The Great Pandemic Rewrite Comes for Vaccines
John Ioannidis and the Order-of-Magnitude Problem That Reveals the Game
COVID revisionism has been with us almost since the pandemic began. With the elevation of COVID minimizers to leadership positions at HHS, however, those efforts have shifted into overdrive. What began as an attempt to downplay the severity of the disease itself has evolved into something broader: an effort to recast the pandemic as a public-health overreaction, in which interventions mattered little and vaccines accomplished far less than widely believed. Under RFK Jr.’s vaccine-hostile leadership, that rewrite has now set its sights on the most consequential intervention of all—COVID vaccination.
In a previous post, I examined the public-health benefits of the current U.S. vaccine schedule, estimating cases of acute and chronic disease avoided and deaths prevented. Taken together, the 15 routine vaccines on the schedule prevent roughly 80,000 deaths per year in the United States.
That analysis deliberately excluded the most controversial immunization of all: the COVID-19 mRNA vaccine.
Unlike other vaccines on the schedule, COVID vaccination was deployed in response to a rapidly evolving pandemic rather than an endemic disease. As a result, its benefits cannot be assessed in the same way. Instead, several major studies have attempted to estimate how many COVID deaths were prevented during the pandemic itself. Three such analyses converge on estimates ranging from 100 to 500 lives saved per 100,000 people per year, depending on geography and time period. Even the most conservative of these estimates implies that COVID vaccination prevented between one and three million deaths in the United States alone—more than thirteen times the mortality reduction associated with all other routine vaccines combined.
Taken together, these three analyses point to a clear conclusion: the COVID mRNA vaccines have produced an immense net public-health benefit. (A fourth analysis with a radically different estimate bears closer scrutiny and is discussed below.)
These numbers provide important context as the FDA, under the direction of Vinay Prasad and Tracy Hoeg, scours federal databases for possible adverse effects. Both were authors of deeply flawed cost-benefit analyses that helped drive the narrative of net harm from COVID vaccination. Meanwhile, a large British study of 42 million people found that young men had a substantially higher risk of myocarditis following SARS-CoV-2 infection than after vaccination and that the rare immunization-associated risk varied by vaccine type:
Current leadership at HHS has cited rare adverse reactions as justification for minimizing use of COVID mRNA vaccines and slashing federally supported mRNA vaccine research. This reasoning is backwards. When a vaccine saves millions of lives, the appropriate response to safety signals is more research: to refine formulations, adjust dosing, and improve targeting. Ending research because a highly effective vaccine has rare adverse effects is the opposite of evidence-based public health.
Yes, vaccine risks matter. But they must always be weighed against the very real risks of the diseases that vaccines prevent. (To see the cumulative public-health impact of vaccination, explore the vaccine schedule game and observe how quickly preventable illness, disability, and death accumulate.)
The Ioannidis Outlier Problem
Against this backdrop, the recent estimate by John Ioannidis et al. stands out.
It is an order of magnitude smaller than every other major analysis, despite covering a longer time period. That fact alone warrants scrutiny.
Rather than engage with the scale of this discrepancy, the paper relies on a set of assumptions: strong and durable infection-derived immunity, rapidly declining infection-fatality rates independent of vaccination, and limited validity of excess mortality as a proxy for COVID-attributable deaths. These assumptions are asserted rather than convincingly demonstrated, and each acts to reduce estimated counterfactual mortality in the absence of vaccination.
Rather than cataloguing the paper’s flaws in detail, consider a single number.
Ioannidis et al. estimate that COVID vaccination saved only 299 children’s lives worldwide.
For comparison, UNICEF estimates approximately 17,400 COVID deaths globally among people under 20 years of age. Combining UNICEF’s estimate with Ioannidis’ figure implies that COVID vaccination reduced global childhood COVID mortality by less than 2%.
Even after accounting for incomplete pediatric uptake and conservative estimates of vaccine effectiveness against mortality, the implied reduction in childhood COVID deaths is on the order of tens of percent, not 2%—placing Ioannidis’ estimate another full order of magnitude below plausible values, consistent with the pattern seen in his earlier pandemic predictions.
In sum, this paper reveals far more about the assumptions driving Ioannidis’ framework than about the real-world impact of COVID vaccines. Its conclusions can reasonably be set aside. That judgment is reinforced by the broader pattern of his pandemic-era estimates.
That pattern is not new. Ioannidis’ first specific prediction was that COVID-19 would cause approximately 40,000 deaths in the United States during the first year of the pandemic. The actual number exceeded 520,000, an error of more than an order of magnitude. That estimate, in turn, appears to have been based on his estimate of an infection fatality rate that disagreed with a team of global experts by a factor of 7.5. Those early misses foreshadowed this latest estimate, suggesting a pattern of repeated understatements of the risks of the disease and the benefits of intervention.
Throughout this period, the intimation has been that the consensus was wrong. But each one of those consensus estimates came from a different team of recognized experts using different methods. The extreme outliers, on the other hand, all came from the same source. At some point, repeatedly being an outlier ceases to suggest unique insight and steadily raises the probability of simply being wrong. When consistent divergence from the evidence becomes a pattern, that pattern itself becomes evidence.
Method note: Plausible range of pediatric COVID deaths averted
Estimates of pediatric COVID deaths averted must account for both incomplete vaccine uptake and vaccine effectiveness against mortality. A simple, conservative approximation of population-level impact is:
Population Mortality Reduction = vaccine coverage x vaccine effectiveness against death
U.S. pediatric COVID vaccination uptake varied substantially by age and over time, with higher uptake in adolescents and lower uptake in younger children. At the same time, multiple studies estimate vaccine effectiveness against severe disease and death on the order of 70–90% among vaccinated individuals, even accounting for waning and variant effects.
Using conservative ranges for both parameters yields the following plausible population-level reductions in pediatric COVID mortality:
Even under conservative assumptions at the low end of these ranges, the implied reduction in pediatric COVID mortality is on the order of tens of percent, not ~2%. This places the estimate implied by Ioannidis et al. another full order of magnitude below plausible values, consistent with the pattern observed in earlier pandemic-era estimates.