This article was co-authored by Bert Popping1, Carmen Diaz-Amigo1 and Richard Fielder2
1FOCOS ; 2BioCheck UK
How safe is safe? How clean is clean? And whether to label: it contains or ‘may contain’? These are the questions that are being hotly debated at present. Why? Because three European member states committees have published food allergen references doses, which not only do not match, but they actually differ substantially, in some cases as much as 100 times. The reason for the differences is that the reference doses are based on different levels of protection (based on the proportion of reactions to an eliciting dose in a sensitive population) and correspond to either the ED01 or the lower limit of the 95% confidence interval of the ED05. Furthermore, our understanding of these ED values is being challenged by more recent clinical studies (Hourihane et al., 2017).
Back in 2014, the German enforcement authorities published reference doses (Waiblinger and Schulze, 2018), derived from VITAL 2.0, and, together with action values as well. These action values are concentrations above which enforcement authorities would assume the label is incorrect and will take action. This can be either to determine if an ingredient has been forgotten from the label, or, if it may be a cross-contamination.
The values appear acceptable to many food manufacturers, as well as allergic consumers. For the food industry, food allergen control requires the implementation of many different measures and factory- and product-dependent strategies. For example, they include ingredient segregation, proper identification of rework products as well as rework scheduling, production workflow, equipment design and sanitation procedures. Cleaning procedures are a critical point in the control of adventitious contamination of products with food allergens. In food production, ensuring production lines are 100% clean is a major challenge. Equipment assembled in production lines may include parts that favor the accumulation of residues or inaccessible sections, posing a challenge for an effective cleaning. In a zero-tolerance enforcement environment, implementing stricter sanitation procedures and additional allergen control measures will result in a major burden for the food industry, with limited additional protection for allergic consumers. Therefore, acceptable allergen tolerance levels are crucial. They should be achievable for the industry with proper control measures, and, at the same time, should be low enough to prevent allergic consumers from experiencing severe and life-threatening reactions when exposed to low amounts allergens. Zero tolerance has not been demanded by advocacy groups of allergic consumers since this would restrict even further the choices of products they can safely purchase. For this purpose, the VITAL concept of risk management and recommendations for precautionary allergen labeling (PAL) was developed. German control authorities base their action values on this concept, except where VITAL 2.0 reference doses are so low that no current routine analytical method would be able to detect at those levels. One example for this scenario is egg, which has a very low reference dose of 0.03 mg protein. Here, the German authorities adopted the Limit of Detection of available egg detection systems (e.g. ELISA). At this point, it is important to mention that VITAL does not cover all regulated in Europe, and validated reference materials are lacking for most of them, impacting in turn on the accuracy of analytical results.
In 2016, a Dutch scientific committee chose not to adopt the same VITAL reference doses as Germany, instead, they set significantly lower reference doses for the allergens. How these can be practically monitored and enforced with current analytical tests is not discussed in their statement.
Most recently, the Belgian scientific committee of the Federal Agency set their own reference doses, which differ from the German as well as from the Dutch. In particular, reference doses for peanut, egg, and milk are about 10 times higher as those set by Germany (see Table 1), and about 100 times as those set by the Dutch committee.
Table 1. Action limits proposed for peanut, milk and egg in Germany, Belgium and The Netherlands.
Figure 1 shows an overview over all references doses proposed by the committees of Germany, Belgium and The Netherlands.
Figure 1. Proposed food allergen reference doses in Germany, Belgium and The Netherlands.
Clearly, EFSA has not recommended Europe-wide reference doses and is unlikely to do so while the ThrAll project is still running (until about 2022), with the often-heard statement that no sufficient data to base a recommendation upon are available at the time. Because of lack of progress on this issue at EFSA level, individual countries are taking the lead, each with a different direction. This is generating a difficult and confusing environment for industry and consumers in the EU. So, the question is if over the coming years we will see different reference doses set by each European member states, of which at present, there are still 28.
What does this mean for the food manufacturer and ‘may contain’ labeling? Would the same product be labeled ‘may contain’ in the Dutch language? Since Belgian products also require labeling in Dutch (flemish) language, would there be a Dutch label for Dutch citizens with ‘may contain’ and a Dutch label for Belgian citizens without? Are affected consumers in The Netherlands safer than those in Germany or are the affected Dutch consumer just restricted even more for choice without good reason?
What is clear is that all who have a vested interest in this area cannot wait for another four years or more to come up with harmonized reference doses. This situation poses increased complexity for the food industry especially with products on the market in several different EU countries which will lead to greater potential for forced errors and arguably product recalls, and it is definitely confusing and unhelpful for affected consumers who must decide every day what is an acceptable level of risk.
