homeContact us

In this section:

Risk assessment

The purpose of an allergen risk assessment is to determine the risks due to unintentional presence of allergens. Based on the result of the risk assessment you can decide whether or not allergen advisory labelling is appropriate.

Important factors to cover in the risk assessment are:

  • Which allergenic foods or ingredients that unintentionally could get into contact with your food products
  • The amount of the allergenic food generally needed to provoke a reaction in allergic people
  • How common adverse reactions are to a particular food
  • Whether subgroups of the population have an increased risk
  • The relative allergenicity of your ingredient and whether processing changes the allergenicity. In general there is no effective way to process out allergens.But clearly if no or very little protein is present as could be the case for e.g. refined nut oils, the risk of an allergic reaction is very low.
  • The physical form of the ingredient e.g. whether air-borne cross-contact is a possibility
  • How cross-contact could happen and how likely it is to happen

In practice it is difficult to assess the risk with our current knowledge. Recent research has concluded that there exist lower limits under which food allergens will not cause any symptoms (called a threshold). Researchers use data from the diagnostic method double blind placebo controlled food challenge (DBPCFC) to obtain knowledge on thresholds for individuals. However, current available data do not permit estimation of population thresholds with adequate certainty. The amount of food allergen that may trigger allergic reactions visible for others range from a tenth of milligram in rare cases up to grams, and sometimes tens of grams, with considerable variability between individuals as well as between allergens. The data also indicate that, at the level of the allergic population, the lower the dose of food, the less severe symptoms, and the lower the proportion of allergic people with symptoms. However, in diagnostic challenges with milk, egg, peanut, soy and wheat patients who experienced more severe reactions tended to react to a lower dose than those experiencing milder reactions. A review of diagnostic DBPCFCs conducted since the 1970s found that the majority of food-allergic individuals tested needed to eat more than 500 mg of the offending food to provoke allergic reactions, but a significant minority responded to lower doses.

To date, the most reliable data on individual threshold levels come from studies performed in peanut allergic patients. Sufficient DBPCFC data do likely exist for cows’ milk, eggs and peanuts to be able to make a statistical evaluation to predict a dose of those foods that it will be safe for the majority of allergic individuals to eat. However, it is only recently that scientists and regulatory authorities have started discussions on what is the best way to make statistical evaluations of the individual data. So it still remains a matter of debate how to use the individual threshold data to predict the amount of food safe to eat for most allergic patients. The Australian Food and Grocery Council is to our knowledge the first organisation to develop and recommend the use of an allergen risk assessment tool to harmonise the application of allergen precautionary labelling.

Even though the risk is difficult to assess you can decrease it if you decrease how likely it is for allergen cross-contact to happen on your premises. That is one of the major objectives of allergen risk management.

More information


  • EuroPrevall state of the art paper: R. W .R. Crevel, B. K. Ballmer-Weber, T. Holzhauser,  J. O'B. Hourihane, A. C. Knulst, A. R. Mackie, F. Timmermans, S. L. Taylor (2008). Thresholds for food allergens and their value to different stakeholders. Allergy 63, 597-609.
  • For detailed information (e.g. data from DBPCFC studies) on foods, which may cause allergic reactions see "The InformAll Database"

Australian risk assessment tool:

British guideline on allergen management: