My default position when I started looking into peanut allergies was that exposure to potential allergens (particularly the ones outside our evolutionary heritage) would sensitize children to that allergen in the future. Eliezer once described changing your mind as the penultimate technique of rationality, and I always enjoy the opportunity to stretch that particular muscle. But I’m getting ahead of myself here…
Peanut allergies are one of the most common, and yet also most severe, food allergies. It tends to begin very early (unlike seafood allergies), and persist throughout life (unlike milk and soy allergies). The incidence is hard to measure, but potentially up to a few percent of the population will suffer from them over time. This may not seem large in an absolute sense, but it is one of the more common diseases, and is potentially fatal, requiring extensive lifestyle modification to avoid exposure. Allergies themselves, in one form or another, are much more common and often annoying, so if any underlying mechanism can be revealed and averted, so much the better.
I am involved with a small rationality and parenting mailing list, and in one thread I casually mentioned that we planned to exclusively breastfeed through 6 months, in part because of allergies. The benefits of (in some cases exclusive) breastfeeding are well established, and the WHO recommends exclusive breastfeeding up through 6 months. Furthermore hunter-gatherers don’t seem to supplement food until at least 6 months of age. That part is not so controversial. But I offhandedly said I’d heard the advice to avoid allergenic foods until two years of age. That prompted a response. Particularly a link to this paper.
To summarize the methodology, the researchers sent out surveys to a few thousand Jewish families in Israel and the UK, asking them about their weaning behavior and incidence of various allergies and other atopic disease. They had some very stark findings.
[Update: the randomized controlled trials I mentioned were coming have finally started to arrive – and the LEAP study from the UK finds a dramatic 81% decrease in peanut allergies from early exposure.]
It is unlikely genetics can account for the entire difference. They were not morally allowed to ask Israelis about Ashkenazi or Sephardic heritage, but they did test a small subsample of each in the UK and did not find significant differences between the groups. That said, the UK sample undoubtedly leaned more Ashkenazi, so this could possibly account for some of the difference. This suggests, however, that the cause is primarily environmental.
The first thing worth noticing is that the UK had higher rates of all allergies and atopic disease than Israel across the board, and unfortunately the study did not try to address this much. This suggests that whatever is going on is happening across all domains, it’s not specific to peanut allergies. From that perspective we still don’t have an answer – something about the UK lifestyle predisposes these infants to atopic disease, period. Eczema rates, for instance, are five times higher in the UK. I am skeptical that peanut consumption alone could provide this huge difference. In terms of overall feeding behavior, the following trends were significantly different between the groups: higher (exclusive) breastfeeding of UK infants, higher peanut, sesame, and cow’s milk consumption in Israeli infants – egg, soy, tree nuts, and any solids, were similar.
Now, these atopic diseases tend to be correlated with one another. If you are allergic to one food, you’re much more likely to be allergic to another. If you have eczema, you’re much more likely to develop food allergies later. And so on. For this reason, the researchers do adjust for existing atopic disease in the paper, which does change the food allergy findings somewhat. Milk and egg allergies are now insignificantly different between the groups, but still higher in the UK. But even after adjustment, the allergy rates are three times higher for sesame (not significant), almost six times higher for peanuts, and over eight times higher for tree nuts. I am suspicious that the incidence of tree nut allergies is the most extreme, when both groups consume very low amounts of them. The authors of the paper argue that seed storage proteins are highly conserved across different plants, and thus consuming any nut/seed should induce tolerance to all of them. This seems plausible on its surface, but I have not yet done further investigation.
That is a very significant finding, and enough for me to take notice. Not just me, as a matter of fact – there are now at least two randomized, controlled trials being conducted with high-risk infants. One of them in the UK is expected to wrap up this year [update: trial is complete, results released], and the other in the US is due in the summer of next year… but for those of us with babies right now we have no choice but to look at the best evidence and take a guess. All of our actions carry responsibility, for ourselves and others, and we almost always must act under imperfect information.
The paper itself does summarize the existing research of the time. Without linking to all of the papers, it’s worth a quick summary. Notably, avoiding allergens during pregnancy and breastfeeding has no effect, despite recommendations to the contrary. Next, there is significant evidence that allergen exposure to the skin is more likely to result in sensitization. Peanut allergies increase with severity of eczema (an allergic skin condition), and using peanut oil on infants with eczema results in almost seven times the risk. Environmental exposure results in greatly increased risk, apart from consumption. In mouse models at least, which are an imperfect approximation of humans, oral exposure to allergens results in tolerance (even after a single exposure). There is some evidence of this channel in humans too though: kids who get nickel braces are much less likely to have allergic reactions to nickel earrings later in life. Furthermore, since that paper was released we have new biochemical evidence that the type of exposure matters – the researchers isolated memory T-cells from individuals with and without peanut allergies, and found skin-exposure markers in allergic patients and oral-exposure markers in the controls. This convergence of epidemiological data, animal models, and verified biochemical mechanisms represents the highest standard of evidence apart from double-blind RCTs.
[Update: the LEAP trial in UK showed the expected result: a drastic 81% decrease in peanut allergies in response to early oral exposure.]
What are my specific, actionable recommendations?
1) Based on food intake data from Israel, it seems like a couple of grams of peanut butter a few times per week is sufficient to produce this effect.
2) Given the sensitizing effects of cutaneous and inhaled exposure, I do not recommend having peanuts or peanut butter open inside the house. Furthermore, until your baby is able to reliably swallow food put inside her mouth, I would not try feeding her any potential allergens! She needs oral exposure before any cutaneous exposure, so spitting it out could potentially be actively harmful. Certainly don’t introduce peanut butter with baby-led weaning, where she has to grab it before putting it in her mouth and consuming it.
There is still a lot we don’t know about atopic disease. My biggest questions, in order:
A) What is causing the vastly higher rates of all forms of atopic disease in the UK? This is potentially a more important question than how to specifically avoid peanut allergies. The hygiene hypothesis is one potential line of investigation here.
B) Does this mechanism work for other food allergens? Other allergens in general? Maybe we do want to have kids crawling around in the dirt and eating leaves and grass and flowers and things, if that could reduce other environmental allergies. Also, what is going on with severe food allergies that develop later in life, like shellfish?
C) Does exposure to any nut result in tolerance to all of them? Or is this a legitimate warning sign about this study?
When I do more research on this question, you will be the first to know!
