On January 11, 2018, in a room at the Golden Plough Inn in New Hope, Pennsylvania, an 85-year-old physician sat for a videotaped deposition that ran until 6:42 in the evening. The underlying case was a custody dispute in Oakland County, Michigan.
A divorcing couple disagreed about whether their daughter should be vaccinated, and Stanley Plotkin, MD, had volunteered to testify for the father. He had not reviewed the child's medical records and did not know her age. The attorney across the table, Aaron Siri, questioned him for roughly nine hours.
Out of those nine hours came a few minutes of video that have circulated ever since, posted in full to The Highwire and promoted by the Informed Consent Action Network and by Robert F. Kennedy Jr. in the years before he ran the US Department of Health and Human Services. The clips purport to show the most credentialed vaccinologist alive admitting under oath that hepatitis B vaccines were tested for safety for only five days.
The material resurfaced on December 5, 2025, when Siri delivered a 76-slide presentation on the childhood schedule to the reconstituted vaccine advisory committee of the Centers for Disease Control and Prevention (CDC), a presentation I examined at the time, and it anchors part of the book he published last year. The version that prompted this piece was smaller: a reply to a recent thread of mine, pointing to the measles, mumps, and rubella (MMR) vaccine package insert, which lists Guillain-Barré syndrome (GBS), as proof that the vaccine causes it.
These claims belong to a single genre: reading a package insert aloud and presenting every event it lists as a harm the document has proven. The genre has a founding document. The founding document, however, read in full, refutes it.
The man in the chair
Plotkin developed the rubella vaccine strain, RA 27/3, that is still in use worldwide. Before it, the 1964 to 1965 rubella epidemic left roughly 20,000 American infants with congenital rubella syndrome, deaf, blind, cognitively disabled, or all three, because their mothers had encountered a circulating virus during pregnancy.
In 2004, rubella and congenital rubella syndrome were declared eliminated from the United States.
This is the man the clips purport to show conceding that vaccine safety testing lasts five days.
Plotkin also contributed to vaccines against rabies, rotavirus, and cytomegalovirus, and he is the founding author of the field's definitive textbook, now known simply as Plotkin's Vaccines. When Siri asked him during the deposition whether he could supply peer-reviewed evidence that vaccines are safe and effective, Plotkin suggested the textbook, which contains thousands of references.
This is the man the clips purport to show conceding that vaccine safety testing lasts five days.
Origin of the ‘5 days’
The famous exchange begins at page 149 of the 400-page transcript (on file with the author), and what precedes it matters more than what follows. Two pages earlier, Siri asks Plotkin to define the difference between solicited and unsolicited reactions in vaccine trials. Plotkin explains that solicited reactions are the ones investigators specifically ask about, and unsolicited reactions are everything participants report on their own.
The distinction is elementary trial methodology. Solicited reactions are what a parent records on a diary card, the standardized checklist investigators hand out for noting specific symptoms each day after a dose: soreness, redness, fever. They are tracked for a few days. because acute inflammatory responses happen quickly or they do not happen.
Everything else is captured through separate reporting that runs for weeks to months, and through follow-up that, in the hepatitis B efficacy trials, ran one to two years. The first of those trials, published by Wolf Szmuness and colleagues in the New England Journal of Medicine in 1980, was a randomized, double-blind, placebo-controlled study that followed more than 1,000 participants for 24 to 30 months.
Having elicited the distinction, Siri hands Plotkin the Recombivax HB insert, marked as Plaintiff's Exhibit 10, and asks how long Section 6.1 says safety was monitored. Plotkin reads the number: five days. Siri asks whether five days is long enough to detect reactions that occur after five days. Plotkin begins to answer, "No. They would be," and Siri interrupts. Plotkin finishes anyway: "They would be reported separately as observed in the clinic."
That completed sentence appears in none of the circulating clips. Neither does what happened when the questioning moved to GSK's Engerix-B insert, marked as Plaintiff's Exhibit 11. Plotkin reads that subjects were monitored for four days after administration and immediately adds that this does not necessarily mean reactions were not collected afterward, since four days of monitoring "means active monitoring as opposed to collecting reports later on," which is standard in trials.
Siri asks whether four days is long enough to detect an autoimmune issue or a neurologic disorder arising later, and Plotkin answers no, such events "would be reported later." Siri then demands proof. Plotkin, in a hotel conference room with no trial records in front of him, says he is willing to bet the company collected reactions beyond four days and that the Food and Drug Administration (FDA) would not have allowed otherwise, and when pressed he concedes that this is "speculation based on experience."
On page 163, Siri asserts the conclusion he has been building toward, that efficacy was followed for 12 or 18 months while safety was done in four or five days, and Plotkin answers, "I do not agree with that statement."
The speculation concession is the strongest moment in the clips' favor, and it is also where the record turns. Plotkin's bet was right. The current Engerix-B prescribing information describes a clinical trial in which subjects were monitored for solicited adverse reactions for four days after each vaccination and for serious adverse events through 30 days after the last. This is the exact two-window structure Plotkin described, and the efficacy trials followed participants for a year or more, collecting safety data throughout.
An expert asked to reconstruct trial records from memory called it speculation based on experience, and the experience was correct in every particular. The label says so.
What a package insert actually is
The insert-reading genre depends on an assumption its practitioners never state: that a package insert is a catalog of proven side effects, which is not so. It is a legal disclosure document, and its own text says so.
Consider the label from that reply. The current M-M-R II prescribing information, revised in November 2025, lists GBS in Section 6, in a run of nervous system terms between subacute sclerosing panencephalitis and acute disseminated encephalomyelitis. The same list includes cough, sore throat, rhinitis (runny nose), and diarrhea.
One sentence introduces all of it: “The following adverse reactions include those identified during clinical trials or reported during post-approval use of M-M-R II vaccine or its individual components.” The section offers no rates, no comparison to background incidence, and no causality assessment. It pools nearly five decades of reports across hundreds of millions of doses, including reports that followed the old standalone measles, mumps, and rubella vaccines. The label lists GBS the same way it lists a sore throat.
In short, an entry in this section means the event was reported after vaccination. It does not mean, and was never meant to mean, that the vaccine caused it.
Other labels state the caveats outright. The Fluzone insert introduces its postmarketing list, which also includes GBS, by explaining that the events were reported voluntarily from a population of uncertain size, that their frequency often cannot be reliably estimated, and that a causal relationship often cannot be established. It then gives the inclusion criteria: severity, frequency of reporting, or strength of causal evidence, any one of which is sufficient. A condition can appear on a vaccine label because it is serious, with no causal evidence behind it at all.
An entry in this section means the event was reported after vaccination. It does not mean, and was never meant to mean, that the vaccine caused it.
The threshold is low by regulation and by design. Even the Warnings and Precautions section, the most prominent real estate on a label, requires only reasonable evidence of a causal association under federal labeling rules, which state that a causal relationship need not have been definitely established.
Every tier of the document operates below certainty, deliberately, so that clinicians hear about possible signals early. Reading any tier as a registry of proven harms subverts the document's purpose.
The studies the list never mentions
The question of whether the MMR vaccine causes GBS has an answer, and it does not live in Section 6. Finnish investigators in 2001 linked their country’s national hospital discharge register to individual vaccination records during its MMR campaign: about 630,000 vaccine recipients, 900,000 doses, and 189 GBS hospitalizations nationwide over four years.
Vaccination produced no increase over the background incidence and no clustering at any interval, and every case among vaccinated people occurred more than six weeks after the dose, with gaps ranging from 80 days to years. The 2011 Institute of Medicine review placed MMR and GBS in its "inadequate to accept or reject" category and took pains to explain that the phrase means exactly what it says, that the evidence is insufficient to establish an association in either direction.
GBS does have well-documented triggers, and they are infections, most prominently Campylobacter and influenza. For influenza the comparison has been quantified directly: an Ontario study spanning 1993 to 2011 estimated roughly 17 GBS admissions per million medically attended influenza illnesses against about 1 per million vaccinations. A person whose goal is to avoid GBS has an argument for vaccination.
The label grades its own evidence
The strongest refutation of the insert-reading genre is what labels do when the evidence changes. This March, the FDA added a febrile seizure warning to the same Fluzone insert after the agency's own controlled analyses, run across two recent seasons, found an increased risk in young children on the day after vaccination.
That entry arrived with attributable risk estimates and explicit language that the results suggest a causal relationship, and it sits in the Warnings section. In January 2025, the FDA required a GBS warning on the RSV vaccine labels after a mandated postmarketing study in older adults suggested an increased risk, while stating plainly that the evidence could not yet establish causation. In 1999, when the intussusception signal after the first rotavirus vaccine held up under investigation, the product came off the market within months.
GBS does have well-documented triggers, and they are infections, most prominently Campylobacter and influenza.
Real signals move up the document, acquire numbers, and sometimes end products. The GBS entry on the MMR label has sat in the undifferentiated pool of reports for decades, unmoved, because study after study looked for the pattern and found nothing. The two kinds of entry sit on the same pages, and the label tells any careful reader which is which.
When honesty is contorted into a confession
The deposition technique has a consistent structure, as Siri demonstrates. Establish a technical distinction, then ask questions that depend on erasing it. Read narrow label language literally and interrupt the explanation of what it means. Demand documentary proof from a witness who has no documents at hand, then brand his experience as speculation. Shift between rare specific outcomes and the entire safety database when it’s convenient. Treat the ethical constraints on trial design, such as the impossibility of randomizing children to no protection once a vaccine is proven, as evidence of deficiency.
What remains after misleading editing is a distinguished scientist appearing to concede things he explicitly denied, sometimes in the same sentence.
The insert-reading version of the technique is gentler in form and broader in reach, but it works on the same principle. It takes the least-filtered list in the document and presents it as the most definitive. It succeeds precisely because it looks like diligence, like citing primary sources. And so it falls hardest on the people trying hardest to do their homework: parents who pull up the label, find GBS in black and white, and reasonably conclude that someone has been keeping this from them.
Nobody has. The insert lists everything, because the system hides nothing. Yet the insert-reading genre teaches people to read that honesty as a confession.
A system hiding harms would not print them on the label.
That reversal is the real cost, and it compounds. Kindergarten MMR coverage has slipped below the threshold that keeps measles from circulating. This year's CDC count stands at 2,777 confirmed cases as of August 20, the highest since 1991, and every parent persuaded that the label is a suppressed truth becomes harder to reach with the actual evidence, which is voluminous, public, and printed, in its properly graded form, on the very document being misread.
But a system hiding harms would not print them on the label. The documents have been telling the truth all along, about the reports they contain and about the limits of what those data mean, and the only deception in this story is in the reading.
Dr. Scott is a clinical associate professor of infectious diseases at Stanford University School of Medicine, and a coauthor of "Updated evidence for COVID-19, RSV, and Influenza Vaccines for 2025-2026" in the New England Journal of Medicine.
The opinions voiced in CIDRAP Op-Ed pieces are the authors' own and do not necessarily represent the official position of CIDRAP.
