Wednesday, February 6, 2013

Can we establish safe allergen thresholds in foods? What does the current science say?

In light of the open request from the US Food and Drug Administration (FDA) seeking public feedback on whether food manufacturers can set a “safe” threshold for major food allergens (peanut, milk, egg, etc.), I wanted to reason through this question with the existing scientific evidence regarding food allergy.

Disclaimer:  Although I am a scientist and thus I try very hard to shut my emotional brain off to take a hard look at the data, I also happen to be the mother of child with life-threatening food allergies.  I may be biased.  That said, I will address this question to the best of my knowledge in light of what the scientific data can say and also where scientific data may be lacking.

In addition to the FDA query, FARE (previously Food Allergy and Anaphylaxis Network and Food Allergy Initiative) is requesting information from food allergy sufferers and their families in a short survey.  With this information and scientific/medical advice on thresholds from their medical advisors, FARE will formulate their response to the FDA’s request.  I guarantee that the FDA will listen to FARE.  (Here is my first ping to you, to at least take the FARE survey – 5 minutes, promise!).

Now that the nitty gritty details are out of the way, what is the business about the FDA setting thresholds for major food allergens, who are the players, and what does it mean for the food allergy community?

The players: 
1.       Consumers affected by food allergy/intolerances
2.       Food manufacturers
3.       FDA (the regulatory “man in the middle”)
4.       Consumers not affected by food allergy (the overwhelming majority)

It seems obvious from the FDA’s query that the food manufacturers and the FDA are currently in the midst of a risk assessment.   

Risk assessment is the determination of quantitative or qualitative value of risk related to a concrete situation and a recognized threat (also called hazard). Quantitative risk assessment requires calculations of two components of risk (R):, the magnitude of the potential loss (L), and the probability (p) that the loss will occur.

Whether you realize it or not, we the consumers affected by food allergy are doing our own qualitative risk assessment with the threshold question (a risk assessment associated with quality, not quantity/ numbers).  Those us who have experienced anaphylaxis directly or live in fear from the threat of anaphylaxis have been told from the instant of diagnosis to “avoid the allergen at all cost,” which includes “made on shared equipment” with the allergen.  “Made on shared equipment” involves trace amounts of allergen that can send a person into an anaphylactic/life-threatening allergic reaction.   Initially after my son’s diagnosis, I was naively skeptical (“but, it must be such a small amount.  No way it could happen!”) until my son had a strong allergic reaction (hives, throat constriction, vomiting) after eating a food with safe ingredients that was “made on shared equipment” with egg.

Parent risk assessment:  Risk = death, Probability of occurring = who cares?!  It's death!
With the possibility of death involved, our instant reaction as food allergy parents is “Threshold?  Why is this even a discussion?!"

Just to play devil’s advocate –
Let’s say that a safe threshold can in fact be determined (which it probably can with enough data, resources, and money).  The benefit would be enormous for many food allergy sufferers and their friends and family.  My heart breaks when my son can’t enjoy the foods that most of his friends do, and I would love every now and again, a worry-free break from making every darn meal/dessert known to man.  So, just how good are we humans at assessing risk qualitatively?  Quite bad, actually.  

“Our emotions push us to make snap judgments that once were sensible (my own words - from an evolutionary point of view)— but may not be anymore.  Fear also strengthens memory.  So catastrophes like plane crashes and terrorist attacks (or anaphylaxis – my own words) stay with us.  As a result, we overestimate the odds of dreadful but infrequent events and underestimate how risky ordinary events are. The problem is, this leads to bad decisions," the author writes. (Psychology Today and NYT)

Given that humans are terrible at qualitatively assessing risk, I ask you to entertain this threshold idea from a more scientific/quantitative perspective, regardless of what we, the food allergy affected may conjure up to explain the food industry’s real motivation behind a threshold.  Just be sure to take the bias hat off for a minute and dream of a world where a safe minimum threshold can and does exist.  It would revolution our lives!  

Currently, I do not support a minimum threshold of major food allergens based on current scientific data-

After taking a look at some of the current scientific data, I unfortunately cannot swallow setting a safe minimum threshold [in scientific jargon – some point between the Lowest Observed Adverse Effect Level (LOAEL) and No Observed Adverse Effect Level (NOAEL)].  In spite of this, I do believe thresholds may be quite useful in the future if scientific data on food allergy thresholds improve.  Here are my reasons –

Part of the problem with setting a threshold is that we need a lot of good quality, highly controlled food challenge studies known as the double-blind placebo-controlled food challenge in order to get the necessary data to determine a safe threshold.  Essentially, we need a large population of allergic individuals who are willing to submit themselves to increasing doses of their allergen up to the point of having concrete symptoms (something observable to a “blinded” test administrator – e.g. hives on skin, vomiting, etc.) because “subjective” symptoms (e.g. itching mouth, stomach cramps) cannot be verified by the “blinded” test administrator.

If this can successfully be accomplished, the data are put into a plot, and something called a dose-response curve is fit.  From this curve, we should in theory by able to determine a “minimum” dose.  In fact you can make two plots while you’re at it – one for observable reaction and one for subjective reaction.

 The curve for subjective reaction will show a lower threshold dose than the objective curve.  I realize that there were some critisisms floating around about why the FARE survey would ask questions such as, “would you be willing to eat “x” if you knew no adverse reaction would occur, but minor reaction would be acceptable (e.g. mouth tingling)?”  I realize the question seems strange, but I believe it is entirely based on the difference between the dose-response curves for observable vs. subjective reactions.

But, I digress…

Fundamentally, what the data of a dose-response curve can tell is that at __mg of peanut (whatever that hypothetical may be), we can say with 95% certainty (or whatever probability you choose to set) that at any one given time 2 people out of 100 will have a minor reaction.  This is a quantitative risk assessment in a nut shell (pun intended).  The greater the certainty desired in your threshold (e.g. 99% instead of 95%), the more test subjects you will need in your study.  Unfortunately,these studies are quite time-consuming and expensive.

So let's say we establish our threshold dose with a dose-response curve.  This seems fantastic except for a few major problems:
1.      Very few studies of the appropriate type with enough study participants for EACH individual allergen currently exists as of February 2013 –
a.       Many small studies have been done (e.g. 30 or even fewer subjects), yes, but their methods may differ drastically from one another - so much so that scientists would be comparing the proverbial “apples to oranges” by trying to combine the data from smaller studies.  A small study is not adequate by itself to get the level of confidence we need in our threshold to keep food allergic individuals safe.
b.      The study design must be a food challenge, preferably the double-blind placebo- controlled food challenge, because there is no test (e.g. skin prick test or blood serum IgE levels) that can predict with any accuracy at what level an allergic individual will respond (the response part of our dose-response curve). See publication for 5 different curves published just last month!  Can you believe it?!  This data is just coming out Jan 2013?!.  Have I mentioned that these studies are expensive?  Who is going to pay for them?  Moreover, how many allergy-sufferers are going to sign up for a test, knowing that the test unequivocally involves an observable allergic reaction (unless the person’s allergy has miraculously gone away, which would be awesome!).
c.       Thus, enrolling study participants is a major problem.  There is no way in hell I’m going to force my 4 year old child to consume an allergen (previously anaphylactic) to the point of an observable reaction just for the sake of “getting data.”  The only way I will food challenge him is if we think the clinical allergy is going away or we are enrolled in some kind of clinical trial for the treatment of food allergy.  Either one of these scenarios could seriously flaw interpretation of data if these are the only individuals willing to participate in a threshold study.  What’s more is that data may be less reliable at the lowest doses near the threshold.  It turns out that individuals who are likely to be the most sensitive (e.g. react at very small doses), tend to react the most severely.  It just may not be ethical for a person that sensitive to be included in a threshold study.  So if we don’t sample an entire population of allergic individuals, what does the threshold actually mean?  The data are flawed.
d.      Each allergen needs to be evaluated independently.  You cannot get minimum dose thresholds for peanut, which seems to have the most solid clinical data, and expect an egg allergy to behave the same.  Have I mentioned how few of these studies exist on a large scale and how expensive they are?!
2.      Current studies evaluate a reaction threshold in a population at ONE GIVEN TIME –
a.       This threshold would work really well, except for one glaring flaw, REACTION THRESHOLDS CHANGE IN A SINGLE PERSON FROM TRIAL TO TRIAL! 
                                                              i.      Let’s say that at one given trial, patient 22 and patient 89 (out of 100) react.  If you retest the same 100 patients, patient 22 and 89 do not react, but rather patient 3 and 44 react. 
                                                            ii.      Studies retesting the same individual from trial to trial are very much lacking in the literature– see recent publication, where a re-trial has many of the same individuals responding quite differently (many with over a 10x difference in threshold dose from trial one to trial two – note the non-linear scale - 10x difference in dose between 1-2 for example).

An external file that holds a picture, illustration, etc.
Object name is pone.0053465.g003.jpg Object name is pone.0053465.g003.jpg
Threshold doses of peanut in the same child at the two peanut challenges.
The amount peanut eaten were divided into 5 dose steps: Dose 1 = 0.001 g; Dose 2 = 0.01 g; Dose 3 = 0.1 g; Dose 4 = 1 g; Dose 5 = 3.6–5 g.  Source:

3.      A final consideration is how reliable are allergen detection methods that industry will use to measure the amount of allergen in a food?  How will it be measured?  Who will measure it?  How often will products be measured?  Allergenicity of a given protein can change depending on how the food is processed (e.g. cooked vs. uncooked). And on and on I could go with more questions.


Many questions remain.  Ultimately what is at stake is that all of this testing will get passed along to the consumer (non-allergic majority and allergic minority alike) in the form of increased prices on our grocery store shelves.  My question is – is it worth it?  Given all of the questions and problems (e.g. a major lack of solid data!) surrounding establishing thresholds for allergic individuals, it seems unlikely most allergic individuals and their families are going to trust a threshold any time soon. 
Instead of investing a ton of very limited research dollars to acquire the necessary data for thresholds that we could trust, I would much rather see those research dollars going toward reversing food allergies and preventing them from happening in the first place.

As to the food industry and the FDA, please be more transparent with labeling practices.  For example, what does “made in a facility that also processes ___” actually mean?  Made on the same equipment?  Made in the same building, but on separate lines?  Given that I have a child who has many allergies outside of the “top 8,” life would be so much easier if catch-all categories like “spices” and “natural flavors” did not exist.  Tell me what it actually is and how it is manufactured, and gasp… I just may purchase the item.  Everyone would win.


  1. Excellent post breaking down the many flaws in the proposal. My initial reaction to the concept of "thresholds" was that they were an arbitrary system set up to give cover to the food industry at the expense and endangerment of food allergic individuals. You know what? I think my first answer was correct.

    1. Thanks so much! It's hard to know what motivations are, so I personally try to avoid speculating :) The safe threshold would be awesome if there was more data to back it up. It just isn't there right now. In all honesty, I would have no problems with a threshold if the packaging actually states that a product tests below ... parts per million of allergens x,y, and z. With more knowledge about how individuals actually respond to allergens (yet to come from the scientific data), I could see this as a really powerful tool for doctors to give specific advice to their patients. Clearly they would advise really sensitive folks to stay clear, but perhaps it would be helpful to a lot of people who are not as sensitive. As long as the FDA uses thresholds to enforce the food industry to clearly label it on packaging, I'm completely ok with it. What is not ok in my opinion would be for food companies to use allergen testing to avoid putting any precautionary labels on packaging. It's all in how the threshold info gets used. It could be really helpful or be an absolute nightmare for all of us dealing with food allergy.

  2. I appreciate this post, and understand the reasoning completely. However, I'm concerned that if we ask for a 0% threshold, that I will not be able to find ANY groceries to buy for my family. I think that most companies will choose to say that they are unsafe, so that they don't have to worry about the extra cleaning/choosing suppliers, and I think that it might lead to a very difficult time buying even the staples - flour, sugar, rice, etc, much less any processed food.

    1. Thanks, Lori! I have had the exact same thought as you. See my response above. I really like the idea of thresholds if the FDA enforces the food industry to label it on the packaging. What I fear, and what I think you are getting at, is that they will use the threshold info to do away with any precautionary labeling whatsoever, making virtually all food untrustworthy. I'd really like to see an effort to say that this product tests below __ parts per million of allergen x, y, or z. This way doctors could hopefully advise patients on an individual basis how concerned they should be once there is more solid clinical data on specific allergens. Everyone would win.

  3. Do you know about the studies done in Europe regarding threshold? It is my understanding that the Europeans are more aggressive with conducting food challenges and establishing thresholds in patients so as to increase their quality of life. Meaning, of course, that certain patients would be able to eat foods labeled as "produced in a facility...." So obviously this wouldn't apply to everyone, but it would definitely set a benchmark for labeling.

    I was just thinking about what it would mean to be able to include such testing on an individual level, but that really means nothing, if the companies don't label...

    And I am definitely with you about labeling- "spices" means nothing to me, having a child with a sesame allergy......