CIDRAP Op-Ed: The report that proves the point—ACIP, COVID vaccines, and predetermined conclusions

Two weeks ago in a CIDRAP Op-Ed, I argued that the reconstituted Advisory Committee on Immunization Practices (ACIP) was preparing to spend its March meeting relitigating COVID-19 vaccine safety concerns that the evidence had already resolved—that the myocarditis signal detected with the original 2021 vaccines was real, was acted on, and has since returned to background levels with every formulation on the market today, and that an agenda organized around that historical signal in 2026 was not a serious engagement with current data. 

It was, I wrote, a fixation on a historical grievance at the expense of present-day public health priorities. 

Three days ago, US District Court Judge Brian Murphy arrived at the same core finding through a legal route: that this ACIP had abandoned the evidence-based process it was created to follow. In a preliminary ruling, he found that the reconstitution of ACIP by Health and Human Services (HHS) Secretary Robert F. Kennedy Jr., and the changes made to the childhood vaccine schedule in January, were likely illegal, violating the Administrative Procedure Act. 

“There is a method to how these decisions historically have been made,” Murphy wrote in his 45-page opinion, “a method scientific in nature and codified into law through procedural requirements. Unfortunately, the Government has disregarded those methods and thereby undermined the integrity of its actions.”

There is a method to how these decisions historically have been made.

The ACIP meeting scheduled for yesterday and today has been postponed. The committee, as currently constituted, cannot vote.

In the meantime, a leaked work group report prepared for that meeting has been circulating since the New York Times obtained it earlier this week. I have now read it in full. It deserves a careful response, not because every concern it raises is wrong, but because the document as a whole reveals something important about the process that produced it.

A 300-fold prevalence range

The report introduces a new clinical entity it calls post-acute COVID-19 vaccination syndrome, or PACVS: heterogeneous, prolonged symptoms involving multiple organ systems persisting at least 12 weeks after vaccination and unexplained by another condition. It proposes ICD-10 diagnostic codes, new clinical guidelines, and a national network of specialized treatment centers.

These are significant policy proposals. The evidentiary foundation beneath them is not.

The report’s own opening summary states that “existing prevalence estimates range from 0.003% of the general population to 0.9% of young and middle-aged persons.” That is a 300-fold difference in a single sentence, drawn from a single cited paper. A clinical entity whose prevalence cannot be estimated within three orders of magnitude is not ready for ICD-10 codification. It is ready for more research.

The diagnostic framework has a more fundamental problem. The report recommends a battery of tests for evaluating suspected PACVS patients, then states in Appendix IA that some of those tests “did not go through comprehensive clinical validation and cost-effectiveness evaluation.” Elsewhere it argues that normal routine laboratory results “do not rule out immune, microvascular, autonomic, or mitochondrial pathology.” 

A diagnostic system built on unvalidated tests where normal results cannot exclude disease is not a clinical pathway. It is a framework that cannot be falsified, one designed so that a clinician can never tell patients they do not have the condition being evaluated.

A 3-page commentary vs years of surveillance data

The report’s most consequential claim appears on page 39—that “an analysis of a Japanese database of 18 million people showed that people who received COVID-19 vaccines had a significantly higher risk of death in the first year after vaccination compared to the unvaccinated, and the risk increased with each additional dose.”

The citation behind this claim does not support it. The cited publication is a three-page opinion piece in a Japanese journal that presents no original data, no database analysis, and no study of 18 million people. It is a narrative argument that juxtaposes vaccination rates and mortality trends without any statistical analysis—what epidemiologists call an ecological fallacy. 

A formal critique published in the same journal three months before this ACIP report was finalized identified the paper as committing textbook ecological fallacy, misusing government data that Japan’s own Ministry of Health had explicitly warned must not be used to evaluate vaccine effectiveness, and citing a study on cancer mortality that had already been formally retracted due to serious methodological flaws. The ACIP report acknowledges none of these problems.

The cited publication is a three-page opinion piece in a Japanese journal that presents no original data, no database analysis, and no study of 18 million people.

In Japan’s vaccination rollout, elderly and high-risk people received priority access, meaning the vaccinated population may have been older and sicker at baseline than the unvaccinated comparison group—a form of confounding (when a variable causes a false association) that could produce spurious excess mortality in the vaccinated cohort without any causal relationship to the vaccine. The opinion piece also ignores well-documented alternative explanations for Japan’s 2023 excess deaths, including record-breaking summer heat waves, an unusually severe influenza season, and the country’s rapidly aging population. Whether any of this was addressed is unknowable, because the cited paper contains no methodology to evaluate.

What the report does not cite is the Vaccine Safety Datalink (VSD), an active surveillance system embedded in major US health systems covering millions of patients with verified clinical diagnoses in both vaccinated and unvaccinated populations. VSD data show post-vaccination mortality consistently below background rates across every age-group analyzed. The observed-to-expected ratio is less than one in every cohort, from six months through age 65 and older. Separate VSD analyses confirm no increased risk of all-cause mortality, cardiac mortality, or non-COVID mortality following vaccination across both Pfizer and Moderna products.

A three-page opinion piece that invents a database and ignores its own confounders, and years of domestic active surveillance data, are not equivalent forms of evidence. Presenting one while ignoring the other is not scientific balance.

The report’s evidentiary baseline is worth noting. Before citing any peer-reviewed literature, the document invokes a Rasmussen Reports survey, conducted by a political polling firm, finding that 24% of survey participants reported knowing someone who died from a COVID vaccine. Rasmussen polls are not tools to detect adverse vaccine outcomes. They do not establish causation, biological plausibility, or accurate recall. 

Using such a survey to anchor a federal policy document’s case for widespread vaccine mortality tells you something about the standards at work throughout the work group report.

The work group’s own members pan the report

None of this analysis requires outside expertise. Three members of the work group itself expressed similar reservations about the report’s conclusions, in writing, in Appendix IV.

Henry Bernstein, DO, MHCM, wrote that “it is challenging to make specific recommendations based upon the imbalanced presentation and discussion of the data.” Mitchell Miglis, MD, wrote that he “does not agree with the utilization of some of the assays mentioned in the Appendices IA and IB that are currently not validated for clinical use or supported by consensus expert opinion.” Stanley Perlman, MD, PhD, declined to sign the document entirely, stating that “PACVS needs to be better defined before ICD-10 Diagnostic Codes should be formulated.”

These are not outside critics. They were inside the process, reviewed the same evidence, and concluded that the document’s recommendations outran its evidence. When a report’s own authors call it imbalanced, object to its diagnostic framework, and decline to sign it, the footnote count does not fix the problem.

Three members of the work group itself expressed similar reservations about the report’s conclusions.

Judge Murphy made a related observation about the committee members themselves. Reviewing their qualifications, he found in multiple instances that “there is no evidence in the record” that new ACIP members have “any relevant vaccine-related experience or expertise.” The work group chair, Retsef Levi, PhD, is a professor of operations management at MIT, not a physician, vaccinologist, epidemiologist, or infectious disease specialist. 

In 2023, before his appointment, he wrote publicly that the evidence was “mounting and indisputable” that mRNA vaccines cause serious harm, including death, and should be removed from the market immediately. That statement predates by two years the safety review he was appointed to lead. The question of whether this constitutes genuine scientific inquiry or a review organized around a predetermined conclusion is fair to ask, and the document itself provides evidence toward an answer.

What legitimate concern looks like

The specific, serious adverse events associated with COVID vaccination are documented, were identified by active surveillance, and were acted on. Myocarditis following mRNA vaccination was real, characterized, and used to update clinical guidance. Vaccine-induced immune thrombotic thrombocytopenia (VITT)—a type of blood clotting—is real. Rare anaphylaxis is real. 

These are well-defined conditions with established diagnostic criteria and measurable incidence rates, not a heterogeneous symptom cluster spanning twelve organ systems with a 300-fold prevalence range.

The call for better ICD-10 documentation of known post-vaccination adverse events is defensible. The call for continued active surveillance is defensible. What is not defensible is building a federal diagnostic and treatment infrastructure around a syndrome that has no validated case definition, no validated diagnostic tests by the report’s own admission, and a ridiculously broad prevalence estimate drawn from a single paper.

The patients who experienced myocarditis, VITT, or anaphylaxis after COVID vaccination deserved and received clinical attention grounded in real evidence. The proposal in this report does not extend that standard to a new population. It lowers the standard, then asks the federal government to fund a national treatment network built on it.

Why the methodology issue matters most

The postponed meeting had a third agenda item that has received less attention than it deserves: a review of the GRADE evidence framework, the internationally recognized system ACIP has used since 2010 and which was formally expanded with the Evidence to Recommendations framework by unanimous vote in 2018. 

The same framework is used by the World Health Organization’s Strategic Advisory Group of Experts on Immunization and national immunization advisory bodies worldwide. It exists to ensure that vaccine recommendations are built on the totality of evidence, systematically evaluated, rather than on selective citation of individual studies or unverified reports.

What replaces GRADE matters more than anything else on the agenda. A methodology that treats a Rasmussen poll and a VSD analysis as equivalent sources of evidence does not just affect one COVID vote. It changes the evidentiary standard for every future recommendation on every vaccine this committee touches: measles, hepatitis B, polio, and everything else it is statutorily charged with evaluating.

The data on myocarditis (inflammation of the heart muscle) illustrate exactly why this matters. VSD detected elevated myocarditis rates with the original 2021-22 monovalent COVID vaccines, peaking at approximately 38 cases per million doses after the second dose in males aged 16 to 17 years old, a genuine signal that guided real regulatory responses. 

Then the vaccines changed. With the 2023-24 formulations, the VSD rate was approximately five per million. With the 2024-25 formulations, approximately two per million, at or below the background myocarditis rate in any unvaccinated population of the same age. The same surveillance system sensitive enough to find the original signal is not finding excess harm from the vaccines on the market today. Reaching that conclusion requires looking at all of this evidence together, over time, across populations. 

A framework that permits any single data point to be evaluated in isolation, or weighed against unverified reports as though they carry equivalent value, will produce worse policy across the board.

The context the report ignores

As of March 12, the United States has recorded 1,362 confirmed measles cases across 30 states and New York City, on pace to surpass last year’s total of 2,284 before summer. Last year was already a 35-year high. Measles was eliminated in this country in 2000. 

Twenty-two percent of this year’s cases are in children under five. Ninety-two percent of case-patients are unvaccinated or have unknown vaccination status. The Pan American Health Organization, which had scheduled a session in April to review US elimination status, postponed that review to November—after the midterm elections—at HHS’s request.

This is the public health environment in which the reconstituted ACIP planned to spend its March meeting on a document that opens with a political poll, recommends unvalidated diagnostics, attributes a fabricated database analysis to a paper that contains no such thing, and carries written dissents from three of the work group’s own members, while a measles outbreak accelerates and the committee’s evidentiary framework sits on the chopping block.

Judge Murphy’s ruling buys time. It does not solve the underlying problem, which is that the institutions responsible for vaccine policy have been reorganized around conclusions rather than evidence, and the consequences are already visible in case counts, in professional society withdrawals, and now in a 45-page federal court opinion finding that the process was likely illegal.

My March 3 CIDRAP Op-Ed ended by asking whether the committee would follow the evidence or confirm that the process had been subordinated to a political project. The meeting didn’t happen. But the report did.

The question has been answered.

Dr. Scott is a Clinical Associate Professor of Infectious Diseases at Stanford University School of Medicine and a co-author of “Updated Evidence for Covid-19, RSV, and Influenza Vaccines for 2025-2026“ in the New England Journal of Medicine.

The opinions voiced in CIDRAP Op-Ed pieces are the authors’ own and do not necessarily represent the official position of CIDRAP.