The Full Risk–Benefit Balance Behind the First Covid-19 Vaccine

Health Secretary Robert F. Kennedy, Jr.’s recent decision to terminate the Covid-19 Emergency Use Authorization (EUA) declarations¹ invites reconsideration of the evidence that launched this extraordinary regulatory chapter. The FDA’s first EUA for a Covid-19 vaccine rested on its determination that the known and potential benefits of the Pfizer–BioNTech vaccine outweighed its known and potential risks. Yet the pivotal trial report did not bring those benefits and risks together within a common quantitative framework.

Better late than never: more than five years later, I did—and what I found was startling. 

At first glance, the famous 95% efficacy against protocol-defined, laboratory-confirmed symptomatic Covid-19—based on eight versus 162 cases in a trial that randomized 43,548 participants—seems highly promising. But when this finding is considered alongside the broader set of participant-relevant outcomes (dispersed across the trial publication, its Supplementary Appendix, and the contemporaneous FDA review materials), the overall clinical picture looks markedly different. 

Depending on the counting period used, the vaccine group had 2–8 fewer cases of severe Covid-19, whereas the safety data showed numerical excesses of 4–101 participants across clinically consequential adverse-event categories. These categories are not directly comparable, and some may overlap. Even so, the evidence did not establish that the vaccine’s clinically consequential benefits outweighed its potential harms. Nor did the trial establish that vaccination reduced the overall burden of Covid-like symptoms, let alone that it reduced person-to-person transmission. Crucially, this conclusion does not rest on hindsight: the relevant evidence was already before the FDA when it issued the EUA.

The full basis for this conclusion is presented in my comprehensive reassessment of the evidence, currently available as a preprint.² That reassessment is necessarily detailed because the relevant data are complex, scattered across multiple documents, and subject to important methodological and interpretive qualifications. The present column is not intended to replace that analysis, but to distill several of its central findings and implications for a broader readership.

1. What Did the 95% Efficacy Figure Actually Measure?

The gap between the headline figure and the overall clinical picture begins with the narrow endpoint and brief followup behind the 95% efficacy estimate. That figure applied only to protocol-defined, laboratory-confirmed symptomatic Covid-19 over an average of approximately 44 days. The trial did not assess asymptomatic infection or person-to-person transmission, and the publicly available record allows SARS-CoV-2 testing to be reconstructed for no more than 8.6% of participants. It therefore could not establish whether vaccination reduced the overall risk of acquiring or transmitting the virus, or how long its protection against symptomatic disease would last.

This gap later attracted considerable public attention. During a hearing in the European Parliament, a Pfizer executive acknowledged that the vaccine had not been tested for its effect on transmission before entering the market, explaining that the company had needed to “move at the speed of science.”³ Yet the facts discussed in that exchange had already been documented before the rollout. The trial’s narrow endpoint and brief followup were clear from the published record, and the FDA explicitly acknowledged the absence of direct evidence on asymptomatic infection and transmission when it issued the EUA.

These facts had far-reaching implications. At the time, vaccine mandates and passport systems were presented as tools to reduce transmission in public spaces and protect others—especially older and vulnerable people. Yet those claims could not be grounded in the trial’s central finding, which concerned only short-term protection against laboratory-confirmed symptomatic Covid-19. The trial therefore provided no direct basis for using vaccination status as a marker of reduced transmission risk, let alone for treating that status as valid for six months, as some passport systems did.

We should not have been so surprised, then, when it became clear after the EUA that vaccinated people continued to become infected in large numbers and that major waves of transmission persisted even after most adults had been vaccinated. This waning protection against infection was soon documented empirically as well. For example, a nationwide study from Qatar reported that effectiveness against any documented SARS-CoV-2 infection was negligible during the first two weeks, peaked at 77.5% during the first month after the second dose, and then declined progressively.⁴ 

Unfortunately, the consequences of this evidentiary gap were concrete: vaccine-passport policies restricted individual rights and placed considerable pressure on people to undergo a novel medical intervention, even though the evidence available at authorization had not established that vaccination reduced transmission or protected others. 

2. Did Vaccination Reduce the Overall Burden of Symptoms?

Not only did the trial fail to establish that vaccination prevented infection or transmission; it did not even establish that vaccinated participants experienced fewer Covid-like symptoms overall.

That may sound surprising. Let me explain.

The famous 95% efficacy figure rested on just 170 laboratory-confirmed cases of symptomatic Covid-19: eight in the vaccine group and 162 in the placebo group. Yet another 3,410 suspected symptomatic cases were recorded without laboratory confirmation—1,594 in the vaccine group and 1,816 in the placebo group. When these much larger numbers are considered, the contrast between the groups becomes far less dramatic: the vaccine group recorded only 12.2% fewer suspected symptomatic cases.

Furthermore, even this modest difference cannot, by itself, be interpreted as evidence of a net symptomatic benefit. Vaccination itself produced a substantial burden of symptoms, including fatigue, headache, fever, chills, and muscle pain—the same kinds of symptoms that can resemble Covid-19. When these reactions are taken into account, the imbalance is striking: adverse events judged related to the intervention were reported in 4,484 vaccine recipients, compared with 1,095 placebo recipients—more than four times as many.

Finally, case detection may not have been fully symmetrical. During the first week after each injection, investigators were instructed to use clinical judgment in deciding whether symptoms overlapping with expected vaccine reactions (e.g., fever) warranted PCR testing. This introduced a risk of functional unblinding and may have caused some early infections in vaccine recipients to go undetected. Although those cases fell outside the primary efficacy window, the missed infections could still have influenced the later comparison by providing short-term immunity that was then attributed to vaccination.

Taken together, the relatively small difference in suspected cases, the substantial burden of vaccine reactions, the overlap between those reactions and Covid-like symptoms, and the possibility of unequal case detection mean that the trial did not establish a reduction in participants’ overall symptomatic burden.

3. Did the Trial Show a Favorable Balance on Severe Outcomes?

With all due respect to the reported efficacy against symptomatic Covid-19 (challenged above), the most consequential risk–benefit question before authorizing a novel vaccine for use on such a vast scale was whether it would prevent enough severe disease and death to outweigh the serious harm it might cause. This basic principle of medical decision-making is even more important today, now that we know the vaccine did not provide durable protection against infection.

“Based on the totality of scientific evidence available,” the FDA explained in its EUA Review Memorandum, “it is reasonable to believe that the Pfizer-BioNTech COVID-19 vaccine (BNT162b2) may be effective in preventing such serious or life-threatening disease or condition that can be caused by SARS-CoV-2.”⁵ Sharp-eyed readers may notice the three layers of epistemic qualification in this standard regulatory formulation: “reasonable,” “believe,” and “may be.” But what was the actual evidence? After all, no Covid-19-related deaths occurred in either trial group.

The trial researchers emphasized that severe Covid-19 occurred in one vaccine recipient and nine placebo recipients, calling this “preliminary evidence of vaccine-mediated protection against severe disease.”⁶ But much of that apparent protection arose before participants were considered fully vaccinated, including cases that occurred too early to be reasonably attributed to vaccine-induced immunity. This was precisely why the same early period was excluded from the primary efficacy analysis of symptomatic Covid-19: including it would have substantially lowered the reported 95% efficacy estimate.

To find out what happened during that prespecified efficacy period, I had to turn to the FDA review documents. There, the picture was far less impressive: severe Covid-19 occurred in one vaccine recipient and three placebo recipients—a difference of only two cases among more than 43,000 participants. Unsurprisingly, this difference was not statistically significant. Nor did the later six-month followup⁷ show any mortality benefit (if anything, the subsequent open-label period raised a possible safety signal that could no longer be evaluated within a randomized comparison).

On strictly evidentiary grounds, the discussion could have stopped there: the trial did not demonstrate that vaccination reduced either severe disease or death. Those who argue that the study was simply too small to detect such benefits must confront three further points:

First, even the observed effect was tiny: approximately 9,200 people would have needed to be vaccinated to prevent one severe case during the trial’s brief followup.

Second, the trial did not show that vaccination reduced the risk of severe disease among participants who had already become infected. This conditional question became especially important later, once it was clear that the vaccine provided poor protection against infection and the claim shifted to protection against severe disease (see also my previous Brownstone columns on this topic).^8,9 Yet the relevant comparison provided no clear evidence of such protection—and during the prespecified efficacy period, the numerical direction was actually reversed.

Finally, the small-sample argument must be applied symmetrically to potential harms. If a larger sample might have rendered a small benefit statistically significant, it might also have rendered the observed safety imbalances significant.

What, then, happened on the safety side?

The numerical imbalances pointed in the opposite direction. Compared with the placebo group, the vaccine group included four more participants with serious adverse events judged related to the intervention, seven more with related events leading to withdrawal, 15 more with serious adverse events over the full reported period, 22 more with serious adverse events during the prespecified safety window, and 101 more with severe adverse events.

Of these imbalances, the prespecified safety-window comparison is especially important: 103 vaccine recipients experienced serious adverse events, compared with 81 placebo recipients. Yet these figures were absent from the main trial article and its Supplementary Appendix, appearing only in Table 23, on page 87 of Pfizer’s 92-page FDA submission.¹⁰

4. What Can the Trial Tell Us about the Emergency Itself?

The safety findings also invite a broader look at the public-health context in which the vaccine was authorized. A randomized clinical trial cannot determine whether a public-health emergency existed or measure the overall severity of a pandemic. But if its findings are meant to guide decisions for the wider public, the trial population must provide a meaningful basis for generalization. Within that population, we can therefore ask how severe Covid-19 compared with the broader burden of serious illness observed during followup.

Across both groups, 379 participants experienced at least one severe adverse event and 237 experienced at least one serious adverse event, whereas only ten cases of severe Covid-19 were reported after the first dose—several of which were classified as severe solely because of borderline oxygen-saturation readings. These categories may overlap and cannot be treated as clinically equivalent. Still, among this predominantly healthy trial population, severe Covid-19 represented only a small part of the broader burden of serious illness observed during followup.

This does not deny the seriousness of the pandemic or the existence of a genuine public-health emergency. It does, however, caution against a form of public-health tunnel vision in which one disease is isolated from the wider landscape of human illness and allowed to override every competing concern. Such tunnel vision can have far-reaching consequences, especially when it drives sweeping measures such as lockdowns, prolonged isolation, and the shutdown of social and economic life.

Conclusions: The Risk–Benefit Comparison the Trial Never Made

The emergency context may explain why regulators were willing to act quickly under uncertainty. It does not answer the question on which the EUA ultimately depended: did the pivotal trial establish that the vaccine’s clinically consequential benefits outweighed its potential harms? 

The present analysis revealed a risk–benefit balance far less favorable than the headline 95% efficacy figure suggested. As noted at the outset, depending on the counting period used, the vaccine group experienced 2–8 fewer cases of severe Covid-19, whereas the safety data showed numerical excesses involving 4–101 participants across clinically consequential adverse-event categories. These categories are not interchangeable, some may overlap, and they cannot simply be added together or equated one-for-one. Even so, the overall quantitative picture is difficult to reconcile with a clearly favorable balance. This is the comparison the original trial report never presented.

Some may argue that a larger trial would have detected benefits that this trial was too small to establish. That possibility cannot be ruled out—but it must be applied symmetrically. If uncertainty on the efficacy side is counted as a potential benefit, unresolved numerical excesses on the safety side must also be counted as potential risks. By the same logic, model-based claims that vaccination saved millions of lives cannot legitimately extrapolate such life-saving effects from the trial’s primary efficacy result alone. On the evidence actually observed, the trial provided no affirmative empirical basis for concluding that the overall clinical balance favored vaccination.

Crucially, this conclusion does not depend on hindsight. It emerges from reconstructing the evidence available when the first EUA was issued—evidence scattered across the trial publication, its Supplementary Appendix, and the FDA review documents, but never brought together within a common quantitative framework.

More than five years later, revisiting this evidence is essential if we are to avoid repeating the same mistakes when the next public-health emergency arrives. An emergency may justify speed and action under uncertainty, but it cannot justify counting uncertain benefits generously while discounting uncertain harms.

Acknowledgments

The author is grateful to Dr. Yaffa Shir-Raz and Dr. Shay Zakov for their help in locating and navigating the trial-related materials, including the relevant FDA documents, and for their thoughtful discussions on the interpretation of the evidence.

The present reassessment also builds on earlier critical analyses of the pivotal trial (e.g., ^11-16). These works examined complementary aspects of the evidence (e.g., the interpretation and clinical scope of the efficacy findings, the distinction between relative and absolute effects, and the safety and mortality outcomes) and helped inform the present analysis.

Bibliography

1. HHS secretary Kennedy signs COVID-19 emergency use authorization declaration terminations. https://www.hhs.gov/press-room/hhs-ends-covid-19-emergency-use-authorizations.html. Updated 2026. Accessed Jul 7, 2026.
2. Ophir Y. A comprehensive reassessment of the risk-benefit balance in the pivotal Pfizer-BioNTech COVID-19 vaccine trial. 2026:1–42. https://www.researchgate.net/publication/410683450_A_Comprehensive_Reassessment_of_the_Risk-Benefit_Balance_in_the_Pivotal_Pfizer-BioNTech_COVID-19_Vaccine_Trial. doi: 10.13140/RG.2.2.19594.84169.
3. Chung F. Pfizer did not know whether covid vaccine stopped transmission before rollout, executive admits. news.com.au. 2022. https://did.li/QVEgT. Accessed August 7, 2025.
4. Chemaitelly H, Tang P, Hasan MR, et al. Waning of BNT162b2 vaccine protection against SARS-CoV-2 infection in Qatar. N Engl J Med. 2021;385(24):e83.
5. FDA. Emergency use authorization for Pfizer-BioNTech COVID-19 vaccine: Review memorandum. US Food and Drug Administration (FDA). 2020. https://www.fda.gov/media/144416/download. Accessed December 19, 2025.
6. Polack FP, Thomas SJ, Kitchin N, et al. Safety and efficacy of the BNT162b2 mRNA covid-19 vaccine. New England Journal of Medicine. 2020;383(27):2603–2615. https://www.nejm.org/doi/full/10.1056/NEJMoa2034577. Accessed Jun 29, 2025. doi: 10.1056/NEJMoa2034577.
7. Thomas SJ, Moreira ED, Kitchin N, et al. Safety and efficacy of the BNT162b2 mRNA covid-19 vaccine through 6 months. N Engl J Med. 2021;385(19):1761–1773. https://doi.org/10.1056/NEJMoa2110345. doi: 10.1056/NEJMoa2110345.
8. Ophir Y. Did covid vaccines really save millions? Brownstone Journal. 2025. https://brownstone.org/articles/did-covid-vaccines-really-save-millions/. Accessed Sep 6, 2025.
9. Ophir Y. The final brick in the vaccine efficacy narrative. Brownstone Journal. 2023. https://brownstone.org/articles/final-brick-vaccine-efficacy-narrative/. Accessed Jun 20, 2025.
10. Pfizer-BioNTech. Pfizer-BioNTech COVID-19 vaccine (BNT162, PF-07302048): Vaccines and related biological products advisory committee (VRBPAC) briefing document (meeting date: December 10, 2020). U.S. Food and Drug Administration (FDA). 2020. https://www.fda.gov/media/144246/download. Accessed Feb 4, 2026.
11. Doshi P, Godlee F, Abbasi K. Covid-19 vaccines and treatments: We must have raw data, now. BMJ. 2022;376:o102. https://www.bmj.com/content/376/bmj.o102. Accessed Sep 18, 2025. doi: 10.1136/bmj.o102.
12. Fraiman J, Erviti J, Jones M, et al. Serious adverse events of special interest following mRNA COVID-19 vaccination in randomized trials in adults. Vaccine. 2022;40(40):5798–5805. https://www.sciencedirect.com/science/article/pii/S0264410X22010283. doi: 10.1016/j.vaccine.2022.08.036.
13. Doshi P. Will covid-19 vaccines save lives? current trials aren’t designed to tell us. BMJ. 2020;371:m4037. http://www.bmj.com/content/371/bmj.m4037.abstract. Accessed August 7, 2025. doi: 10.1136/bmj.m4037.
14. Olliaro P, Torreele E, Vaillant M. COVID-19 vaccine efficacy and effectiveness—the elephant (not) in the room. The Lancet Microbe. 2021;2(7):e279–e280. https://www.thelancet.com/journals/lanmic/article/PIIS2666-52472100069-0/fulltext?utm_source=chatgpt.com. Accessed Jul 15, 2026. doi: 10.1016/S2666-5247(21)00069-0.
15. Benn CS, Schaltz-Buchholzer F, Nielsen S, Netea MG, Aaby P. Randomised clinical trials of COVID-19 vaccines: Do adenovirus-vector vaccines have beneficial non-specific effects? Iscience. 2023;26(5). doi: 10.1016/j.isci.2023.106733.
16. Zakov S. Pfizer’s covid-19 vaccine clinical trial review. 2022. https://pandata.org/covid-19-vaccine-trial-review/. Accessed Jun 9, 2025.

---

Join the conversation:

[Sassy_Follow_Icons social_networks="x=https://x.com/brownstoneinst,facebook=https://www.facebook.com/brownstoneinst,telegram=https://t.me/brownstoneinst, truth_social=https://truthsocial.com/@brownstoneinst, linkedin=https://www.linkedin.com/company/brownstone-institute/, gettr=https://gettr.com/user/brownstoneinst, gab=https://gab.com/brownstoneinst" shape="round" style="background-color:#ac3900;]

---

Published under a Creative Commons Attribution 4.0 International License
For reprints, please set the canonical link back to the original Brownstone Institute Article and Author.