Did a Celebrated Researcher Obscure a Baby’s Poisoning? [Archive]
A long read. Summary quotes —
After a newborn died of opioid poisoning, a new branch of pediatrics came into being. But the evidence doesn’t add up.
A long read. Summary quotes —
The coroner’s office asked one of Canada’s leading pediatricians and toxicologists, Gideon (Gidi) Koren, to examine Tariq’s file. For the past two decades, Koren had been running a program at the Hospital for Sick Children called Motherisk, which provided guidance for pregnant women and new mothers about drugs and breast-feeding. He was widely considered to be among the most capable research scientists in the field. Koren met with Rani’s physician and quickly ruled out foul play. “There was no evidence of psychiatric issues,” he later wrote. Instead, Koren interpreted the toxicology report as a scientific revelation: if mothers with a certain genetic predisposition took even a mild dose of codeine, the amount of morphine that ended up in their breast milk could kill their children.
he published his findings in The Lancet, one of the world’s top medical journals. Some women, like Rani, have a genetic predisposition to convert codeine into morphine faster and in higher quantities than the rest of the general population. Therefore, the authors concluded, “codeine cannot be considered as a safe drug for all infants during breastfeeding.”
Few academic-journal articles have had so abrupt an effect on the daily practice of medicine. Prior to its publication, the American Academy of Pediatrics had listed codeine as generally compatible with breast-feeding. “After we published it in Lancet, the F.D.A.—the Food and Drug Administration—said, ‘This is enough for us to change labelling,’ ” Koren said. Canadian and European health regulators soon followed suit.
Bateman told Juurlink that when he first read the Lancet report he’d thought, This can’t be true. “The science on metabolism—codeine to morphine—was beautiful,” Bateman said. But the numbers were off. Ultra-rapid metabolizers are generally exposed to around fifty per cent more morphine than the average person. And yet, though Rani had been taking only a fraction of her prescribed dose, Tariq had died with a concentration of morphine in his blood which was more than fifty times higher than the midpoint of the expected range.
Before his visit to Scotland, Juurlink had never questioned Koren’s findings. He had even used the Lancet paper to teach medical students how individual genetic variations can pose unexpected risks in the use of certain drugs. But, if Bateman was right, the implications went far beyond the revelation of a possible crime. The opioid crisis was taking off in North America. With codeine now considered unsafe for breast-feeding, millions of new mothers were being prescribed stronger, more addictive opioids instead.
Bateman had detailed his concerns about Tariq’s death in a letter he’d sent to The Lancet in 2007. Koren, using arguments from the Jamiesons’ lawyer, had tried to block its publication by insinuating both that it was defamatory and that Bateman was a paid shill for the manufacturer of Tylenol-3. (A different subsidiary of Johnson & Johnson had named Bateman on an unrestricted educational grant that it awarded to the University of Edinburgh, where he was a professor at the time.)
Koren wrote that Bateman’s critique was “frivilous and based on errors”; Bateman revised and resubmitted. When his letter finally ran, in August, 2008, it did so alongside a derisive rebuttal from Koren and his team. “Nicholas Bateman and colleagues’ comments seem to stem from fundamental flaws in their understanding of perinatal toxicology,” they wrote. They went on to attribute the high concentration of acetaminophen in Tariq’s blood to “postmortem redistribution”—the phenomenon of drugs migrating through the body after death, potentially skewing toxicological results—then cited a study that did not support their conclusion.
Back in Toronto, Juurlink searched the scientific literature for clues. His specialty was complex drug interactions and poisonings in adults—not breast-feeding, or infants—and it had never occurred to him that Koren could be so wrong, so publicly, about a matter that was so consequential, and so squarely within his longtime professional focus. And yet, the more closely Juurlink studied the issue, the less confidence he had in Koren’s interpretation.
Koren and his team kept publishing on codeine, repeating their warning that the drug, if taken by new mothers, could pass through breast milk at levels that might kill infants. There was almost no real pharmacokinetic data to back up this assertion, only extrapolation and anecdotes.
Only one of Koren’s cases stood out as “extremely compelling,” Juurlink told me, for the clarity and quantity of its medical data. “A one-week-old boy was seen in the emergency department with a two-day history of poor feeding and increasing lethargy,” Koren and a colleague named Michael Rieder wrote, in Paediatrics & Child Health, the journal of the Canadian Paediatric Society. His breathing was slow and shallow. “This baby had the classical combination of lethargy and bradypnea associated with opiate overdose,” Koren and Rieder reported. They referred to him as Baby Boy Blue.
Years after publication, Juurlink shared a taxi with Rieder, Koren’s co-author on the paper, while they were attending a professional meeting in Ottawa. By then, Juurlink had been studying the death of Tariq Jamieson for a decade, and had found no other credible case of an infant dying from breast-feeding. The only data point in the scientific literature that had shaken his theory of the case was the near-death of Baby Boy Blue. He asked Rieder about the case.
“Oh, we made it up,” Rieder replied.
every detail was fiction. Koren and Rieder had even invented Baby Boy Blue’s siblings, a five-year-old sister, who was born in Sri Lanka, and a three-year-old brother, who was “born in Canada by caesarean section because of failure to progress.” The morphine concentration was implausibly high because it was fabricated. No life was jeopardized; no life was saved.
according to an e-mail that Rieder sent Juurlink years later, the case was created as “a cautionary tale,” for teaching purposes. No such disclaimer appeared in print. Meanwhile, the paper has been cited in at least one court case and in a doctoral thesis. “Pathologists and forensic toxicologists have come to accept the idea of ‘death by breast milk’ based upon terribly sloppy work that began in Gidi’s lab,” Juurlink wrote to Rieder. “Experts and the courts are being misled by this belief. Unfortunately, your case work contributes to that misconception.” Rieder said that the paper would “likely” be updated with a disclaimer in 2024, some fourteen years after it was published. But this has not happened.
In July, Rani sent me a copy of Koren’s original draft of the Lancet paper, in order, she said, to demonstrate that there was never an attempt to conceal Tariq’s blood-codeine level, which Juurlink had treated as a kind of smoking gun. She was right: the codeine and acetaminophen levels were present. The draft was twelve pages long, including references, and The Lancet had asked Koren to shorten it to a page. It is unclear why, during that process, Koren removed the codeine and acetaminophen readings. But, shortly after publication, he promoted his young graduate student Parvaz Madadi as a new expert on opioids in breast milk.
“Gidi had a tendency to just submit papers without the co-authors being advised or consulted, and let the peer-review process handle all the shitty nitty-gritty, you know?” a person who worked with the Motherisk program for several years told me. “Rather than sit back and be, like, ‘Let’s make sure, before we go forward with this paper, that we have all our ducks in a row.’ ”
They conducted a study of some hundred and seventy thousand new mothers, to see if infants of women who were prescribed opioids shortly after birth were at an increased risk of harm. What they found, instead, was that many women who are prescribed opioids postpartum appear to avoid breast-feeding, in order to “protect” their children—and thereby deprive them of immunological and other benefits. “The number of infants affected by this globally is now easily in the millions,” Juurlink told me.
“The case against science is straightforward: much of the scientific literature, perhaps half, may simply be untrue,” Richard Horton, the editor of The Lancet for the past thirty years, wrote in the journal, in 2015. “In their quest for telling a compelling story, scientists too often sculpt data to fit their preferred theory of the world. Or they retrofit hypotheses to fit their data. Journal editors deserve their fair share of criticism too. We aid and abet the worst behaviours.”
In 2020, Juurlink, Zipursky, Naylor, and several other experts—including Nick Bateman, the Scottish toxicologist; Thomas Hale, one of the world’s leading scientific authorities on breast milk; and Ronald Cohn, who had replaced Naylor as president and C.E.O. at the Hospital for Sick Children—wrote to The Lancet, asking the journal to retract Koren’s paper. “For more than a decade now, Koren either knew or ought to have known that his original interpretation of the case was flawed yet he took no corrective action,” the letter read. “He instead continued to publish work citing this case as foundational to consideration of the risks of opioid use while breastfeeding.”
The journal referred the request back to the Hospital for Sick Children, which established an internal research-integrity review. Since Cohn had signed the letter asking for retraction, the lead inquirer did not report to him.
What followed was an exercise in obfuscation through institutional review. There was no meaningful effort to assess whether Juurlink’s analysis disproved Koren’s; instead, the reviewers narrowed the scope of their inquiry to the matter of whether Koren’s team had demonstrated such a “lack of rigour” that retraction was unavoidable.
Koren sent the review team a bizarre eight-page document, which contained factual errors, non sequiturs, multiple font sizes, and what appeared to be copied-and-pasted correspondence with Rani. The hospital’s research-integrity office chalked it up to a scientific dispute. The editors of the Canadian practitioners’ journals offered to provide The Lancet with the unambiguous findings of their own review, but The Lancet declined. An independent expert I contacted told me that, some years ago, The Lancet had asked him to weigh in on the subject—then ignored his advice to retract. Koren’s case report remains on the journal’s website, with no notes or corrections appended, and serves as the core basis for regulatory guidance on the subject of breast-feeding and codeine all over the world.
In the decades since Koren’s first warnings concerning codeine and breast milk, public-health authorities and patient-advocacy services have issued guidance to new mothers that ranges from scientifically incoherent to potentially dangerous. A clinical report published by the American Academy of Pediatrics in 2013 cites the Lancet case as a reason to avoid prescribing codeine to breast-feeding mothers but notes that morphine—its metabolite—“appears to be tolerated by the breastfeeding infant.” The same guidelines also recommend the use of hydromorphone, which is about forty times more potent than codeine and can be highly addictive. Meanwhile, the U.K.’s National Health Service categorically warns against taking codeine while breast-feeding but allows for the use of fentanyl.
