Anything But Vaccines

An Essay on the Alternative-Cause List in “It Happened to Audrey”

Author’s note. The phrase “anything but vaccines” I owe to Sasha Latypova. It names the streetlight effect at its endpoint, where the excluded question is structurally unavailable. I am writing about Audrey Edmunds now because a woman in Idaho named Andrea Shaw was arrested on July 1, 2026 and charged with two counts of first-degree murder over the deaths of her twin toddlers, Dallas and Tyson, who died eight days after receiving DTaP, hepatitis A, and flu vaccines at an 18-month visit.²⁵ Shaw and her mother-in-law had warned the pediatrician about a family history of flu-shot reactions and been dismissed. The emergency room diagnosed the twins with “post-immunization reaction, initial encounter” the day after the injections. A week later both twins were found unresponsive. The Edmunds case is thirty years old. Audrey Edmunds and her co-author Jill Wellington laid it out in It Happened to Audrey in 2012. This essay reads that book and asks one question of its record. The case is not old news.

A close reading of one exoneration.

Perloff’s presentation

Six weeks after seven-month-old Natalie Beard died in Audrey Edmunds’s home daycare on October 16, 1995, Dr. William Perloff, head of the University of Wisconsin-Madison Pediatric Intensive Care Unit, presented a case to a continuing education class for emergency medical technicians.¹ He did not name the case. He told the students that the babysitter had claimed the baby choked on milk, but no milk was found in the airway.

Lorraine Endres was in the audience. She had responded to Audrey’s house that morning. She recognized the case immediately. Her own EMT report had documented a milky substance in Natalie’s nose and mouth. The police chief’s report had documented the same. The transport physician’s observation from Med Flight had documented the same.² Perloff was Wisconsin’s expert witness on shaken baby syndrome. His trial testimony had convicted Audrey. Endres wrote to Audrey’s defense attorney: “Why did Dr. Perloff lie about that? If he lied about the milk, what else did he lie about?”³

The state’s diagnostic construct required no milk in the airway. The triad the prosecution rested on, subdural hematoma, retinal hemorrhage, and cerebral edema, was held out at trial as definitive proof of violent shaking. Milk in the airway pointed at choking. Choking pointed away from shaking. The state’s expert edited the physical record to fit the diagnostic story.

The construct that required the edit eventually collapsed. Audrey came home in 2008 because appellate physicians persuaded a Wisconsin court that the triad had many possible causes.⁴ This essay is not about the construct that collapsed. It is about the architecture that surrounds it. What Perloff did to preserve one diagnostic story is what a professional culture does at scale to preserve others. Endres’s question is the frame this essay adopts. If Perloff lied about the milk to protect one construct, what does the defense’s own alternative-cause list look like when we ask what it left out?

This work stays free because paid subscribers make it possible. They get the full book library, the Deep Dive Audio Library, and the Questions for Your Doctor, Before You Consent, and Package Insert series. No grants, no gatekeepers — your subscription is what keeps it that way.

The list

Four alternative explanations for Natalie’s collapse reached the record in Audrey Edmunds’s case. Natalie choked on her formula, and the milk in her airway triggered the collapse. An ear infection, treated for four days with amoxicillin, tracked to the brain. A subdural hematoma sustained during birth had become chronic, formed a membrane, and rebled. A rebleed from ordinary handling of an infant with a pre-existing injury produced the triad of findings the state’s experts had read as shaking.⁵

Each explanation was supported by testimony from a credentialed physician. Each attributed causation to a natural event, to Natalie’s own biology, or to an ordinary domestic occurrence. None named a product. None named a manufacturer. None named a schedule.

Natalie was almost seven months old. She had been on antibiotics for four days for an ear infection.⁶ The morning she collapsed, her mother arrived at Audrey’s door exhausted, saying the baby had been fussy through the night and had refused her bottle.⁷ What appointments Natalie had attended in the preceding weeks, what injections she had received on the standard childhood schedule, does not appear in the 351 pages of It Happened to Audrey, the 2012 memoir Audrey Edmunds wrote with journalist Jill Wellington documenting her arrest, conviction, and exoneration.

The word vaccine does not appear once.

The absence is not a memoir editing choice. It is the record. The trial did not raise the question. The 2007 post-conviction hearing that produced the vacatur did not raise the question. The Stanford pediatric neuroradiologist who testified to multiple possible causes of the triad, listing birth trauma, prior infection, clotting disorders, short falls, and rebleeds from earlier injuries, did not raise the question.⁸ Everyone with an interest in freeing Audrey Edmunds selected causes from a permitted range. The question this essay pursues is what defines the range.

What the record shows

Dr. Patrick Barnes, called by the defense at the post-conviction hearing, walked the court through what mainstream medicine had come to accept in the decade after Natalie’s death.⁹ The triad of subdural hematoma, retinal hemorrhage, and cerebral edema, once treated as definitive proof of violent shaking, had many possible causes. He named them.

He described birth trauma producing chronic subdural bleeds that form membranes and rebleed weeks or months later from ordinary handling. Natalie’s own birth records showed a bruise on the back of her head, consistent with the birth process itself.¹⁰ He described infection tracking from the middle ear to the brain, a mechanism known for centuries. Natalie’s CT scan showed fluid or thickening in the ear.¹¹ Asked directly whether a simple ear infection a week before could be a cause of what happened, Barnes said yes.¹²

Dr. Janice Ophoven, a pediatric forensic pathologist, had earlier reviewed the case for the defense and concluded that the timing of Natalie’s injuries could not be established with certainty, and that the injuries could have occurred before Natalie arrived at Audrey’s home that morning.¹³ Dr. John Plunkett, a coroner with the Minnesota Regional Coroner’s Office, had reviewed the trial transcript and reported no scientific data supporting the shaken baby conclusions the prosecution had presented at trial.¹⁴ Dr. John Galaznik, a pediatrician at the University of Alabama, testified at the post-conviction hearing that Natalie’s symptoms were consistent with choking rather than trauma, and that the timing of her respiratory failure matched a choking event.¹⁵

The physical evidence for the choking hypothesis was already in the record. The milk in Natalie’s nose, mouth, and airway had been documented by first responders and the transport physician on the morning of the collapse. Perloff had denied it at trial and again at the EMT continuing education class. The defense had the physical evidence and had a credentialed pediatrician willing to build the timing argument on it. What the defense did not have, on the record the memoir preserves, was an inquiry into what else had happened to Natalie in the weeks before the ear infection that put her on amoxicillin.

What the record does not show

The six-month visit on the 1995 American childhood vaccination schedule called for the third dose of the diphtheria-tetanus-pertussis vaccine, the third dose of the oral polio vaccine, and the third dose of the Haemophilus influenzae type b vaccine.¹⁶ Natalie was almost seven months old on the morning she collapsed. Her most recent scheduled visit, on the standard American schedule, would have fallen in the weeks preceding the ear infection that put her on amoxicillin.

Whether Natalie attended that visit, what she received if she did, and what, if any, adverse reactions followed, does not appear in Audrey’s memoir.

This is not because the memoir is thin on medical detail. The book documents the antibiotic: amoxicillin, pink liquid, four days.¹⁷ It documents the birth bruise on the back of Natalie’s head.¹⁸ It documents the CT scan findings on ear fluid.¹⁹ It documents Natalie’s slow motor development, her difficulty rolling over at nearly seven months, her inability to hold her own bottle.²⁰ The memoir names Natalie’s pediatrician, Dr. Susan Padberg, and records that she was called by the prosecution.²¹ What the memoir does not do, and by extension what the record it draws from did not do, is ask what was in the pediatric file besides the antibiotic prescription.

The pediatric file existed. Padberg was called to the stand. The defense had access to Natalie’s medical history. The question that established the birth trauma theory was: what does the birth record show? The question that established the infection theory was: what does the CT scan show? The question that established the choking theory was: what did the EMTs find in the airway? Each question, asked, produced an alternative the court accepted. The question that did not get asked was: what does the vaccination record show?

The book records no such inquiry. The trial produced no such testimony. The Wisconsin Court of Appeals, in vacating the conviction in 2008, considered no such theory. The single most concrete piece of medical history that a defense would examine for any infant with a sudden neurological collapse, in the paradigm the defense declined to enter, was systematically absent from the record.

The shape of the list

Every alternative that made it into the exoneration case shares a structural feature. It points at nature, at the mother’s body, at the baby’s own biology, or at ordinary domestic accident. Choking is a natural airway event. The ear infection was Natalie’s own, tracking through her own anatomy. The chronic subdural formed during her own delivery. The rebleed arose from routine handling of a fragile pre-existing injury. Nothing on the list points at a manufactured product. Nothing on the list points at a licensed medical intervention. Nothing on the list points at anyone who could be sued, regulated, or defunded by the alternative being credited.

The list is coherent. Coherence of this kind, across a decade of appellate work involving multiple experts and defense attorneys, does not appear by accident.

What separates the vaccine hypothesis from the ear-infection hypothesis, on the evidence available to the Edmunds defense, is not evidentiary weight. It is where the causal arrow lands.

The ear infection hypothesis lands on Natalie. The choking hypothesis lands on the feeding. The birth trauma hypothesis lands on delivery. The rebleed hypothesis lands on ordinary handling. The vaccine hypothesis lands on a scheduled pharmaceutical product manufactured by a small number of corporations, mandated for school attendance, promoted by the pediatric establishment, and administered by the same pediatrician on whose testimony Natalie’s parents relied. The other hypotheses require an appellate court to revise its understanding of biomechanics. The vaccine hypothesis requires the court to revise its understanding of who is being protected by the diagnosis it has been asked to overturn.

Why the list has that shape

In 1986 the United States Congress passed the National Childhood Vaccine Injury Act.²² The Act removed vaccine injury from ordinary civil courts and reassigned it to a specialized administrative program, the Vaccine Injury Compensation Program, funded by an excise tax on each dose. The Program adjudicates injury claims under a schedule of presumed vaccine-attributed conditions. Manufacturers of scheduled childhood vaccines are shielded from civil liability for design defects.

The Act did more than create a compensation route. It created a jurisdictional partition. Vaccine causation now had its own courthouse. The consequence for every other courthouse followed by professional culture rather than by statute. A defense attorney raising vaccine causation in a criminal court after 1986 was raising a question the legal system had already assigned to a different venue. The formal barrier is thin. The reputational barrier is not.

Consider the defense attorney’s problem. The client faces a conviction that will end her life as she knows it. The judge is receptive to alternative causes for the triad, provided the alternative comes from a credentialed source and does not exceed what the court will code as legitimate medical dissent. The available alternative-cause menu includes short falls, rebleeds, clotting disorders, infections, and birth trauma. Each of these has published pediatric literature behind it. Each has expert witnesses who will testify without being marked as fringe. The defense selects from this menu.

The vaccine hypothesis is not on this menu. It sits in a separate legal jurisdiction. Its expert witnesses are professionally marginalized. Its published literature circulates in journals the mainstream declines to cite. A defense attorney who raises it in a criminal courtroom is not making a technically prohibited argument. She is making an argument that comes into the courtroom already coded as fringe, with an expert whose credentials the prosecution will attack, in a factual domain where the state has arranged for authoritative pronouncements to come from a different agency. The rational defense attorney does not raise it. Not because she has been told not to. Because the calculation of acquittal probability instructs her not to.

Over time and cases, the shape of the alternative-cause menu reinforces itself. Junior attorneys observe what senior attorneys argue. Defense expert witnesses maintain their standing by staying within the accepted range. The set of thinkable alternatives narrows to the set of successfully argued alternatives, which is the set the courts have already coded as legitimate. The excluded set becomes not merely difficult to argue but difficult to consider.

This is what an epistemically captured environment produces at the level of professional practice. Nobody instructs the defense bar to leave vaccination alone. The instruction is not needed. The 1986 Act made the instruction structural.

The obvious objection

The objection to this reading is that the Edmunds defense may simply have found nothing in the vaccination record worth raising. Natalie may not have received an injection in the relevant window. The shot record may have been examined and set aside. The absence in the memoir may reflect the absence of a finding rather than the absence of an inquiry.

This objection assumes the defense looked. The memoir provides no evidence that they looked. The birth records were examined, and produced the head bruise Barnes used to argue chronic subdural. The CT scan was reinterpreted, and produced the ear-fluid finding. The pediatrician was called to the stand. What the memoir does not record is anyone requesting the vaccination record, examining it, and finding nothing worth raising. The record is not silent on the shot record because the shot record was examined and dismissed. The record is silent because the shot record was not the object of inquiry.

The argument does not depend on this specific case. Even accepting that Edmunds’s defense examined the vaccination record and found nothing, the pattern in the shaken baby literature is that the vaccination record is not the standard object of defense inquiry. A single case may exit that pattern for individual reasons. The pattern itself is not the accumulation of independent choices not to look. It is the shape of what professional culture has made unavailable to raise.

The Edmunds memoir is one document in that pattern. Its silence is data whether the silence was chosen at the trial table or produced structurally by the culture the trial table sat inside.

What was returned and what was not

Audrey Edmunds went home in February 2008 to three daughters, Carrie, Allison, and Jennifer, who had grown up while she was in prison.²³ She had raised no other child in the eleven years she was gone. The court that freed her did not rule that she was innocent in the abstract. It ruled that the science on which her conviction rested had moved sufficiently to warrant a new trial. The state declined to retry.²⁴

The exoneration was internal to the establishment. The experts held mainstream credentials. The mechanisms they proposed were published in mainstream journals. The court was persuaded by a critique of the shaken baby construct produced by physicians trained inside the same medical schools that produced the doctors who convicted her. Nothing about the exoneration required stepping outside the licensed range of medical doubt.

The list of alternatives that produced her freedom is a document about what the legal and medical systems will accept. Choking. Ear infection tracking to the brain. Birth trauma producing chronic subdural that rebled from ordinary handling. Anything the courts could code as natural, biological, or accidental. Anything but a scheduled product manufactured by a shielded industry.

Perloff kept a lie about milk in the airway going for years because the diagnostic construct required it. What Perloff did is small. He edited one physical finding to fit one story in one continuing education class. The Act of 1986 does the same thing at the scale of a professional culture. It edits an entire causal domain out of every criminal courthouse in the country by relocating the question to a specialized administrative program that no jury will ever hear from.

Audrey Edmunds spent eleven years in prison because a lie about milk succeeded at trial. She came home because the diagnostic construct that required the lie eventually cracked. Which cause was Natalie’s, none of them will ever know. The one cause the courts would not consider was the one that would have named a manufacturer. Somewhere in the same court system, in every year the vaccination schedule has expanded since 1986, another caregiver is serving the same years for the same reason. The alternative that would free her sits in a different courthouse.

---

References

1. Edmunds, A. and Wellington, J. It Happened to Audrey: A Terrifying Journey from Loving Mom to Accused Baby Killer. TitleTown Publishing, Green Bay WI, 2012. Account of Perloff’s continuing education presentation, drawn from the letter Lorraine Endres wrote to defense attorney Dean Strang.

2. Ibid. The milky substance in Natalie’s nose, mouth, and airway is recorded in the EMT report, the police chief’s statement, and the observation of the Med Flight transport physician on October 16, 1995.

3. Ibid., Endres’s letter to Strang, quoted in the memoir.

4. State v. Edmunds, 2008 WI App 33, 308 Wis. 2d 374. Wisconsin Court of Appeals granted a new trial on the ground that a shift in mainstream medical opinion regarding shaken baby syndrome constituted newly discovered evidence. The Dane County District Attorney declined to retry the case and Edmunds was released.

5. Edmunds and Wellington. The four alternative causes are presented across the trial narrative and the post-conviction proceedings.

6. Ibid., Chapter One. Cindy Beard reported to Audrey on the morning of October 16, 1995 that Natalie had been on the antibiotic for four days.

7. Ibid., Chapter One.

8. Ibid., testimony of Dr. Patrick Barnes at the 2007 post-conviction hearing.

9. Ibid. Barnes, a pediatric neuroradiologist at the Lucile Packard Children’s Hospital at Stanford University, had used the triad diagnosis in his own practice before revising his position on the strength of the developing literature.

10. Ibid., Barnes testimony on the birth-record head bruise and the mechanism of chronic subdural hematoma with rebleed.

11. Ibid., Barnes testimony on CT scan findings of ear fluid or thickening.

12. Ibid., direct examination of Barnes by Keith Findley of the Wisconsin Innocence Project.

13. Ibid., case review by Dr. Janice Ophoven, pediatric forensic pathologist, submitted to defense attorney Dean Strang during the first appeal.

14. Ibid., case review by Dr. John Plunkett, coroner with the Minnesota Regional Coroner’s Office, submitted to the defense during the first appeal. Plunkett had published on the topic in the journal Child Abuse and Neglect.

15. Ibid., testimony of Dr. John Galaznik at the post-conviction hearing.

16. Centers for Disease Control and Prevention, Recommended Childhood Immunization Schedule, United States, 1995. Standard six-month interventions in this period were DTP dose three, OPV dose three, and Hib dose three.

17. Edmunds and Wellington, Chapter One.

18. Ibid., trial narrative and Barnes testimony.

19. Ibid., Barnes testimony.

20. Ibid., Chapter One.

21. Ibid., trial narrative.

22. National Childhood Vaccine Injury Act of 1986, Pub. L. No. 99-660, codified at 42 U.S.C. § 300aa. Established the Vaccine Injury Compensation Program administered by the U.S. Court of Federal Claims, funded by an excise tax on covered vaccines, and providing manufacturers with liability protection for injuries arising from unavoidable adverse effects. Supreme Court affirmed the shield’s near-categorical reach in Bruesewitz v. Wyeth LLC, 562 U.S. 223 (2011).

23. Edmunds and Wellington, dedication page and closing chapters.

24. State v. Edmunds, 2008 WI App 33.

25. Baletti, B., “Breaking: Mother of Twins Who Died 8 Days After Vaccinations Charged With Murder,” The Defender (Children’s Health Defense), July 1, 2026. Reporting the arrest of Andrea Shaw of Payette, Idaho, on two counts of first-degree murder in the deaths of her 18-month-old twins, Dallas and Tyson Shaw, who died on May 1, 2025, eight days after receiving DTaP, hepatitis A, and flu vaccines. Emergency room diagnosis on April 24, 2025 was “post-immunization reaction, initial encounter.”