New Analysis Reveals Potential Safety Data Analysis Flaws in FDA’s COVID-19 Vaccine Monitoring

A report released by US Senate Permanent Subcommittee on Investigations suggest that a senior scientist at the Food and Drug Administration (FDA) developed and presented a more effective data analysis method in early 2021 that could have identified additional statistically significant safety signals for adverse events related to COVID-19 vaccines. Despite these findings being shared with FDA officials responsible for vaccine safety surveillance, these analyses were reportedly overlooked, and the scientist was eventually instructed to halt her data mining efforts.

The research in question was conducted by Dr. Ana Szarfman, who was a senior medical officer and safety data mining developer at the FDA. Collaborating with Dr. William DuMouchel, the inventor of the data mining algorithm that underpinned the FDA’s existing system, Dr. Szarfman employed an updated methodology. This newer approach aimed to mitigate a known issue called “masking,” a limitation of the FDA’s current method that can obscure the detection of safety signals for adverse events.

Dr. Szarfman’s initial analysis, presented in March 2021, identified 49 instances of extreme masking. After adjusting for this masking effect, over twenty adverse events showed statistically significant safety signals that had not been previously detected by the FDA’s standard methodology. These included serious conditions such as sudden cardiac death, Bell’s palsy, and pulmonary infarction. Dr. Szarfman continued to share similar findings with FDA officials throughout April, May, and July 2021, with subsequent analyses revealing even more statistically significant safety signals for adverse events like acute myocardial infarction, embolism and thrombosis, dementia, and even death, associated with the various COVID-19 vaccines.

However, records indicate that these statistically significant safety signals, uncovered by the improved methodology, were largely disregarded by Biden administration health officials. Instead of issuing public warnings or initiating further investigations into these potential risks, the focus appeared to be on Dr. Szarfman’s work itself. Internal communications reveal that FDA officials expressed concern about reaching out to Dr. Szarfman, preferring to discuss matters internally first.

By May 2021, senior officials within the FDA’s Center for Biologics Evaluation and Research (CBER), the division responsible for COVID-19 vaccine safety, reportedly directed Dr. Szarfman to cease generating and distributing her data mining reports and analyses. Dr. Peter Marks, who led CBER at the time, even voiced concerns that Dr. Szarfman’s data mining could “create erroneous conflicts that feed into anti-vaccination rhetoric.” Despite these directives, Dr. Szarfman continued her analyses, asserting that her and Dr. DuMouchel’s method was more effective, particularly in reducing confounding factors and identifying higher safety signals for adverse events like acute myocardial infarction.

The findings also touch upon the detection of a safety signal for myopericarditis in specific age groups in May 2021. Despite this signal being acknowledged internally, public warnings about the risks of cardiac events associated with COVID-19 vaccines were reportedly delayed.

Furthermore, the analysis suggests a broader effort to restrict access to and distribution of the FDA’s COVID-19 vaccine data mining analyses. In August 2021, an FDA official informed CDC colleagues that the agency planned to limit the distribution of these reports, citing “data security reasons.” Later, in September 2022, it was suggested that this decision to cease distributing weekly data mining reports to the CDC might have been influenced by Freedom of Information Act (FOIA) requests from organizations and a U.S. Senator.

The documents reviewed also indicate that FDA officials acknowledged the limitations of their existing empirical Bayesian (EB) data mining system, including the masking effect, in internal discussions and draft publications. However, despite these acknowledgments and Dr. Szarfman’s persistent advocacy for adopting a more advanced methodology, the FDA reportedly continued to use its established system without making the recommended adjustments. It wasn’t until September 2024 that FDA officials, in an internal email exchange, appeared to acknowledge that a newer model, RGPS (Regression-Adjusted Gamma Poisson Shrinker), which Dr. Szarfman had promoted, showed more sensitive results compared to the current MGPS (Multi-item Gamma Poisson Shrinker) model for certain adverse events.

This development highlights how Dr. Szarfman’s proposed methodology, which she had been advocating for since early 2021, was eventually seen as potentially more effective by FDA officials years later. The report notes that FDA officials had ample opportunity to involve Dr. Szarfman and Dr. DuMouchel, given their expertise, in safety surveillance efforts, but instead, their analyses were reportedly set aside.

Article by Ken Buckler, based upon information from US Senate Report “Unmasked: How Biden Health Officials Purposely Turned a Blind Eye Toward COVID-19 Vaccine Safety Signals” — April 29, 2026. Ken was injured by the COVID-19 vaccine and continues to suffer adverse effects today.

---