What do peanuts, the cause of deadly allergies, and breastmilk, the purported “elixir of life,” have in common?
Both have been the subject of medical recommendations that are wrong, based on weak science, implemented without adequate studies, involving failure to consider risks as well as benefits.
Peanuts do not cause peanut allergies, and breast is not best.
Peanuts do not cause peanut allergies, and breast is not best.
Medical researchers recently announced that they had made a big, life threatening mistake. The advice they provided regarding deadly peanut allergies, was 100% wrong. After years of insisting that keeping babies away from peanuts prevented peanut allergies, experts have reversed themselves and announced that giving babies peanut protein prevents deadly peanut allergies.
It’s hard to be more wrong than that.
As pediatrician Aaron Carroll notes in the NYTimes piece How to Prevent Whiplash from Ever-Changing Medical Advice:
…[T]he National Institute of Allergy and Infectious Diseases expert panel changed course and recommended that we start giving babies peanut powder or extract in food before they are six months old rather than make sure they go nowhere near it. The panel said this is good advice, especially if the babies are at higher risk for developing an allergy…
But it’s important to remember that the earlier recommendation wasn’t made in the Dark Ages. As recently as 2000, the American Academy of Pediatrics declared that children at risk for allergies be given no peanuts until they were 3 years old. It’s not unrealistic to think that this might have increased the number of children with peanut allergies, not decreased them.
This is hardly the first healthcare reversal. I’m old enough to remember when mothers were told to place their babies on their stomach to sleep … which turned out to increase the risk of death from SIDS (sudden infant death syndrome). I’m old enough to recall when doctors were told to give all post menopausal women hormone replacement therapy … which turned out to increase the risk of breast cancer. And I’ve lived through a plethora of dietary injunctions … which paradoxically increased the incidence of problems they were designed to prevent.
How could researchers and physicians have made such terrible mistakes?
Carroll carefully elucidates the reasons.
1. Concluding causation from research that merely demonstrates correlation.
A great deal of epidemiological research, particularly research about nutrition, suffers from this problem.
…[M]ost often, recommendations fail because they aren’t supported by high-quality research. They have only observational trials behind them, not the full weight of randomized controlled trials. Too many times, organizations have been burned when the associations we see don’t translate to causal changes in the more strict randomized controlled trials.
Correlation is not causation. That is Statistics 101, yet researchers, in their enthusiasm to address serious medical problems, often ignore that most critical caveat.
The recommendations to restrict peanut exposure were based on observational studies that erroneously concluded causation. Similarly, nearly all recommendations about the purported benefits of breastmilk are based on small observational studies — weak, conflicting and riddled with confounders — that erroneously conclude causation.
2. Extrapolation from high risk groups to everyone.
If something works for one group, we tend to believe that it should work for more. People at high risk for breast or prostate cancer may need to be screened for the diseases. Expanding that screening to people at low risk, however, yields more false positives than true revelations. Hormone replacement therapy most likely benefits some women, especially younger women and those who have had a hysterectomy. Others get the same side effects or harms with little benefit.
Premature babies benefit from breastmilk. Premature babies are at risk for nectrotizing enterocolitis and breastmilk seems to reduce the risk as compared to formula. But just because breastmilk is beneficial in high risk situations does not mean that we ought to extrapolate to all babies.
3. Failure to study, in many cases to even contemplate, the downsides of recommendations as carefully as the upsides.
This would apply to recent sleep recommendations, which may (and I stress may) benefit babies but also might hurt them and parents in other ways.
This especially applies to breastfeeding research, which suffers from a severe case of white hat bias.
‘White hat bias’ (WHB) [is] bias leading to distortion of information in the service of what may be perceived to be righteous ends… WHB bias may be conjectured to be fuelled by feelings of righteous zeal, indignation toward certain aspects of industry, or other factors…
Breastfeeding researchers are so sure that breastfeeding is beneficial and are so angry at historic malfeasance of the formula industry, they exaggerate findings that place breastfeeding in a positive light and ignore findings that the benefits of breastfeeding in industrialized countries are actually trivial (approximately 8% of breastfed infants have one fewer cold or diarrheal illness in the first year). They never consider the potentially deadly side effects of the relentless promotion of breastfeeding including starvation, dehydration and even death. Indeed, the Fed Is Best Foundation now exists to warn women about the very real risks and downsides of breastfeeding as well as the benefits.
Breastfeeding researchers “know” that breast must be best and therefore always begin and end any breastfeeding study with the claim that the benefits of breastfeeding are well proven when the truth is that they are not. The purported benefits are based almost exclusively on small observational studies, assumed rather than demonstrated causation, and unjustified extrapolation from high risk studies.
Researchers on peanut allergies were forced to reverse the recommendations that they had based on weak science. Breastfeeding researchers have based their claims on science that is even weaker still.
It is inevitable that “breast is best” will eventually be replaced by “fed is best.” The only question is how many infants and mothers will be hurt in the meantime.