Two chocolate bars sit on a shelf. One says “may contain traces of nuts”. The other says nothing. A shopper managing a nut allergy reasonably assumes the second bar was tested and found clean, and the first was tested and found borderline.
That assumption is wrong in almost every respect. In most jurisdictions, the declaration of an ingredient deliberately added to a food is tightly regulated, standardised in wording, and legally enforceable. The precautionary statement about what might have got in accidentally is neither standardised nor generally required, and there is usually no rule stating what level of risk justifies printing it or what evidence must exist before it is left off. The silent bar may have been made on a line that has never handled nuts, or on a line that handles them daily by a company with a different legal opinion.
Understanding why the labels behave this way means understanding what an allergen test can and cannot measure, and where in a factory allergens actually travel.
Key takeaways
- Allergen tests detect protein, not the food itself, and always through a proxy target chosen by the kit maker.
- Processing can destroy the shape a test antibody recognises while leaving the protein capable of triggering a reaction.
- Cross contact happens mostly through shared equipment, rework, airborne dust and people, not through raw ingredients.
- Precautionary “may contain” labelling is voluntary and unstandardised in most markets, so its absence is not evidence of absence.
- A negative swab or product test means below the detection limit of one method on the portions examined, nothing more.
How Allergen Protein Detection Works
The thing that causes an allergic reaction is a protein, or more precisely a small region on a protein’s surface that the immune system has learned to recognise. That region is called an epitope. Allergen testing works by mimicking the immune system: it uses antibodies raised against the same proteins to capture and report them.
This immediately sets the limits of the method. A test does not detect peanut. It detects one or a few peanut proteins, and only in a form its antibodies can still bind. Everything that happens to food between the field and the packet can change that form.
Extraction comes first and is more consequential than it appears. The sample is ground and mixed with a buffer designed to pull target proteins into solution, and some matrices release them readily while others do not. Chocolate contains polyphenols that bind proteins and hold them back, which is why allergen kits for chocolate include additives to counter that effect. High-fat products and dense baked goods recover poorly compared with a simple flour or liquid, and a protein that never leaves the matrix cannot be detected however sensitive the assay.
Then there is heat. Roasting, baking, extrusion and retort processing denature proteins, unfolding them and causing them to aggregate or bind irreversibly to sugars and other food components. Some epitopes are linear stretches of amino acid sequence and survive this reasonably well. Others are conformational, meaning they exist only while the protein holds a particular three-dimensional shape, and they are destroyed by heating. The uncomfortable consequence is that a heat-treated protein can lose the shape a test antibody needs while retaining enough structure to provoke a reaction in a patient, so a low or negative result on a processed food does not reliably indicate low allergenicity.
DNA-based testing using PCR sidesteps some of this, because DNA survives processing better than protein conformation does. But DNA is not the allergen, and the two can come apart: a refined oil may carry detectable plant DNA with negligible protein. PCR answers whether a species passed through, which is a proxy for the thing that matters.
ELISA Kits and Their Target Proteins
The workhorse of allergen analysis is the enzyme-linked immunosorbent assay. In the common sandwich format, antibodies fixed to the wall of a well capture the target protein from the extract, a second antibody carrying an enzyme binds to the captured protein, and a substrate is added that the enzyme converts into colour. More target means more colour, and the intensity is read against a calibration curve.
The design choices inside the kit determine what it reports.
The first choice is the target. Different manufacturers raise antibodies against different proteins from the same food, and some kits target a total protein fraction rather than a single named protein. Results are then reported in different units, sometimes as the target protein, sometimes converted into an equivalent quantity of the whole food using an assumed protein content. Two kits can therefore analyse the same sample and produce numbers that differ severalfold without either being defective.
The second is cross-reactivity. Antibodies bind by shape, and related species share shapes. Kits for tree nuts frequently show some response to related nuts, legume kits can respond to related legumes, and gluten kits respond to related cereal proteins by design. A positive therefore identifies a family more reliably than a species.
The third is the calibrator and the matrix. A kit calibrated with unprocessed material will under-report the same allergen after roasting. Kit makers publish recovery data across matrices for this reason, and a laboratory choosing a kit for a chocolate line makes a different choice from one testing a cereal.
| Approach | What it measures | Main strength | Main weakness |
|---|---|---|---|
| Laboratory ELISA | Specific allergen protein, quantified | Sensitive, quantitative, widely accepted | Recovery falls sharply in processed and fatty matrices |
| Lateral flow strip | Same targets, yes or no at a cut-off | Fast, usable on the line by trained staff | Not quantitative; cut-off fixed by the kit |
| PCR | DNA of the source species | Survives heat processing well | Detects species, not allergenic protein |
| Mass spectrometry | Specific peptide sequences | Multiple allergens at once; robust to shape loss | Costly, slow, needs specialist expertise |
| Total protein swab | Any protein residue on a surface | Cheap, immediate, good cleaning indicator | Not allergen-specific at all |
Mass spectrometry is the interesting outlier. Because it identifies short peptide sequences rather than folded shapes, heat denaturation troubles it far less. Cost, throughput and the absence of an agreed reference method keep it confirmatory rather than routine.
Where Cross Contact Actually Happens
Cross contact is the industry term for the accidental transfer of an allergen into a food that does not list it. Understanding where it occurs explains why testing finished product is a weak control.
Shared equipment is the dominant route. Most factories run several recipes through the same mixers, hoppers, conveyors, extruders and fillers. Anywhere product can lodge, it will: dead legs in pipework, the corners of hoppers, screw flights, the inside of augers, seals and gaskets. When the next recipe runs, that lodged material is gradually released, so contamination is heaviest at the start of a run and declines. Production scheduling exploits this, running allergen-free products first after a full clean and allergen-containing products later in the sequence.
Rework is the route that surprises people. Off-specification product, trimmings and start-up material are often reintroduced into later batches to reduce waste. If rework containing an allergen enters a product that does not declare it, the result is contamination with no equipment fault at all. Disciplined rework rules, meaning like into like only and full traceability, are among the most effective allergen controls a plant can have.
Airborne dust matters wherever powders are handled. Flour, milk powder, soya and nut meal generate fine dust that settles across a wide area and travels on air currents through open doorways. A line that never handles peanut can be contaminated by a peanut-handling line on the other side of the room.
People and tools complete the picture. Gloves, aprons, scoops, brushes and hands carry residues between areas, which is why zone-specific colour-coded tools and controlled staff movement exist.
Finally, ingredients arrive with their own histories. A supplier’s spice blend or bulk powder may itself have been produced on shared equipment, so a manufacturer’s risk assessment is only as good as the supplier information behind it.
Cleaning Validation and Swab Testing
If cross contact is mostly an equipment problem, the control is cleaning, and the evidence that cleaning works is validation.
Validation is a one-off study proving that a defined cleaning procedure, performed as written, reduces allergen residue below a defined limit on a defined piece of equipment. It usually involves deliberately soiling the line with the allergen-containing product, applying the cleaning procedure exactly, and then sampling extensively, with particular attention to the worst-case locations that are hardest to clean. Once the procedure is validated, routine production relies on verification, meaning lighter checks confirming that the validated procedure was followed and produced the expected result.
Sampling technique determines the answer more than the test kit does. A swab pressed firmly and worked over a defined area in overlapping passes recovers far more residue than a light wipe, and a swab from an easy flat surface says nothing about the inside of a valve. Wet surfaces dilute residue and produce falsely reassuring results, so swabbing is done on dry equipment. Where equipment cannot be swabbed, rinse water can be sampled instead, though the dilution raises the effective detection limit considerably.
Two kinds of test are used in parallel. Allergen-specific lateral flow devices answer the direct question at a fixed cut-off. Generic protein swabs, which change colour in the presence of any protein residue, answer a broader question cheaply: is this surface visibly and chemically clean. A generic protein swab cannot identify an allergen, but a surface that fails it has not been cleaned properly, and that is often the more actionable finding.
The record matters as much as the result. Where a batch is later questioned, the cleaning records, validation study and verification swabs are the evidence that risk was controlled, and a swab result with no record of what was cleaned, when and against which procedure proves very little.
Why May Contain Labels Are Voluntary
Here the food safety picture collides with the legal one. In most major markets, the rules covering allergen labelling apply to ingredients. A defined list of allergenic foods must be declared clearly whenever they are used deliberately, and the emphasis, wording and placement are prescribed. This part of the system works well.
Precautionary allergen labelling, the “may contain” family of statements, sits almost entirely outside that framework. It is generally voluntary. The wording is unregulated, which is why shelves carry a proliferation of phrasings, from “may contain” to “made in a facility that also processes” to “not suitable for”, with no defined difference in meaning between them. Crucially, there is usually no legal requirement to apply a precautionary statement when a real risk exists, and no requirement to justify one when it is applied.
The incentives that follow are predictable. Applying a warning costs almost nothing and reduces liability exposure. Omitting one requires a manufacturer to be confident enough in its own controls to accept the consequences of being wrong. The result is over-application: warnings appear on products where the risk is negligible, and occasionally do not appear on products where it is not.
That over-application has a genuine cost to the people it is meant to protect. When a large fraction of ordinary products carries a warning, allergic consumers face a choice between an extremely restricted diet and ignoring the warnings. Many end up ignoring them selectively, based on brand trust or past experience, which is exactly the situation labelling was supposed to prevent. A warning that carries no information trains people not to read warnings.
Regulators and expert bodies in several regions are working toward standardised precautionary labelling anchored to measured risk, so that a statement appears only when a defined threshold could plausibly be exceeded. Progress has been slow and uneven, and until it is complete the phrase on the packet describes a company’s risk appetite rather than the food.
Threshold Doses and Reference Amounts
The scientific groundwork for fixing this exists, and it rests on the observation that allergic reactions are dose-dependent. There is a quantity of allergen protein below which a given individual reacts either not at all or only trivially, and that quantity varies enormously between people.
Clinical challenge studies, in which consenting allergic patients consume carefully escalating quantities under medical supervision, allow this variation to be described as a distribution across a population. From that distribution, expert panels derive a reference dose: an amount of allergen protein chosen so that only a small proportion of the allergic population would be expected to react, and reactions in that minority would be expected to be mild. These reference doses are expressed as a quantity of total protein from the allergenic food, and they are strikingly small, far below what most people imagine when they think of a trace.
Turning a reference dose into a labelling decision takes two further steps. The dose is divided by the amount a consumer might reasonably eat at one sitting, converting it into a concentration comparable with an analytical result. Then the manufacturer estimates how much allergen its process could realistically introduce, which is a modelling exercise on worst-case assumptions rather than a measurement.
Quantitative frameworks built on this logic are in use, and they let a manufacturer state, with reasoning, whether a precautionary label is warranted. They also expose the limits of analysis, since the concentrations involved often sit close to what routine methods can reliably quantify in a difficult matrix.
What a Negative Test Genuinely Means
A negative allergen result is a statement with several conditions attached, and every one of them narrows it.
It applies to the portions actually tested, and contamination from shared equipment concentrates at the start of a run, so a sample from mid-batch can be clean while the first minutes were not. It applies to the target proteins the kit recognises, not every allergenic component. It applies to the extraction efficiency achieved in that matrix, which in a fatty or heated product may be far from complete. And it applies only down to the method’s limit of quantification, which in a hard matrix can sit above the concentration a highly sensitive individual would react to.
None of this makes testing pointless. It makes testing a verification tool rather than a safety mechanism. A well-run allergen programme controls risk through segregation, scheduling, validated cleaning, rework discipline and supplier assurance, then uses testing to confirm those controls are behaving as expected and to investigate when something changes.
Frequently asked questions
Does no warning on the label mean the product is free from that allergen?
No. Precautionary labelling is voluntary in most markets, so the absence of a statement can mean the manufacturer assessed the risk as negligible, or that it decided not to label, or that the question was never formally considered. Deliberately added ingredients must always be declared, so the ingredient list is reliable in a way the precautionary statement is not. For a serious allergy, contacting the manufacturer about their allergen controls gives more information than the packet does.
Is “made in a facility that also handles” safer than “may contain”?
There is no defined difference. The phrases are not standardised in most jurisdictions, and companies use them inconsistently, so one manufacturer’s “may contain” and another’s “made in a facility” can describe identical situations, or very different ones. Reading a gradation of risk into the wording is not supported by anything in the labelling rules. Allergy organisations generally advise treating all precautionary phrasings as carrying the same meaning.
If a laboratory can detect tiny amounts, why not just test every batch?
Because testing samples a small portion of a batch in which contamination is unevenly spread, and because sensitivity collapses in difficult matrices. A cocoa-rich or heavily baked product can hide allergen protein from an antibody-based assay through poor extraction and heat damage to the epitopes. Testing every batch would add cost while providing weaker assurance than controlling the process, which is why regulators and standards bodies emphasise prevention with testing as verification.
Why do two laboratories report different allergen levels for the same sample?
Chiefly because they used different kits. Kits target different proteins, use different calibrators, and convert results into whole-food equivalents using different assumptions, so the numbers are not directly comparable. Extraction differences and matrix effects add more variation. This is why allergen results should be interpreted alongside the method used, and why comparing a figure from one report against a threshold derived using another method can mislead.
Are allergen thresholds the same everywhere?
No. Gluten is the notable exception, with a widely harmonised limit for foods described as gluten-free. For other allergens, reference doses derived from clinical challenge data are broadly similar across the expert groups that have published them, but jurisdictions translate them into labelling rules differently and several have adopted no formal thresholds at all.
The practical position for someone managing an allergy is unsatisfying but clear. Read the ingredient list, because it is legally controlled and reliable. Treat precautionary statements as information about the manufacturer rather than about the product, and take all wordings as equivalent. Where the allergy is severe, ask companies directly what their controls are, since a manufacturer with genuine segregation and validated cleaning will usually be able to describe it in detail, while one that applies warnings reflexively will not.
For the industry, the direction of travel is toward statements that mean something: a precautionary label applied because a quantitative assessment found a plausible risk, and omitted because it did not. Until that is the norm, the gap between what the sentence appears to say and what it actually says keeps falling on the people least able to absorb the risk.
This is education, not medical advice. Laboratory results only carry meaning alongside your symptoms, history and examination. Talk to a qualified clinician about your own results before changing anything about your care or supplements.
