Chapter 28 — Food Allergies and Intolerances: Six Mechanisms, One Word, and the Tests That Don't Work

The Hook: Six people at a table

A dinner party. Six people. Every one of them tells the host they have a problem with a food.

⚠️ Six mechanisms. One word. And the word is doing all six jobs.

⚠️ The cost of that falls almost entirely on person 1.

When five of six stated food allergies at a table are something else, restaurants stop treating the word as urgent, hosts assume flexibility, and a teenager with a genuine peanut allergy is asked whether "a little bit" would be alright.

⚠️ It kills people. Not often, and reliably enough that every allergy organization in the world says the same thing.

And the cost runs the other way too. Person 5 has eliminated twenty-three foods on the basis of a test that measures the wrong thing, and nobody has told her.

⚠️ So this chapter is about distinguishing six things that the language collapses into one — because getting it wrong is dangerous in one direction and expensive, restrictive and demoralizing in the other.

⚠️ And a number worth holding throughout: self-reported food allergy runs several times higher than challenge-confirmed food allergy. Something in the region of a fifth of adults report one; the confirmed prevalence is a small single-digit percentage. That gap is not people lying. It is this chapter.

🏃 Fast Track: §28.1 (the taxonomy), §28.3 (⚠️ anaphylaxis), §28.6 (⚠️ test before removing gluten), §28.10 (⚠️ the tests that don't work). Thirty-five minutes.

🔬 Deep Dive: §28.4 (sensitization versus allergy), §28.8 (⚠️ non-coeliac gluten sensitivity), §28.13 (immunotherapy).

28.1 The taxonomy

⚠️ Everything in this chapter follows from getting this right.

💡 ⚠️ Look at the dose column, because it separates the two most consequential rows.

Coeliac disease and IgE allergy are the ones where small amounts matter. Almost everything else is a threshold effect — and threshold effects mean the answer is usually "how much," not "never."

⚠️ Which is why person 3 at the dinner table can eat yoghurt, and person 2 cannot have a crumb.

⚠️ And nocebo deserves its own sentence, because it is real and it is not "imaginary."

In double-blind food challenges, a substantial proportion of people who reliably report symptoms to a food develop those symptoms on placebo. ⚠️ The symptoms are genuinely experienced — nausea, pain, bloating, headache are produced by expectation through real physiological pathways. Saying "it's nocebo" is not saying "you're making it up." It is saying the trigger isn't the food.

28.2 IgE-mediated food allergy

The immune system produces IgE antibodies against a food protein. On re-exposure, those antibodies trigger mast cells to release histamine and other mediators. ⚠️ Rapid, reproducible, potentially fatal.

The common allergens: ⚠️ cow's milk · egg · peanut · tree nuts · soy · wheat · fish · shellfish · sesame — the last of which was added to mandatory US labelling relatively recently.

⚠️ Natural history matters and is often not explained:

Which means a child with a milk allergy has a good chance of losing it, ⚠️ and should be reassessed periodically rather than avoiding indefinitely by default.

Presentation: urticaria, angioedema, vomiting, abdominal pain, wheeze, stridor, hypotension — ⚠️ and the defining feature of anaphylaxis is that it involves more than one body system, or cardiovascular or respiratory compromise.

28.3 ⚠️ Anaphylaxis

The section with the highest stakes in this book.

⚠️ ADRENALINE (EPINEPHRINE), INTRAMUSCULAR, INTO THE ANTEROLATERAL THIGH, IS THE FIRST-LINE TREATMENT. THERE IS NO ABSOLUTE CONTRAINDICATION. DELAY IS ASSOCIATED WITH DEATH.

⚠️ Antihistamines do not treat anaphylaxis. They may help urticaria and itching. ⚠️ They do not reverse airway compromise or hypotension, and reaching for one instead of adrenaline is the most common fatal error.

What to do:

1. ⚠️ Give adrenaline. Early. If you're unsure whether it's anaphylaxis, give it. 2. Call emergency services. 3. ⚠️ Lie the person flat with legs raised — or sitting up if they are struggling to breathe, or on their side if vomiting or unconscious. ⚠️ Do not stand them up or walk them anywhere. Sudden posture change has been associated with deterioration. 4. ⚠️ A second dose may be needed after about five minutes if there's no improvement. 5. Everyone who has had anaphylaxis needs observation, because biphasic reactions — a recurrence hours after apparent resolution — occur.

⚠️ Risk factors for fatal outcome, which is the list worth knowing:

• ⚠️ Asthma, particularly poorly controlled — by some distance the most important
• ⚠️ Adolescence and young adulthood — risk-taking, not carrying the device, eating out
• Peanut and tree nut allergy
• ⚠️ Delayed adrenaline administration
• Being upright or standing during the reaction

⚠️ The practical instruction that matters most: carry two devices, carry them everywhere, and replace them before they expire.

An adrenaline auto-injector at home is a device you don't have.

⚠️ Recognizing it

Because hesitation is the failure mode, and hesitation comes from uncertainty about whether it "counts."

⚠️ The rule that removes the hesitation: airway, breathing or circulation involvement — or two or more systems — after a likely exposure. Give adrenaline.

⚠️ And skin signs are not required. A substantial minority of fatal anaphylaxis cases had NO rash. Waiting for hives is waiting for something that may not come.

⚠️ Two more practical points that save lives and are rarely taught:

1. ⚠️ Reactions can be worse when a cofactor is present — exercise, alcohol, NSAIDs, infection, and around menstruation. A food that was tolerated last week can produce anaphylaxis after a run or a drink. Food-dependent exercise-induced anaphylaxis is a recognized entity, most classically with wheat.

2. ⚠️ And an infant's presentation differs. They cannot report throat tightness. Sudden floppiness, pallor, persistent crying, drooling or vomiting after a new food is the presentation, and it is easy to attribute to something else.

28.4 ⚠️ Diagnosis: sensitization is not allergy

The most consequential misunderstanding in allergy practice, and it produces enormous amounts of unnecessary restriction.

⚠️ A positive skin prick or specific IgE test means you have antibodies. It does NOT mean you will react.

Many people are sensitized to foods they eat without any problem at all.

⚠️ Which produces the single most important rule in this section:

The history drives the testing. The testing does not drive the diagnosis.

⚠️ Testing a panel of foods in someone with no suggestive history generates false positives, and each false positive generates an unnecessary elimination.

⚠️ This is why "let's just test for everything" is actively harmful. A child tested against a broad panel because of eczema will very likely show sensitization to several foods, most of which they tolerate — and each result carries a real risk of removal from the diet, ⚠️ which in the case of foods like peanut and egg can INCREASE the chance of developing true allergy (§28.14).

Component testing helps. ⚠️ For peanut, sensitization to certain specific components is much more predictive of clinical reaction than whole-peanut IgE — which can separate someone genuinely at risk from someone cross-reacting to pollen.

28.5 Coeliac disease

⚠️ Not an allergy. Not an intolerance. An autoimmune disease.

In genetically susceptible people — carrying HLA-DQ2 or DQ8 — gluten triggers an immune response that damages the small intestinal lining. Prevalence is around 1% in many populations, ⚠️ and a large proportion remain undiagnosed.

⚠️ The presentation is frequently not gastrointestinal, which is why it gets missed:

⚠️ Anyone with unexplained iron-deficiency anaemia should be screened for coeliac disease. It is one of the highest-yield tests in general practice and it is frequently not done.

Diagnosis: serology — tissue transglutaminase IgA, with total IgA measured alongside, ⚠️ because IgA deficiency is more common in coeliac disease and produces false negatives. Duodenal biopsy usually confirms in adults; some paediatric pathways allow diagnosis without biopsy at very high antibody titres.

Treatment: ⚠️ strict, lifelong gluten avoidance. Not "mostly." ⚠️ Small amounts cause immune activation and intestinal damage, frequently without symptoms — which is why "a little bit doesn't affect me" is not evidence that it's safe.

Untreated coeliac disease is associated with osteoporosis, persistent anaemia, infertility, and — rarely — small bowel lymphoma.

28.6 ⚠️ Test before you remove gluten

The single most practically important instruction in this chapter.

⚠️ COELIAC TESTING IS ONLY VALID WHILE YOU ARE EATING GLUTEN.

Both serology and biopsy depend on active immune response and active damage. Remove gluten and both normalize.

⚠️ What happens if you remove it first:

1. Your antibody test comes back negative — regardless of whether you have coeliac disease. 2. ⚠️ To be tested properly you must then undergo a GLUTEN CHALLENGE — eating a meaningful quantity of gluten daily for several weeks, deliberately, while symptomatic — before retesting. 3. ⚠️ Many people, understandably, refuse. 4. Which leaves them on a lifelong restrictive diet with no diagnosis — and without a diagnosis they get no follow-up, no bone density monitoring, no dietitian, no prescribed products where those exist, and no answer for their relatives, ⚠️ who have a substantially elevated risk and should be screened.**

⚠️ If you suspect gluten is a problem: get tested first. It takes a blood test and it takes a fortnight. Then remove it.

This is the instruction I would most want a reader to act on from this chapter, and it is the one most often reversed.

⚠️ The gluten-free diet in practice, which nobody explains

Diagnosis is the beginning. ⚠️ The diet is the treatment, and it is harder than it sounds.

⚠️ The single most common error after diagnosis is replacing every wheat product with a gluten-free packaged version. That produces a diet that is technically compliant, lower in fibre, more processed and substantially more expensive.

The better move is Chapter 22's slot swap: rice, potatoes, buckwheat, quinoa, oats where certified, beans and lentils — food that was never going to contain gluten in the first place.

28.7 Lactose intolerance: the global norm

⚠️ A framing correction that changes how the whole thing should be discussed.

All mammals lose lactase activity after weaning. ⚠️ It is the human ability to KEEP it into adulthood — lactase persistence — that is the unusual, derived trait, arising through mutations that spread with dairying.

⚠️ The global majority of adults are lactase non-persistent — on the order of two-thirds worldwide, with enormous variation by ancestry. Very low rates of non-persistence in populations of northern European descent; very high in much of East Asia.

⚠️ Which means lactose intolerance is not a disease. It is normal adult human physiology, and the persistent phenotype is the exception.

Practically, and this is the part that matters:

1. ⚠️ It is strongly dose-dependent. Most people with lactose malabsorption tolerate something in the region of a cup of milk's worth of lactose — around 12 g — particularly when taken with other food and spread through the day.

2. ⚠️ Fermented and aged products are usually fine. Hard cheeses contain very little lactose. Yoghurt contains bacterial lactase, which does some of the work for you.

3. Lactase supplements work.

4. ⚠️ And the important one: don't cut dairy entirely. Calcium matters — Chapter 25's bone story runs across sixty years, and eliminating a major calcium source over a threshold effect is a poor trade.

⚠️ Secondary lactose intolerance — after gastroenteritis, or with coeliac disease or IBD damaging the brush border — is temporary and resolves as the gut heals. People frequently conclude they are permanently intolerant on the basis of a fortnight after a stomach bug.

28.8 ⚠️ Non-coeliac gluten sensitivity: the uncomfortable answer

I promised in Chapter 27 that this one is genuinely uncomfortable. Here is why.

A large number of people report that gluten causes them symptoms, have coeliac disease and wheat allergy excluded, and feel better without it.

⚠️ Their symptoms are real. What is contested is what causes them.

What the rechallenge studies found:

⚠️ Double-blind, placebo-controlled rechallenge in people self-reporting gluten sensitivity has repeatedly failed to reproduce symptoms with gluten specifically — once background FODMAP intake was controlled. Nocebo response rates in these studies have been substantial.

And a crossover study published in Gastroenterology in 2018 (Skodje and colleagues) tested gluten, fructan and placebo separately in people with self-reported gluten sensitivity. ⚠️ Fructan — the FODMAP in wheat — provoked symptoms. Gluten did not.

⚠️ Which points at three candidate explanations:

1. ⚠️ FODMAPs. Wheat is a major dietary fructan source. Removing gluten removes wheat, which removes fructans — Chapter 27's mechanism, and Chapter 17's bundled intervention.

2. ⚠️ Amylase-trypsin inhibitors (ATIs) — other wheat proteins with proposed immune-activating properties. A live hypothesis, not established.

3. ⚠️ Nocebo, which the blinded studies show is a substantial contributor and which is not the same as "imaginary."

🔬 Verdict: 🟡 Unclear / it depends — and I want to state both halves plainly.

⚠️ Something real is happening in at least some people. Gluten is probably not the agent in most of them. Fructans and nocebo appear to account for a great deal. And a residual group who reproducibly react to gluten under blinding may exist.

⚠️ Telling someone "there's nothing wrong with you" is both unkind and unsupported. Telling them "you have a gluten problem" is also unsupported, and it commits them to a lifelong restriction that may be aimed at the wrong molecule.

⚠️ The correct pathway, and it is specific:

1. ⚠️ Exclude coeliac disease FIRST, while still eating gluten (§28.6). 2. Exclude wheat allergy if the history suggests it. 3. ⚠️ Then a structured FODMAP process with reintroduction (Chapter 27 §27.11) — which will frequently identify fructans, at a threshold, rather than gluten absolutely. 4. ⚠️ Which usually means eating some wheat, rather than none.

28.9 The other intolerances

Briefly, because they're common and frequently misattributed.

⚠️ Oral allergy syndrome (pollen-food syndrome). Cross-reactivity between pollen proteins and similar proteins in raw fruit, vegetables and nuts. Itching and tingling of the mouth and lips, usually mild, usually confined to the mouth. ⚠️ The proteins are heat-labile, so the cooked form is generally tolerated — someone who reacts to a raw apple can usually eat apple pie. Common, reassuring once explained, ⚠️ and it does occasionally progress to more significant reactions, so it's worth assessing rather than dismissing.

Histamine intolerance. 🟡 ⚠️ Proposed as reduced diamine oxidase activity, with symptoms from histamine-rich foods. The entity is contested, the available tests are unreliable, and it is a common self-diagnosis. Something may be there; the diagnostic tools are not.

Sulphites. ⚠️ Real, particularly in people with asthma, where sulphites can provoke bronchoconstriction. Wine, dried fruit, some processed foods. Labelling is mandatory above thresholds in many jurisdictions.

⚠️ Food colours and hyperactivity. A UK study published in The Lancet in 2007 found that mixtures of certain artificial colours with sodium benzoate were associated with increased hyperactivity in children in the general population. ⚠️ The effects were modest, the finding was contested, and it led to warning labelling requirements in the EU. 🟡 to 🟢 — real, small, and much less dramatic than its reputation.

⚠️ MSG. ❌ The syndrome attributed to it originates in a 1968 letter to a journal, acquired a name referencing a specific cuisine, and has not survived double-blind testing. Glutamate is present in tomatoes, parmesan, mushrooms and human breast milk. ⚠️ The persistence of the belief, and the fact that it attached to one cuisine and not to parmesan, is worth sitting with.

Tyramine — ⚠️ a genuine and dangerous interaction with MAOI antidepressants, requiring specific dietary restriction of aged cheeses, cured meats, fermented products and some others. A real one, and a useful reminder that a small number of food restrictions are genuinely non-negotiable while most are not. ⚠️ Chapter 16 §16.6's interaction material belongs here too — the restriction is driven by the drug, not by the food.

⚠️ A note on the language, and who pays for it

The MSG entry deserves following up, because it illustrates something the rest of the chapter depends on.

The syndrome was named after a cuisine. ⚠️ Not after a compound, not after a symptom — after the restaurants of a particular ethnic group. Glutamate occurs naturally and abundantly in parmesan, tomatoes, mushrooms, soy sauce and human breast milk, and none of those acquired a syndrome.

⚠️ The belief survived decades of failed blinded testing, and it survived because the framing did work the evidence never had to.

⚠️ Which connects back to the dinner table. Language allocates seriousness.

When "allergy" is used for six different things, it stops functioning as a warning — and the person it protects loses the protection. When a syndrome is named after a cuisine, the name does the persuading.

⚠️ In both cases the vocabulary is doing work that the evidence isn't, which is Chapter 17's subject arriving in a place where the stakes include a teenager's adrenaline device.

⚠️ The practical version, and it's a small ask: use "allergy" for allergy, "coeliac disease" for coeliac disease, and "I don't tolerate it well" or "it doesn't agree with me" for everything else. It costs nothing, and it keeps the word working for the person who needs it.

28.10 ⚠️ The tests that don't work

🔬 Claim → Evidence → Verdict

The claim: "This test will tell you which foods you're intolerant to."

⚠️ IgG food antibody panels — the most widely sold, and the most consequential.

What they measure: IgG antibodies to food proteins. ⚠️ IgG to food is a normal, expected marker of EXPOSURE — and possibly of tolerance. People who eat a food regularly tend to have IgG against it. The test reliably identifies what you eat.

⚠️ Major allergy and immunology bodies have issued explicit position statements advising against the use of food-specific IgG testing for diagnosing food allergy or intolerance.

The others: ⚠️ hair analysis · applied kinesiology (muscle strength testing) · electrodermal and VEGA testing · cytotoxic and leucocyte activation assays · iridology · pulse testing. None has demonstrated validity.

📉 Evidence quality: For IgG, the biology is well characterized and points the other way. For the rest, no validity demonstrated.

Verdict: ❌ Not supported.

⚠️ And the harm is not the fee. It is: unnecessary elimination of multiple foods · nutritional inadequacy · social and psychological restriction · cost · and — the serious one — DELAYED DIAGNOSIS of coeliac disease, IBD, or a genuine allergy while someone follows a list.

28.11 ⚠️ The restriction pattern, fourth instance

Chapter 26 §26.11b named it in cancer and kidney disease. Chapter 27 found it in IBS. ⚠️ Here is the fourth, and it's the purest form.

⚠️ A person with real, distressing, unexplained symptoms is unusually motivated, unusually receptive, and unusually likely to be handed a restriction by someone offering an answer.

⚠️ What makes this instance different from the others: the restriction is generated by a test.

Which supplies three things a hunch cannot:

1. ⚠️ Apparent objectivity. A printed list, with numbers. 2. ⚠️ Specificity. Not "eat better" — twenty-three named foods. 3. ⚠️ And an explanation for years of symptoms, which is what the person came for.

And the elimination frequently helps, at first — ⚠️ because removing twenty-three foods removes most ultra-processed food, most alcohol, most eating out, and usually several genuine FODMAP triggers (Chapter 17's bundled intervention, Chapter 22's slot problem).

⚠️ Which means the test appears to be validated by the outcome. It isn't. The bundle is.

⚠️ The costs, which accumulate quietly: nutritional gaps · social withdrawal · food-related anxiety · a shrinking list, because when symptoms recur the response is to remove more · and the delayed real diagnosis.

28.12 Living with a genuine allergy

⚠️ Practical, and it belongs in a nutrition text because most of it is about food.

⚠️ And a note on precautionary labelling that people find genuinely difficult: "may contain" is applied inconsistently, is not risk-graded, and appears on a very large number of products. The result is that people either avoid everything so labelled — which is severely restrictive — or learn to ignore it, which is unsafe. That is a labelling failure, not a patient failure.

28.13 Immunotherapy: what has changed and what hasn't

⚠️ The most significant development in this chapter in a decade.

Oral immunotherapy (OIT) — giving gradually increasing doses of the allergen under supervision, to raise the reaction threshold. ⚠️ A standardized peanut product has been approved in several jurisdictions for defined age groups.

⚠️ What it achieves — and the distinction is the whole thing:

⚠️ DESENSITIZATION is not TOLERANCE.

Desensitization means the threshold rises while treatment continues. It reduces the severity of an accidental exposure.

⚠️ It does not mean the allergy is gone, it generally requires ongoing daily dosing, and the protection is lost if dosing stops.

⚠️ What it costs:

• Adverse reactions during treatment are common, and some are significant
• ⚠️ Some analyses have found that OIT increases the rate of anaphylaxis DURING the treatment period compared with avoidance, while achieving desensitization — a genuine trade-off that should be discussed explicitly
• A minority develop eosinophilic oesophagitis
• ⚠️ It demands sustained daily adherence, for years

⚠️ What it does not change: you still avoid the food, you still carry adrenaline, and you still read labels.

⚠️ Which makes the decision harder than the headlines suggest, and it is worth setting out how it actually gets made.

⚠️ There is no right answer here, and I want to be explicit about that. This is a values decision made under uncertainty, and a family that declines is not being negligent.

The thing that makes it a decision rather than a leap is the distinction in the box above: desensitization, not cure. ⚠️ A family told "this will fix the allergy" is being set up for a misunderstanding that will surface in a supermarket aisle three years later.

Also emerging: epicutaneous (patch) immunotherapy, and ⚠️ anti-IgE therapy — omalizumab was approved in the US in 2024 to reduce allergic reactions to multiple foods following accidental exposure. Which is genuinely new, is not a cure either, and changes the risk profile of accidental exposure rather than removing the allergy.

28.14 Prevention

Chapter 25 §25.5's material, because it belongs here too.

✅ Early introduction of allergenic foods reduces the risk of developing food allergy — the LEAP trial and the guideline changes that followed. ⚠️ Introduce peanut and egg in age-appropriate forms from around six months, and KEEP THEM IN THE DIET REGULARLY.

⚠️ Two additional points:

1. ⚠️ Regular ongoing consumption matters, not just introduction. A food introduced and then dropped for months can be lost.

2. ⚠️ Infants with severe eczema or existing food allergy should be assessed before peanut introduction — the LEAP population was high-risk and was screened first.

And the eczema connection is worth knowing: ⚠️ the dual-allergen-exposure hypothesis proposes that sensitization can occur through inflamed skin while oral exposure promotes tolerance — which would explain why early oral introduction protects and why aggressive eczema management matters. Plausible, influential, and not fully established.

28.14b Why has allergy become more common?

⚠️ The cross-cutting question, because it is the one people actually ask and because the candidate answers illuminate the rest of the chapter.

Food allergy prevalence appears to have risen substantially in industrialized countries over recent decades. ⚠️ Some of that is better recognition and better diagnosis. Not all of it.

The main hypotheses, none established:

💡 ⚠️ The fifth row is the uncomfortable one, and it should be said plainly.

For years, guidance advised delaying peanut and egg introduction in high-risk infants. That advice was reasonable given what was known, it was followed conscientiously, and the best current evidence suggests it was probably counterproductive.

⚠️ Chapter 17's stale-guidance category again — and this instance is a reminder that stale guidance is not just a failure to update. Sometimes the advice actively caused harm while it stood.

Which is an argument for humility about current recommendations, including the ones in this chapter.

28.15 Who this chapter is about

⚠️ How firmly I hold these

🧾 What it costs

⚠️ This chapter has an unusual economics, because two of its items are genuinely expensive and unavoidable.

⚠️ Two observations.

The gluten-free premium and the auto-injector price are the two costs in this chapter that fall on the people with the real diagnoses — and neither is a nutrition problem. Both are policy ones.

⚠️ And the expensive tests generate the expensive diets. A $250 IgG panel that removes twenty-three foods commits someone to specialty products indefinitely — which is the business model working exactly as designed.

What we don't know

⚠️ Whether a distinct gluten-reactive non-coeliac entity exists, and how large it is. Why food allergy prevalence has risen — the dual-allergen-exposure hypothesis is influential and unproven. Whether OIT can produce sustained unresponsiveness rather than desensitization. ⚠️ And how to identify the residual group in §28.8 who may genuinely react to gluten, which is the question that would settle the most arguments.

28.16 What to actually do

1. ⚠️ If you have an IgE allergy: two devices, always, in date. Adrenaline early. Never antihistamines instead. 2. ⚠️ If you suspect gluten: test first, while eating it. §28.6. 3. ⚠️ If you have unexplained iron-deficiency anaemia: ask for coeliac serology. 4. If dairy troubles you: ⚠️ find your threshold rather than eliminating. Yoghurt, hard cheese, lactase, smaller amounts with food. 5. ⚠️ Don't buy an IgG panel, and if you have one, §28.10. 6. ⚠️ Get a diagnosis before you get a diet. In that order. Every time. 7. If you have an infant: introduce allergens early and keep them in. 8. ⚠️ And if your list of avoided foods has grown over time rather than shrunk, that is the finding.

9. ⚠️ Use the word "allergy" for allergy. It costs nothing and it keeps the word working for the person who needs it.

⚠️ A closing observation about the shape of this list.

Look at what most of it is: get a diagnosis, find your threshold, test before you restrict, keep the food in.

⚠️ Six of the nine items are about NOT eliminating something, or about eliminating less than you thought. In a chapter about food reactions, the dominant error is over-restriction — and the single exception, the person with a genuine IgE allergy or coeliac disease, needs the opposite advice delivered with total seriousness.

Getting those two groups the right advice, and not each other's, is the entire clinical problem.

Spaced Review

1. (Chapter 27) Why does §28.8's pathway end with "eat some wheat" rather than "avoid gluten"?

⚠️ Because the likely agent is fructans — a FODMAP — and FODMAP reactions are THRESHOLD effects, not absolutes. A structured reintroduction (Chapter 27 §27.11) typically finds a dose at which wheat is tolerated, ⚠️ which is a completely different life from lifelong exclusion. And Chapter 27 CS2's lesson applies: two triggers out of seven, one dose-dependent.

2. (Chapter 25) Why does §28.7 warn against eliminating dairy entirely?

⚠️ Calcium, and Chapter 25 §25.10b's bone story across sixty years. Lactose intolerance is a dose-dependent enzymatic threshold, not an immune absolute — so eliminating a major calcium source over a threshold effect is a poor trade, particularly in adolescence and around the menopause transition.

3. (Chapter 2) Why does testing a broad panel in someone with no suggestive history cause harm?

⚠️ Because the tests measure SENSITIZATION, not allergy, and false positives are common. In a population with low pre-test probability, most positives are false — and each one generates an elimination. ⚠️ Worse, removing foods like peanut and egg can INCREASE the chance of developing true allergy (§28.14). The history drives the testing; the testing does not drive the diagnosis.

Project Checkpoint: Your Reaction Log

Component twenty-eight. ⚠️ Before you eliminate anything.

Step 1 — What do you currently avoid, and why?

⚠️ The last column is the exercise. "I don't know" is a common and useful answer.

Step 2 — ⚠️ The urgent screen, first.

• [ ] ⚠️ Have I ever had a rapid reaction involving breathing, throat, or faintness? → This is an allergy assessment, today
• [ ] ⚠️ Do I have unexplained iron-deficiency anaemia? → Coeliac serology
• [ ] ⚠️ Am I about to remove gluten? → TEST FIRST (§28.6)

Step 3 — Log properly, for two weeks. ⚠️ Not "I think dairy bothers me" — record it.

⚠️ The delay column is diagnostic. Minutes to two hours suggests IgE. Hours suggests enzymatic or osmotic. Days suggests something else or nothing.

⚠️ And the last column is the one people omit, which is why single-food attributions are so often wrong.

Step 4 — The dose question.

⚠️ "Have I ever tried a SMALLER amount?" ____

For everything except IgE allergy and coeliac disease, the answer is usually a threshold — and most people have never looked for theirs.

Step 5 — ⚠️ The trajectory question:

Number of foods I avoided two years ago: _ · Now: _

⚠️ If that number has grown, that is the finding (§28.11), and Chapter 34 is the chapter.

Step 6 — ⚠️ The language audit, which takes thirty seconds.

What word do I use? ____

⚠️ If it's "allergy" and the mechanism in Step 1 isn't IgE or coeliac disease — change it. "I don't tolerate it well" is accurate, gets you the same food, and leaves the stronger word available for the person who needs a restaurant to take it seriously.

Step 7 — And if you're holding a test result:

⚠️ What did it measure? _ Was it IgE, IgG, or something else? _

⚠️ If IgG: it measured what you eat.

Next checkpoint (Chapter 29): the clinical picture — what changes when nutrition becomes treatment rather than prevention.

Chapter Summary

⚠️ Six mechanisms, one word. IgE allergy · non-IgE immune · autoimmune (coeliac) · enzymatic · pharmacological · functional/osmotic — plus nocebo and preference. ⚠️ Self-reported food allergy runs several times higher than challenge-confirmed allergy, and the gap costs the person with the real one.

⚠️ §28.3's rule that removes hesitation: airway, breathing or circulation involvement — or two or more systems — after a likely exposure. Give adrenaline. And note the infant presentation: sudden floppiness, pallor, persistent crying, drooling or vomiting, because they cannot report throat tightness.

⚠️ §28.14b's uncomfortable row: decades of guidance advised DELAYING allergenic foods, it was followed conscientiously, and the best current evidence suggests it was probably counterproductive. Stale guidance is not just a failure to update — sometimes the advice actively caused harm while it stood. ⚠️ Which is an argument for humility about current recommendations, including this chapter's.

⚠️ §28.1's dose column separates the two that matter: coeliac disease and IgE allergy are where small amounts count. Almost everything else is a threshold — so the question is usually "how much," not "never."

⚠️ §28.6 is the most actionable instruction in the chapter: test for coeliac disease BEFORE removing gluten. Remove it first and serology and biopsy both normalize, and the only way back is a deliberate weeks-long gluten challenge that many people refuse — leaving them on a lifelong restriction with no diagnosis, no follow-up, no bone monitoring, and no answer for relatives who should be screened.

⚠️ §28.11 is the restriction pattern's fourth instance, and the purest: the restriction is generated by a TEST, which supplies apparent objectivity, specificity and an explanation — and the elimination frequently helps at first, because removing twenty-three foods removes a bundle. The test appears validated by the outcome. It isn't. The bundle is.

The one thing to remember: ⚠️ Five of the six people at that table did not have an allergy — and the sixth is the one who pays for it.

What's Next

Chapter 29 closes Part V with ★ clinical nutrition — what happens when food stops being prevention and becomes treatment.

Malnutrition screening in hospital, which is more common and more consequential than most people realize. Enteral and parenteral nutrition. ⚠️ Refeeding syndrome — the one place where feeding someone too enthusiastically can kill them. Nutrition in critical illness, where almost every intuition is wrong. Drug–nutrient interactions properly. Perioperative nutrition and prehabilitation.

⚠️ And the honest through-line: this is where Chapters 24, 25 and 26's inversions all arrive at once, and where "eat less" becomes the most dangerous sentence in the book.