How common is mustard allergy?
This ought to be one of the easiest questions in this guide.
It isn't.
If somebody asks how many people in Britain have mustard allergy, the most accurate current answer is:
We don't know.
Anaphylaxis UK describes mustard allergy as thought to be rare in the UK, but explicitly says that the number of people affected here is unknown.
That might seem surprising.
Britain has a sophisticated healthcare system. Mustard is one of the 14 regulated allergens. People have been diagnosed with mustard allergy for decades.
Surely somebody has counted them?
The difficulty is that counting people with food allergy is much harder than counting something like births or prescriptions.
First, we have to decide what exactly we are counting.
And depending on how we answer that question, we can produce spectacularly different numbers.
"I have a food allergy" is one measurement
One way to estimate food allergy is simply to ask people.
Do you have a food allergy?
This is useful.
People know their own experiences, and large questionnaires can reach thousands of participants relatively cheaply.
But self-reported food allergy includes several different things.
Some people have a specialist-confirmed IgE-mediated allergy.
Some have experienced convincing reactions but have never been investigated.
Some have an intolerance.
Some experienced food poisoning and attributed it to a particular ingredient.
Some have gastrointestinal disorders.
Some were told in childhood that they were allergic and have never been reassessed.
Some received a result from an unvalidated commercial "food intolerance" test.
And some simply feel consistently unwell after eating a food.
Those experiences can all be genuine.
They are not necessarily the same medical condition.
Consequently, self-reported food allergy tends to produce much higher prevalence figures than carefully confirmed food allergy.
A large European systematic review illustrates just how dramatic the difference can be.
Across the studies it analysed, the pooled point prevalence of self-reported food allergy was about 13%.
The pooled prevalence based on a positive food challenge was about 0.8%.
Those figures should not be interpreted as saying that 12 people out of every 13 reporting a food problem are imagining it.
They demonstrate that the phrase food allergy means different things when measured in different ways.
A positive blood test gives us another number
Instead of asking people, we could test their blood.
Now we are measuring something more objective.
But we encounter the problem from the previous section.
Specific IgE tells us about sensitisation.
It does not automatically establish clinical allergy.
The same European review estimated point prevalence of sensitisation measured by specific IgE at about 16.6% — actually higher than its pooled estimate of self-reported current food allergy.
That initially looks bizarre.
How can more people have the allergy antibody than say they have an allergy?
Because the antibody isn't the disease.
Some people have detectable IgE to foods they eat without any problem.
Cross-reactive proteins can also produce positive test results.
So if we define "food allergy" as "anybody with detectable food-specific IgE", we create a population containing people who are perfectly capable of eating the food.
The test has answered:
Does the immune system recognise this?
It has not necessarily answered:
Does eating it make this person ill?
Skin testing produces another number again
We could instead perform skin-prick tests.
The same European analysis found a pooled point prevalence of positive skin-prick testing to foods of approximately 5.7%.
Now we have three very different figures:
Self-reported current food allergy: around 13%.
Specific-IgE sensitisation: around 17%.
Skin-prick sensitisation: around 6%.
And challenge-confirmed allergy: below 1%.
They are all results from legitimate forms of research.
They are simply measuring different things.
This is why an article stating:
"X% of the population has food allergy"
without telling you how allergy was defined has omitted one of the most important pieces of information.
A percentage can look wonderfully authoritative while concealing a very messy denominator.
The closest thing we have to asking the body directly
This is one reason the controlled oral food challenge is so important in allergy research.
Instead of asking:
"Do you think this food causes symptoms?"
or:
"Do you have antibodies capable of recognising it?"
researchers actually observe what happens when the food is eaten under controlled conditions.
That gets us much closer to clinical reality.
But it creates a practical problem.
You can email a questionnaire to 50,000 people.
You cannot easily bring 50,000 people into allergy clinics, feed them a series of suspected allergens under supervision and monitor them for reactions.
Food challenges are time-consuming, expensive and potentially capable of provoking serious reactions.
So the diagnostic method that gives us some of the best evidence is also one of the hardest methods to use for enormous population studies.
Epidemiology is full of compromises like this.
The easiest thing to measure is not always the thing we most want to know.
Mustard has an additional numbers problem
For peanut, milk or egg allergy, researchers have relatively large populations to study.
Mustard allergy is less common in many countries.
That means individual mustard studies can be small.
A study might include a few dozen mustard-sensitised people.
Another might examine children attending a specialist allergy clinic.
Another might recruit people specifically because mustard allergy is already suspected.
Those studies can tell us extremely useful things about people with mustard allergy.
They cannot automatically tell us what percentage of the entire British population has mustard allergy.
This is a classic problem called selection bias.
If you stand inside an allergy clinic and count allergies, you are not looking at a miniature version of Britain.
You are looking at a building specifically populated by people who are unusually likely to have allergies.
It would be like standing inside a tyre shop and concluding that an extraordinary percentage of British cars have punctures.
How mustard became a major childhood allergen in French studies
This is where some older French research becomes fascinating.
In one substantial series of French children with challenge-confirmed food allergy, mustard accounted for approximately 6% of the food allergies recorded.
Another French paediatric series reported mustard among five foods responsible for the great majority of confirmed food hypersensitivity, with mustard implicated in nearly 9% of the affected children in that selected clinical population.
Those are striking numbers when viewed from Britain, where mustard allergy is regarded as rare.
But notice what those percentages do not mean.
They do not mean 6–9% of French children are allergic to mustard.
They mean mustard represented that proportion of allergies among selected children who already had confirmed food hypersensitivity in those clinical studies.
That is an enormous difference.
If 100 children with food allergy attend a specialist service and six have mustard allergy, mustard accounts for 6% of the food allergies.
It tells us nothing directly about the other thousands of children outside the clinic.
This is one of the easiest statistical mistakes to make when reading allergy research.
Even positive mustard tests can greatly overestimate actual allergy
One prospective study provides an almost perfect demonstration.
Researchers investigated children with positive mustard skin-prick tests.
There were 36 of them.
If we stopped there, we might call all 36 "mustard allergic".
Then the researchers performed food challenges.
Only 15 were confirmed as allergic.
The other 21 had negative challenges.
In another prospective study involving people sensitised to mustard, only about 23% reacted to a routine dose during controlled challenge.
Different study designs and populations mean those percentages should not be directly compared as though they were one experiment.
But the principle is unmistakable.
Mustard sensitisation can substantially outnumber challenge-confirmed mustard allergy.
That is exactly why population estimates based solely on positive allergy tests need careful interpretation.
Geography genuinely matters
Food allergy is not distributed evenly across the world.
Even within Europe, large research programmes have found substantial geographical differences both in overall food-allergy prevalence and in the foods responsible.
In a multicentre European adult study, estimated probable food allergy varied markedly between participating cities.
And the dominant foods differed.
Hazelnut, peach and apple were important in some European centres.
Fish and shrimp featured more prominently in others.
That makes biological sense.
You cannot easily become allergic through ordinary dietary exposure to a food nobody around you eats.
Different populations eat different foods, prepare them differently and encounter different pollens and environmental exposures.
Genetic backgrounds also vary.
So do infant-feeding practices, agricultural environments, lifestyles and potentially many other influences.
Food allergy emerges from the interaction between biology and environment.
Mustard provides an unusually visible example.
France and mustard: familiarity changes exposure
Mustard has a much more established culinary presence in France than it traditionally has in Britain.
It appears not merely as a condiment placed beside food but as an ingredient in sauces, vinaigrettes and numerous preparations.
Greater consumption creates greater opportunity for exposure.
Greater exposure can make an allergen more important at population level.
It can also make clinicians more likely to recognise and investigate it because they encounter cases more frequently.
This creates a feedback loop.
A food commonly eaten in a country produces more opportunities for reactions.
Doctors become more familiar with the allergy.
Researchers study it more.
Diagnostic testing becomes more established.
More cases are identified.
The allergy becomes more visible.
None of that means French immune systems have a special dislike of Dijon mustard.
It means exposure patterns matter.