Health Library › News › Health Drive › Peanuts and babies: the advice reversed completely
Peanuts and babies: the advice reversed completely
For twenty years parents were told to keep peanuts away from babies. Then somebody finally ran the trial, and the answer pointed the other way entirely.
Transcript
Welcome to the sofidex Health Drive podcast series. Today we are talking about peanuts and babies, and about a piece of official advice that did not just change over the years but reversed, completely, into its own opposite. Maybe you raised children under the old rule, keeping peanuts away like a careful parent should. Maybe you are weaning a baby right now and the clinic has told you the opposite of what your own mother remembers being told. And somebody at the table, lovingly, thinks somebody else is about to make a terrible mistake.
And the answers on offer come in the usual two unhelpful sizes. Either the experts clearly cannot make up their minds, so ignore all of it. Or the new advice is recited with no explanation of why anyone believed the old advice at all. Neither settles the argument at the table. So we want to ask the question properly. Not just what the new advice says, but what the old advice rested on, what finally got measured, and how the same profession came to recommend a thing and then its exact opposite within one generation of parents.
We will follow one invented family and one jar of peanut butter. Along the way: a teething snack from Israel that became the best clue in modern allergy research, the landmark trial that settled it, and the risk-tiered advice that replaced the old blanket rule. Plus the one group of children for whom nothing tonight changes. And to say it before the worry grows: whichever rule you followed, you followed the best advice of your era, in good faith. Nothing tonight asks anyone to feed a baby anything without their own clinician in the loop, where the guidance asks for one.
One thing before we start. This is health information, not medical advice, and nothing in it is a substitute for talking to your own doctor about your own treatment. What we can tell you is where all of it came from — 6 published sources, all listed with the episode. I want to start at a kitchen table. With a family we'll invent, so nobody's real baby is in this story. Call the baby Noah. And he's six months old, with bad eczema, and on the table between the adults who love him sits a jar of peanut butter. Because his mother has been told, by the clinic, to start giving him peanut now. And his grandmother is horrified. Because when she raised Noah's mother, the rule was the exact opposite. No peanuts before three. She didn't mishear it. Because that really was the rule.
And that's what makes this episode different from most reversals. Because usually the old advice was folklore, and the new advice is evidence. But Noah's grandmother got her rule from doctors. Official bodies, in several countries, told a generation of parents to keep peanuts far away from babies. And then, within about fifteen years, the same profession came back and said: actually, do the opposite. Early. Deliberately. Keep going. So tonight is the story of the fastest full reversal in modern dietary advice. And of the strange snack-food clue that triggered it.
Start with the old rule, then. Where did no-peanuts-before-three come from? Because somebody presumably had a reason.
Well, a reasonable-sounding one. Which is the trap this show keeps finding. Because the logic went like this. Allergy is the immune system overreacting to something it's met. And babies' immune systems are immature. So don't let the immature system meet the dangerous protein. Then delay the meeting until the child is older and sturdier, and you'll get fewer allergies. It's clean. Intuitive. And parents could do something, which always helps advice spread. And through the late nineties and early two-thousands it hardened into official guidance. Because the United Kingdom recommended avoidance in infancy. So did Australia. So did North America. And pregnant women were even advised to go easy on peanuts, in some versions. Whole maternity wards ran on it.
And I notice what's missing from that story. A trial. Anybody checking whether the clean, intuitive thing was true.
Nobody checked. Nobody. And that's not me being harsh with hindsight. The documents themselves barely pretended otherwise. Because the advice rested on mechanism and caution, not on a test. Which might have been fine, if it were costless. The era's thinking went: if avoidance helps, we win. If it does nothing, we've lost nothing. And peanut butter isn't a food group. But that if-it-does-nothing clause was never checked either. And there was a third possibility nobody priced in. That avoidance was actively causing the thing it was meant to prevent.
Hold the causation for a second. First, what actually happened to peanut allergy while everyone was avoiding?
It went up. Sharply. The LEAP investigators put it in the first sentence of their trial report. The prevalence of peanut allergy among children in Western countries doubled in ten years. Doubled. And during the avoidance era, mind. Now, one line like that doesn't prove the advice caused the rise. Because allergies of all kinds were rising, for reasons still argued about. But hold the shape of it in your head. A rule is issued to prevent a disease. And the disease doubles. So, at minimum, the rule isn't working. And a few people started asking the darker question. What if it's backwards?
Before the clue. Why would avoidance backfire? You said there was a third possibility nobody priced in. What's the mechanism that makes backwards make sense?
Well, there's a hypothesis. And Noah's own skin is at the centre of it. Because the researchers kept noticing that the babies who became peanut-allergic were so often the eczema babies. And eczema isn't just itchy. It's a broken skin barrier. Cracks in the wall, literally. So picture a house where peanut is eaten by everyone else. Because there's peanut protein in the dust, on hands, on the high-chair tray. A baby with eczema meets those traces through inflamed, leaky skin. And the immune system, meeting a protein through a damaged wall, in the company of inflammation, tends to read it as a threat. Now suppose that same baby never eats peanut, because the leaflet said not to. The one route that teaches the immune system the other lesson, the gut, where food is met calmly and filed as food, stays silent. So the threat lesson is the only lesson. So skin says danger. And gut says nothing. Avoidance didn't close the danger route. It closed the safety route and left the danger route open.
Which would explain the cruel pattern. The more carefully an eczema family avoided the jar, the more one-sided the baby's education got.
And that's the hypothesis, stated fairly. It's mechanism, so I hold it as lightly as I've been telling you to hold mechanisms all night. But it made a prediction. If the gut route teaches tolerance, then deliberately feeding peanut early, especially to eczema babies, should prevent the allergy. That prediction is what LEAP went and tested. And mechanism that survives a randomised test is a different creature from mechanism that was never asked.
Which brings us, I think, to the snack food. Tell the Israel story, because it's the best clue in this whole saga.
Well, it's one of the great natural experiments. So. A team of allergy researchers noticed something odd about who gets peanut allergy. Because in Israel, babies eat peanut. There's a hugely popular teething snack, a puffed thing made of peanut, and infants gum it down from a few months old, by the bagful. But in the UK, under the avoidance advice, babies ate essentially none. And here's what made it a fair comparison. Because the researchers didn't compare all Israelis with all Britons, which would tangle genetics and environment. They compared Jewish children in Israel with Jewish children in the UK. Similar ancestry. Different kitchens. One variable.
And the numbers?
Stark. Because the prevalence of peanut allergy among the UK Jewish schoolchildren was 1.85 per cent. In Israel, 0.17 per cent. That's a 10-fold difference. Tenfold. Ten. And the feeding data behind it was just as clean. Israeli infants aged eight to fourteen months were eating a median of 7.1 grams of peanut protein a month. The UK median was zero. So, grams of peanut: lots versus none. Allergy: rare versus ten times commoner. And the direction of that result is the scandal. The country following the protective advice had the allergic children. The country cheerfully ignoring it had barely any.
An observation, though. You'd be the first to say it. Different countries, different sunshine, different infections, different everything. What did it actually prove?
Nothing, on its own. And the authors said so. Almost word for word. Their paper ends by raising the question of whether early introduction, rather than avoidance, prevents peanut allergy. Raising the question. And that's exactly the right weight. But notice the date. 2008. And notice what else happened that year. The American Academy of Pediatrics formally walked back its old position. Their updated report replaced the earlier policy statement, and it said plainly that there's little evidence that delaying foods beyond four to six months prevents allergic disease. So by 2008 the old rule was officially an orphan. Because avoidance wasn't commanded any more. But early introduction wasn't commanded either. The advice went quiet. And quiet, for an anxious parent, means keep doing the cautious thing. So most people kept avoiding.
So the question was raised, the old rule was retired, and everyone needed the thing nobody had done in the first place. The trial.
And this time somebody did it properly. Because it's called LEAP for a reason. Learning Early About Peanut allergy. Run in London. Published in the New England Journal, 2015. And the design went straight at the hard case. They took 640 infants at high risk of peanut allergy. Babies with severe eczema, egg allergy, or both. The worst hands. Noah, in other words. And four to eleven months old. And then they did the thing that sounds almost reckless after twenty years of avoidance advice. They randomised them. So: one group assigned to eat peanut regularly, from infancy until the age of five. The other, strict avoidance. Same five years. And then everyone got formally tested for peanut allergy.
The grandmother in me tightens at that. Feeding peanut to the highest-risk babies on purpose. What did it show?
Well. The cleanest result in this entire field. Take the babies who started with no measurable sensitivity on a skin-prick test. Among those who avoided peanut, 13.7 per cent were allergic by age five. Among those who ate it, 1.9 per cent. So the avoiders had seven times the allergy. And the already-sensitised babies, the ones whose skin test had started to flicker? Avoidance group, 35.3 per cent allergic. Eating group, 10.6. So even in babies already on the road to allergy, early eating cut the destination rate by two-thirds. And safety, the thing the grandmother is tight about: there was no significant difference between the groups in serious adverse events. The eating was supervised and structured. But it wasn't carnage. It was protection.
You said the trial sorted babies by a skin test before it even started. Walk me through that, because it's the same test Noah may be sent for, and the phrase skin-prick sounds worse than I suspect it is.
Well, it's almost disappointingly gentle. Because here's the whole procedure. A drop of peanut extract goes on the forearm. A tiny lancet presses through the drop, barely breaking the surface. And then you wait a few minutes and measure. If the immune system has made antibodies against peanut, a little raised bump appears, a wheal, like a midge bite. And the size of the bump is the information. All of it. In LEAP, babies with no measurable wheal went into one cohort. Babies with a small one, 1 to 4 millimetres, went into the other. Already sensitised, but not yet reacting to food. And babies above that, likely already allergic, weren't randomised to eat peanut at all. Which is worth underlining. Because even the boldest trial in this field didn't feed peanut blindly to everyone. It measured first. Then fed. That's the shape the guidelines copied.
Say the shape of that out loud, because it's the whole reversal in one sentence.
The thing parents were told to do, avoidance, was the thing producing allergic children. And the thing parents were warned against was protective. Not neutral. Protective, sevenfold, in a randomised trial. The old advice hadn't been merely useless. It had been pointing backwards for twenty years, while the allergy rate doubled. That's why this story deserves an episode. Not because doctors changed their minds. Because of what it took to find out, and how completely the answer inverted.
What were the eating babies actually eating? Because I'm trying to picture a five-month-old with a jar of crunchy.
Not the jar, and certainly nothing whole. In the trial it was baby-safe forms. The famous Israeli puffed snack, mostly, or smooth peanut butter worked into food. And the dose wasn't heroic. The sister trial measured the relationship and found that around 2 grams of peanut protein a week was associated with the protection. Two grams. That's roughly a couple of teaspoons of peanut butter across a whole week. Not a regime. A habit. A taste that keeps showing up on the spoon, week after week, while the immune system takes notes.
Two obvious worries, then. First: does it last? Five years of deliberate peanut, and then the family relaxes. Does the protection survive?
They tested exactly that. And it's a lovely follow-up. LEAP-On. Same children. One more year. At the end of the main trial, everyone, both groups, was asked to avoid peanut entirely for twelve months. And if the protection were a fragile thing that needed topping up, a year off should've eroded it. It didn't. At all. After the year of avoidance, allergy stood at 18.6 per cent in the original avoiders and 4.8 per cent in the original eaters. So the gap held. And within the eating group, the year off caused no significant rise, 3.6 to 4.8 per cent, which is statistical noise. So early eating isn't a subscription. Because it looks like genuine immune training. The system met peanut early, filed it under food, and the filing stuck.
Second worry. LEAP was high-risk babies. Eczema, egg allergy. Noah's cohort. What about ordinary babies, the ones with clear skin and calm parents?
Well, there the evidence is softer. And I want to be honest about its shape. A sister trial called EAT took 1303 breast-fed babies from the general population and randomised them to early introduction of six allergenic foods, peanut among them, against standard UK practice. And on the headline analysis, the strict intention-to-treat one, it missed. Food allergy in 5.6 per cent of early introducers versus 7.1 on standard timing. Not significant. Honestly. But the families who actually managed the early feeding tell a different story. In that per-protocol analysis, peanut allergy was 2.5 per cent on standard timing and zero in the early eaters. Zero. Of the children who actually ate it.
And you've taught me to raise an eyebrow at per-protocol. The families who manage the regime differ from the families who don't. That's the whole lesson of half our episodes.
Raise it. And keep it raised. Because per-protocol compares completers, and completers are special people. So EAT, on its own, proves much less than LEAP. What it gives you is consistency. Every finger points the same way. The randomised high-risk trial, decisively. The follow-up, durably. The population trial, suggestively, in the same direction. The Israel comparison, observationally, tenfold. When four different kinds of evidence disagree about strength but agree about direction, the direction is usually real. And no finger, anywhere, points back toward avoidance. Because the old advice has no surviving evidence at all. And that's rare. Most reversals leave a rump of defenders. This one left nothing.
And the pregnant women? The grandmother will remember being told the avoidance started before birth. Some versions had mothers off peanuts while breastfeeding.
And that arm of the old advice collapsed at the same time. Because when the American pediatricians rewrote their position in 2008, they went through the pregnancy evidence too. Their conclusion: current evidence does not support a major role for maternal dietary restrictions during pregnancy or lactation. In plain words, a mother eating peanut butter in pregnancy is not seeding her baby's allergy. She never was. The same report, by the way, kept the genuinely supported piece. Breastfeeding for at least four months has evidence behind it for preventing or delaying eczema and cow's milk allergy in early childhood. So the rewrite wasn't a demolition of everything old. It kept what had data. It dropped what had only the mechanism story. Which is exactly what a rewrite should do.
Does early peanut tangle with breastfeeding advice, though? The UK line, last I knew, was exclusive breastfeeding to around six months. Peanut at four months breaks that.
It brushes against it, and the EAT trial lived exactly in that tension. Its control group was, in the trial's own words, the current practice recommended in the United Kingdom of exclusive breast-feeding to approximately six months. And its intervention group started allergenic foods from three months alongside the breast milk. Alongside is the operative word. Nobody was weaned early. The milk continued either way. So the practical reconciliation, and where the guidelines have drifted, is that around six months remains the centre of gravity for starting solids generally, and the allergy evidence argues for not drifting late, and for peanut being among the early arrivals rather than a food you save for last. The old instinct, save the scary foods for later, is the precise instinct the whole evidence base now points against.
So the guidelines turned. Walk me through what officialdom actually says now, because that's what the clinic told Noah's mother.
Well, in the United States it turned formally in 2017. The National Institute of Allergy and Infectious Diseases, with some twenty-five professional organisations alongside, issued addendum guidelines specifically to address preventing peanut allergy. Prompted, their own text says, by a recent landmark clinical trial. And that's LEAP, wearing its official citation. And the structure of the new advice is the interesting part. Because it's not one rule. It's three, by risk. For babies at the highest risk, the severe-eczema, egg-allergy babies like Noah, the guidelines cover testing first, and the timing and approaches for introduction, sometimes in the clinic rather than the kitchen. And for milder eczema, earlier and simpler. For babies with no risk factors, peanut-containing food gets introduced freely with the family's ordinary diet. So the one-size rule died. And risk tiers replaced it.
And notice what the tiers quietly concede. The highest-risk babies get a doctor involved first. So this episode can't end with go home and feed everyone peanut butter.
It absolutely can't. So let me put the guard rails in myself. First guard rail. Because everything tonight is about preventing an allergy from developing. A child who already has peanut allergy is in a different universe. And for them, avoidance remains the rule, prescribed, with a plan for accidents. Nothing here says otherwise. Second. A baby like Noah, with severe eczema or an existing food allergy, should have the conversation with their own doctor before the first taste, because the guidelines themselves build in that step, sometimes with a skin test first. Third, and small but real: whole peanuts don't go anywhere near babies. Because that's a choking rule, not an allergy rule. Smooth peanut butter thinned down, or peanut puffs. And if any of this will keep a parent up at night, that's precisely what the appointment is for. The point of tonight is that the era of reflexive avoidance is over. Not that the kitchen table replaces the clinic.
You keep saying peanut. The grandmother's next question is obvious. Eggs. Milk. Sesame. Is this a peanut story or an allergy story?
Honestly, mostly a peanut story, with strong hints beyond. Because the EAT trial fed its babies six allergenic foods. Peanut, cooked egg, milk, sesame, whitefish, wheat. And in the families who managed the schedule, the egg result echoed the peanut one. Egg allergy, 1.4 per cent in early eaters against 5.5 on standard timing. Same direction, decent size. But milk, sesame, fish and wheat showed no significant effect either way. So the honest summary runs: peanut, proven, randomised, durable. Egg, likely, same logic, softer data. The rest, undetermined, but no signal anywhere that early introduction harms. Which is why the modern posture is permissive across the board. The fences came down generally. The guard tower only ever had real evidence around peanut, and that evidence says open the gate.
Give me the stakes, while we're being official. Because a sceptical listener could say: it's a food allergy, carry an antihistamine, calm down.
And the guideline's own opening paragraph answers that listener. Food allergy is an important public health problem, it says, because it affects children and adults, can be severe and even life-threatening, and may be increasing in prevalence. Life-threatening is not padding. Peanut is one of the leading causes of fatal food reactions. It's the allergy of adrenaline pens and ambulance rides, of school lunchrooms with peanut bans, of parents reading every label of every biscuit for fifteen years. And unlike egg and milk allergy, which most children outgrow, peanut allergy usually stays for life. So preventing one case isn't sparing a child some hives. It's deleting a lifelong, occasionally lethal condition from one person's biography. Multiply that by the doubling we started with, and you see why twenty-five organisations signed the reversal.
Now the part I find almost painful. The twenty years in between. A generation of parents did the careful thing, the recommended thing. And some of their children have the allergy, plausibly because of it. What do you say to those parents?
That they did nothing wrong. And I mean that in the strict sense. They followed the best official reading of a plausible theory. Because the failure wasn't theirs. And it wasn't really individual doctors' either. The failure was structural. A mechanism-based rule went out to millions of kitchens for years before anyone ran the test that would've taken five of those years to run. And that's the lesson this episode files next to the others. Because plausible plus official doesn't equal measured. And when the stakes are a generation of immune systems, measured is the only standard that should ever have been allowed to write the leaflet. And the system has learned some of that. LEAP exists. The guidelines cite it line by line. The grief of the in-between families is the price of learning it late.
Let's land it, then. Monday morning. Noah's kitchen. The jar on the table, the grandmother's folded arms. The take-homes.
Three things. And they're all small. First, for the grandmother: the rule you remember was real, you followed it in good faith, and it's gone. It was tested, at last, and the test said the opposite. Because early, regular peanut, started in infancy, prevents peanut allergy. That's the strongest single result in this field. Second, for the mother: which path Noah takes depends on his risk. Because with his eczema, the first portion may belong near a clinician. And that's not bureaucracy. It's the guideline doing its job. Ask your own doctor the one precise question: given his skin, how should we start peanut? And third, for every other kitchen, the low-risk babies with no eczema and no allergies: the fence is down. Peanut-containing food, in a baby-safe form, belongs in the ordinary adventure of first foods. Not as medicine. As dinner, plainly.
One loose end. Noah has a cousin, and the cousin already has diagnosed peanut allergy. Same family, different baby. Does tonight change anything for him?
Nothing. Nothing at all. And the cleanest way to say it is that tonight's evidence is about prevention, and his diagnosis means prevention's moment has passed. For him the old rules genuinely still apply. Strict avoidance, a written plan for accidental exposures, the pen if one's been prescribed, and specialist follow-up, because treatments for existing allergy are their own developing field with their own trials. So what his parents shouldn't do is hear this episode and experiment at home. And his existence changes one more thing. It nudges the family's new babies up the risk ladder, which makes the before-you-start conversation with their own doctor more important for them, not less. So: one family, three babies, three different right answers. That's not confusion. That's what risk-tiered advice looks like when it's working.
And the bigger take-home, above the jar.
When advice reverses this completely, don't conclude that experts know nothing. But conclude something more useful. Advice that was never measured can point anywhere, and official is not the same as tested. Because the fix, both times in this story, was the same boring magnificent thing. Somebody counted. So when the next confident rule arrives in your kitchen, ask the question Noah's grandmother never got the chance to ask. Has anyone actually checked? And if the answer is yes, a trial, with children like mine — that's the advice worth folding your arms over.
Three things to ask your doctor
- My baby has eczema or an egg allergy — how should we introduce peanut, and should there be a test first?
- My child already has a peanut allergy. Does any of the new advice change their avoidance plan?
- We avoided peanuts with our older children. Should we do anything differently with this baby?
Sources
- 1Du Toit G, Roberts G, Sayre PH, et al; LEAP Study Team. Randomized trial of peanut consumption in infants at risk for peanut allergy. N Engl J Med. 2015 Feb 26;372(9):803-13. doi:10.1056/NEJMoa1414850. PMID 25705822.In 640 high-risk infants (severe eczema, egg allergy or both) randomised to eat or avoid peanut until 60 months, peanut allergy occurred in 13.7 percent of avoiders versus 1.9 percent of consumers (skin-test negative) and 35.3 versus 10.6 percent (sensitised), with no significant difference in serious adverse events
- 2Du Toit G, Sayre PH, Roberts G, et al; Immune Tolerance Network LEAP-On Study Team. Effect of avoidance on peanut allergy after early peanut consumption. N Engl J Med. 2016 Apr 14;374(15):1435-43. doi:10.1056/NEJMoa1514209. PMID 26942922.After 12 months of peanut avoidance following the primary trial, allergy remained far lower in the original consumption group (4.8 versus 18.6 percent), with no significant increase within the consumption group
- 3Perkin MR, Logan K, Tseng A, et al; EAT Study Team. Randomized trial of introduction of allergenic foods in breast-fed infants. N Engl J Med. 2016 May 5;374(18):1733-43. doi:10.1056/NEJMoa1514210. PMID 26943128.In 1303 breast-fed infants from the general population, early introduction of six allergenic foods did not reduce food allergy on intention-to-treat (5.6 versus 7.1 percent), but per-protocol peanut allergy was 0 versus 2.5 percent and egg 1.4 versus 5.5 percent, with about 2 g a week of peanut protein associated with protection
- 4Du Toit G, Katz Y, Sasieni P, et al. Early consumption of peanuts in infancy is associated with a low prevalence of peanut allergy. J Allergy Clin Immunol. 2008 Nov;122(5):984-91. doi:10.1016/j.jaci.2008.08.039. PMID 19000582.Despite avoidance guidelines, peanut allergy was 10-fold higher in UK Jewish schoolchildren (1.85 percent) than in Israeli Jewish children (0.17 percent), where infants eat a median 7.1 g of peanut protein a month against 0 g in the UK — raising the question of whether early introduction prevents the allergy
- 5Greer FR, Sicherer SH, Burks AW; American Academy of Pediatrics Committee on Nutrition; Section on Allergy and Immunology. Effects of early nutritional interventions on the development of atopic disease in infants and children. Pediatrics. 2008 Jan;121(1):183-91. doi:10.1542/peds.2007-3022. PMID 18166574.The 2008 AAP clinical report replaced the earlier policy statement, finding little evidence that delaying complementary foods beyond 4 to 6 months prevents atopic disease and no major role for maternal dietary restriction in pregnancy or lactation, while keeping the evidence for breastfeeding at least 4 months
- 6Togias A, Cooper SF, Acebal ML, et al. Addendum guidelines for the prevention of peanut allergy in the United States: report of the National Institute of Allergy and Infectious Diseases-sponsored expert panel. J Allergy Clin Immunol. 2017 Jan;139(1):29-44. doi:10.1016/j.jaci.2016.10.010. PMID 28065278.Prompted by the landmark trial, NIAID and 25 professional organizations issued 2017 addendum guidelines providing 3 risk-tiered recommendations for early introduction of peanut-containing foods, covering testing and the timing and approaches for introduction in the office or at home
How this was made
Information, not medical advice. Every claim in this episode has a published source, listed above. It is not a substitute for advice about your own treatment — for that, talk to your doctor. Published 2026-11-02.