Time to ACT-UP: Update on precautionary allergen labelling (PAL)^☆

Introduction

Ten years ago, the World Allergy Organization (WAO) published a proposal to establish an international framework for the use of Precautionary Allergen (“may contain”) Labelling (PAL) to communicate unintended allergen presence (UAP) in foods (usually, but not limited to prepacked foods).¹ At the time, it was already clear that PAL use was increasing but often applied inconsistently and without regulation, which reduced its usefulness to consumers with food allergy and those purchasing food for them.²

In 2019, the World Health Organization (WHO) and the Food and Agriculture Organization (FAO) of the United Nations convened an expert consultation to address the issue,³ the outputs of which are now being considered by the Codex Committee on Food Labelling (CCFL) for implementation into global food standards. We summarise here the latest data to inform PAL and review the FAO/WHO Expert Consultation recommendations. We also identify areas of uncertainty which might hinder international agreement on a regulated framework for precautionary labelling of food allergens. A definition of key terms and concepts used throughout the paper appears in Box 1.

Food allergy – the patient experience

The gold-standard for establishing the prevalence of food allergy is a longitudinal population-based study in a defined birth cohort, evaluating the presence of FA through double-blind, placebo-controlled oral food challenges (OFC). However, there are considerable challenges and costs involved. Thus, many studies use surrogate measures of clinical reactivity (rather than OFC), such as the presence of IgE-antibody to a food allergen with or without a recent clinical history suggestive of reaction. The use of surrogate measures can overestimate prevalence by over ten-fold.⁴ In children, FA prevalence has been estimated to be between 0.1% and 9.3% with large geographical variations, which suggests a larger disease burden in the developed world.⁵ The prevalence of food allergies has been reported at 1.5% in Hong Kong, 0.87% in Russia, 0.21% in mainland China, and 0.14% in rural India.⁶ Data are more limited for adults. A population-based survey of US adults estimated a prevalence for FA of 10.8%, of which around half were adult-onset.⁷ In the United Kingdom, FA prevalence in adults has been estimated to be between 1.0 and 1.8% (population-based cohort of ∼7 million individuals)⁸ and 5.7% for IgE-mediated FA in a community survey, some of whom underwent confirmatory OFC.⁹

In terms of the most common triggers for food-induced allergic reactions, a systematic review recently evaluated the most common foods causing anaphylaxis presenting to Emergency Departments globally (Fig. 1).¹⁰ As reported by the authors, the most common triggers generally corresponded to stated “priority” allergens for any given region (see Section 2).

Living with FA, or being a caregiver for an affected individual, has a major impact on socialisation and household finances. Food allergies impact several domains in quality-of-life, including attendance at daycare/school/college; social interactions, including birthday parties; family relationships; finances; shopping for safe foods; eating out and travelling; and mental health.¹¹ This results in a decreased quality-of-life, primarily driven by the impact on social interactions and dietary diversity.¹² Children/young people and adults report increased stress and anxiety relating to fear over accidental reactions (which can exacerbate social isolation); concern about one's confidence and ability to deal with reactions (self-efficacy); adverse influence on interpersonal and social relationships; restrictions on career options; and fears for the future. Food-allergic consumers move between balanced vigilance and appropriate dietary avoidance (which promotes effective coping strategies) and periods of hypo/hypervigilance, putting physical health or psychological wellbeing at risk.¹³ Ambiguity over the interpretation of allergen food labels is a key driver of the impairment in quality-of-life reported by food-allergic individuals and their carers.^2,14

Current legislation for food allergen declaration

In order to promote safe food choices, relevant information must be communicated to consumers in a transparent and accessible manner. In addition, food-allergic consumers must be able to trust the information provided.¹⁵

The Codex Alimentarius (often abbreviated to “Codex”) is a collection of internationally adopted food standards and related texts presented in a harmonized document, overseen by WHO and FAO. The Codex standards aim to protect consumer health and ensure fair practices in international food trade, and include provisions in respect of food hygiene, food additives, food contaminants, labelling and presentation, import/export inspection and certification. Since 1985, Codex has included food allergens, with the General Standard for the Labelling of Prepackaged Foods (GSLPF)¹⁶ requiring mandatory declaration of 8 “priority” allergens.

• •

  Cereals containing gluten; ie, wheat, rye, barley, oats, spelt or their hybridised strains and products of these;

• •

  Crustacea and products of these;

• •

  Eggs and egg products;

• •

  Fish and fish products;

• •

  Peanuts, soybeans and products of these;

• •

  Milk and milk products (lactose included);

• •

  Tree nuts and nut products; and

• •

  Sulphite in concentrations of 10 mg/kg or more.

Strictly speaking, sulphite is not an allergen but a chemical preservative that can elicit respiratory symptoms consistent with an allergic reaction. For this reason, sulphites were not included as an allergen in the recent FAO/WHO Expert review.¹⁷ CCFL is currently revising the GSLPF based on the recommendations of the FAO/WHO Expert review, and it is likely that in the future, oats will be considered separately from gluten-containing cereals. Soy and peanut will be considered as separate allergens, and tree nuts will need to be specified by name.

Currently, almost 100 countries have legislation requiring mandatory declaration when an allergen is intentionally present as an ingredient. The Codex standards have been adopted by most countries, either by reference to Codex itself or its inclusion in local legislation (see Table 1). Many countries have modified the list of “priority” allergens to meet local needs.

In Europe, most countries are aligned with legislation from the European Union (EU), which mandates the declaration of 14 allergens (the Codex-8, with peanut and soya named separately, plus celery, mustard, sesame, lupine, molluscs).¹⁸ The most recent relevant legislation is EU Regulation 1169/2011,¹⁸ commonly known as Food Information to Consumers (FIC), which requires mandatory declaration for these 14 allergens for both prepacked and non-prepacked foods (including those sold by catering outlets such as restaurants). FIC also stipulates how ingredients should be listed in terms of font size on labels. “Contains” statements are prohibited; instead, allergens must be declared within the ingredients list, and emphasised (e.g. in bold print). Specific disclosure of individual gluten-containing cereals and tree nuts is required.

Japan, which was the first country to regulate both intentional and unintended allergen presence (UAP), uses a localised list of allergens, divided into those for which disclosure is mandatory (wheat, buckwheat, egg, milk, peanut, shrimp, crab and walnut) and allergens where disclosure is recommended but not required (almond, abalone, squid, salmon roe, oranges, cashew, kiwifruit, beef, sesame, salmon, mackerel, soybean, chicken, banana, pork, macadamia nut, peach, yam, apple and gelatin).¹⁹ A similar approach has been adopted in South Korea, although the list of allergens is different.²⁰ The US has implemented mandatory allergen declaration through the Food Allergen Labelling and Consumer Protection Act 2004 (FALCPA).²¹ FALCPA includes the same allergens as Codex, with the notable exception that wheat is the only named cereal (other gluten-containing grains are not included). Subsequently, the 2021 Food Allergy Safety, Treatment, Education, and Research (FASTER) Act amended FALCPA to add sesame as an allergen.²² For tree nuts, the specific nut must be declared; for crustacea and fish, the species must be declared.²³ “Gluten-free” is a voluntary claim; if used on a food label, gluten must not exceed 20 parts per million (ppm). Priority allergens in Canada include the Codex-8, but also molluscs and mustard. In both US and Canada, priority allergens must be declared in the ingredients list, or optionally using a separate “Contains” statement for priority allergens^23,24 — something prohibited in the European Union.¹⁸ A “contains statement” is mandatory for priority allergens in Australia and New Zealand.²⁵

Precautionary allergen labelling

Food production has become increasingly complex, involving multiple supply chains across many countries to try and drive efficiencies and lower costs to the consumer. Despite adherence to “Good Manufacturing Practices” (GMP), allergens may be present because of unintentional cross-contact (also referred to as cross-contamination) between foods either during harvesting, storage, transport or processing (Fig. 2). The Codex standard is widely interpreted as being applicable only to intentionally-added components – and not to allergens that may inadvertently be added during the supply chain or production (ie, UAP). In many countries, this distinction is made explicit. For example, in Australia/New Zealand, the European Union and United Kingdom, local regulations make it explicit that labelling applies only to allergens (or their derivatives) present as a result of intentional use as an ingredient, and not due to UAP.

Many food businesses mitigate against the risk of UAP using PAL, although this does not replace the need for compliance with GMP and risk management protocols such as Hazard Analysis Critical Control Point (HACCP). PAL emerged in the 1980s as a risk-communication measure to allow food businesses to provide information to allergic consumers that a product might pose risk to them (ie, trigger an allergic reaction) due to allergen cross-contact or cross-contamination, because of production on shared equipment or allergens being present within the same facility. Examples of PAL statements are shown in Fig. 3.²⁶

In almost all countries, PAL statements are not regulated or standardised,^26,27 although guidance in some countries (United States, Canada, European Union, and United Kingdom) states that PAL cannot be misleading. This has resulted in very inconsistent application by food businesses, with consumers often unaware of the reason why a PAL is present (or absent), and whether a risk assessment has been conducted. (Fig. 4). Consumers frequently perceive that PAL are used by manufacturers as a “safety net” to convey an unspecified risk of possible cross contact, rather than actual risk.²

In 2014, WAO published a summary of the inconsistencies in PAL application, and the legislative “grey zone” that exists.¹ A few countries stipulate rules over the use of PAL (see Table 1), although these rules are not always consistent nor helpful in communicating actual risk to consumers. In the European Union, PAL are not regulated; however, the Czech Republic has instituted legislation relating to UAP since 2018, while other member states (eg, Belgium, the Netherlands) have instituted local guidelines and intent-to-legislate notices to establish “action levels” for PAL – although there have been significant variations in the action levels proposed,²⁷ something which increases the potential for inconsistent application of PAL. Further confusion arises from the different regulatory approaches to the use of “Contains …” statements for declaring intentionally-added ingredients that are allergens (voluntary in some countries such as the United States and Canada, mandatory in others eg, Australia, prohibited in the European Union).

PAL is now used for the majority of supermarket products in Australia (65%, 2011),²⁸ Canada (72%, 2021;²⁹ 67%, 2022³⁰), United Kingdom (55%, 2014),³¹ Latin America (33–63%, 2020),³² and a large percentage in the United States (17% across all food categories, >50% for candy and cookies, 2009).³³ The latter is almost certainly an underestimate, reflecting the frequent use of “minor ingredient labelling” in the United States, a practice now appearing in other countries too. This scenario occurs when manufacturers declare an unintended allergen as an actual ingredient (appearing as 1 of the last 3 ingredients on the label), perhaps to serve as a stronger deterrent than PAL.³⁴ However, this practice has seen a resurgence with the intentional addition of small amounts of sesame to food products by manufacturers to circumvent the need to risk-assess and manage potential UAP due to sesame — something now required in the United States due to sesame being added as a priority allergy in the FASTER Act.³⁵

The high proportion of products with PAL increases the perception that PAL are applied indiscriminately.^2,26 The use of multiple PAL statements, each with different wording, adds to this perception. For example, a 2006 US survey of supermarket products reported 25 different PAL statements on 3442 different products,³³ while a similar survey in Latin America found 33 different types of PAL statement on food products.³² Shopping for a food-allergic person takes longer and costs more, which has obvious implications on consumer choice.³⁶ This is likely to cause an even higher burden in families with low or no income, exacerbating existing socioeconomic disparities in allergy care provision. Social food schemes (such as food kitchens) can struggle to provide safe food options for people with food allergies, due to how ingredients are sourced.

Do foods with PAL cause allergic reactions?

A number of studies have evaluated whether prepacked foods with or without PAL actually contain a priority allergen not listed in the ingredients. These data are summarised in Fig. 5. For peanut, the majority of prepacked foods with PAL do not contain detectable amounts of peanut, particularly for non-confectionery items.³⁷ This holds true for other allergens such as cow's milk, egg, hazelnut, or soya in those countries where data are available, although there are some important exceptions, particularly with chocolate products where UAP to cow's milk is common.³⁸ These data are not entirely surprising. UAP can occur in 2 forms: particulate and non-particulate contamination. Non-particulate allergens are ingredients such as powders or liquids, which are usually distributed relatively uniformly.³⁹ In contrast, particulate contamination (see Fig. 6) is infrequent and results in non-uniform distribution in the final food product. This poses significant difficulties with risk assessment, and results in the common scenario where most products with PAL do not have UAP — because the risk and distribution of particulate contamination is sporadic.

These same surveys have also shown that sometimes, prepacked foods without PAL can contain undeclared allergen.40, 41, 42, 43, 44 For example, Pele et al reported an analysis of 544 cookie biscuits and chocolates sourced from 10 European countries: peanut was detected in 108/333 (32%) of prepacked foods with PAL, but also in 52/211 (25%) of products without PAL.⁴⁴ Hazelnut was detected in 76% of chocolates with PAL, and 50% of chocolates without PAL. These data show that PAL does not always indicate actual UAP, but nor does absence of PAL imply no or even low risk of contamination. Furthermore, the wording of the PAL statements bear no relationship with the risk posed and can mislead consumers. These studies have highlighted that products with PAL phrases which may be interpreted as conveying a lower risk of UAP (eg, “made in an environment … ”) (see Table 2) frequently UAP of a magnitude similar or even greater than those labelled “may contain … ”.^43,44

Remington et al. undertook an analysis of how often foods labelled with PAL could cause a reaction in allergic consumers, using data from the United Kingdom. While the vast majority of foods did not contain UAP, those that did would have caused a reaction in >1% of the allergic population, and for some foods (chocolate), the risk could have been up to 50%.⁴⁵

However, few studies have investigated how common accidental reactions are in real life, due to UAP. Sheth et al reported 651 food-allergic patients in a Canadian registry who experienced an allergic reaction due to inadvertent allergen exposure; 37% attributed their reaction to a failure to read or heed PAL, and one-third of these thought their reaction was due to UAP not declared with PAL.⁴⁶ In Australia, 6.7% (58/864) of responders to a survey run through a patient support organisation reported anaphylaxis due to undisclosed allergen in prepacked foods, although PAL to the allergen was present in 81% of cases.⁴⁷ A retrospective survey of 100 children in Switzerland who regularly consumed foods with PAL found that 18% reported a previous reaction to a food with PAL (the most common products were chocolates, cookies and cakes). Reported reactions were all mild (skin and/or gastrointestinal symptoms).⁴⁸

In a prospective study in 157 adults with physician-confirmed food allergy in the Netherlands, 153 reactions were reported by 73 patients;⁴⁹ 62 (41%) of these were due to prepacked foods; in one-third of cases, the suspected allergen(s) were not mentioned either as an ingredient or on PAL. Allergen analysis was conducted in 51 samples (67% were prepacked foods): UAP was detected in 19, of which 10 (53%) did not have a corresponding PAL statement.⁵⁰

Irrespective, the surveys also indicate that aside from “risk anomalies” (for example, the production of dark chocolate on the same line as milk chocolate results in a relatively high frequency of UAP to milk³⁸), the amounts of potential UAP would not be expected to cause anything other than mild allergic symptoms in the vast majority of food-allergic individuals.^51,52 In support of this observation, low-level milk contamination in foods (corresponding to an ingestion of 0.5–2 mg of milk protein) did not cause reactions in a cohort of 29 milk-allergic children with a history of prior anaphylaxis.⁵³ Similarly, there are very few reports in the literature of potentially life-threatening reactions due to UAP,⁵⁴ despite evidence suggesting that many or even most food-allergic individuals eat foods with PAL (see Table 2). There are 2 reports from the United States where it was suggested that consumption of foods with PAL caused a fatal reaction: 1 over 20 years ago following consumption of cake, and another around 10 years ago in a peanut-allergic teenager who consumed a snack bar (of dried fruits, nuts and seeds) which did not contain peanut as an ingredient but did have a PAL to peanut.⁵⁵ There are no reports of fatal reaction due to UAP in a prepacked food with PAL in the UK Fatal Anaphylaxis Register which has collected data for over 25 years, although there has been at least 1 fatality due to consumption of a prepacked-for-direct-sale vegan wrap which contained undeclared cow's milk but no PAL.

How do food-allergic consumers, and those advising them, interpret PAL?

Given the proliferation of products with PAL, it is not surprising that surveys report that allergic consumers (or those purchasing food for them) often ignore PAL (see Table 2). There is a common theme: the public often assumes that the wording of the PAL statement conveys useful information as to the magnitude of potential UAP, that an item with a "may contain" PAL poses a greater risk than other PAL statements that may be perceived as more neutral (such as “manufactured in a facility which also processes X"). Consumers rely on instinct to make decisions whether or not to consume a product with PAL.⁵⁶

Clinicians may not discuss how to interpret PAL with patients, nor is there a consensus amongst as to what advice to provide. Turner et al surveyed 239 healthcare professionals in the United Kingdom providing dietary advice to allergic consumers or their caregivers: 38% recommended complete avoidance of foods with PAL to nuts (but no nut listed in the ingredients), while 22% advised no avoidance was necessary, with the remainder taking a permissive approach to PAL depending on the patient's health and access to rescue medication.⁵⁷ Clearly, there is variability in the advice healthcare professionals provide with respect to PAL, ranging from strict allergen avoidance to ignoring PAL altogether (Fig. 7).⁵⁸

Both patients/families and clinicians are confused as to whether PAL are required by law. Marchisotto et al surveyed 6684 participants in North America (22% with a food allergy themselves, 84% caregivers): 46% were either unsure or incorrectly believed that PAL is required by law, and 37% thought the wording used for PAL reflected how much UAP might be present.⁵⁹ In a survey of healthcare professionals in the United Kingdom and Australia, 32% wrongly believed that PAL use was subject to a standardised risk assessment and 13% believed that PAL was regulated by law.⁶⁰ Only 50% considered that PAL statements such as “may contain X″ or “X may be present” indicated a real risk of UAP. There was a strong consensus for the introduction of a single and simple PAL statement, something also reported by allergists in Canada.⁶¹ Some regulators now recommend specific phrases: both Health Canada⁶² and the UK Food Standards Agency⁶³ recommend the simple/unequivocal phrase “May contain X” without further qualification.

Although official guidance is welcome, our conclusion from 2013 remains valid today: that “the current situation [with respect to PAL] does not benefit either the allergic consumer or food manufacturers, who are potentially liable for an allergic reaction resulting from cross-contamination”.¹

Attempts to rationalize the use of PAL

Madsen et al reviewed different approaches to risk assessment in food allergy.⁶⁴ Risk assessment processes applied to UAP involves 4 distinct steps: identification of the hazard, characterisation of the hazard, characterisation of the risk, and then assessment of risk mitigation measures. The difference between a hazard and a risk is shown in Fig. 8. Risk considers both the hazard, and the likelihood of exposure to that hazard causing an adverse outcome. This can be assessed in a qualitative manner, or a quantitative risk assessment applied (Fig. 8).

Quantitative or qualitative RAs are structured science-based processes that estimate the probability and severity of illness from consuming food containing biological, chemical or physical contaminants. They also identify where in the production chain controls and interventions will have the greatest impact on reducing risk. With respect to UAP, a qualitative RA essentially involves answering a YES/NO as to whether there is a risk of any UAP, while quantitative RA provides a more pragmatic assessment of the degree of risk posed by UAP to allergic individuals, to guide whether a PAL is needed (see Fig. 9).

Remaining obstacles to a regulatory approach for PAL

Arguably, there is now a global consensus that the status quo for PAL is no longer sustainable. Currently, application of PAL is widespread, inconsistent and causes confusion. A “zero-risk” approach might be initially attractive to allergic consumers, but in practice would result in even more products with PAL and therefore of little benefit (indeed, many would see this as a significant step backwards, particularly given that there has been a consensus for over a decade that “as with other risks in society, zero risk for food-allergic people is not a realistic or attainable option.”⁹² Therefore, the way forward has to be a regulated framework for application of PAL, which utilises a cut-off or reference dose below which PAL would not be required (Fig. 17).

The FAO/WHO Expert consultation has provided much “food for thought” in an evidence-based format, to underpin the global regulation of PAL. The reports generated by the consultation are currently being considered by CCFL as well as by industry stakeholders, and potential implementation will need buy-in from all stakeholders. Notwithstanding, at least 1 European Regulator (in the Netherlands) has already announced they will adopt the approach proposed by the FAO/WHO Expert group as an interim measure, with immediate effect.⁹³ However, some outstanding issues need to be considered.

• i)

  Is a system based on ED₀₅acceptable to all stakeholders, and what measures might be needed to help more sensitive allergic consumers who react to ED₀₅exposures to avoid reactions?

ED₀₅ levels of exposure will still result in objective symptoms in 5% of the allergic population, and up to one-third may experience (typically mild) subjective symptoms. These sub-populations can be identified through single-dose challenges,^83,90 and diagnostic blood tests may soon be able to achieve the same result, at least for peanut.⁹⁴ There are 3 key considerations.

• •

  Could individuals who might react to ED₀₅-levels of exposure be readily identified in current healthcare systems, in a way that does not exacerbate existing disparities in healthcare due to socioeconomic factors?

• •

  How best to protect the most sensitive 5% of the allergic population who are at risk of objective symptoms of UAP ≤ ED₀₅, as they would not know if a food product without PAL poses a risk to them or not.

• •

  When allergic individuals begin to experience mild symptoms following an allergen exposure, they do not know whether their only symptoms will be mild and transient oropharyngeal itch (likely with very small exposures) or whether their reaction may progress to anaphylaxis.

One solution discussed, but not supported, at the FAO/WHO Expert Consultation was to apply a specific statement to indicate UAP ≤ RfD (Fig. 18). The rationale is that consumers would still be alerted as to the potential of risk, but the wording would communicate that consumption is “safe” for 95% of allergic consumers. If a consumer knew that they did not normally react to ED₀₅ levels of exposure but developed oral itch, they could be reassured that their reaction would be unlikely to progress. Ultimately, this option was not recommended, primarily because it would not lead to less use of PAL for negligible UAP which was a desired advantage for a risk-based PAL approach, ie. there is a risk that many food businesses might default to a “low risk” PAL. However, the Expert Group acknowledged that further work and stakeholder engagement is needed to understand what strategies would be acceptable to allergic consumers.

At risk is the possibility that a small proportion of food products might have no PAL but cause (typically mild) objective reactions in 5% of the most sensitive allergic individuals, and mild subjective symptoms in up to 30% — as this could undermine consumer confidence in this approach. This risk – and potential solutions — needs to be considered by all stakeholders, including food businesses who may be concerned as to potential adverse impact on brand reputation were individuals to react to food products lacking PAL.

• ii)

  Should UAP only ever be communicated through PAL?

In some countries, UAP may be communicated through minor ingredients labelling or in certain countries, if UAP exceeds an action level (e.g. Japan). The inclusion of allergens in ingredients lists which may or may not be present due to UAP can cause confusion, reduces consumer trust in the value of ingredients lists, and may encourage risk-taking (if consumers ignore allergens listed as ingredients).

• iii)

  What wording or symbol should be applied for PAL?

The FAO/WHO Expert Group recommended the use of a single standardised statement which would convey that a product with PAL poses a health risk to individuals with an allergy to that particular food, and is thus not suitable for them. DunnGalvin et al. surveyed 1560 European adults with FA or parents of food-allergic children, and found that the most popular PAL phrasing was “not suitable for” (46%), closely followed by “may contain X” (44%).⁹⁵ However, amongst members of the Expert Group, the wording “not suitable for” was not considered to accurately describe the risks from UAP present at the ED₀₅ level but would be tolerated by (and thus “suitable” for) up to 95% of allergic consumers.⁸⁷

A 2009 survey undertaken by the US FDA in 1243 individuals (of whom 739 made food purchases for food-allergic consumers) reported that while “may contain” was the most preferred wording, this choice was only expressed by one-third or respondents.⁹⁶ A more recent US survey of food-allergic consumers reported that 29% preferred “Not suitable for …” while 22% preferred “May contain”.⁹⁷ A recent survey from the United Kingdom also reported a preference for the simplest expression of PAL such as “May contain X”.⁹⁸ Use of the term “traces” may lead consumers to believe that a food poses less of a risk “as it is only a trace”; likewise, people who experience symptoms to foods labelled “may contain traces” may believe they are more at risk of severe reactions as they reacted to a “trace".⁹⁹ The Expert Group concluded that the term “trace” is usually ill-defined, and does not adequately convey that a risk assessment has found UAP which poses an appreciable risk. Therefore, the use of “traces” was not recommended. Whichever type of phrase or statement is chosen, it should effectively convey the message that the product may contain UAP at levels above appreciable risk, and should therefore not be consumed by allergic individuals. The precise wording of the single phrase for PAL should be decided by CCFL in conjunction with all relevant stakeholders and should consider local linguistic nuances.⁸⁷

The Expert Group recommended that the application of PAL should be regulated.⁸⁷ However, aware that this might not be adopted by CCFL, the Expert Group recommended that there should be some form of indication (such as a specific symbol) on packaging so that consumers would know that a formal risk assessment had been conducted to inform the need for PAL (something that would be unnecessary if all PAL were applied as part of a regulated framework).

• iv)

  What role would healthcare professionals have?

There is already a need for healthcare professionals to help allergic consumers how to interpret allergen labelling, including PAL. If a consistent and regulated approach were to be applied to PAL, then healthcare professionals would have a critical role in advising their patients, particularly those allergic individuals who might react to exposures below the RfD. Given the potential medicolegal concerns (which may be of greater relevance in some countries), there is a need for a consensus in terms of the advice these individuals should receive with respect to consuming food products with potential UAP ≤ RfD (and how to interpret a two-level PAL system (such as that proposed in Fig. 18), if such a scheme were implemented. Given the existing wide variation in advice given by healthcare professionals to patients,^57,60 consistent education of allergic consumers and other stakeholders will be critical.

Whilst the majority of allergic consumers (and those choosing food on their behalf) desire more consistent and meaningful PAL, many are reluctant to purchase foods that might contain even a small amount of UAP. The potential for UAP to occur but not be declared on packaging may not be acceptable for many allergic consumers, partly because it is contradictory to previous public health communications that exposure to even a very small amount of allergens can be potentially fatal.

• v)

  Should the potential frequency of UAP be a factor in the application of PAL?

An important consideration is the frequency of the cross-contact problem that might lead to UAP. This will depend on the type of food being produced, and the equipment involved. UAP can vary from highly sporadic to frequent (Fig. 19). In contrast, the frequency of UAP due to particulates (e.g. whole seeds, chunks of nut) is much lower, and can be difficult to evaluate because it is usually non-uniform and occurs sporadically (Fig. 6). The current proposals being considered by CCFL assume that PAL should be applied whenever there is a risk of UAP > RfD; it does not take into consideration whether application of PAL should also depend on the frequency at which UAP might occur.

If application of PAL becomes regulated, then it will be necessary for regulators to include guidelines as to how to conduct the risk assessment (and the use or not of sampling plans to verify the assessment together with how to deal with analytical uncertainties). These guidelines will need to be developed for each industrial sector (e.g. bakery products, meat processing, etc) in close collaboration with food businesses. Justification of PAL use should be based on evidence that cross-contact does occur, and can exceed the action level (such evidence can include visual inspection, weighing of ingredients, analysis of allergens, etc). While the principles of the risk assessment should be the same across industry sectors, recommendations for implementation may need to be specific and therefore not straightforward. Given that zero risk is not possible, there is potential uncertainty over the legal consequences of a regulated system where UAP might cause an allergic reaction, but the regulated framework did not require PAL.

• vi)

  How might “free-from” claims fit into any regulated framework?

“Vegan” and “Free from” allergen labelling presents a different challenge to PAL since “free” is often taken to mean completely absent, or not detectable, rather than below a numerical threshold. However, in most regions, from the legal perspective, these are considered “marketing” statements. “Vegan” and “Free-from” labels exist in a legal grey-zone: while products labelled in this way should not contain an ingredient that contradicts the label (and thus comply with marketing claim), the “free from” or “vegan” claim does not extend to UAP. Thus, in many countries, "free-from" foods may also have a PAL to the allergen the product is “free-from”, Notably, in Germany, this scenario is prohibited, which might contribute to the relative paucity of “free-from” foods in this country. “Vegan” claims only indicate that no animal-derived ingredients are used in the food. It is for this reason that vegan or “free-from” products often have PAL, including allergens (like egg or milk) that are derived from animals. Consumers are generally not aware that “free-from” and “vegan” are considered marketing statements, and their use is not informed by food-safety legislation.

However, consumers often rely on vegan or “free-from” labels to guide allergen-free food choices. Vegan foods are often considered to be “safe” by those purchasing products where there is a need to avoid animal proteins (meat, poultry, fish/seafood, mammalian milk and eggs). The term "vegan" may be used as a shortcut for consumers who are less confident in explaining their food allergies to staff. A Canadian survey of consumers allergic to egg or cow's milk reported that 86% would consider “vegan” products to be “safe” for them.¹⁰⁰ Surveys in both Canada and the United Kingdom have demonstrated that prepacked foods carrying vegan and/or “plant-based” labels can pose a significant risk, particularly to milk-allergic consumers.^100,101 At least 1 fatality has been reported in the United Kingdom to a vegan wrap which contained “vegan” yoghurt with detectable cow's milk protein present.¹⁰²

Further confusion can arise from the use of “gluten-free” claims: in many countries, these are regulated and subject to a maximum gluten presence of 20 ppm.¹⁰³ Products containing 20 ppm or less can be labelled as gluten-free. However, a wheat-allergic person consuming a 200 g food serving which contains 20 ppm gluten (and can therefore be labelled as “gluten-free”) would consume around 5.5 mg total wheat protein – an exposure to which 2–6% of wheat-allergic individuals would be expected to have an objective allergic reaction, and 10–15% would be expected to experience mild subjective symptoms.⁷⁶ Thus, “gluten-free” foods may not be appropriate for a proportion of wheat-allergic individuals who react to lower doses.

The fact that “gluten-free” claims have a regulated cut -off (and thus permit the presence of gluten≥20 ppm) should not be seen as setting a precedent for applying existing action levels to other allergens. At the inquest relating to the UK fatality due to “vegan” yoghurt in a wrap (referred to above), the manufacturer asserted that they could declare their product to be “milk-free” as it would not have had cow's milk protein present above an ED₀₅ level, analogous to “gluten-free” labelling. However, the Coroner rejected this argument, clearly stating that “foods labelled in this way must be free from that allergen, and there should be a robust system to confirm the absence of the relevant allergen in all ingredients and during production when making such a claim.”¹⁰² The FAO/WHO Expert Consultation upheld the position stated by VITAL®, that Reference Doses are “not appropriate, nor intended, to be used to define ‘allergen-free’ labelling”.⁸

Conclusion

We have summarised developments in the application of PAL, including the recent FAO/WHO Expert Consultation, for healthcare professionals. CCFL are currently reviewing these recommendations. It is essential for healthcare professionals to provide input into this process, given their key role in supporting consumers with food allergy. We hope this review will provide a foundation for discussion and help build a consensus on how PAL can be better used to support safe food choices for allergic consumers.

Abbreviations

Act-up, World Allergy Organization Consensus on the Use of PAL; CCFL, Codex Committee on Food Labelling (CCFL); ED, Eliciting Dose; ED01, Eliciting Dose in 1% of patients; ED05, Eliciting Dose in 5% of patients; ED10, Eliciting Dose in 10% of patients; FA, food allergy; FALCPA, Food Allergen Labelling and Consumer Protection Act; FAO, Food and Agriculture Organization; FARRP, Food Allergy Research & Resource Program; FASTER, Food Allergy Safety, Treatment, Education, and Research Act; FDA, Food & Drug Administration; FDEIA, Food-Dependent Exercise-Induced Anaphylaxis; FDNIA, Food-Dependent Non-steroidal anti-inflammatory drug-Induced Anaphylaxis; GA2LEN, Global Allergy and Asthma European Network; GMP, Good Manufacturing Practices; GSLPF, General Standard for the Labelling of Prepackaged Foods; HACCP, Hazard Analysis Critical Control Point; LOAEL, Lowest Observed Adverse Effect Level; OFC, Oral Food Challenges; PAL, Precautionary Allergen Labelling (PAL); RA, Risk Assessment; RfD, Reference Dose; UAP, Unintended Allergen Presence; VITAL®, Voluntary Incidental Trace Allergen Labelling; VSEP, VITAL® Scientific Expert Panel; WAO, World Allergy Organization; WHO, World Health Organization.

Funding

ACT-UP! is a project of the World Allergy Organization.

Author contribution

AF, SS, JOBH, PE, and SLV drafted the first version of the manuscript. PT deeply revised and harmonized it. Each author contribute to revise all sections of the paper.

Ethics approval

Not applicable.

Consent for publication

All authors approved the final version and its submission.

Informed consent

Not applicable.

Declaration of competing interest

Paul J Turner: Grants from UK Medical Research Council, UK Food Standards Agency, JM Charitable Foundation, NIHR/Imperial Biomedical Research Centre and End Allergies Together, outside the submitted work; personal fees from UK Food Standards Agency, DBV Technologies, Aimmune Therapeutics, Allergenis and ILSI Europe outside the submitted work.

Antonio Bognanni: No conflicts to disclose.

Stefania Arasi: Advisory board member, consultant, and/or speaker for Novartis, DBV, Ferrero, and Ulrich outside the submitted work.

Ignacio J Ansotegui: Advisory board member, consultant, and/or speaker for Bayer, Bial, Cipla, Eurodrug, Faes Farma, Gebro, Glenmark, Menarini, MSD, Roxall and Sanofi outside the submitted work.

Sabine Schnadt: Speaker honoraria and advisory panel consultancy outside the submitted work for Aimmune and DBV.

Sébastien La Vieille: Employment: Health Canada (Government of Canada) - no conflict of interest.

Jonathan O’B Hourihane: Research funding from Johnson& Johnson, DBV Technologies; Travel support Stallergenes; Speaker fees Nutricia; Consultancy Camallergy, Stallergenes; Board membership Clemens von Pirquet Foundation and Irish Food Allergy Network.

Torsten Zuberbier: Institutional funding for research and/or honoria for lectures and/or consulting from Amgen, AstraZeneca, AbbVie, ALK, Almirall, Astellas, Bayer Health Care, Bencard, Berlin Chemie, FAES, HAL, Henkel, Kryolan, Leti, L'Oreal, Meda, Menarini, Merck, MSD, Novartis, Pfizer, Sanofi, Stallergenes, Takeda, Teva and UCB, Uriach; in addition, he is a member of ARIA, DGAKI, ECARF, GA2LEN and WAO.

Philippe Eigenmann: Speaker and advisory board honoraria: DBV technologies, Novartis, ThermoFisher Scientific, Nestlé Health Sciences, Synlab, GSK,; Stocks and Stock options: DBV technologies.

Motohiro Ebisawa: No conflicts to disclose.

Mario Morais-Almeida: No conflicts to disclose.

Julie Barnett: Funded research from Food Standards Agency, UK. Member of Food Standards Agency Advisory Committee for Social Science.

Bryan Martin: No conflicts to disclose.

Linda Monaci: Speaker honoraria and funded research project by Ferrero outside the submitted work.

Graham Roberts: No conflicts to disclose.

Gary Wong: Advisory panel consultancy outside the submitted work for Haleon, Nestle, Novartis, OM Pharma, Ferrero.

Ruchi Gupta: Research support from the National Institutes of Health (NIH) (R21 ID # AI135705, R01 ID # AI130348, U01 ID # AI138907), Food Allergy Research & Education (FARE), Melchiorre Family Foundation, Sunshine Charitable Foundation, Novartis, and Genentech. Medical consultant/advisor for Genentech, Novartis, Food Allergy Research & Education (FARE), OWYN, Kaléo, Aquestive Therapeutics, and Byrn Pharma. Ownership interest in Yobee Care, Inc.

Sophia Tsabouri: No conflicts to disclose.

Clare Mills: No conflicts to disclose.

Simon Brooke-Taylor: Consults to and receives payment from the Allergen Bureau of Australia & New Zealand.

Consults on food regulation and compliance with Australian and New Zealand food standards to the food industry.

Joan Bartra: Speaker honoraria outside the submitted work for Thermo Fisher Scientific, Novartis, Menarini.

Michael Levin: Speaker honoraria and advisory panel consultancy outside the submitted work for Viatris, Novartis, Organon, Sanofi, Pfizer.

Marion Groetch: No commercial interests to disclose. Royalties from UpToDate and Academy of Nutrition and Dietetics and consulting fees from Food Allergy Research Education; serves on the Medical Advisory Board of IFPIES, as a Senior Advisor to FARE, as a Health Sciences Advisor for APFED.

Luciana Tanno: Speaker honoraria of Sanofi, DBV Technologies, Research grant from ANS (Agence Numerique de Santé), AllerGos.

Elham Hossny: No conflicts to disclose.

Barbara Ballmer Weber: Speaker honoraria and advisory panel consultancy outside the submitted work for Thermo Fisher Scientific, Novartis, ALK, Allergopharma, Menarini, Sanofi, MSD, Aiummune.

Vincenzo Fierro: Speaker honoraria for Stallergenes, Sanofi, GSK.

Benjamin C Remington: No conflicts to disclose.

Jennifer Gerdts: Food Allergy Canada receives unrestricted funding support for educational programming from Pfizer, DBV, Sanofi, American Peanut Council and consultant fees from Novartis.

M Hazel Gowland: Speaker honoraria: Food industry, technical and regulatory conferences, academic lecture fees Advisory panel consultancy: UK Food Standards Agency, Safefood, industry food technical and public health organisations.

Funded research: FSA, UKRI, Safefood funded studies, UK Fatal Anaphylaxis Registry.

Derek Chu: No conflicts to disclose.

Marjan Van Ravenhorst: Employee, Allergenen Consultancy BV.

Jennifer Koplin: No conflicts to disclose.

Alessandro Fiocchi: Speaker honoraria and advisory panel consultancy outside the submitted work for Nutricia, Abbott, Danone, Stallergenes, DBV, Novartis. Funded research (Institution) from Sanofi, Novartis, Ferrero, DBV, GSK, Astrazeneca, Hipp GmBDH, Humana SpA.