinated (1221).

This is just age group 60-69, and it’s just one month - but the same picture is present in the other nearby age groups 50-59 and 70-79; and in the other months around the same time.
Again, these are death rates per 100,000 population - reflecting real risk for each population (unvaxed, 1 dose etc), irrespective of how many are in the population.
(BTW, I chose non-covid deaths to get away from these strange, coinciding peaks in the covid deaths. Though I think there is a real peak in the first few weeks of the covid deaths, I now think the peaks coinciding reflect an admin glitch. I think the main issue is with the non-covid deaths.)

You can see this playing out in the whole data set, in another great blog post by Excess Burden (what a name!):

It’s well explained, so doesn’t take very long to read.

Basically the vaccinated data showed good overall results for a few months after Jan 2021, but soon waned and then went into negative territory - with very bad outcomes. This is shown by the first two graphs in the blog post - which show the two ‘January’ months 2021 and 2022.
I’ll not show these, but the difference is pretty shocking.
These are all-cause mortality - so includes any covid benefit.

The point I made re the snippet of data - ie vaccinated ‘First dose at least 21 days ago’ mortality vs unvaccinated - can be seen to startling effect in the next graph in the blog post (the first line graph):
image

In the remainder of the blog post the writer breaks it down by age, showing this is a consistent pattern (but as you go down the ages, even the brief period when the vaccine seems to do well vanishes) As Excess Burden notes,

"This data is all very alarming. A poorly functioning vaccine should still have at least a small positive effect. A non-functioning vaccine should have no effect. Yet we see a negative effect in all age groups for both 1 or 2 doses taken ‘at least 21 days ago’ , and it is most cases the negative effect is quite large. The fact that the pattern is consistent and predictable, meaning it moves smoothly from month to month and age bracket to age bracket, gives even more credibility to the pattern. "

It does look grim.

It may look like the culprit is the one category ‘First dose, at least 21 days ago’ but a look at the other vaccinated categories gives the impression that alll of them probably overtake the unvaccinated in time.

Finally, a look at the 18-39 age group (in the blog post) seems to show that the mortality in the jabbed groups was always higher than in the unvaccinated. See below

image

One final thing - if these interpretations are correct, what is happening inside the 18-39 group. Is it masking a very bad outcome among 18-29’s?

And what does all this mean for the vaccination of children?

1 Like
#8

Hi ED

I agree. I don’t think that the ONS make data up generally, and I would also agree that this is usually the best data available in the UK. In this case, something weird is going on between table 2,3 etc and table 9. It makes me wonder what exactly is being represented by table 9, and why tables 2 etc are so detailed on vaccination status, and table 9 is so vague. Don’t really get it… and that makes me uneasy about trying to decipher what the patterns might or might not mean. I hope that we get more clarity on this dataset at some point.

This is an interesting observation. So is the observation basically that folks in the groups you mention have an elevated risk of a non-covid death in the time window more than three weeks after the first vaccine shot, but before the second one? Something to mull over for sure.

Could you elaborate a bit more so that I can understand your thinking here? I’m not sure I follow.

As for the new blog on the all-cause mortality, there are some interesting bits there. In particular I am interested in the shape of the mortality curves. It would be interesting to correlate them against the vaccine rollout - more vaccine → more death would be a strong signal to look for. As it is, the signal of deaths 3 weeks post first shot is already interesting and worth thinking more about.

One has to be a little careful evaluating a vaccine’s efficacy using all-cause mortality, though (as we have discussed before). The best a vaccine does is to prevent someone dying from that particular disease, effectively removing that disease as a cause of death. It leaves all other causes of death at baseline. Basically if all-cause mortality simply goes back to what it looked like without Covid, then the vaccines were very successful.

Things like this are important, because if one reason someone was an early candidate for vaccination was because they had a lot of comorbidities, then those comorbidities remain, and remain dangerous even if the risk of death by covid has been effectively removed. As one would expect people with more comorbidities to die (generally) at a higher rate than those without, this would then be reflected in the numbers more in the vaxxed populaton. This is not to say that I think this is in any way an explanation for the data presented - just that there is a lot to take into consideration when teasing this stuff out.

Lots to think about here. Cheers
PP

1 Like
#9

Hi PP

I don’t see that Table 2 and Table 9 should have the same level of detail, as Table 2 has several vaccination statuse to contend with, and Table 9 only one.
But OTOH, I’m not clear on what ‘vaccination’ Table 9 is referring to. Is it time since last jab, or what. It can’t be ‘any jab’, can it - you’d have to adjust for the number of jabs.

I wondered about this as well - does it mean people who only had the first shot. Or does it mean everyone who had the first shot. I presumed the former, as if it was the latter you wouldn’t be able to have vaccinated totals without multiply counting people who had more than one jab. But if it’s everyone who only had the first shot you’d have to consider that that might be setting up the count to increase artificially due to people that happened to die before their next shot.
Y’know PP, I think we’re going to end up with a load of good questions and zero answers…

PP: Could you elaborate a bit more so that I can understand your thinking here? I’m not sure I follow.

I meant that some admin-related glitch is causing the pile-up after exactly 3 weeks in the covid deaths. The Covid deaths in the vaccinated are probably indeed elevated in the first few weeks, with the glitch making it highest at 3 weeks. There is an issue there but I think it’s dwarfed by the apparent extra deaths in the non-covid deaths.

I’m a bit uneasy with the uncertainty in the meaning of the data categories. Maybe waiting for someone who knows the data is an idea…or an enquiry to ONS.
I agree some caution is needed before pronouncements…though TPTB cannot claim at the moment to have illustrated using data that the vaccine is a good idea. The onus should be on them - not on those with concern to ‘prove’ things are as bad as are suggested by the best data they can produce.

Cheers

1 Like
#10

Hahahaha! You might be right… On a more practical note, I have no idea how to ask for clarification on a dataset from ONS, do you?

If I get some time I’ll have a look at the time correlation of vaccine rollout and the deaths after the first shot. You never know, something interesting might very well turn up…

Good luck on your investigations. Let us know if you come to a conclusion.

Cheers

#11

There’s usually an email contact on the main page where the link to the datasets is
See “Contact details for this dataset …”

However, Excess Burden is on it. There’s now a note at the top of the blog page, expressing doubt at the meaning of the data - as you expressed PP - and they’ve contacted ONS. EB has tried to soldier on with the uncertainties meantime, but as it’s getting confusing I think I’ll wait for the mist to clear. Maybe @Twirlip will come out of it :slightly_smiling_face:

Excess Burden says their April 15 post still stands

But I’m going to ‘self-isolate’ from this one until I get the all-clear about the data :smile:
Good luck with your correlation if you pursue it
Cheers

1 Like
#12

Hi ED

That’s great, and I think it speaks very highly of Excess Burden’s bona fides that they are trying to get to the bottom of this subject and are holding off on making any claims about the data until then. Bravo - I hope they get some clarity.

Lol! Nicely put. It would be lovely if that were to happen. I’ll let you know if I get anywhere with the correlation stuff.

Cheers
PP

1 Like