What The ACIP Wasn’t Shown

Introduction by: Dr. Robert Malone

Investigative journalist Dr. Maryanne Demasi and colleagues have discovered and reported what appears to be data analysis and reporting irregularities in the information presented at the June 2025 ACIP meeting in support of a recommendation that Merck’s new RSV monoclonal antibody, Clesrovimab, be administered to all healthy newborns born to mothers not recently vaccinated for RSV.

To recap, the eventual ACIP recommendation on this topic (as outlined above) was approved by a vote of 5 to 2. The dissenters on that vote noted the consistent trend, described by the CDC consultant presenting the data as not statistically significant and determined not to be related by the sponsor to product administration, of excess deaths in the related clinical trial treatment groups relative to controls.

Based on this new information reported by Dr. Maryanne Demasi and then further expanded by Dr. Yaffa Shir-Raz, it appears that this decision was based on manipulated data analyses. For context, the ACIP members were presented with these data and all other data and presentations the day before the meeting, and had little chance to review let alone investigate all of the information provided. The presumption was that the information had been thoroughly vetted by the CDC and the outgoing ACIP RSV subcommittee (one member of which was present as a member of the reconstituted new ACIP) prior to the presentations. We had no reason or opportunity to question. Still, a minority subset of members were uncomfortable with the data trends. I voted in favor of the resolution based on the information and logic presented. That trust in the data presented now appears to have been ill-advised. Going forward, speaking for myself, based on these findings, I will no longer be able to trust that what is presented in CDC summaries to the ACIP is transparent, accurate, and unbiased. Whether or not this decision will be reconsidered is speculative, and I am embargoed from discussing future ACIP plans and non-public discussions or information.

I previously covered this RSV monoclonal antibody decision here:

Dr. Yaffa Shir-Raz, PhD, is a risk communication researcher and a teaching fellow at the University of Haifa and Reichman University. Her area of research focuses on health and risk communication, including Emerging Infectious Disease (EID) communication, such as the H1N1 and the COVID-19 outbreaks. She examines the practices used by the pharmaceutical industries and by health authorities and organizations to promote health issues and brand medical treatments, as well as censorship practices used by corporations and by health organizations to suppress dissenting voices in the scientific discourse. She is also a health journalist, and the editor of the Israeli Real-Time Magazine and a member of the PECC general assembly.

She is highly qualified and credentialed to perform the analyses that she presents below.

This publication represents my opinions and those of the author, Yaffa Shir-Raz. They should not be construed as representing the position of the US Government, the Centers for Disease Control and Prevention, or the Advisory Committee on Immunization Practices (ACIP).

Among many other things, Dr. Malone is a medical advisor to the MAHA alliance, a Special Government Employee (unpaid), and a sitting member of the CDC ACIP. He has been determined by the US Government as not having any conflict of interest related to this topic.

What The ACIP Wasn’t Shown

Analysis and Summary Article By Dr. Yaffa Shir-Raz

In June 2025, the Advisory Committee on Immunization Practices (ACIP) held its first meeting under the new leadership appointed by Health Secretary Robert F. Kennedy, Jr. The public expectation was clear: that this newly appointed committee would restore rigor, independence, and critical examination of evidence before recommending routine use of new pharmaceutical products.

One of the most significant items on the agenda was whether to recommend Merck’s new RSV monoclonal antibody, Clesrovimab, for routine use in healthy newborns. Though marketed as a new product, it is nearly identical in structure and function to Sanofi–AstraZeneca’s nirsevimab, approved in 2023.

The committee ultimately voted 5 to 2 in favor of the recommendation. That vote followed a CDC presentation, which framed the safety data as reassuring, leading most members to conclude there were no outstanding safety concerns.

But was that reassurance justified? And on what exactly was it based?

The Seizure Signal, and How It Was Presented

During its June 2025 meeting, ACIP members were shown a safety slide from the CDC’s Vaccine Safety Datalink (VSD), focusing on seizures after administration of nirsevimab. The data were split into two age groups: infants aged 0-37 days and those aged 38 days to under 8 months. Each group showed elevated risk ratios for seizures (3.50 and 4.38, respectively), but both were labeled “not significant.” No pooled analysis was displayed.

However, as Dr. Maryanne Demasi later reported, combining the two groups into a single cohort yields a very different picture: a nearly four-fold increase in seizure risk (RR 3.93, 95% CI 1.21–12.79, p=0.02), a result that is statistically significant. That consolidated signal was never presented to the committee.

The decision to stratify at 38 days – precisely the point in US schedules when routine infant vaccinations begin – had no clear biological justification, and by dispersing the signal across two smaller groups, it effectively erased the statistical significance.

A second design choice compounded the problem. The CDC’s analysis applied a self-controlled risk interval with only the first 7 days designated as “risk” and days 8-21 treated as the “control” period. Any seizure occurring on day 8 or later was thus counted against the background rate, even though such timing could plausibly reflect a product-related effect. Standard pharmacovigilance practice calls for testing multiple windows, not a single narrow cutoff.

These analytical decisions mattered. The vote to recommend clesrovimab passed 5-2. Had members been shown the pooled seizure risk alongside the consistent trial-level imbalances in nervous system events, shifting just two votes would have changed the outcome.

Finally, as Demasi emphasized, the concern is not confined to one brand. Given the structural similarity between nirsevimab and clesrovimab, the seizure risk is likely a class effect. This means the omission of the pooled analysis did not just obscure a statistical detail. It withheld information with direct implications for every RSV monoclonal antibody now in use.

These findings emerged only through independent reanalysis. Without Dr. Demasi’s work, they may have remained unknown – not only to the public, but even to ACIP members casting their votes.

The Mortality Picture ACIP Did Not Weigh

The CDC’s presentation to ACIP did not include any integrated review of mortality data from the clinical trials of either RSV monoclonal – not Merck’s clesrovimab, and not Sanofi-AstraZeneca’s nirsevimab. This omission is striking, given that across both product lines, trial results show a consistent and notable imbalance in deaths between treatment and control arms.

Nirsevimab: Deaths by Arm

FDA’s Integrated Review for nirsevimab explicitly flagged an “unexpected imbalance” in deaths observed across pediatric trials. The data are as follows (Table 49, p.117):

• Trial 03: 2 deaths among 968 nirsevimab recipients; 3 among 479 controls.

• Trial 04 (MELODY): 4 deaths among 1,998 nirsevimab recipients; 0 among 996 placebo.

• Trial 05 (MEDLEY): 5 deaths among 613 nirsevimab recipients; 1 among 304 given palivizumab.

• Trial 08: 1 death among 60 nirsevimab recipients; no concurrent control arm.

In total: 12 deaths among 3,710 nirsevimab recipients versus 4 deaths among 1,797 controls – a mortality rate of 0.32% in the treatment arms compared to 0.22% in the control arms. The imbalance may appear small in absolute terms, but it was unexpected, and it runs consistently in one direction.

Clesrovimab: Deaths by Arm

The FDA’s 2025 risk review for clesrovimab – the product under ACIP consideration – shows a similar trend across its two main trials:

• CLEVER (MK-1654-004): 7 deaths among 2,409 clesrovimab recipients; 3 among 1,202 placebo.

• SMART (MK-1654-007): 8 deaths among approximately 500 clesrovimab recipients; 4 among approximately 500 receiving palivizumab.

Across both studies: 15 deaths in the treatment arms versus 7 in controls.
While FDA reviewers did not attribute the deaths to clesrovimab after case review, they explicitly acknowledged the numerical imbalance.

Deaths Hidden in Footnotes

In the 2023 update to the MELODY trial, the published manuscript in the New England Journal of Medicine reported four deaths in the nirsevimab arm and zero in the placebo arm, concluding that the product remained safe because those deaths were judged unrelated to treatment.

But a closer look at the trial’s Supplementary Appendix tells a different story. Beneath the CONSORT flow diagram, a footnote records a fifth death in the nirsevimab arm. The note explains that four deaths up to Day 361 were included in the safety analysis, while one additional death on Day 440 was excluded.

That exclusion does not align with the trial’s own protocol, which prespecified safety follow-up of approximately 510 days after dosing. By that definition, a death occurring on Day 440 falls inside the intended safety window.

A similar ambiguity appears in Merck’s CLEVER trial. Seven deaths in the clesrovimab arm and three in the placebo arm were reported within the 365-day observation period, all dismissed as “unrelated.”

Yet the CDC presentation also included a footnote about an additional death on day 487, after the infant had formally discontinued participation at the physician’s instruction. It remains unclear whether this case was counted among the seven or treated separately.

The fact that this out-of-window case was highlighted in detail, while the seven in-window deaths were only presented as totals with no breakdown of causes or timing, points to a selective approach to transparency. Such reporting practices prevent independent reviewers from assessing whether mortality patterns were plausibly due to chance or warranted closer investigation.

The same pattern is seen again in Merck’s SMART trial. There, eight deaths occurred among clesrovimab recipients versus four among infants given palivizumab. Once again, investigators concluded that none of the deaths were “related,” and no detailed breakdown by timing or cause was provided.

The key issue here is not causality but transparency. Readers and advisers should be able to see every death in the main dataset when totals are this small. Instead, the published article reports one number, while the supplementary materials reveal another.

This selective reporting leaves advisers without the ability to fully assess mortality risks. And when all the trials are considered together, a troubling consistency emerges. None of the individual trials was powered to detect differences in mortality, and the total numbers are small. Nonetheless, when four independent randomized comparisons – across two products and multiple geographies – all show more deaths in the treatment arms than in controls, the consistency is hard to ignore.

As Prof. Retsef Levi, one of only two ACIP members who voted against approval, pointed out: “Four different trials all show deaths going in the same direction.”

For a product intended for routine administration in healthy full-term newborns, even modest safety signals should trigger close scrutiny. That did not happen in this case, and the full mortality picture was never placed on the table.

Missing Transparency on Causes of Death

A complete and transparent briefing to ACIP should have included not only the raw death counts by trial arm but also a structured table listing cause of death, timing, and arm assignment for every case. That level of detail is essential according to current methodological and regulatory standards. The CONSORT Harms 2022 extension (which integrates into the main CONSORT checklist) emphasizes the need for complete, prespecified reporting of harms in randomized clinical trials. Similarly, the ICH E9(R1) guideline underscores the importance of defining estimands (in plain terms: the exact outcome the trial claims to measure) and conducting transparent analyses that allow independent scrutiny, rather than relying only on narrative judgments.

Nevertheless, the FDA’s public summaries largely rely on narrative statements that deaths were “not related,” without presenting arm-level breakdowns that would allow independent reviewers to check for clustering by time, syndrome, or comorbidity. Neither the CDC nor the product sponsors supplied ACIP with such a side-by-side accounting.

This gap is not theoretical. In the phase 2b trial of nirsevimab, for example, two deaths in the treatment arm were attributed to gastroenteritis in otherwise healthy infants – one on Day 143 and another on Day 338. Such outcomes are rare. Without a transparent arm-level table of causes and timing, together with even a basic statistical check of the overall imbalance, advisers are left unable to judge whether these deaths reflect random variation or a meaningful safety signal warranting further investigation.

One Surveillance Source, No Triangulation

At its June 2025 meeting, the CDC’s safety briefing to ACIP drew exclusively on the Vaccine Safety Datalink (VSD), an active surveillance system linking electronic health records across 13 US health systems. No parallel analysis was shown from VAERS or FDA’s MedWatch, even though federal guidance explicitly splits reporting for RSV monoclonals: when the antibody is given alone, adverse events are to be reported to MedWatch; when given together with vaccines, reports go to VAERS.

By limiting its analysis to a single source, the CDC presented ACIP with a one-system view of safety. This narrow lens risks overlooking signals that might emerge first in another surveillance stream, precisely the reason why triangulation across systems is considered a baseline expectation in pharmacovigilance.

That selectivity extended beyond US borders. Independent real-world data from France, presented by researcher Hélène Banoun, further underlines the importance of a comprehensive viewpoint. During the first nationwide rollout of nirsevimab in autumn 2023, deaths occurring in infants aged 2-6 days exhibited a striking temporal pattern:

• September 2023: 55 deaths (statistically significant increase)

• October 2023: 62 deaths (statistically significant increase)

When distribution was limited and supply-rationed in November 2023, death counts dropped sharply to 26. Later, as distribution resumed, mortality rose again to 50 in December and 52 in January 2024, both representing statistically significant peaks.

These shifts closely aligned with the pattern of rollout and supply constraints – suggesting a temporal clustering that, while not proving causation, constitutes a meaningful signal. Such real-world patterns – with accompanying methodological caveats – should have been presented alongside US VSD data, yet were omitted from the ACIP briefing.

Taken together, the omission of both domestic and international signals meant that ACIP members were shown only a reassuring slice of the available evidence, not the full picture.

A Built-In Reporting Blind Spot

The issues are not limited to what ACIP showed in June. The very classification of RSV monoclonals creates a built-in blind spot in US safety reporting. These are biologic drugs, yet for liability purposes, they have been slotted into the childhood vaccine schedule, which affords manufacturers immunity under the National Vaccine Injury Compensation Program. For billing, they are treated as drugs. For safety reporting, they are split: given alone, they are directed to MedWatch; co-administered with vaccines, they are routed to VAERS.

This dual identity creates what experts call a “reporting blind spot.” Providers often default to VAERS when treating infants, but VAERS has no dedicated field for nirsevimab or clesrovimab. Reports can end up misfiled under “unknown vaccine type” or siloed in the FDA’s drug database, invisible to the CDC’s vaccine safety analysts. Events can therefore fall through the cracks entirely, undermining the very system meant to capture early warnings.

Beyond the Data: Confidence in ACIP

The omissions in June 2025 were not technical footnotes, but decisions that shaped how evidence was framed for those tasked with safeguarding public health. ACIP members were shown partial analyses that downplayed safety concerns, while broader and more troubling patterns remained off the table.

If an advisory committee rebuilt under promises of independence can still be steered by incomplete presentations, the issue goes far beyond one antibody. What is at stake is whether ACIP can live up to its role as a truly independent arbiter of risks and benefits – and whether the public can trust that its youngest members are being protected with full transparency.

This article is being republished under a Creative Commons license granted by Brownstone Institute.