Stanford led study: at least 20,000 global deaths from COVID jabs

New COVID Vax Paper Breaks with the Narrative

By trialsitenews , republished by permission

TrialSite Staff | Making Biomedical Research Evidence Accessible to All

Dec. 23, 2025, 5:30 a.m.

In a high-profile analysis published in JAMA Health Forum, a team from Stanford University, Università Cattolica del Sacro Cuore and the Fondazione Policlinico Universitario A. Gemelli IRCCS led by John P. A. Ioannidis set out to answer one of the most contested questions of the pandemic era: How many lives did COVID-19 vaccination actually save worldwide? The study estimates that COVID-19 vaccination averted approximately 2.5 million deaths globally between 2020 and 2024, with a wide uncertainty range of 1.4 to 4.0 million, and saved about 15 million life-years, though results depend heavily on modeling assumptions. Nearly 90% of deaths averted occurred in adults aged 60 years and older, while children, adolescents, and young adults contributed a negligible share of the total benefit. Compared with earlier pandemic models, these findings suggest a far more limited, age-concentrated mortality benefit, particularly during the Omicron period and among those vaccinated before first infection.

A Quiet but Explosive Admission

Dr. Ioannidis and colleagues make an unusually candid concession for a high-profile vaccine-benefit paper: they do not separate deaths averted by vaccine effectiveness from deaths caused by vaccine-related harms. In fact they explicitly acknowledge that randomized trial data are insufficient to quantify vaccine-associated mortality and that estimates derived from registries and observational sources carry “substantial uncertainty.”

This is not a minor caveat—it is a structural limitation of the analysis. The authors further note that, depending on ethical framing and risk aversion, a death caused by an intervention may not be considered equivalent to a death averted by it, particularly when harms cluster in specific subpopulations.

To put bounds around this uncertainty, the authors turn to eAppendix 2 (Supplement 1), where they restrict analysis to widely recognized and accepted fatal adverse events: thrombosis with thrombocytopenia after adenovirus-vector vaccines, myocarditis following mRNA vaccines (primarily in younger males), and deaths temporally associated with vaccination in highly debilitated nursing-home residents.

Using global administration data (~13.64 billion doses) and conservative risk assumptions, they estimate roughly 20,000 vaccine-associated deaths worldwide, while an independent extrapolation from Qatar’s national mortality review suggests a broader range of approximately 16,000 to 48,000 deaths.

Compared with the study’s central estimate of ~2.5 million lives saved, this supports the authors’ statement that vaccine-related deaths were “probably” about two orders of magnitude lower than benefits at the population level.

The critical word is “probably.” The authors explicitly state that these adverse-event death estimates carry “very large uncertainty” and emphasize that the margin between benefit and harm may be substantially smaller—or even reversed—in specific subgroups where risks are concentrated and benefits are limited, such as younger males or frail elderly residents.

In effect, the paper concedes that while COVID-19 vaccination likely reduced mortality overall, net benefit was neither uniform nor guaranteed across populations. A bombshell acknowledgment—confined to the supplement rather than the abstract—directly challenges any absolutist claim that vaccine harms were negligible or irrelevant and represents a rare moment of methodological and ethical candor in the COVID-19 vaccine literature.

What the Study Did—and Did Not—Do

Importantly, this study did not measure deaths directly. It did not compare vaccinated and unvaccinated people in real-world cohorts over time. Instead, it used a counterfactual modeling approach: estimating how many people might have died if vaccines had never existed, based on assumptions about infection rates, fatality risks, and vaccine effectiveness against death.

This distinction matters. The results are not observations; they are conditional estimates. Change the assumptions, and the outcomes change—sometimes dramatically.

The authors are transparent about this. Vaccine effectiveness against death was assumed to be 75% before Omicron and 50% during Omicron, based largely on observational studies rather than randomized trials. Infection was assumed to become nearly universal during the Omicron period in the absence of vaccination. Life expectancy adjustments relied on a correction factor meant to account for underlying illness—an area of active scientific debate.

None of these choices are unreasonable. But together, they create a structure where small errors compound into large numerical differences.

The Most Robust Finding: Age Matters—A Lot

Where the paper is strongest is not in its global totals, but in its age-stratified results.

Roughly 90% of all deaths averted occurred in people aged 60 and older. Children, adolescents, and young adults contributed virtually nothing to the total lives or life-years saved—often less than one-tenth of one percent. Long-term care residents, despite extremely high mortality risk, contributed relatively few life-years saved because of limited remaining life expectancy.

This finding is consistent across sensitivity analyses and aligns with what is already known about COVID-19 risk. It also quietly raises uncomfortable policy questions. If benefits were so concentrated by age, should vaccination strategies—and mandates—have been equally concentrated?

The authors do not answer that question directly, but the data point clearly in that direction.

John P. A. Ioannidis, Corresponding Author

Source: Stanford University

Where Confidence Drops

The weakest link in the analysis is the assumed vaccine effectiveness against death. Randomized trials were never powered to detect mortality differences, and observational studies are vulnerable to biases—particularly the well-documented “healthy vaccinee” effect. Even modest overestimation of effectiveness causes the modeled benefit to shrink rapidly.

Another unresolved issue is natural immunity. While the model adjusts for vaccination timing relative to infection, it does not symmetrically model protection in unvaccinated people who were previously infected—despite growing evidence that natural immunity provides substantial protection against severe disease and death.

Finally, although the authors acknowledge vaccine-related harms, adverse events are not incorporated into the primary equation. The study estimates gross benefit, not net benefit—especially relevant for younger, low-risk populations.

On vaccine harms:

“Our estimates do not separate deaths averted from VE vs deaths caused from vaccination-related harms. Some may argue that, depending on risk aversion and regret considerations, a death caused because of harm may not carry the same weight as a death averted because of efficacy. Adverse events from COVID-19 vaccines remain a contentious topic. Randomized trial data are very limited.²⁷ Estimates of risk from registries and other observations carry substantial uncertainty. However, as shown in eAppendix 2 in Supplement 1, the number of deaths due to widely recognized and accepted adverse events (thrombosis, myocarditis, deaths in highly debilitated nursing home residents) are probably approximately 2 orders of magnitude smaller than the overall benefit. Still, these harms are important to weigh against benefits in specific subpopulations where there they have the highest frequency and where risk-benefit may change or even get reversed.”

Critical Point of View

Covid Vaccine critic and scholar Raphael Lataster, BPharm, PhD formerly with University of Sydney argues that the Ioannidis et al. analysis represents a step in the right direction because it decisively reins in the inflated claims of earlier COVID-19 vaccine modeling studies—most notably those asserting that nearly 20 million lives were saved in a single year—by correcting a central error: overstated infection fatality rates (IFRs).

By applying lower, more empirically grounded IFRs, Ioannidis and colleagues arrive at far more conservative estimates, with sensitivity analyses suggesting as few as approximately1 million lives saved globally over four years. Lataster highlights that the study’s most robust contribution is its clear age stratification—showing that roughly 90% of any mortality benefit accrued to people aged 60 and older, with negligible benefit in children and young adults—thereby implicitly challenging blanket vaccine mandates in low-risk populations and aligning policy closer to biological reality.

While Lataster remains critical that even these reduced figures may still be overstated due to unresolved issues around vaccine effectiveness assumptions, “with vs from COVID” death classification, and healthy-vaccinee bias, he views the paper as an important corrective: it narrows the plausible benefit window, exposes the fragility of prior models, and moves the debate away from certainty toward measured, evidence-aware skepticism.

So, What Can We Say with Confidence?

This study does not prove that COVID-19 vaccines saved exactly 2.5 million lives. It does something more subtle—and arguably more important.

It shows that:

• Any mortality benefit was real but far smaller than early models claimed.

• Benefits were overwhelmingly concentrated in older adults.

• For younger populations, mortality benefit was minimal to negligible.

• The precise magnitude of benefit remains highly uncertain and assumption-dependent.

In that sense, the paper functions less as a final verdict and more as a course correction—a reminder that precision in numbers should never be confused with certainty in evidence.

The Bottom Line

For readers across the spectrum—from policymakers to clinicians to skeptics—the takeaway is not a single number, but a reframing:

COVID-19 vaccination likely reduced deaths, especially among older adults. But the scale of that benefit is far narrower than advertised, more age-dependent, and more uncertain than many public narratives have suggested.

That conclusion may not satisfy advocates on either extreme. But in science, restraint is often the most credible position of all.

---

TrialSite Evidence Strength Indicator™ (ESI 2.0)

Category

Score (0–10)

Rationale

Methodological Rigor & Risk of Bias: 5 / 10

Fully transparent but assumption-dependent modeling with no direct mortality validation.

Consistency & Effect Size: 5 / 10

Age-gradient is stable; absolute magnitude is highly parameter-sensitive.

External Validity & Applicability: 6 / 10

Global scope with uneven data quality and limited country-level actionability.

Human Consequence Index (HCI): 7 / 10

Strong ethical and policy relevance due to extreme age stratification of benefit.

Pluralism Index (PI): 8 / 10

Explicit engagement with uncertainty and critique of prior inflated models.

Transparency & Disclosure: 8 / 10

Assumptions, equations, and sensitivities clearly disclosed.

Net Benefit Assessment: 4 / 10

Gross benefits modeled; adverse events and natural immunity not integrated.

Total Evidence Strength (Weighted)

6 / 10 (≈ 60%)

Moderate-low confidence: directionally informative, numerically uncertain, and strongest for older populations.

Source: Ioannidis JPA, Pezzullo AM, Cristiano A, et al. Global Estimates of Lives and Life-Years Saved by COVID-19 Vaccination During 2020–2024. JAMA Health Forum. 2025;6(7):e252223.

© 2025 TrialSite News™. All rights reserved.

---