The FDA Is (Finally) Applying Appropriate Standards of Evidence for COVID‑19 Vaccines in Children

Dr. Vinay Prasad has enacted new policy that will hold vaccines to the same standard of evidence on safety as drugs. This is #RevolutionaryReform at its best.

James Lyons-Weiler, PhD for Popular Rationalism

Key Takeaways

The FDA now acknowledges that at least ten U.S. children died because of COVID-19 vaccination, admitting that regulators failed to identify these fatalities in real time. An internal memo by vaccine chief Vinay Prasad details years of methodological and cultural failure inside the agency, including reliance on unstable observational designs and surrogate immune markers in place of direct clinical evidence. Under the new standards, pediatric and pregnancy-related vaccine approvals must be supported by randomized data demonstrating meaningful clinical benefit, and longstanding practices — from immunobridging to annual flu-shot immunogenicity — can no longer justify authorization. The “Kennedy Bar” now governs all indications, a shift already reflected in the termination of multiple mRNA programs and the cancellation of a major BARDA contract for inadequate evidence. The FDA frames this pivot as essential to restoring credibility: vaccines will be regulated with the same rigor as any high-risk medical product, and assumptions will no longer substitute for proof.

WASHINGTON — In a decisive break from years of inertia, the Food Administration’s top vaccine official has acknowledged that federal regulators failed to detect — or act upon — evidence that COVID‑19 vaccines caused the deaths of American children. An internal memo from the director of the Center for Biologics Evaluation and Research states that at least ten pediatric deaths were attributable to the vaccines, marking the first such admission by the agency. The disclosure has triggered a sweeping overhaul of vaccine approval standards in the United States.

For the first time, the FDA is admitting that children died because of vaccination — and that the agency did not recognize these deaths in real time. That recognition is reshaping the nation’s vaccine regulatory framework more fundamentally than any action since the introduction of recombinant vaccine technologies.

The memo’s author, FDA vaccine CBER chief Vinay Prasad, describes a methodical review undertaken by agency staff that examined nearly one hundred pediatric death reports. The reviewers concluded that no fewer than ten were likely or probably caused by vaccination. The language is unequivocal: the deaths were “after and because of” vaccine administration. He further notes that passive reporting, physician reluctance, and form‑completion burden suggest that the true count is higher.

That frank assessment is paired with an equally direct critique of the agency’s past. For several years, federal leaders insisted that safety signals were unsupported, delayed meaningful acknowledgement of myocarditis risk in adolescent boys, and failed to enforce post‑market commitments. The memo recounts that the United States was not the first to detect myocarditis linked to mRNA vaccines — Israel was — yet senior officials publicly claimed no signal existed. The document characterizes that episode as a turning point, one that undermined internal confidence and contributed to the departure of long‑serving vaccine regulators.

What emerges from this internal communication is an unflinching account of procedural failure. Cases of children aged seven to sixteen who died following vaccination were not systematically reviewed. Years passed between the first pediatric fatality reports and the first internal adjudication. According to the memo, the delay reflected cultural and structural weaknesses: a misplaced belief that vaccines should be treated with a presumption of benefit, an overreliance on observational studies with unstable denominators, and a tendency to favor surrogate immune markers over direct clinical outcomes.

The implications of that critique are sweeping. Prasad outlines a regulatory doctrine anchored in evidence rather than assumption, reversing practices that expanded vaccine indications based on antibody titers rather than clinical benefit. Under the new standards, manufacturers seeking approval in children will be required to provide randomized evidence that vaccination reduces meaningful outcomes — not simply that it raises antibody levels. Populations can no longer be added through immunobridging alone; they must be represented in trials. Surrogates cannot serve as the sole basis for authorization in pregnant women, a population that has long been excluded from direct evidence generation and then asked to rely on inference.

The new framework also directly challenges longstanding reliance on immunogenicity as sufficient evidence for influenza updates. Annual flu‑shot approvals will be re‑evaluated through a process that demands transparent, reproducible evidence rather than laboratory proxies. The memo characterizes the existing flu system as built on poor assays, weak designs, and effect estimates that lack credibility. The FDA will now demand data that reflect real‑world performance, not abstract correlates.

The changes extend further. The agency will revisit how it approaches simultaneous vaccine administration, a practice rooted in convenience rather than robust testing. In the past, regulators accepted combination safety and efficacy claims based on studies too small to detect rare harms or subtle interactions. Under the revised approach, these assumptions will be replaced by prospectively designed trials with adequate power, clear endpoints, and rigorous monitoring.

This regulatory shift is animated by a core question raised in the memo: did COVID‑19 vaccination save more healthy children than it harmed? Prasad argues that the evidence needed to answer that question was never developed. Randomized pediatric trials were underpowered for severe outcomes, and observational studies were compromised by differential exposure, biased denominators, and inconsistent case ascertainment. He notes that many children had negligible risk from infection and that the decline of multisystem inflammatory syndromes further reduced anticipated benefits. In this context, meaningful assessment required data the agency did not collect.

The memo also challenges a widely repeated argument that infection poses a greater myocarditis risk than vaccination. Prasad contends that comparisons based on clinical‑care populations misrepresent true incidence, because most infected individuals never seek care. He argues that rigorous comparison must reflect vaccine plus virus exposure versus virus alone in age‑specific cohorts, a standard not met by existing studies.

Taken together, the findings present a sobering narrative: the United States’ vaccine regulatory apparatus did not meet the evidentiary standards expected of modern biomedical oversight.

Historical Arc of Increasing Stringency and the Emergence of “The Kennedy Bar”

For years, mRNA vaccines were treated inside federal agencies as a platform whose theoretical advantages justified accelerated pathways and reliance on immunologic surrogates. But long before the current review of pediatric deaths, the scientific record showed that mRNA technologies did not uniformly succeed when subjected to classical clinical endpoints. The first large‑scale mRNA COVID‑19 vaccine to falter was CureVac’s CVnCoV, which achieved only 47 percent efficacy in a pivotal Phase 2b/3 trial — a result that failed to meet prespecified success criteria and led the company to withdraw its European application entirely. Sanofi and Translate Bio followed soon after, terminating their MRT5500 program once early data failed to demonstrate advantage over existing options. These were early reminders that mRNA platforms were not intrinsically self‑validating; they had to demonstrate benefit the same way other vaccines do: through clear, reproducible clinical outcomes.

That pattern continued after the pandemic emergency faded. In 2025, Moderna’s congenital cytomegalovirus vaccine — mRNA‑1647, touted as the company’s most important maternal‑child health product — failed its Phase 3 CMVictory trial in seronegative women, the very population that would need to benefit to prevent congenital infection. The trial’s top‑line efficacy estimates ranged from 6 to 23 percent, far short of clinically meaningful thresholds. Moderna discontinued the program. The failure underscored a principle that will now govern all pregnancy‑adjacent indications: surrogates and mechanistic optimism cannot substitute for demonstrated protection in the specific population at risk.

The shift became unmistakable when the Department of Health and Human Services, under Secretary Robert F. Kennedy Jr., formally wound down twenty‑two federally funded mRNA vaccine projects across influenza, COVID variants, and emerging avian strains. The agency concluded that these candidates — some operated by major federal contractors — did not produce evidence of reliable protection against upper respiratory infections. The termination included cancellation of a more than $700 million Biomedical Advanced Research and Development Authority (BARDA) contract for Moderna’s H5N1 avian‑influenza mRNA vaccine after internal review determined that the program did not meet scientific or safety expectations. Whatever the political context, the administrative action established a precedent: mRNA vaccines could no longer advance through federal pipelines on the strength of neutralizing‑antibody surrogates alone.

Within industry, too, performance has varied. Moderna’s own seasonal influenza vaccine, mRNA‑1010, missed key non‑inferiority criteria against influenza B strains in early analyses and had to re‑enter later‑stage trials with revised expectations. Pfizer and BioNTech’s combined COVID‑flu candidate failed one of its two primary endpoints because the influenza component underperformed. These stumbles reinforced what the early CureVac and Sanofi programs had already revealed: mRNA vaccines are not exempt from the ordinary demands of clinical science, and they must earn each indication through direct evidence rather than platform enthusiasm.

Taken together, these episodes form the regulatory and scientific backdrop against which current pediatric decisions must be judged. The “Kennedy Bar” — a requirement that vaccines deliver measurable clinical benefit in the population to whom they are recommended — is not theoretical. It has already been invoked to end programs, cancel contracts, and terminate development lines where benefit was uncertain, inadequate, or unproven. With the FDA now acknowledging at least ten child deaths attributable to COVID‑19 vaccination, it follows that pediatric recommendations must be held to the same evidentiary standard. Immunobridging to adult titers can no longer justify authorizations for young children. Observational studies with distorted denominators cannot substitute for randomized evidence of reduced severe disease. Surrogates cannot sustain a recommendation when the underlying risk to children is low and the margin for harm is narrow.

If the United States is to rebuild vaccine credibility — and protect children with full transparency — the pediatric schedule must conform to the same rigor that is already reshaping adult, respiratory, maternal, and pandemic‑preparedness programs. The era of assumption‑based authorizations is over. Evidence, not narrative, now defines the bar.

The new policy direction seeks to break that pattern. Regulators will enforce long‑neglected post‑market commitments, including studies of subclinical myocarditis. Surrogate‑endpoint approvals for pregnancy will end unless direct clinical evidence is provided. Combination vaccines, annual strain updates, and pediatric schedules will be subjected to evaluation consistent with the rigor applied to gene therapies, antisense drugs, and monoclonal antibodies. Third‑party or “community” benefits will not be assumed; they must be demonstrated.

The memo outlines that vaccines will be treated no differently — and no more gently — than any other medical class. Enforcement will be stricter, evidentiary thresholds higher, and assumptions grounded in data rather than precedent. Staff who disagree with this philosophy are invited to step aside.

In practical terms, this marks the first meaningful pivot in American vaccine regulation in decades. It is a recognition that public health credibility depends on intellectual honesty and methodological integrity, not on narrative protection or political alignment. For families seeking clear guidance, clinicians working to balance benefit and harm, and researchers building the next generation of vaccines, the shift represents a long‑delayed return to first principles.

Whether this course correction will restore confidence remains uncertain. But for the first time since the pandemic began, the nation’s chief vaccine regulator has acknowledged the scope of the problem — and placed evidence, not assumption, at the center of the FDA’s mission.