Senator Ron Johnson’s Report Exposes Critical Failures in COVID-19 “Vaccine” Monitoring
The FDA allegedly ignored known flaws in VAERS, suppressed safety signals, and delayed warnings, raising serious questions about transparency and accountability
Introduction
On April 29, 2026, Ron Johnson, Chairman of the Senate Permanent Subcommittee on Investigations (PSI), released an interim majority staff report titled Unmasked: How Biden Health Officials Purposely Turned a Blind Eye Toward COVID-19 Vaccine Safety Signals. Supported by around 600 pages of internal FDA and HHS communications obtained through congressional oversight, the report alleges that federal health officials were aware as early as March 2021 that their standard VAERS data-mining system was obscuring statistically significant safety signals tied to the COVID-19 “vaccines.” Rather than adopting more advanced analytical methods that would have revealed these signals, officials are accused of dismissing the concerns and halting further inquiry.
The report was complemented by a Senate hearing that expanded on these findings (you can find the full video and summary further down in this post).
Please consider supporting our ongoing work to restore human rights, uphold informed consent, and care for those who have been harmed. You can help by making a one-time donation or by becoming a paid Substack subscriber.
Early Signals Identified
The Vaccine Adverse Event Reporting System (VAERS) is a U.S. government-run passive surveillance database, jointly managed by the FDA and CDC, which collects reports of potential adverse events following vaccination. In March 2021, FDA safety officer Dr. Ana Szarfman applied a revised data mining method to VAERS data designed to correct for known statistical distortions.
Using the Regression-Adjusted Gamma Poisson Shrinker (RGPS), a more advanced version of the FDA’s standard empirical Bayesian model, Dr. Szarfman identified 49 examples of “extreme masking” and uncovered over 20 statistically significant safety signals that had not been detected by the FDA’s existing system. These included sudden cardiac death, pulmonary infarction, and Bell’s palsy. Follow-up analyses conducted between April and July 2021 revealed additional signals with even greater statistical strength, including acute myocardial infarction associated with Pfizer and Moderna products, embolism and thrombosis across multiple manufacturers, dementia, “death,” and “sudden death.” In each case, RGPS consistently produced stronger signal values than the FDA’s standard method.
Despite repeated internal circulation of these findings, the report shows that FDA officials did not escalate them publicly, nor did they adjust policy, labeling, or rollout strategy in response. Instead, the internal response shifted toward containment. By May 2021, senior officials instructed Dr. Szarfman to stop generating and distributing further reports, effectively halting independent signal detection using the more sensitive method.
The MGPS System and the Masking Effect
At the center of the report is the FDA’s continued reliance on its so-called “gold standard” method: empirical Bayesian data mining using the Multi-item Gamma Poisson Shrinker (MGPS) algorithm, implemented through Oracle’s Empirica Signal software. MGPS is used to detect unusual patterns in reported adverse events. It flags a potential safety signal when a statistical threshold (EB05 > 2.0) is exceeded, indicating that a specific event is being reported more often than expected. In principle, this helps identify possible safety concerns. In practice, however, the method has a critical vulnerability: masking.
Masking occurs when a dominant product or group of products floods the database with reports, artificially inflating the baseline against which all signals are measured. When this happens, meaningful differences between products are suppressed. That is exactly what occurred in 2021. VAERS became overwhelmingly dominated by COVID-19 “vaccine” reports. Instead of being compared against a diverse set of unrelated medical products, each COVID injection was effectively being compared against other COVID injection variations with similar adverse event profiles, raising the baseline across the entire dataset.
When multiple products in a set share elevated adverse event rates, it becomes significantly harder for any one of them to exceed the EB05 > 2.0 threshold. Internal FDA communications explicitly acknowledged this, noting that when “most reports received involve COVID-19 vaccines,” disproportionality scores are “driven toward the null.”
The RGPS Model: A More Sensitive Alternative
The RGPS algorithm was designed to solve this exact problem. Developed by statistician Dr. William DuMouchel (who also created MGPS), RGPS incorporates regression adjustments that account for confounding variables and correct for masking effects. It produces a more sensitive metric (ER05) and significantly reduces false negatives.
In a March 1, 2021 briefing, Dr. Szarfman presented RGPS to FDA leadership as a “state of the art” alternative, explicitly highlighting its ability to unmask signals that MGPS would miss. Direct comparisons confirmed this, with RGPS consistently generating higher signal values and detecting adverse events that MGPS failed to identify, particularly for cardiovascular outcomes.
Internal communications show that officials raised concerns about introducing a “new” or “possibly unvalidated” method during the emergency use period. Yet this reasoning is difficult to justify. RGPS was not an untested or experimental tool; it was an evolution of the FDA’s own methodology, developed by the same statistician and designed specifically to address a known flaw. The decision effectively ensured that the FDA would continue relying on a system that its own experts knew would miss signals.
Confirmed Signals and Delayed Public Response
Even within the limitations of MGPS, some signals still emerged. By May 24, 2021, internal records confirmed that VAERS was signaling for myopericarditis in individuals aged 16–24, demonstrating that the underlying safety issues were strong enough to break through even a masking-prone system.
Yet detection did not lead to immediate action. Public warnings were delayed, and official label changes acknowledging myocarditis risk were not implemented until June 25, 2021, long after initial signals were identified. Other signals, including ischemic stroke in older populations, were similarly detected but not promptly communicated.
Restricting Analysis and Limiting Data Access
By September 2021, the suppression became explicit. Dr. Szarfman was formally instructed to “cease and desist” her data mining work using RGPS. Senior officials described her analyses as a “distraction” and warned that they could “create erroneous conflicts that feed into anti-vaccination rhetoric.”
At the same time, the FDA began restricting its own internal reporting. Weekly VAERS data mining reports that had previously been shared with the CDC were gradually reduced and then discontinued entirely by July 2022. Internal discussions proposed shifting to a model where only “clinically relevant” signals would be shared, rather than providing full analytical transparency.
Emails indicate that FOIA concerns influenced this decision. Officials acknowledged that continuing to share these reports could make internal findings accessible through public records requests, leading them to limit distribution. This was a failure in both transparency and oversight.
Known Limitations Without Correction
What makes this situation particularly significant is that the limitations were not hidden. Internal FDA and CDC documents repeatedly acknowledged that empirical Bayesian data mining has “blind spots.” Draft manuscripts explicitly stated that “absence of a disproportionality alert does not rule out presence of a safety problem.”
Further analysis later confirmed that masking was substantially more likely, estimated to be up to eight times more frequent, in datasets dominated by COVID-19 “vaccine” reports. Even as late as 2024, internal evaluations continued to confirm that RGPS was “way more sensitive” than MGPS.
Despite this, there is no evidence that the FDA adjusted its methodology during the critical period of 2021. The same flawed system remained in place, and the more accurate alternative was never implemented.
Structural Constraints and Accountability
The report ultimately describes more than a technical failure. It outlines a convergence of known methodological limitations, deliberate institutional decisions, and restricted information flow. A surveillance system dependent on a masking-prone algorithm was maintained despite internal warnings. A superior method capable of revealing hidden signals was identified, tested, and then suppressed.
The result was predictable. Safety signals were delayed, minimized, or missed entirely, not because they did not exist, but because the system used to detect them was structurally incapable and the effort to correct it was actively blocked.
This is not a minor procedural issue. It represents a systemic failure in pharmacovigilance and raises serious questions about regulatory accountability, public safety, and the integrity of decision-making during the COVID-19 response.
Read the Report
Unmasked: How Biden Health Officials Purposely Turned a Blind Eye Toward COVID-19 Vaccine Safety Signals Link↗
Full Hearing
Ron Johnson held a Senate hearing built around the report findings, bringing the report’s internal evidence into public view and subjecting federal health agencies to direct scrutiny. The proceedings focused on how officials within the FDA and CDC handled early warnings about limitations in VAERS data analysis, particularly the issue of statistical masking. Witnesses, including medical researchers and data experts, testified that more advanced analytical methods were available but not adopted, while internal communications pointed to concerns over transparency and public disclosure. The hearing expanded the scope of the report, reinforcing its central claim that key safety monitoring failures were known yet not adequately addressed.
This hearing contains important insights and revelations. Below is a concise summary highlighting the key takeaways for your convenience.
Senator Ron Johnson introduced the hearing alongside a report alleging deliberate inaction on safety signals. He stated that internal FDA and HHS communications show officials were aware of risks related to COVID-19 “vaccines” yet failed to act, calling the findings “as accurate as it is incriminating.”
(In the Report) FDA safety officer Dr. Ana Szarfman identified a critical flaw in the agency’s VAERS data analysis system. She determined that the standard MGPS algorithm could not reliably detect safety signals when the database became dominated by COVID-19 “vaccine” reports, due to a statistical effect known as masking.
Masking occurs when similar products generate overlapping adverse event patterns, distorting the baseline. Dr. David Wiseman explained that under these conditions, real safety signals can be “drowned out or masked,” appearing insignificant despite large numbers of adverse events.
(In the Report) Szarfman applied an improved method that exposed signals the standard system missed. Using the Regression-Adjusted Gamma Poisson Shrinker (RGPS), she identified 49 instances of extreme masking and more than 20 statistically significant safety signals, including sudden cardiac death, myocardial infarction, thrombosis, and death.
The issue was not a lack of signals, but a failure in the analytical framework. Dr. Wiseman emphasized that the FDA had the tools and expertise to detect these signals but failed to apply them appropriately.
(In the Report) FDA leadership declined to adopt the improved method despite internal validation. Witnesses pointed to this decision as a critical moment where known limitations were left uncorrected.
(In the Report) Internal communications show that further analysis using the improved method was halted. The work on masking was described as a “major distraction,” and the statistician applying the method was instructed to cease further analysis.
(In the Report) Internal emails indicate efforts to limit written records and reduce FOIA exposure. Officials suggested holding sensitive discussions verbally, raising concerns about transparency and accountability.
VAERS data showed a dramatic rise in adverse event reports and deaths during rollout. Senator Ron Johnson pointed to the scale of reporting as an obvious signal that warranted immediate investigation.
A significant proportion of reported deaths occurred shortly after injection. Johnson highlighted the temporal clustering as a pattern that required urgent scrutiny.
(In the Report) Serious safety signals still emerged despite the masking-prone system. By May 2021, VAERS data signaled myopericarditis in individuals aged 16–24, yet public warnings and label updates were delayed until late June.
(In the Report) Other signals followed a similar pattern of delayed communication. Conditions such as ischemic stroke were identified internally but not immediately disclosed to the public.
Testimony highlighted the real-world consequences of these delays. Maria Young’s experience reflected how patients faced prolonged illness without clear acknowledgment or timely guidance.
Officials publicly downplayed VAERS while privately relying on it. This contradiction was raised during the hearing as a breakdown between internal analysis and public messaging.
The V-safe monitoring system revealed substantial rates of post-injection disruption. Johnson emphasized that millions reporting health impacts should have triggered a far stronger safety response.
Clinical testimony reinforced that these outcomes extend beyond initial survival. Maria Young described prolonged recovery and lasting health consequences, emphasizing that survival alone does not capture the full impact of adverse events.
The Vaccine Safety Datalink (VSD) was available but underutilized for deeper analysis. Dr. Wiseman pointed out that more robust methods existed within the system but were not consistently applied.
Experts testified that masking was a known and solvable problem. The hearing reinforced that the FDA had both the tools and expertise to correct the issue but failed to act.
Testimony characterized the failure as a collapse of pharmacovigilance. Dr. Karl Jablonowski concluded, “Simply put, during the largest pharmaceutical roll out in history, pharmacovigilance did not exist. That is the betrayal of our time, so vast that we cannot even count casualties.”
Gaps in pre-authorization testing and oversight were emphasized. Dr. Jablonowski noted: “Had they actually turned out to be safe, it would have been serendipitous. A happy accident.”
Contrasting testimony defended the “vaccine” rollout, but this position was directly challenged by the broader evidence presented. Senator Richard Blumenthal argued the data did not support a cover-up, despite the hearing highlighting suppressed safety signals, flawed detection methods, delayed warnings, and systemic failures in safety monitoring.
A breakdown in coordinated surveillance across agencies was identified. Johnson pointed to fragmented responsibility between agencies, resulting in gaps where no single system functioned as a reliable early warning mechanism.
Statistical thresholds were criticized as insufficient for real-world signals. Johnson argued that observable increases in adverse events should have prompted action regardless of algorithmic criteria.
(In the Report) The central failure is defined as one of intentional decision-making. Known limitations were not corrected, improved tools were set aside, and transparency was reduced, leading to delayed or missed safety signals with real-world consequences.
It would be instructive to know what Health Canada knew and when. What data were they using to make the recommendations that they forced on all Canadians. Great report.
Thanks Mark for this excellent and succinct report and explanation. It’s a win.