The Pfizer Vaccine Trial Was Not Double Blind

But it's actually worse than it sounds

As people being digging into the detailed records of the Pfizer/BioNTech COVID-19 vaccine trial released by the FDA, there is something they need to understand about how the trial was run.

Recently, people like Ed Dowd have pointed out that, in light of the scandalous malfeasance at one of the Pfizer vaccine trial sites exposed in the British Medical Journal, it would have made it quite easy for Pfizer to commit fraud by re-jiggering results coming from that site. The reason is that the identities of vaccine and placebo recipients was pretty much “out in the open” at that site. It wasn’t “double blind,” meaning an experiment where neither the researchers nor the participants know who is getting the treatment and who is getting the placebo.

What Ed and others haven’t seemed to realize yet is that it wasn’t just a problem at an isolated site: the entire trial wasn’t even designed to be double blind in the first place. It was billed as an “observer blinded” study, meaning ostensibly that the researchers don’t know who is getting the treatment but the recipients do or might. In this case, Pfizer said that the appearance of the liquid in the vaccine vials was different than the placebo vials, so the nurses delivering the injection would know the difference, and the trial subjects might get tipped off or figure it out on their own. The nurse wasn’t supposed to tell any of the other researchers who was who. And if none of the researchers know, then they can honestly call it an “observer blinded” study.

Except knowledge of who was in the treatment and control groups went far beyond the nurses administering the shots. According to the Pfizer/BioNTech study protocol, the study manager and the clinical research associate at each study site were also unblinded, along with a team of researchers, including people responsible for reviewing adverse events and protocol deviations (see p.49):

[UPDATE MAY 25, HERE IS A PAGE FROM ONE OF THE PFIZER DOCUMENTS RELEASED BY THE FDA AS PART OF THE FOIA REQUEST THAT SHOWS THE SAME THING:]

Why does that matter? Recall that there was a much higher rate of “protocol deviations” among the treatment group than the placebo group (331 in treatment vs. 61 in placebo):

That kind of lopsided outcome is likely to happen by chance less than one in a million times (I did the math). What Pfizer will tell you is that “most” of the protocol violations are due to mistakes with product administration, for example administering a higher dose. But even if that is true, “most” is not all. And the fact that the key decisionmakers at each site were not blinded, along with the staff responsible for reviewing protocol violations and adverse events, opens the study up to all kinds of opportunities for fraud.

For example, they see that somebody in the treatment group got COVID or had a severe adverse event, so they decide to remove them and chalk it up to a protocol violation. Or the reverse: they see somebody in the placebo group had a protocol violation but they got COVID or had a severe adverse event, so let’s keep them in the study because, well, this particular violation wasn’t really that bad. We can imagine other types of fraud: a person’s adverse event is minimized or re-classified if it is known they received the vaccine, and/or the opposite if it is known they are in the placebo group. The sky’s the limit, really.

Yes, the protocol makes all kinds of promises about how this unblinded analysis team will be insulated from the other researchers and their reason for being is just to share data with the unblinded data monitoring committee (DMC). But it would be naïve in the extreme to take the word of perhaps the world’s most prolific corporate career criminal, Pfizer, especially since we saw the way they’ve misreported and tried to cover-up 12-year-old Maddie de Garay’s severe reaction to the vaccine.

One excuse given for this flagrant violation of double-blind studies seems to be that since the trial had to be done at “warp speed,” they needed to analyze the data simultaneously. They couldn’t wait until after the study was completed. This contradicts what we’re told, which is that no corners were cut even though they developed and tested a vaccine in record time. (Of course studies like this have unblinded DMC’s for safety monitoring during the study, that is common practice, but the extent of unblinding here is not.)

Here they discuss how the “unblinded statistical team” would perform interim analyses at various points throughout the trial:

One rationale for this is that they wanted to be sure they were hitting certain pre-defined early signals of efficacy based on differences in rates of COVID cases between the treatment and control groups. Failure to hit those targets along the way would be an indication that the vaccine was not working and they would need to go back to the drawing board. This is part of the time pressure they were under. It’s hard to believe this pressure was not felt by the unblinded staff making crucial decisions along the way.

Another excuse is that they were worried about antibody-dependent enhancement (ADE), where the vaccine might cause people to have more severe COVID, as in this section on page 64:

This is why they say that the unblinded person reviewing AE’s was supposed to look for cases of severe COVID. But of course anyone reviewing a patient’s records would be able to see all of their AE’s, not just cases of severe COVID.

We are taught that double-blind studies are important, because if the researchers know who is in the treatment group, they might act differently towards people in the treatment and placebo group, thereby subtly affecting the results. Likewise, if the subjects know, they might behave in subtly differently ways that could affect the outcome, and also it opens up the results to be confounded by the placebo effect.

That all points to subtle, unconscious behaviors and effects. But given the outright and well-established criminality of pharmaceutical companies, we actually need true double-blind studies as a buffer against outright fraud. And even though they are masters of manipulating data from even properly run double-blind trials, some protection is better than none.