Twenty Years of Unnecessary Pain
If you are not familiar with the case, check Dotan’s cronology at the post: https://99nicu.org/forums/topic/3009-the-codeine-and-breastfeeding-controversy-2006-2026-how-a-flawed-case-report-reshaped-neonatal-guidelines-and-why-the-lancet-retracted-it-20-years-later/
The Lancet has retracted the codeine case report that shaped two decades of postpartum analgesia. The retraction took twenty years. Mothers paid for every one of them.
On 21 November 2026, The Lancet retracted a 2006 case report that blamed a newborn's death on morphine passed through breast milk. An institutional investigation found that key toxicology data had been left out of the paper, and concluded that the omission amounted to falsification.
A retraction is supposed to close a chapter. This one should open a reckoning. For twenty years, a single case report told women who had just given birth that relieving their pain could kill their babies. Many believed it. Many of their doctors believed it too. And many women went home sore, sleepless and afraid, trying to breastfeed through pain that did not need to be there.
One baby, one paper, one global rule
The case is familiar to anyone who has worked in a postnatal ward since 2006. A baby boy died in Toronto in 2005, at under two weeks of age. His mother had been prescribed acetaminophen with codeine (Tylenol 3) after the birth. The authors proposed that she was an ultra-rapid CYP2D6 metabolizer, that she turned codeine into excess morphine, and that the morphine reached her son through her milk.
The story was elegant, frightening and easy to teach. It was also built on an incomplete record. The SickKids Research Integrity Office found that the paper left out the infant's blood codeine and acetaminophen concentrations, and the finding of codeine in his stomach contents. Those are not footnotes. In 2020, Zipursky and Juurlink showed that the baby's blood codeine level was roughly 100 times higher than breastfeeding could explain. Codeine in an infant's stomach points to a different route of exposure altogether.
The doubts were not new. The Lancet published a letter questioning the conclusions in 2008. Canadian Family Physician and the Canadian Pharmacists Journal retracted the related columns in 2020. The Lancet issued an expression of concern only in February 2026, after a New Yorker investigation and new allegations from a former member of the lead author's own lab.
Meanwhile, the case travelled. It was cited hundreds of times and underpinned warnings from the FDA, the European Medicines Agency, Health Canada and ACOG. When the FDA recommended against codeine for breastfeeding women in 2017, the lead author of the retracted report was an author on seven of the 19 papers cited in support.
How many women hurt for nothing?
We will never have a precise number. No registry counts the perineal tears left undertreated, the cesarean wounds managed with "just paracetamol", or the mothers who chose to stop breastfeeding because they were told they had to choose. But codeine had been taken by millions of nursing mothers, and the warning reached all of them. It also spilled far beyond codeine. Fear of "opioids in the milk" became fear of analgesia in general. A lactating woman in pain became, first and foremost, a risk to be managed.
This matters for the very thing the warnings claimed to protect. Pain is not a neutral bystander in lactation; it works against it.
Pain inhibits the milk-ejection reflex. Stress and pain blunt oxytocin release, and classic physiology studies showed that distress can inhibit let-down. A mother bracing against pain at every feed is not well placed to release milk.
Pain predicts early weaning. Breast and nipple pain are among the most common reasons women stop breastfeeding in the first weeks. Higher pain during labour has also been linked to breastfeeding cessation at 6–10 weeks.
Pain predicts depression and chronic pain. In a prospective cohort of 1,288 women, the severity of acute pain after birth, not the mode of delivery, predicted persistent pain and depressive symptoms at 8 weeks. Each point of acute pain was associated with an 8.3% increase in depressive symptoms. The authors already noted, in 2008, that postpartum pain was undertreated partly because of breastfeeding fears.
So the message "endure the pain to protect your baby" was not only unproven. It was counterproductive. It pushed women towards worse pain, worse mental health and shorter breastfeeding. Twenty years of pain, and twenty years of fear that the act of feeding their children might harm them.
The patriarchy of evidence
It is tempting to file this under individual misconduct and move on. We should not. One man's paper could only do this much damage because the system around it was ready to believe it. Lactating women have been excluded from drug research for generations, in the name of protecting them. The result is not protection; it is an evidence vacuum. Into that vacuum, a single dramatic case report from a prestigious institution became global policy. No trial was needed. No one asked how many mothers would be harmed by the alternative.
Notice, too, where the burden of caution fell. Every warning was addressed to the mother. She had to monitor her baby for sedation. She had to choose between her own comfort and her child's safety. Her pain was treated as an acceptable cost; her baby's hypothetical risk was not. That asymmetry is not science. It is an old idea about women, that their suffering is natural, especially around childbirth, and that a good mother puts up with it. The same researcher founded Motherisk, a hospital laboratory whose hair-testing results were later found inadequate and unreliable. Those tests had been used in child-protection cases, against mothers. The pattern is hard to miss: authority concentrated in one expert, women's bodies treated as a source of danger, and mothers carrying the consequences.
Women's health suffers when research is done about women rather than with them. It suffers when lactation is treated as a contraindication rather than as a physiological state that deserves its own pharmacology. And it suffers when the correction of an error takes twenty years, while the error itself took about a year to become a regulatory warning.
Who profits from what we prescribe
There is another actor in this story, and it is not a single author. Pain is a market, and analgesic prescribing has never been shaped by evidence alone. The North American opioid epidemic is the starkest example. Purdue Pharma launched OxyContin in 1996 with a nationwide campaign of sales visits, sponsored talks and payments to prescribers; within five years it was the most prescribed Schedule II narcotic in the United States. Later analyses found that the more opioid-related marketing payments doctors in a county received, the higher that county's prescription-opioid overdose deaths the following year. When Purdue pulled back under litigation, competitors stepped up their own promotion to the same prescribers.
The codeine warnings did not land in a neutral space. Once codeine was off the table for nursing mothers, guidance pointed towards "strong" opioids instead, in a country where those were already being pushed hard. An unreliable case report and an aggressive industry pulled in the same direction: away from simple, cheap, well-known drugs.
Brazil offers an instructive contrast. Here, dipyrone (metamizole) is one of the most used analgesics, sold over the counter and found in almost every home and maternity ward. In a national survey, non-opioid analgesics were used by 18.5% of respondents and opioids by only 0.5%. Dipyrone was never approved in the United States or Canada. My own view is that this cheap, off-patent, non-addictive drug probably helped spare Brazil an opioid tragedy on the American scale. We should not be complacent, though. Prescription opioid sales in Brazil rose between 2009 and 2015. And dipyrone's own lactation labelling is restrictive: the EMA advises against it during breastfeeding, while lactation services such as e-lactancia consider occasional single doses compatible. In other words, a drug that has served Brazilian women well is itself caught in the same thin evidence on lactation.
The point is not that one molecule is good and another evil. It is that, when lactating women are left out of research, the gap is filled by whoever has the loudest voice: a charismatic expert, a regulator acting on a single case, or a company with a sales force.
What we owe mothers now
This retraction does not mean codeine is the ideal postpartum analgesic. Its effect varies with genotype, and better options exist. The lesson is not "codeine was fine after all". The lesson is that a lactating woman's pain is a clinical problem, and that it deserves the same rigour we give any other.
For those of us in neonatal and perinatal care, that means a few concrete things:
Treat maternal pain as part of lactation support. Ask about it at every feed assessment, not only about latch and weight.
Use multimodal analgesia without apology. Paracetamol and NSAIDs are compatible with breastfeeding; short courses of appropriate opioids, with sensible monitoring, can be too.
Retire the fear-based script. "Your medicine could hurt your baby" should never be the default sentence. Counsel with numbers, not dread.
Check the foundations of what we teach. Ask, for any guideline we repeat, how many cases and trials actually sit underneath it.
Demand that lactating women be included in research. Protection by exclusion has failed them.
The Lancet has done the right thing, late. The rest is up to us. Twenty years of mothers were told that their pain was the price of their children's safety. It never was. The least we can do is make sure the next twenty years sound different.
References
The Lancet. Retraction: Pharmacogenetics of morphine poisoning in a breastfed neonate of a codeine-prescribed mother. Lancet 2026; 408 (21 Nov 2026).
Koren G, Cairns J, Chitayat D, et al. Pharmacogenetics of morphine poisoning in a breastfed neonate of a codeine-prescribed mother. Lancet 2006; 368: 704. Retracted.
Retraction Watch. Lancet retracts report by former SickKids pediatrician on infant's death from opioid poisoning. 20 Nov 2026.
Retraction Watch. Lancet flags long-scrutinized report of infant poisoned by opioids in breast milk. 4 Feb 2026.
Zipursky J, Juurlink DN. The implausibility of neonatal opioid toxicity from breastfeeding. Clin Pharmacol Ther 2020; 108: 964–970.
Pimlott N, Tsuyuki RT. Risks of maternal codeine intake in breastfed infants: a joint statement of retraction. Can Fam Physician 2020.
US FDA. Drug Safety Communication: restricting codeine and tramadol in children; recommending against use in breastfeeding women. 20 Apr 2017.
Eisenach JC, Pan PH, Smiley R, et al. Severity of acute pain after childbirth, but not type of delivery, predicts persistent pain and postpartum depression. Pain 2008; 140: 87–94.
Investigating the association between labor pain and cessation of breastfeeding. DOAJ.
Breastfeeding-related pain among postpartum women: a mixed-methods study. J Psychosom Obstet Gynaecol 2019.
Newton M, Newton NR. The let-down reflex in human lactation. J Pediatr 1948; 33: 698–704.
Purdue Pharma's development and marketing of OxyContin (review of Purdue's OxyContin marketing). Indian J Med Ethics 2019.
Hadland SE, et al. Association of pharmaceutical industry marketing of opioid products with mortality from opioid-related overdoses. JAMA Netw Open 2019 (coverage).
University of Washington. Prescription opioid companies increased marketing after Purdue Pharma lawsuit. 9 Oct 2023.
Analgesic use in Brazil: a population-based study (PNAUM). PLoS ONE 2019.
Real world evidence of the use of metamizole (dipyrone) by the Brazilian population. Einstein (São Paulo) 2022.
Krawczyk N, Greene MC, Zorzanelli R, Bastos FI. Rising trends of prescription opioid sales in contemporary Brazil, 2009–2015. Am J Public Health 2018; 108: 666–668.
European Medicines Agency. EMA recommends aligning doses of metamizole medicines and their use during pregnancy and breastfeeding. 2018.
e-lactancia. Metamizole. Updated Sept 2024.
LactMed. Dipyrone. NCBI Bookshelf.
Recommended Comments