FDA officials knew COVID vaccine safety system was flawed, suppressed whistleblower, investigations reveal
The landscape of public health oversight in the United States faces a profound reckoning following the publication of two major investigative reports in November 2026. These findings, stemming from exhaustive inquiries by The BMJ, attorney-led Freedom of Information Law (FOIL) requests, and U.S. Senate investigations, suggest that federal health authorities knowingly utilized a compromised vaccine safety surveillance system for mRNA COVID-19 vaccines while actively suppressing internal efforts to correct the mechanism. The resulting lack of transparency, according to the documents, created a persistent information vacuum that allowed official assurances of safety to continue even as medical examiners and internal scientists expressed growing private alarm.
The revelations center on the U.S. Food and Drug Administration (FDA) and the Centers for Disease Control and Prevention (CDC), spanning a critical period from October 2020 through July 2024. The documents, obtained through subpoenas and public records requests, detail a systemic failure to detect potential adverse events—including myocarditis, pericarditis, Bell’s palsy, and tinnitus—because the mathematical algorithm tasked with identifying these risks was inherently incapable of functioning during the mass rollout.
The Mathematical Flaw: A "Masking" Effect
At the heart of the controversy is the "empirical Bayesian" data mining algorithm, a statistical tool used by the FDA to monitor the Vaccine Adverse Event Reporting System (VAERS). Under normal circumstances, this system is designed to trigger "alerts" when specific adverse events occur at a frequency higher than expected. However, the unprecedented scale of the COVID-19 vaccination campaign rendered the algorithm ineffective.
Because more than 90% of all VAERS reports during the initial rollout were related to COVID-19 vaccines, the data became saturated. When both Pfizer and Moderna vaccines exhibited similar risk profiles, the "observed" frequency of adverse events matched the "expected" frequency, effectively neutralizing the signal. This "masking" phenomenon meant that despite thousands of reports flowing into the system, the automated surveillance software remained silent.
Dr. Ana Szarfman, a seasoned FDA medical officer with extensive experience in pharmacovigilance, identified this critical failure as early as early 2021. Collaborating with renowned statistician William DuMouchel, Dr. Szarfman developed a refined analytical model capable of filtering out the background noise and correctly identifying safety signals. Her efforts to integrate this fix were met with institutional resistance. According to internal communications, she was directed by superiors to "cease and desist" her efforts to implement the updated methodology. This decision effectively maintained a status quo where public health agencies could truthfully, if misleadingly, claim that no "safety signals" had been triggered by their monitoring systems.
A Chronology of Suppressed Warnings
The timeline of these events paints a picture of a disconnect between internal technical awareness and external public messaging.
In February 2021, as the national vaccination campaign gained momentum, internal warnings about the inadequacy of the monitoring tools began to circulate within the FDA. By May 2021, the concern reached the CDC. On May 13, 2021, CDC Deputy Director Tom Shimabukuro sent a internal email expressing confusion regarding the absence of alerts for myocarditis, noting, "I’m perplexed that myocarditis isn’t alerting for either of the mRNA vaccines." Yet, just 24 hours later, the CDC publicly described myocarditis as an "extremely rare" side effect, a statement that relied on the absence of signals from the very system the officials knew was compromised.
This pattern of behavior continued throughout 2022. When the CDC finally conducted a Proportional Reporting Ratio (PRR) analysis—an alternative method that did not rely on the flawed Bayesian algorithm—the results were stark. The analysis revealed that hundreds of adverse events, including the aforementioned heart-related conditions, met the criteria for a safety signal across 13 consecutive weekly reports. Despite this, the FDA’s primary Bayesian algorithm remained silent, and no public advisory was issued to update the safety profiles provided to healthcare providers or the public.
Medical Examiners and the Culture of Silence
While federal agencies grappled with algorithmic failures, local medical examiners were encountering clinical realities that challenged the prevailing narrative of vaccine safety. Emails from the National Association of Medical Examiners (NAME) reveal that forensic pathologists were privately debating cases of sudden death among the young following vaccination.
On June 26, 2021, Dr. James Gill, the Chief Medical Examiner for Connecticut, reached out to a national pathology group regarding a 16-year-old male who had died in his sleep four days after his second Pfizer dose. Dr. Gill noted "striking ischemic changes" in the heart, a finding that prompted him to seek input from peers on whether similar cases had been observed elsewhere.
Earlier that month, on June 4, 2021, Dr. Barbara Sampson, the Chief Medical Examiner for New York City, consulted with Harvard-affiliated experts regarding a case of focal lymphocytic myocarditis in a patient who had received a COVID-19 vaccine two months prior. While she expressed skepticism regarding an association, the act of seeking external consultation signaled that the medical community was quietly grappling with the clinical evidence of adverse events.
These discussions were not always met with open inquiry. By December 2021, the NAME email list became a site of professional friction. When a participant shared information regarding cardiac incidents among athletes—a topic of intense public debate at the time—Dr. Jay Stahl-Herz, a New York City medical examiner, dismissed the reports as "misinformation from a non-medical source." This triggered a rebuttal from Dr. William Oliver, who argued that it was premature to be "dogmatic" and that the medical community had a duty to examine the evidence rather than attack the sources of potential data.
Institutional Implications and Official Response
The implications of these revelations are significant, touching upon the foundational trust between public health agencies and the public they serve. Critics argue that the suppression of Dr. Szarfman’s warnings and the reliance on a flawed algorithm suggest that "vaccine confidence"—a stated goal of the CDC and FDA—was prioritized over the objective, transparent surveillance of medical data.
Legal experts who have reviewed the correspondence, such as attorney Sujata Gibson, suggest that the documents demonstrate a deliberate effort to curate a specific narrative. "The city had the warnings," Gibson stated. "Evidence that didn’t fit the narrative was swept under the rug."
The FDA’s pharmacovigilance chief acknowledged in an October 2023 email that the agency "was aware of this limitation before and during the pandemic," a confession that has fueled calls for a formal congressional inquiry into the decision-making process of senior leadership. When reached for comment, representatives for the FDA and CDC emphasized that vaccine surveillance involves multiple, overlapping systems and that the Bayesian algorithm was only one component of a broader monitoring strategy. However, they have yet to provide a detailed rebuttal to the specific allegations regarding the suppression of Dr. Szarfman’s proposed algorithmic improvements.
The Legacy of the Pandemic Response
The broader impact of these findings is likely to be measured in the erosion of institutional credibility. The fact that medical examiners were privately puzzling over deaths while the agencies responsible for vaccine safety were publicly asserting that no signals existed creates a cognitive dissonance that is difficult to resolve.
Statistician William DuMouchel, the co-creator of the Bayesian algorithm, expressed a somber perspective on the missed opportunity. "If they had paid attention to Ana [Szarfman], they would have done better," he told The BMJ. His assessment underscores the tragedy of the situation: the tools to perform more accurate surveillance were available and suggested by career experts, yet they were discarded in favor of administrative convenience or the maintenance of public messaging.
As the scientific community and policymakers process these revelations, the event stands as a case study in the dangers of centralized, opaque decision-making during a public health crisis. The "real-time clinical trial" that occurred across the United States will now be subjected to intense historical scrutiny. Whether these findings lead to a structural reform of how VAERS and other surveillance systems are managed, or whether they remain a footnote in a long, complex chapter of post-pandemic recovery, will depend on the willingness of government oversight bodies to address the internal failures that allowed the masking of potential vaccine harms to persist for years. The records now in the public domain suggest that the failure was not one of technology, but one of administrative integrity.
