Retracted by Its Title: A Vaccine-SIDS Paper and a Standard That Isn't Science

A retraction is supposed to be the most boring document in science. Somebody finds a concrete defect in a published paper, the defect is real, the conclusion can’t survive it, and the journal pulls the work to keep the record clean. The trigger is a flaw you can point to on the page. That is the whole point of the machinery, and it is why the next part should bother you.

Here is the part that should bother you. A five-year-old paper was just pulled from a peer-reviewed journal after it became a political flashpoint, and the vaccine establishment’s most reliable spokesman defended the move by pointing at the cover. Paul Offit, quoted in MedPage Today this week, put it plainly: “The paper’s title tells you that it shouldn’t be published.” Not the data. Not the model. Not the methods section. The title. That is not how retraction is supposed to work, and the gap between how it’s supposed to work and how it just worked should trouble anyone watching.

Start with the paper itself, fairly, because the dominant framing is mostly right about it. The study was Vaccines and sudden infant death: An analysis of the VAERS database 1990–2019, a single-author paper by Neil Z. Miller, a longtime vaccine critic, published in 2021. Miller pulled 2,605 infant deaths out of the Vaccine Adverse Event Reporting System, found that about 75 percent of them clustered within seven days of a shot, and called that pattern “highly suggestive of a causal relationship.”

You need the mechanism to see why that claim collapses, because the mechanism explains everything. VAERS is a passive, open-submission database. Anyone can file a report, the reports are unverified, and a report sitting near a vaccine date proves only that the two happened close together. Now lay two facts on top of each other. The American infant vaccination schedule is densest in the first six months of life, with shots clustered around two, four, and six months. The peak window for sudden infant death syndrome is also roughly two to four months. Overlay a dense vaccination calendar on the natural age curve of SIDS and deaths will land near shots no matter what, for the same reason most people who die in a hospital were recently near a doctor. The proximity is the inference trap. It is what the calendar guarantees, not what a poison would prove.

The better-designed studies, the ones a VAERS database dredge cannot match, point the other way. A meta-analysis by Vennemann and colleagues pooled the case-control literature and found a summary odds ratio of 0.54 for immunization and SIDS, meaning vaccinated infants in those studies had roughly half the SIDS rate of unvaccinated ones. That is almost certainly a healthy-vaccinee effect, since families who keep up with shots also tend to follow safe-sleep guidance, rather than proof that vaccines guard against SIDS. But it is the opposite of what Miller’s clustering implies, and reanalyses of the case-control record keep landing in the same place. On the narrow scientific question, do these shots cause SIDS, the strongest evidence says no, and Miller’s paper did not lay a glove on that conclusion.

So why does the retraction bother me more than the paper did?

Because if the paper was that flawed, the moment to stop it was peer review in 2021. Toxicology Reports published it anyway and then vouched for it for five years. On April 9, 2026, with editor-in-chief Lawrence Lash signing off, Elsevier finally pulled it, in a notice Retraction Watch documented on May 26, citing the same VAERS limitations that were visible on day one. The notice points to serious methodological flaws in using passive surveillance data to infer causation. It does not allege fraud or fabricated data. Nothing inside the paper changed between 2021 and 2026, in other words. What changed was the weather around it. The paper turned into a national fight the moment vaccine policy did, and only then did the journal find it intolerable enough to remove. A retraction triggered by the temperature outside the paper rather than a defect inside it is not a correction. It is a verdict on the conclusion.

Offit said the quiet part out loud. Judging a paper by its title is a conclusion-first standard: you decide the answer is unacceptable, and the methods critique arrives afterward to justify a decision you already made. And it is worth knowing who is applying that standard. Offit co-invented RotaTeq, the rotavirus vaccine Merck sells, and Children’s Hospital of Philadelphia, where he holds the vaccine chair, sold its royalty interest in that patent for $182 million in 2008, a deal that reportedly paid Offit himself a multimillion-dollar share. The man telling you a vaccine-safety paper should never have seen print made a fortune from a vaccine. Miller’s bias is worn openly; everyone knows he argues against immunization for a living. Offit’s financial history is the part that rarely makes it into the same sentence as his verdict. Both belong on the table.

The sharpest line came from Health Secretary Robert F. Kennedy Jr., who blasted the retraction and asked the question that actually matters: “Americans have a right to know why scientific papers are removed, who made those decisions, what evidence supported them, and whether the same standards are applied consistently.” That last clause is the whole game. The public-health apparatus leans on VAERS as a frontline national early-warning system when it wants to flag a signal, then dismisses it as unverified noise the instant someone reads a signal it dislikes. You cannot run a database as the country’s vaccine-safety smoke detector and also tell the public it means nothing the moment it goes off in a direction you find inconvenient. Pick one.

None of this rehabilitates Miller’s causal claim, and it doesn’t need to. The question was never whether Miller proved that vaccines cause SIDS. He didn’t. The question is whether a journal applies its retraction standard the same way regardless of which direction a result points, and here it plainly did not. Toxicology Reports published a weak paper, vouched for it for five years, then pulled it on a standard no serious scientist would defend in writing: the title offends us. When the gatekeepers reach for that standard, they are not protecting the public from bad data. They are teaching every researcher watching that the safe move is to never ask the inconvenient question at all.

Watch what comes of it next. Kennedy’s HHS says it will rebuild VAERS into a system whose signals can’t be waved away on convenience, which is the easier half of the fix. The harder half is editorial: a retraction standard that fires on demonstrated error and fires the same way no matter which direction the result points. Science earns trust by publishing an uncomfortable finding and then openly dismantling it in full view, receipts on the table. It loses trust the day it starts pulling papers off the shelf because of what’s printed on the cover. This week, in a journal that should have known better, it chose the second path.

Sources

1. Retraction Watch – Elsevier retracts study tying sudden infant death syndrome to vaccinations (May 26, 2026)
2. MedPage Today – “What We Heard This Week,” Paul Offit on the retracted paper (June 21, 2026)
3. ScienceDirect – Miller, Vaccines and sudden infant death: An analysis of the VAERS database 1990–2019 (removed)
4. PubMed – Vennemann et al., “Do immunisations reduce the risk for SIDS? A meta-analysis”
5. Vaccine – Reanalyses of case-control studies on the temporal association between SIDS and vaccination
6. The MAHA Report – Kennedy blasts retraction of the SIDS-vaccine article
7. Wikipedia – Paul Offit (RotaTeq co-inventor; CHOP’s $182M royalty sale)
8. Children’s Hospital of Philadelphia – Vaccines and Sudden Infant Death Syndrome (SIDS)
9. Institute for Vaccine Safety – Do Vaccines Cause Sudden Infant Death Syndrome (SIDS)?
10. Tech ARP – Fact Check: Toxicology Reports infant vaccine study retracted