Toby’s vax efficacy fraud

I found my favourite attempt by mathematicians who should know better on the web today to try, yet again, to undermine vax efficacy. It was this piece which tries to claim the lag in deaths will make the unvaxxed look worse and the vaxxed look better.

It has some mathematical foundation, but the rate of difference does not explain the 10x reduction in case rates, 40x in deaths, that we saw at the start of vaccine rollout. Plus, if it was a placebo, it would not explain why we now have 5–10x lower death rates despite having 2x the mobility.

So it is theory detached from reality, but crucially all data scientist worth there salt adjust for these lags in reporting, so it just wouldn’t happen.

I don’t deny vaxxed vs unvaxxed rates in the UKHSA data have gone from matching trial efficacy, as shown in the plots in these early reports, to parity, as the number of unvaxxed has dropped to <10%, but both these takes ignore the obvious elephant in the room, natural immunity.

Sadly for us the UKHSA has only been publishing the by age data since wk36, so we can’t plot the whole slide of the relative rates of cases, but here is a plot of the ratio of unvaxxed/vaxxed rates for the 7 weeks they were publishing them for.

The interesting age band is <18, where immunity is still very low. In the October spike there was in fact a final epidemic outbreak in this age group with 1/2 the cases being <18. So what this data actually shows is that the unvaxxed got a far faster outbreak than the vaxxed.

Given this effect will be playing out to a lesser degree in the older age groups, it means this background slide, of an increase in efficacy in the unvaxxed, I’ve been talking about will be less prominent in the data.

None the less, it is plain to see.

Where we can really see that this has been a strengthening in unvaxxed efficacy, not a collapse in vax efficacy is in deaths.

Here from the ONS are death rates 80+ by vaccination status, and we can see that between the Jan peak and now, there has been a huge improvement in unvaxxed efficacy, and virtually no change in vaxxed efficacy.

The likes of @toadmeister have used this increase in unvaxxed efficacy to strawman the argument, to claim instead that the vaccine is not working as well, like this piece on the Daily Sceptic, but it is a total misrepresentation of the reality of the data.

So what has really been happening?

Prior to the vax being rolled out, about 30% of the population had developed antibodies through infection. We know this for the ONS infection survey, and this has been observed to reduce the infection rate to 1/10th and deaths to 1/40th.

Similar to the numbers seen in vax efficacy trials for the vax.

Vaccination then charges ahead.

Now, 2/3 of the 30% (20%) don’t know they had C19. We know this as the ONS shows 16 million had antibodies post the Jan peak and pre vaccine effects, but at that point only 5 million people had been confirmed +ve by testing, so pretty accurate.

So of the 30%, 1/3 (10%) do know.

Now do they mainly choose to vax or not?

Seems likely mostly not, along with a whole load who felt they had symptoms and refused to test, true sociopaths.

Now the important part about the vax process that no one discusses, is that having it makes you switch from the unvaxxed cohort, to the vaxxed one. This changes the make up of the unvaxxed group, and we are looking for a something that might cause an increase in efficacy in the unvaxxed.

Likely most of the 20% who don’t know do switch, and most of the 10% that do know, don’t, along with the untested hardliners who know they’ve had it.

When we get to now, with 10% unvaxxed, you end up with most of the unvaxxed, say 3/4, being those that have had it. Add to this, the remaining unvaxxed who were unimmune have carried on having cases at 10x the rate of the vaxxed.

Basically this has made the unvaxxed as immune as the vaxxed.

Which is why they now seem to have similar case and deaths rates.

The ONS antibody survey figures above over the course of the year support the argument that 75%+ of the unvaxxed must have antibodies, as most cohorts 50+ by age peaked at 97–98% antibodies, so 2-3% unimmune, which is 3–4x the number that remain unvaxxed. Bear in mind, these ONS rates don’t account for wane, so the true peak was likely even higher.

Crucially, if you take UKHSA’s case and deaths rates on the vaxxed and apply them to the 75% unvaxxed and immune (as we’re arguing they have similar immunity), then subtract these from the unvaxxed case and death totals, the implied case and deaths rates in the remaining 25% unvaxxed and unimmune are 10x and 40x higher than those with immunity…

… entirely in line with vax efficacy.

You see, at the heart of @toadmeister’s fraud is the simple reality that efficacy trials compared rates in the vaxxed to those without immunity. So to make the comparison of the vaxxed and unvaxxed the same you would need to remove from the unvaxxed cohort all those with immunity by infection, as I’ve just done.

They don’t.

This sleight of hand allows them to massively dilute the case and deaths rates with 3/4 of them being as immune as the vaxxed. Worse, they try to imply we could have got to this place with natural immunity, entirely overlooking this would have incurred an IFR of 0.7%, and about 1 million more admissions.

What a total crock.

I cover this in minute mathematical detail in this walk through video, along with a few other glaring errors in their workings. Bottom line, immunity, whether vax or from infection, reduce case to 1/10th and deaths to 1/40th.