By Eyal Shahar at Brownstone dot org.

We are approaching the fifth anniversary of the trial that set the stage for the vaccination of billions by an experimental product — the Pfizer mRNA Covid vaccine (BNT162b2). There is no other scientific paper that affected so many within a few months of its publication.

Was it a well-designed trial? In many respects, it was not. But was it at least trustworthy as far as the main results are concerned? It is uncertain, and I am not alone in the camp of skeptics. Over the years, I have read various critiques, ranging from testimonies about poor conduct to questionable inference.

For those who are unfamiliar with the design, here is the essence. About 40,000 people were randomized to receive two doses of the Pfizer vaccine, 21 days apart, or two doses of a placebo. Descriptive statistics show well-balanced characteristics in the two arms of the study, with about 20,000 people in each. Participants reported symptoms between the first injection and the end of the follow-up. If they reported at least one of 10 Covid-like symptoms, a PCR test was conducted. If positive, the participant was classified as a Covid case on the date of the first reported symptom.

The main results are shown below.

Notice that we get the correct results from the count of events alone because the denominator (time at risk) was almost identical in the two arms. For example, (1–2/21)x100=90.5%.

My two cents on the results.

I will focus first on a narrow window — eight days in the follow-up time — shortly after the second dose. I call that period The Eight Days of Miracles because what happened at that time was miraculous. Vaccine effectiveness has dramatically increased in the blink of an eye: from 50% to 90%. Too good to be true?

If we accept that the Pfizer vaccine was highly effective in preventing symptomatic infection, we don't need more than one week of follow-up after dose 2. Does it matter whether a risk is cut by 90% or by 95%? Not really. Of course, that's if we trust that estimate of 90% effectiveness.

In those eight days, there were 19 more cases in placebo recipients than in vaccine recipients. All that it takes to change the effectiveness back to 50% is to find 10 more cases or so in about 20,000 recipients of the second dose. Do we have any plausible reason to assume that cases were missed in the vaccine arm of the trial (undercounting) in The Eight Days of Miracles?

We surely do.

Misattributing Covid Symptoms to Side Effects

As we all know, side effects were common, and they were much more common after the second dose than after the first. The table below shows the frequency of three symptoms that were also considered as Covid symptoms in the definition of a case.

We cannot add the percentages because a participant could have reported multiple symptoms. Nonetheless, with almost 20,000 people in the vaccine arm, these percentages translate to thousands of people whose symptoms were attributed to side effects of the second dose ("reactogenicity"). For example, over 2,000 vaccine recipients reported fever after the second dose.

Was Covid ruled out by a PCR test in every case?

No, it was not.

That's what we find in the protocol (section 8.13).

"During the 7 days following each vaccination, potential COVID-19 symptoms that overlap with specific events (ie, fever, chills, new or increased muscle pain, diarrhea, vomiting) should not trigger a potential COVID-19 illness visit unless, in the investigator's opinion, the clinical picture is more indicative of a possible COVID-19 illness than vaccine reactogenicity." (my italics)

In other words, a PCR test is left to the discretion of the investigator, with a clear guideline: it is a priori assumed not to be Covid. Indeed, of thousands of participants who reported such symptoms in those seven days, only a few hundred were tested and classified as "suspected but unconfirmed Covid." All others were not tested.

How do we know how many were tested?

There is an FDA ...