Officials chose not to look, and you cannot know if you do not look.
In the long history of public health, the tension between promoting a treatment and scrutinizing its harms has rarely resolved cleanly. A Senate report now alleges that during the COVID-19 vaccine rollout, federal health officials at the FDA suppressed the work of a senior safety scientist, Dr. Ana Szarfman, who had identified dozens of statistically significant adverse signals — including cardiac death and Bell's palsy — using advanced data analysis techniques. Rather than investigate her findings, officials reportedly expressed concern that doing so might undermine public confidence in vaccination. The deeper question the report raises is not simply one of bureaucratic failure, but of what a society loses when the machinery of reassurance is allowed to override the machinery of inquiry.
- A Senate report backed by HHS documents alleges that FDA officials told a trained safety statistician to stop her analyses after she uncovered dozens of hidden adverse signals in vaccine reporting data.
- The suppression, if accurate, was not accidental — internal communications suggest officials feared that surfacing the data would 'feed anti-vaccination rhetoric,' prioritizing messaging over investigation.
- The signals Szarfman identified — sudden cardiac death, Bell's palsy, pulmonary infarction, acute myocardial infarction — were not minor anomalies but statistically significant findings produced by a novel masking-correction technique.
- Young people and college students, who faced minimal COVID risk, may have borne the greatest relative exposure to undisclosed vaccine risks, as universities mandated vaccination without the benefit of completed safety review.
- Because the investigation was halted before it could conclude, the true significance of those safety signals remains unknown — the report's most unsettling finding may be not what was discovered, but what was deliberately left unexamined.
Senator Ron Johnson released a report last week drawing on documents obtained from the Department of Health and Human Services, raising serious questions about how federal health officials managed vaccine safety data during the COVID-19 rollout. At the center of the account is Dr. Ana Szarfman, a senior FDA medical officer and data mining specialist, who in early 2021 began analyzing the Vaccine Adverse Event Reporting System using an advanced statistical technique developed with a colleague at Oracle.
Her analyses uncovered what researchers call 'extreme masking' — a phenomenon in which common signals in large datasets obscure rarer ones. Correcting for this effect, Szarfman identified roughly 20 to 25 previously hidden safety signals, among them sudden cardiac death, Bell's palsy, and pulmonary infarction. She shared her findings with colleagues responsible for vaccine safety surveillance. According to the report, the response was not further investigation but institutional resistance. One senior official suggested meeting privately before engaging with her, citing considerations 'not suited to email.' Dr. Peter Marks reportedly worried her data mining could generate conflicts that would fuel vaccine skepticism. By mid-2021, she had been told to stop.
In June of that year, Szarfman flagged statistically significant signals for acute myocardial infarction and noted detection of 'clear signals for other similar events' in an email to an FDA colleague. No documented action followed.
The report lands against a backdrop of an aggressive and at times overstated federal vaccination campaign — one in which officials downplayed age-based risk differences, made claims about transmission that proved false, and pursued mandates at universities and large employers. What the report suggests was absent from that effort was any equivalent commitment to investigating potential harms. VAERS has real limitations as a passive reporting system, but those limitations make it more urgent, not less, to investigate when a specialist identifies dozens of anomalous signals — not less.
The population most implicated is also the most troubling: young people and college students, for whom COVID posed minimal threat of severe illness, may have been exposed to vaccine-related risks that officials chose not to fully examine. Whether Szarfman's signals would have proven clinically significant remains unknown. They were never allowed to be tested. What the documents appear to show is that officials made a deliberate choice not to look — and that choice has now become the story.
Senator Ron Johnson released a report last week based on documents obtained from the Department of Health and Human Services that raises questions about how federal health officials handled safety data during the COVID-19 vaccine rollout. At the center of the investigation is Dr. Ana Szarfman, a senior medical officer and safety data mining specialist at the FDA, who conducted statistical analyses of the Vaccine Adverse Event Reporting System in early 2021.
Szarfman's work identified dozens of statistically significant safety signals associated with the vaccines. Using an updated data analysis technique developed with Dr. William DuMouchel, then chief statistician at Oracle, she discovered what researchers call "extreme masking"—a phenomenon where common signals in data obscure rarer ones. Her analysis suggested that roughly 20 to 25 previously undetected safety signals had been hidden by this masking effect. The signals included sudden cardiac death, Bell's palsy, and pulmonary infarction. She shared these findings with other FDA officials responsible for vaccine safety surveillance.
Instead of launching further investigation, according to Johnson's report, those officials largely dismissed her work. One senior FDA official wrote to colleagues that before reaching out to Szarfman, "we should meet internally—many considerations not suited to email." Another expert, Dr. Peter Marks, expressed concern that the data mining could "create erroneous conflicts that feed in to anti-vaccination rhetoric." By mid-2021, as Szarfman continued to surface updated findings about potential cardiac events and other adverse effects, the FDA essentially told her to stop the analyses.
In June 2021, Szarfman emailed an FDA colleague about discussions between the FDA and CDC regarding potential myocardial events linked to the vaccines. She attached data showing statistically significant safety signals for acute myocardial infarction and noted that they had "detected clear signals for other similar events." The FDA took no documented action on this information.
The broader context matters here. During the pandemic, federal health officials pursued an aggressive vaccination campaign. Dr. Anthony Fauci downplayed the differences in risk between age groups, recommending universal vaccination regardless of age or underlying health conditions. Former CDC Director Rochelle Walensky made claims that vaccinated people would not carry or transmit the virus—assertions that proved false. President Biden warned of a "winter of severe illness and death" for the unvaccinated and attempted to mandate vaccines at private businesses with more than 100 employees. Universities imposed vaccine requirements. The push for uptake was relentless.
What appears to have been missing from this effort was a parallel commitment to investigating potential harms. VAERS, the system Szarfman was analyzing, has known limitations—it is a passive reporting system rather than a controlled scientific study, and it can be subject to bias. But precisely because of those limitations, one might expect that when a trained statistician identified dozens of previously hidden safety signals, officials would want to understand whether those signals reflected real risks or statistical artifacts. Instead, the pattern suggests that concerns about how further investigation might affect public confidence in vaccination took priority over the investigation itself.
The report points to a particular vulnerability: young people and college students, for whom COVID posed minimal risk of severe illness, may have been exposed to vaccine-related risks without full awareness of the safety questions that existed. Whether those signals ultimately would have proven significant remains unknown, because the investigation was discouraged before it could proceed. What is documented is that officials chose not to look.
Citas Notables
Before we potentially reach out to Ana, we should meet internally—many considerations not suited to email.— Senior FDA official, quoted in Johnson's report
The data mining could create erroneous conflicts that feed into anti-vaccination rhetoric.— Dr. Peter Marks, FDA official
La Conversación del Hearth Otra perspectiva de la historia
Why would FDA officials actively discourage investigation into safety signals rather than simply investigate them?
The documents suggest they were worried that any investigation—even one that might ultimately clear the vaccines—could be weaponized by vaccine skeptics. They saw the risk to the vaccination campaign as greater than the risk of not knowing.
But doesn't that logic assume the signals were probably nothing? What if they weren't?
Exactly. That's the tension. You can't know if you don't look. And if you don't look, you can't make an informed risk-benefit calculation for different populations.
The report mentions young people specifically. Why would that group matter more?
Because COVID itself posed very little risk of severe illness to young, healthy people. So the risk-benefit math was already tight. If there were real vaccine risks in that population, the calculation flipped entirely.
Did Szarfman's signals turn out to be real problems, or were they statistical noise?
We don't know. That's the point. The investigation was stopped before it could answer that question. We have the signals she found, but not the follow-up work that would have determined whether they reflected actual harm.
Is this about the vaccines being unsafe, or about how officials handled uncertainty?
It's about the second thing. The vaccines may have been safe. But officials chose not to investigate questions about safety because they feared the political consequences of the investigation itself.
What changes if this report is accurate?
It suggests that during a critical moment, the priority was managing public perception rather than understanding risk. That erodes trust, regardless of what the underlying science actually showed.