Scientists on the other side of the argument were quick to call the study flawed and suggest its publication might lead more women to take codeine for pain associated with childbirth.
But even the authors themselves said that the findings shouldn't be seen as an endorsement of codeine use among nursing mothers — or any patient group for that matter.
"There's never any reason to prescribe codeine," said lead author Dr. David Juurlink, a medical toxicologist and scientist at Toronto's Institute of Clinical Evaluative Sciences.
The study was published in the journal Clinical Toxicology. Juurlink did the work with colleagues from the University of Toronto and Queen's University in Kingston, Ont.
The reason Juurlink and his co-authors suggested codeine use doesn't make sense relates to the way the drug acts in the body.
Codeine is what's called a prodrug, a compound that is inactive until it is metabolized in the body. Codeine is turned into morphine in the liver.
But individuals metabolize the drug at different rates, depending on their genes. Some metabolize it at an ultra accelerated rate, others more normally. And in a small percentage of people, codeine does not turn into morphine, so they get no benefit from the drug.
"When a doctor prescribes a known amount of codeine, what they're really giving the patient is an unknown amount of morphine. So it's inherently an irrational drug on that basis alone," Juurlink said in an interview.
"The reason it doesn't make sense to give codeine to somebody with pain is because you don't know whether they're going to get any opioid or not."
In recent years there have been reports in the medical literature that some breastfed infants suffered from depression of the central nervous system — sedation and slowed breathing, among other symptoms — due to codeine use in their mothers. Some of the reports came from Motherisk, a program run at the Hospital for Sick Children in Toronto.
The reports were based on small numbers of infants. So Juurlink and his coauthors looked at Ontario health data to see if they could spot whether there was evidence of an effect.
Using case files from April 1, 1998, to March 1, 2008, they identified 7,804 women who filled prescriptions for codeine shortly after childbirth. They then pulled the files for those women's babies, looking for deaths and hospitalizations within the first 30 days of life.
The researchers then compared the statistics for the infants whose mothers had prescriptions for codeine to a similar number of babies born to women who were not given prescriptions for the drug.
They found no evidence of increased deaths among the babies in the codeine group, nor did they see elevated rates of hospitalizations. And in fact the rate of deaths was higher in the non-codeine group, though there were few deaths in both groups.
"Ninety-six per cent of children did not go to hospital for any reason, and why the four per cent went to hospital isn't clear," Juurlink said.
"But we do know there was no difference between the two arms. Whether Mum got codeine or not was not a determinant of whether the child came to hospital for any reason."
Outside experts asked to comment on the finding were critical of the study, suggesting its limitations — limitations that were acknowledged by the authors — meant it couldn't answer the question it posed.
Those limitations include the fact that the authors don't know whether women who filled the prescriptions took the codeine, or for how long. They cannot rule out the possibility that women who didn't get a prescription for codeine might have been given a few doses of the drug to take home from hospital with them. And they don't know how many, if any, of the women in the two groups were fast metabolizers of codeine.
Because they were looking at billing records, not talking to actual people, the authors also do not know how many women in the two groups breastfed their babies.
Dr. Gideon Koren, director of Motherisk, said using data of this sort was "a very rough technique" that could not bring a clear picture into focus.
"They don't treat kids. They don't specialize in kids. And the study is poorly done," he said.
Koren's concerns were echoed by Gerald Briggs, a clinical pharmacologist at Miller Children's Hospital in Long Beach, Calif., and author of "Drugs in Pregnancy and Lactation."
"It was a good attempt, but the weaknesses limit the value of the study," said Briggs, who volunteered that he and Koren are friends. Briggs pointed to Koren's writings on the issue, calling them "great studies."
"It's not a common toxicity. Does it occur? Yes," he said.
Koren said he knows of three infants — breastfed by mothers taking codeine — who have had to be treated with the antidote for codeine, which he called direct proof of codeine poisoning.
But Juurlink said toxicology studies have shown that the dose of codeine passed through breast milk is low.
"The amount of morphine that a child can ingest through breast milk is measured in micrograms. And it is typically somewhere perhaps 25 to 50 times less than what we might give a child to treat pain. So it's a very, very small amount of morphine that a child can ingest through breast milk."
He said millions of women worldwide who metabolize codeine at high rates give birth every year. If even a small portion of them are taking codeine while breastfeeding, he said, one would expect to see more evidence of harm in babies than has been spotted.