FDA Knew of Better Method to Detect COVID Vaccine Safety Signals — But Refused to Use It
The signals were there. They just looked the other way. Bombshell new documents obtained by Sen. Ron Johnson reveal what really happened.
This article originally appeared on The Defender and was republished with permission.
Guest post by Suzanne Burdick, Ph.D.
New documents obtained by Sen. Ron Johnson reveal that Biden-era federal health officials refused to use a state-of-the-art statistical tool for detecting COVID-19 vaccination signals in VAERS — even though they knew the tool they were using was broken and failed to pick up on safety signals, including sudden cardiac death.
Biden-era health officials rejected a state-of-the-art statistical tool for detecting COVID-19 vaccine safety signals — and instead deliberately continued using a broken method because they didn’t want to “feed in to [sic] anti-vaccination rhetoric,” according to a report released today by Sen. Ron Johnson (R-Wis.).
The report — “Unmasked: How Biden Health Officials Purposely Turned a Blind Eye Toward COVID-19 Vaccine Safety Signals” — includes emails from U.S. Food and Drug Administration (FDA) officials who told an FDA researcher to “cease and desist” using the updated tool to analyze COVID-19 vaccine injury reports in the Vaccine Adverse Event Reporting System (VAERS).
The FDA knew as early as 2020 that the method they were using was inferior and that a better one existed, according to the emails.
Ana Szarfman, M.D., Ph.D., a then-FDA medical officer and safety data mining developer who worked in the FDA Center for Drug Evaluation and Research (CDER), tried repeatedly to get FDA officials to switch over to the improved tool.
Szarfman wanted the agency to use an updated empirical Bayesian (EB) data-mining method that was capable of accounting for masking. Masking occurs when the presence of other vaccines in the system hides safety signals for a specific vaccine.
In a 2022 peer-reviewed paper in Drug Safety, Szarfman and her co-authors reported the discovery of roughly 25 statistically significant COVID-19 vaccine safety signals — including sudden cardiac death, Bell’s palsy and pulmonary infarction — using the improved EB data-mining method that the FDA’s older method missed.
The emails released today show that in 2021, FDA senior officials attempted to stop Szarfman from sharing results of the improved method.
The emails also reveal that FDA leaders insisted on using the older method of EB data mining — even though they knew it suppressed safety signals linked to the COVID-19 vaccines.
During a hearing later today, Johnson will release roughly 600 pages of emails cited in the report that show how Biden-era health officials failed to detect or respond to COVID-19 vaccine safety signals, a source familiar with the investigation told The Defender.
It’s unclear what EB data-mining tool the FDA currently uses to monitor VAERS reports following COVID-19 vaccination.
Children’s Health Defense (CHD) Senior Research Scientist Karl Jablonowski will testify at today’s hearing about VAERS’ failure to detect COVID-19 injuries. CHD.TV will livestream the hearing here, beginning at 2:30 p.m. ET.
Improved tool detected ‘death and sudden death’ linked to Pfizer and Moderna vaccines
Johnson obtained the emails after he subpoenaed the U.S. Department of Health and Human Services in January 2025 for COVID-19 vaccine safety records and pandemic-related communications.
The emails reveal how, in late 2020, Szarfman first told FDA officials about the improved EB data-mining tool. She also warned officials that the FDA’s current tool was likely not picking up on safety signals.
The response by FDA officials was less than enthusiastic. In a May 7, 2021, email, FDA officials told Szarfman to “hold off on creating and sending data mining reports and analyses” using the improved method.
Although Szarfman agreed to “only deliver analyses when I am specifically requested to do so,” she said that she and others would continue to test the improved method, which “strongly reduces confounding.”
Szarfman wrote: “Given the circumstances, it will be good for all to understand its performance with such important data.”
She and William DuMouchel, Ph.D., the chief statistician at Oracle, a data company contracted by the FDA to conduct its EB data mining, continued to try to convince FDA officials that the new method would detect safety signals masked by the older method.
In 1999, DuMouchel developed the original algorithm — Multi-item Gamma Poisson Shrinker (MGPS) — that the FDA used for its EB data mining. In 2012, he developed the improved version — Regression-Adjusted Gamma Poisson Shrinker (“RGPS”) — which controlled for masking effects.
Szarfman and DuMouchel were both data-mining experts who understood why RGPS was better than MGPS — especially when analyzing COVID-19 vaccine injury reports, according to the report.
Still, FDA officials sidelined Szarfman and DuMouchel’s input and refused to use the improved version.
In a July 12, 2021, email, Szarfman sent an FDA official an RGPS analysis by DuMouchel that showed statistically significant safety signals for “death and sudden death” linked to the Pfizer and Moderna COVID-19 vaccines.
Roughly two months later, on Sept. 15, 2021, Peter Marks, M.D., Ph.D., then-director of the FDA’s Center for Biologics Evaluation and Research (CBER), told a CDER official who oversaw Szarfman that her analyses had become a “major distraction.”
Szarfman “has decided on her own to do vaccine analysis using VAERS as part of her work at FDA,” Marks wrote. “She is, however, not doing this in collaboration with our CBER statisticians, and quite to the contrary, has been asked to cease and desist, because the strategy that she is using could create erroneous conflicts that feed in to [sic] anti-vaccination rhetoric.”
Dr. Peter Stein, director of the CDER Office of New Drugs, thanked Marks for “flagging” the issue. “We’ve made it clear to her that she should not be discussing or providing internal analyses externally, and needs to focus on her assigned work. … Hopefully, you won’t have further surprises.”
FDA still refuses to use improved tool after peer-reviewed study shows it works
Szarfman and DuMouchel continued analyzing VAERS reports using the improved tool. In 2022, they and others co-authored a peer-reviewed paper in Drug Safety demonstrating the tool’s effectiveness.
In July 2022, Szarfman emailed the paper to then-FDA Commissioner Robert Califf, noting that the paper describes an advance in EB data-mining methodology “especially in the capacity to unmask hidden signals.”
Upon receiving the paper and another article, Califf responded by writing, “Thanks. These are good.” There’s no evidence that Califf tried to get the FDA to adopt the better method, according to the report.
However, there is evidence that FDA officials continued to disapprove of Szarfman’s work. In August 2022, Richard Forshee, Ph.D., the then-deputy director of CBER’s Office of Biostatistics and Pharmacovigilance, wrote to Marks:
“We have just learned that Dr. Szarfman is a co-author on a recently published paper based on COVID-19 vaccines and VAERS. We believe there are a number of issues with the paper and its findings, and we are discussing how to best respond.”
Szarfman retired from the FDA in 2025.
Johnson tried to contact her, but it is unclear whether she will someday testify before Congress about her efforts on COVID-19 safety signal detection, a source familiar with the matter told The Defender.
FDA likely tried to evade FOIAs by limiting emails about safety signals
The emails contained in Johnson’s report also reveal that FDA officials attempted to evade Freedom of Information Act (FOIA) requests related to COVID-19 vaccine safety signals.
For instance, FDA official Dr. David Menschik told his colleagues in an April 15, 2021, email that they “should meet internally” before agreeing to speak with Szarfman about using the updated data-mining method because there were “many considerations not suited to email.”
In September 2022, the FDA stopped sending weekly reports on its data-mining results to the Centers for Disease Control and Prevention (CDC), citing a need to “reduce email traffic.”
In November 2022, certain CDC officials noted in an email thread how they used to receive the FDA’s data-mining safety signal reports for the COVID-19 vaccine. A CDC official wrote, “I think that because of the FOIAs … we may have asked FDA to stop sending these weekly data mining outputs.”
The report noted that in the months leading up to the FDA’s decision to stop distributing its weekly data-mining reports, Johnson and several organizations — including Children’s Health Defense and the Informed Consent Action Network (ICAN) — had made FOIA requests to the CDC or FDA for data-mining analyses.
CDC, FDA try to publish studies promoting COVID vaccine safety using flawed method
Internal documents obtained last month by Johnson and analyzed by CHD scientists showed that CDC and FDA officials knew that the older EB data-mining method had “blind spots” that rendered it “mostly useless” for picking up on safety signals of COVID-19 vaccines.
Yet, the agencies used the method in analyses and attempted to publish findings from those analyses — including studies that supported the safety of COVID-19 vaccines.
In one case, researchers sought to publish a June 2022 analysis in The Lancet Infectious Diseases using EB data mining on early COVID-19 vaccine data. They dropped the plan only after a reviewer wrote that the likelihood of detecting a safety signal using the method was “likely close to zero.”
An investigation by The Defender also found that the study authors altered the manuscript to downplay deaths following COVID-19 vaccination.
In another instance, CDC and FDA officials used EB data-mining results in an August 2024 paper published in the American Journal of Otolaryngology on tinnitus, or ringing in the ears, following COVID-19 vaccination — even though the authors knew the method failed to accurately detect tinnitus safety signals.
Instead of removing the faulty results from the study, the authors simply added the disclaimer that the findings “cannot definitively exclude the possibility” that COVID-19 vaccines may increase a person’s risk of tinnitus.
Related articles in The Defender
CDC, FDA Admitted to Using ‘Mostly Useless’ Tool to Detect COVID Vaccine Safety Signals
CDC Buried COVID Vaccine Death Data in Lancet Study, Internal Documents Reveal
Breaking: Emails Obtained by CHD Reveal Government’s Failure to Monitor COVID Vaccine Injury Reports
Johnson Subpoenas HHS for COVID Vaccine Safety Records, Fauci Emails
Donate to Children’s Health Defense
