From the 1950s into the 1980s, millions of parents in Britain, the United States, Australia and elsewhere put their babies to sleep lying on their stomachs. They did so because doctors told them to. The most popular parenting manual of the age, Benjamin Spock's Baby and Child Care, which sold tens of millions of copies, advised that babies be placed to sleep on their fronts in editions from the 1950s onward. The advice was wrong, and it cost many lives. Cases like this deserve a hard look.
Untested reasoning
The advice on infant sleep came from a plausible chain of thought rather than a test. Doctors reasoned that a baby lying on its back might choke on its own vomit, and front sleeping seemed safer. No one ran a trial. A 2005 review by researchers led by Ruth Gilbert gathered the studies and found that evidence that front sleeping raised the risk of sudden infant death had been available from around 1970, years before official advice changed. One lesson is to ask whether a piece of expert advice has actually been tested, or only argued for.
Shared blind spots
In November 2008, as banks collapsed around the world, Queen Elizabeth II visited the London School of Economics and asked a simple question: why had nobody seen the crisis coming? A group of economists replied in a letter the next year. Their answer, in substance, was that many clever people, each watching their own part of the financial system, had failed to imagine the risks to the system as a whole. Individual experts were competent in their specialties. The failure lay in what nobody was responsible for seeing, and in a shared belief that risk had been tamed.
Evidence that drifts and money that pushes
In 1980 two doctors, Jane Porter and Hershel Jick, published a letter of a single paragraph reporting that addiction was rare among hospital patients given narcotic painkillers under close supervision. Over the next decades it was cited again and again as if it showed that opioid painkillers were safe for long-term use at home. A 2017 analysis found that the letter had been cited 608 times, mostly in support of the claim that addiction was rare, and usually without mentioning that it concerned hospital patients.
After the painkiller OxyContin was approved in 1995, its maker, Purdue Pharma, marketed it aggressively and played down its risk of addiction, and the company later pleaded guilty to federal criminal charges in 2007 and again in 2020. Experts who wanted to treat pain better, a real and humane goal, were steered by thin evidence and by money. The overdose epidemic that followed has killed hundreds of thousands of Americans.
Trusting the wrong kind of study
For years, large observational studies found that women who took hormone therapy after menopause had fewer heart attacks, and many doctors recommended it partly for that reason. But the women who chose the therapy tended to be healthier and wealthier to begin with. In 2002 a large randomized trial, the Women's Health Initiative, stopped its estrogen plus progestin arm early after finding more breast cancer, heart attacks, strokes and blood clots, not fewer. Later analysis complicated the story, suggesting that risks are lower for women who start soon after menopause, which is a reminder that corrections can overshoot too.
Who caught the errors
Notice who fixed each mistake. Epidemiologists found the danger of front sleeping. A randomized trial overturned the hormone findings. Researchers tracing citations exposed the use of the 1980 letter. In most of these cases the remedy for failed expertise was better expertise, not the absence of expertise. Journalists, grieving families and lawyers also played large parts, especially with opioids, and critics fairly point out that institutions were slow to admit their errors.
What failure does and does not show
Failures are memorable partly because they are exceptions, while quiet successes rarely make the news. Dismissing all expertise because of famous failures would mean throwing away the very methods that found them. The sensible response is to ask which kind of failure a situation invites: untested reasoning, a blind spot shared across a profession, evidence stretched beyond what it showed, money pressing on judgment or a weak study design. Where those warning signs are present, trust should be lower and questions sharper.
Worked example: back to sleep
Picture a new parent in 1980 who is told by a pediatrician to lay her baby on its front. What could she reasonably have done? She had no access to the scattered studies, and her doctor was following the standard advice of the time. Deferring was not foolish. The failure belonged to the system, which lacked the habit of testing reassuring advice.
During the 1980s, studies in New Zealand, Australia, the Netherlands and Britain began to show that babies sleeping on their fronts died more often. Campaigns to put babies to sleep on their backs began in several countries around 1990, in Britain in 1991 and the United States in 1994. Deaths from sudden infant death syndrome fell sharply, in the United States by about half within a decade. The episode teaches three things at once. Experts can be badly wrong. Expert methods found the mistake. And a parent's best question, then and now, is not "Are you an expert?" but "How do we know this works?"
