Thread by @Jikkyleaks

NEW 🧀 ALERT...

Analysis of the antibody testing data from the #Pfizerdata dump shows that their "too good to be true" graph - and the famous "95% prevention of infection" claim cannot be real.

You see, the claim was that 162 people in the placebo group got #covid19 *infection* but only 8 in the BNT162b2 group - a 95% reduction.

So was there another way to test infection rates?

Yes. N (nucleocapsid) antibodies.

So since the #Pfizerdump and #site4444 discovery a few of us have been beavering away looking at their own data - which is a mess - and trying to corroborate it.

Here is the N-antibody data from their "adva" file

[warning - it takes some work to get this data]

Note that both groups are similar (we have checked they are not statistically different) EXCEPT in the group which were NEG for N-antibody at the start of the trial, and POS for N-antibody at Visit 3 (1 month after dose 2)

i.e. they were infected with #covid19 in that time

That group (NEG->POS) reflect the groups that got infected with #SARCOV2 during the study period.

Well that's interesting... because the number in the placebo group is similar to the magical 162, but instead of 8 in the vaccine group - there are 75!

On the face of it the vaccine is still "working" (just) because the vaccine efficacy here is about 53% - nowhere near 95%.

But it's worse, because the vaccinated don't produce N-antibody at the same rate as the unvaccinated.

It's in this paper...
medrxiv.org/content/10.110…

In fact the rate of N-antibody between vaccinated ( with mRNA) and unvaccinated who were known to have #covid19 infection was 40% vs 93%

That is, the vaccinated produce N-antibodies 2.3x less often during infection than the unvaccinated.

Which means we have to adjust the number of patients who tested positive at the 1 month post-vaccination point upwards by 2.3x giving us this:

At best there is NO difference between the groups (the treated group seem to do worse, but it's not significant)

In fact, anything over 130 in the Bnt162b2 group here would mean there was NO significant difference in documented infection rates (chisq p<0.05), so even if the multiplication was a conservative 2.0x instead of 2.3x, there would be no difference
@JesslovesMJK

Presumably they thought nobody would notice. And they could claim that there was a 95% reduction in infection rate - based SOLELY off a PCR test that they controlled in their own lab.

Unfortunately, we did. Their own data says that was false.

Here's the ADVA data file (zipped .csv) for those that are really interested in looking at this for themselves. Converted from the relevant xpt file at ICANdecide.org

files.catbox.moe/i544mb.zip

(h/t @joshg99 @ChrisCottonStat )

Update: Because there are a few people making the same mistake, I'll try and clarify. The sponsor only ever claimed that there was a 95% reduction in "cases" which they defined as being a positive PCR test conducted in their own lab.
@sonia_elijah

Archived archive.ph/9y5od