Do the recent 80k pages of Pfizer documents released really show vaccine efficacy was only 12%?

This week Pfizer released a new stack of documents related to the Pfizer SARS-CoV-2 vaccine, with >80,000 pages of material released. Here is a link to the documents.

In the past couple days there is a viral claim circulating social media that this data dump revealed the actual efficacy of this vaccine was 12%, not 95% as reported, purporting a massive conspiracy. Here are some active tweets repeating this claim:

These tweets and accompanying threads lack documentation of where exactly in the documents it is revealed that the actual vaccine efficacy for the Pfizer vaccine was 12% not 95%. Many of these twitter threads have a few people asking for such documentation, but without response or clarification.

Best as I can tell, this claim originates from a substack post by Sonia Elijah on May 3rd.

Her claims are based on the December 2020 Pfizer FDA briefing document for the Vaccines and Related Biological Products Advisory Committee (VRBPAC) meeting deciding the EUA on December 10, 2020. Incidentally, this document was transparently released to the public the week before the meeting, so is not new to this 80k page document dump this week.

In this SubStack post, she claims that the 95% VE computed based on the fact that 162 of placebo patients had confirmed (PCR+) COVID infections and the vaccine arm only 8 was misleading. She claims the "real VE" should be computed based on the "suspected but unconfirmed" COVID-19 cases in the study, which she claims is "buried" in the document (implying some sort of intentional obfuscation).

From the fact that there were 3410 total "suspected but unconfirmed" cases, with 1594 occurring in the vaccine group vs. 1816 in the placebo group, she estimates that the "true VE" should actually be reported as 1-1594/1816 x 100% = 12.2%. This is seemingly where the viral claim of 12% comes from. Now, if there were indeed 1594 COVID-19 cases in vaccine arm and 1816 in placebo, indeed she would be correct that the VE would be 12%. However, we have to look at what is meant by "suspected but unconfirmed" cases in the Pfizer/BioNTech trial.

If you look through the protocol, which again was transparently posted in its entirety , you see no explicit mention of a category of COVID-19 cases that are "suspected but unconfirmed," but from the protocol the implication is clear.

The only time the idea of "suspected" is used in the protocol is in the context of determining when a unplanned potential COVID-illness visit should be scheduled to give a SARS-CoV-2 PCR test based on reported symptoms. As described in Section 8.13, page 93/146, of the protocol, they did this by instructing subjects to immediately contact the site to set up an unplanned potential COVID-illness visit, ideally within 3 days, if they experienced any of the following symptoms anew that COULD indicate potential COVID-19: fever, cough, shortness of breath, chills, muscle pain, sore throat, loss of taste/smell, diarrhea, and vomiting. Anyone reporting one of these symptoms is a "suspected" COVID-19 case who is to report for an unplanned clinical visit to determine whether it meets the COVID-19 case definition or not.

The protocol clearly states (in the table on Section 1.3.2 on page 23, in Section 8.13.1 on page 95, and in Section 8.15 on page 97) that these unplanned potential COVID-illness visits are to include a nasal swab (PCR test) to determine whether the symptoms experienced by the participant fulfill the COVID-19 test definition. If the test is positive, they are a "confirmed" COVID-19 case, and by context it is clear that the others are "suspected but unconfirmed." Those who are "confirmed COVID-19 cases" are included in the primary efficacy analysis.

So these 3410 "suspected but unconfirmed" cases were people who any symptom mentioned in the list, which of course could come from many causes not just COVID-19 infections, triggering an unplanned COVID-19 visit (for which a PCR test was indicated), and for whom the SARS-CoV-2 test was not positive. Since obviously the SARS-CoV-2 vaccine was not intended to prevent all coughs, fevers, chills, sore throats, muscle pain, shortness of breath, vomiting, etc., from any cause, it would be ridiculous to include all reports of such common symptoms as COVID-19 cases for the purpose of computing Vaccine Efficacy. The only way one might propose this as appropriate is if they assume the false positive rate of the PCR test was >>95%. We of course have plenty of data demonstrating that the false negative rate of these tests is nowhere near that high; it was 4% in this FDA report for the precise PCR test used for the trial. Even if higher, it is not plausible to propose a false negative rate >>95%. And even if they wanted to make that argument, given that the trial was double-blinded, i.e. neither the participants or health care workers knew who received active vaccine and who received placebo, it would be impossible for the vaccinated would have 162 true positives and placebo 8 true positives if indeed the underlying rate of disease was indeed equivalent in the two arms.

So Sonia Elijah is making the same specious argument Peter Doshi did in his editorial in British Medical Journal in January 2021, which I refuted in detail in my blog post that same month. It seems she is making the mistake of not realizing that "suspected but unconfirmed cases" in the protocol refers to individuals receiving PCR tests because that had a symptom from a large pre-specifed list that triggered a PCR test, but for which the PCR test was negative. The authors of the FDA briefing document choosing the term "suspected but unconfirmed cases" for those with negative PCR tests was a poor choice that has contributed to this confusion and led many to misinterpret what it means.

So the argument made by Sonia Elijah suggesting the "real VE" of the Pfizer vaccines is 12% is an old one dating back to January 2021, and an erroneous one. The fact this claim is spreading virally around social media without confirmation or justification, and with very few questioning where it came from or whether it was a legitimate claim, shows how easy it is for unsubstantiated and false claims to quickly spread on social media, where they find a large set of receptive ears ready to believe and promote the claim without any documentation or justification, primarily because it feeds the narrative they believe and are promoting.

As I mentioned, it is not explicitly clear in these viral tweets from whence the claim of 12% efficacy comes. I'll keep my eyes opened and if someone posts a link for documentation other than the substack post I linked here, I'll evaluate and comment on that evidence.