STORY UPDATED: check for updates below.

Number Updated

Did Pfizer under-report the number of deaths in the 40,000-person clinical trial of the company's COVID-19 vaccine? No, that's not true: The claim ignores the fact that, over the course of an ongoing clinical trial for a new vaccine, numbers change when data are updated for multiple reporting deadlines. The claim is based in part on an early and informal report on the clinical trial, from a PDF-sharing site, rather than the peer-reviewed and edited version of the report that was published in the New England Journal of Medicine (NEJM). At the time this fact check was written, no "expression of concern" or other indicator of error had been attached to the study, which had been pored over by Pfizer competitors and credentialled scientific experts for 12 months.

The claim also leaves out the Food and Drug Administration's (FDA) published declaration that it found none of the deaths during the clinical study were attributable to the vaccine. Real-world studies involve volunteers of a wide range of ages and medical backgrounds who die in approximately the same proportion as the population at large, which is why medical experts have warned the public against assuming the coincidence of a post-vaccine death is evidence of vaccine-caused death.

The claim appeared in a November 16, 2021, post (archived here) in the Unreported Truths newsletter by spy novelist Alex Berenson titled "More people died in the key clinical trial for Pfizer's Covid vaccine than the company publicly reported." It opened:

Pfizer told the world 15 people who received the vaccine in its trial had died as of mid-March. Turns out the real number then was 21, compared to only 17 deaths in people who hadn't been vaccinated.

Users on social media only saw this title, description and thumbnail:

More people died in the key clinical trial for Pfizer's Covid vaccine than the company publicly reported

Pfizer told the world 15 people who received the vaccine in its trial had died as of mid-March. Turns out the real number then was 21, compared to only 17 deaths in people who hadn't been vaccinated.

Berenson wrote:

Pfizer somehow miscounted - or publicly misreported, or both - the number of deaths ... Though the absolute numbers are small, overall deaths were 24 percent higher among vaccine recipients. The update also shows that 19 vaccine recipients died between November and March, compared to 13 placebo recipients - a difference of almost 50 percent.

In addition to accusing Pfizer and independent scientists of putting out false data, Berenson criticizes Pfizer for, after the early phases of the study were completed, vaccinating placebo recipients. He wrote that made it impossible to capture long-term comparison of those who were vaccinated and those who received a placebo.

Infectious disease expert: Numbers change in an ongoing study. That's the whole point.

It is routine to have cut offs where the database is locked, nothing in and nothing out, it is audited, and queries are investigated to make sure everything is accurate before going to the fda or some other regulatory agency or when there is a milestone reached.

Incomplete quote of FDA findings

Berenson only used part of a sentence about deaths in an FDA report, calling it "stunning." Here's what Berenson published:

(Source: alexberenson.stubstack.com screenshot taken Mon Dec 6 at ‏‎22:49:25 2021 UTC)

Thomas pointed out that Berenson left out the important next sentence, which notes those deaths were thought to be coincidental to, not caused by vaccination. The following screenshot includes yellow highlighting on the sentence Berenson left out.

(Source: FDA "Summary Basis for Regulatory Action - Comirnaty screenshot taken Mon Dec 6 at 23:08:19 2021 UTC)

Deaths are no surprise in a large-scale, long-term study of a vaccine

Thomas said there will naturally be deaths of study participants over the course of a study. Research experts at the FDA and elsewhere have, since before the vaccine was released, cautioned the public not to fall prey to the correlation/causation fallacy.

When you have a trial enrolling over 40000 volunteers including people up to 85 years old, unfortunately there will be deaths during the trial.

Researchers put volunteer safety ahead of long-term data

Thomas said the research team discussed the ethics of witholding the vaccine -- once the study was concluded -- from placebo recipients. Thomas wrote that after consulting with an ethicist, the team decided to offer the vaccine to placebo recipients:

There is a moral and ethical imperative to do so based on the risk to the placebo recipient and the clinical benefit of the intervention. You should not be at the front of the line as a volunteer but you should also not be at the back.

Pfizer spokesperson responds to Berenson's claims

In a December 23, 2021, email to Lead Stories, Pfizer senior manager for science media relations Kit Longley -- after talking to experts at Pfizer -- explained how data are parsed, and when, in the course of an ongoing study.

Here, with links added by Lead Stories to help readers see key documents, is what Longley wrote:

The data cutoff was 13Mar2021 as disclosed by the FDA summary basis of approval and the NEJM article. Since participants were unblinded before the study was completed, we analyzed the data from dose 1 to the unblinding date to provide a true placebo-controlled analysis as required by FDA and then looked at the data after unblinding participants for subjects originally randomized to vaccine and placebo. To obtain the total number of deaths one needs to account for all the periods we analyzed.

She notes that Berenson's claim that deaths were hidden itself ignores details from the New England Journal of Medicine report on the study, concerning participants who were HIV-positive. Longley wrote in her email:

FDA is correct that as of the cutoff date of 13Mar2021 there were a total of 21 BNT162b2 deaths and 17 placebo deaths which include deaths in HIV+ cases. The protocol-defined analysis excluded HIV+ cases from safety analyses (this is disclosed in the NEJM article) but these deaths (1 BNT162b2 and 1 placebo) were reported in the article in the text for Figure 1 where it was noted that HIV+ deaths would not be discussed in the results section.

Lead Stories notes, for reader convenience, the 21 deaths referred to by Longley in the following passage of her email are recorded on page 23 of the FDA summary of its basis for approval of Pfizer's vaccine.

There were 15 BNT162b2 deaths and 14 placebo deaths in the blinded placebo-controlled period for HIV negative participants as appropriately disclosed in the NEJM article (text and Table S4). After unblinding there were an additional 3 participants originally randomized to BNT162b2 who died after unblinding and 2 placebo participants who subsequently were vaccinated with BNT162b2 and died at some point thereafter. Their 5 cases were disclosed in the NEJM article. Hence for BNT162b2, 15 deaths in blinded period + 1 HIV death + the 5 after unblinding = 21 cases as disclosed in FDA's summary basis of approval. All the BNT162b2 deaths were disclosed in the NEJM article.

Pfizer, Longley wrote, accounts for trial participants' deaths as follows:

For placebo participants we have 14 deaths in the blinded placebo-controlled period, 1 HIV+ death (disclosed in Figure 1) and 2 additional placebo deaths that occurred after unblinding but before receipt of BNT162b2. This adds up to 17 placebo deaths as disclosed in the FDA summary basis of approval. The later 2 placebo deaths were not mentioned in the NEJM article because they occurred after unblinding.