The prevalence of food intolerance

How common is pharmacological food intolerance? Several million people are currently suffering from this problem in the West, and more – perhaps 5-10% – will have experienced it in their lifetime. You might see figures out there, however, that vary widely, from less than a percent to 20% and upwards, but this isn’t so surprising considering that it has also been difficult to get good prevalence estimates for food allergies (1,2). In this article we will sort through the available data and see why 5-10% is a fair estimate for the lifetime prevalence of food intolerance.

What is prevalence?

Prevalence tells us how many people are suffering from a disorder at a single point in time, given as a percentage of the total population. Since it is impossible to ask every person whether they have a certain condition, researchers work instead with a sample of people who will represent the entire population. The size of this sample will dictate the reliability of the results. The best studies also verify a person’s response by having them go through a medical examination or some other diagnostic procedure.

What data is out there on food intolerance?

For conditions like food intolerance, where the prevalence is expected to be only a few percent (or less than a percent), the most reliable data will come from studies where the sample size is in the hundreds (or thousands). From among these studies, we only want to consider the ones that use a double blind placebo-controlled food challenge (DBPCFC) to check whether a person really suffers from food intolerance. It turns out that just a handful of investigations meet our requirements (see Table 1), and four out of five of these focus only on the effects of food additives like artificial colors and sodium benzoate.

Table 1. Large-scale prevalence studies for food intolerance
Country Study type Population Prevalence
Denmark (4)
N = 4274
All children 1-2%
Denmark (5)
N = 379
Atopic children 2-7%
Denmark (6)
N = 1094
22 year olds 0.5%
UK (7)
N = 18,582
Adults 0.01%-0.23%
Germany (8)
N = 4093
& salicylate
Adults 0.78-1.1%

Overall, it appears that food additive intolerance occurs in a few percent of children and a fraction of a percent of adults; when more foods are included, as in the German study, the prevalence increases to about one percent. The studies given in Table 1 also verify what has been observed elsewhere, namely that:

  1. Food intolerance is more common in women than in men.
  2. Children are more likely to suffer from food additive intolerance than adults.
  3. Atopic individuals (people with allergies and often asthma) are more likely to see adverse reactions to food, which often exacerbate flare-ups of their existing conditions.

Unfortunately, the data that we have falls short in several ways, most notably in scope. As mentioned, the studies in Table 1 focused on food additives, not naturally-occurring trigger chemicals in food. Even in terms of additives, each study differed in the number and amounts of additives that were tested, and common preservatives, such as sulphites, were absent in many cases. Madsen (3) pointed out that these studies also differed in their inclusion criteria, that is, the symptoms that people could have and still participate in a study. Furthermore, many symptoms, including migraine, gastrointestinal symptoms, and hyperactivity in children, were not considered. Thus, the reported prevalence rates are slightly lower than what should be expected for pharmacological food intolerance when it is more completely defined.

Towards a more inclusive prevalence rate

Is there any other data that we can use to improve our estimate? Numerous smaller studies have been done on food intolerance in specific groups – like sufferers of chronic skin reactions, respiratory symptoms, and gastrointestinal problems – and it is tempting to extend the numbers found there to the rest of the population. Unfortunately, we cannot rely on studies that were not originally designed to establish prevalence rates because the results may be biased. For example, specialized clinics are likely to have reputations that attract specific types of patients, and these may or may not be representative of all patients with those conditions. This is called spectrum bias, a well-known cause of variation in clinical studies.

Alternatively, we could look at prevalence estimates for sensitivities to individual food chemicals, or at the co-occurrence of food intolerance and certain illnesses, and add it all up. For example, histamine intolerance, defined as two or more non-allergic symptoms that improve through antihistamine use and a low histamine diet, is thought to have a prevalence rate of at least 1% and perhaps higher (9). In the US alone, about 12% of the population suffers from migraines, and food triggers are involved in about half the cases (10). Indirect evidence (11) suggests that irritable bowel syndrome is aggravated by foods that contain or cause the release of histamine in the body – this could affect as much as 5-10% of the population, but this has not been clinically verified. The picture is even less clear for other food triggers and conditions, and we are missing information on sensitivities to naturally-derived salicylate and benzoate, other biogenic amines, etc.

Looking at the numbers that we do have, it seems reasonable, if not conservative, that the overall prevalence of food intolerance would be at least several percent. This consistent with the lifetime prevalence estimate of 5-10% from Loblay and Swain of the RPAH Allergy Clinic in Australia (12); they based this estimate on their experience with thousands of patients, although experience does not carry the same scientific weight as epidemiological studies. Still, we can double check if these numbers make sense by considering lifetime prevalence and self-reported prevalence rates.

Lifetime prevalence

When scientists give prevalence rates, they are actually talking about point prevalence, which is how I defined prevalence at the beginning of this article: the fraction of people suffering from a disorder at a single point in time. Lifetime prevalence is the fraction of people who will suffer from a disorder at some point in their lifetime. For life-long illnesses, the point prevalence and the lifetime prevalence will be equal; for disorders that last for relatively brief periods, the lifetime prevalence can be several or even many tens of times higher than the point prevalence, depending on exactly what it is that predisposes a person to developing a disorder. We can expect that the lifetime prevalence of food intolerance will be higher than the point prevalence values we have been discussing, but we don’t know by how much.

If we have some idea of how long food intolerance lasts, we can put some limits on how many times higher lifetime prevalence could be. Food intolerance is thought of as a transient and individualized problem (13, 14). Different people will experience food intolerance for different lengths of time, and, after undergoing a period of avoiding their personal triggers, will have varying success in reintroducing these food chemicals to their diet. There are limited follow-up reports on individuals who have undergone food intolerance testing. Swain (15) found that 81% of respondents to a follow-up questionnaire still experienced a recurrence of symptoms as much as five years after having identified their trigger chemicals. We can make some assumptions and use a probability calculation to estimate that for this group, food intolerance lasts for 14 years – this means that the lifetime prevalence for adults would be about three times higher than the point prevalence. (16) This keeps us in line with our lifetime prevalence estimate of 5-10%.

Self-reported rates

Doctors and researchers often note that people perceive more adverse reactions to foods than can be verified through double-blind placebo-controlled food challenges. Self-reported prevalence rates from surveys of the general population are as much as 10 times higher than the rates determined through controlled testing – this goes for all types of food reactions, even allergies. For example, in the German study included in Table 1, which also looked at food allergy, the self-reported lifetime prevalence for all food hypersensitivities was 34.9%, while the point prevalence was found to be 3.6% based on food challenges. (8) Surveys conducted in different countries give different self-reported rates (17), which can be explained by different diets, different genetic make-ups, and cultural differences.

The discrepancy between clinical and self-reported rates causes some medical professionals to be suspicious about the existence of food intolerance. But does this discrepancy really mean that food intolerance is, for the most part, ‘all in the mind?’ Let’s consider some possible reasons for the higher self-reported rates:

  1. Depending on the survey questions, the self-reported rates could be representing the lifetime prevalence instead of the point prevalence – we need to be careful not to compare apples and oranges.
  2. The respondents could simply be wrong about what caused their symptoms. Most of the self-reported rates were recorded before the current media focus on food allergies and food intolerance, but, today, heightened awareness is increasing the perception that food is the source of symptoms that are otherwise unexplained.
  3. Food aversion is a real phenomenon – you begin to associate a food with an illness and then any exposure to that food makes you feel ill – and this could be mistaken for food intolerance.
  4. Some people may experience a food sensitivity reaction once or twice in their lives, never to be repeated. Large-scale diagnostic studies like the ones listed in Table 1 do have procedures for screening out such events (for example, see reference 7) and only inviting individuals who have a history of reproducible food-related symptoms to participate in the food challenge phase. Studies that present only self-reported rates may or may not use such procedures, so we need to look carefully at each figure to see exactly what it includes.

Another possibility is that there is something about how food intolerance is verified in clinical studies that leads to an underestimation of the true prevalence. Next we will look at the difficulties involved in diagnosing food intolerance to see if this has any effect on point prevalence.

Diagnostic challenges

While it is possible to diagnose food intolerance by elimination diets alone and by open food challenges, only the double-blind placebo-controlled food challenge (DBPCFC) is viewed by researchers as providing a definitive result. This makes it the ‘gold standard’ among test procedures, but, unlike many other gold standard tests, the accuracy of the DBPCFC is not known. (18) In other words, since the process has not been standardized, the accuracy will vary from study to study depending on how the test was conducted.

The DBPCFC was developed with food allergies in mind, where a single food is expected to produce a fairly rapid reaction. These reactions are often easy to see – an asthmatic patient experiences a decrease in breathing ability or a patient develops hives. However, even with food allergies, the results are not always clear cut. (19-21) Take chronic eczema that has not cleared before the test – how much of a change in skin irritation is needed for a positive result? How long should you wait for a response? How do you grade subjective symptoms like headache? How many times should you repeat the test if the patient responds to the placebo? Should you stop medications during the test? Etc.

The diagnosis of food intolerance shares these uncertainties and more, and it can be difficult to reproduce positive food challenge results even in people with good evidence of food intolerance. (7) Several things come into play here:

  1. It is not uncommon for someone to need to be exposed to a combination of food chemicals before they see symptoms. For example, Di Lorenzo et al. (22) found that few patients with chronic urticaria (hives) reacted to challenges with a single food additive but that many reacted to mixtures. Something like this could result from a true sensitivity to each of the triggers (12, 15) (even ones that are not chemically related) or from the unique effects that mixtures have on the processes that rid the body of foreign chemical substances.
  2. Food intolerance reactions are dose-dependent, so a person will only see symptoms if the accumulated amount of food chemical in their body exceeds their threshold dose. Each person has a different threshold dose and this threshold can change over time. Researchers have trouble addressing these issues in food intolerance studies. For example, high challenge doses do not necessarily guarantee more positive reactions and, in any case, might not be representative of normal consumption. Incremental challenges might seem to mimic real life, but they could also lead a person to become more tolerant of a food chemical.
  3. Similarly, some sufferers will be more sensitive to food triggers when they are also under stress, suffering from an illness, or exposed to certain chemicals (cleaning chemicals, fragrances, etc.) in their environment. (23) These additional factors may not be present at the time of the food challenge. Hormonal changes can also play a role in food sensitivity, but I have yet to see a study account for menstrual cycle, even though more women suffer from food intolerance than men.
  4. Patients with chronic or allergic conditions need to be experiencing symptoms at the time of the food challenge in order for any worsening effect to be observed.
  5. Researchers could simply be testing for sensitivity to the wrong chemicals. While salicylate or histamine intolerance is most often associated with fruit, German researchers reported that patients who failed to react to those substances still reacted to tomato extract, pointing to other unknown trigger chemicals. (24, 25). In another example, Lessof (26) pointed out years ago that a reaction to pesticide residue on unwashed fruit can also mimic food intolerance.
  6. Trying to prove an association between food intolerance and a specific condition is like a shot in the dark if the condition that you are looking at is not uniform. As an example, consider the association between attention deficit hyperactivity disorder (ADHD) in children and food additives. For decades, scientists and the public have gone back and forth over whether behavior in these children is influenced by food chemicals. In 2010, one group of researchers found that food additives affected ADHD-like symptoms in children who possessed a gene that decreased their ability to break down histamine. (27) This result has yet to be replicated, but, if verified, it would go a long way to explaining why different studies had been finding different results – only a subgroup of children with ADHD (those with a certain gene) are sensitive to food chemicals. If a study did not happen to include any of these children, then it would have found no connection between behavior and food.

Right now we do not know the extent to which these diagnostic challenges might impact clinically-determined prevalence rates for food intolerance, but we can say that there is more uncertainty in these rates than is currently understood. From the examples that I have given, you might conclude that point prevalences could only go up if researchers addressed these issues; however, it all depends on how you define food intolerance and its impact on health. Increasing the specificity of the DBPCFC would certainly benefit individuals with hard-to-diagnose food sensitivities; at the same time, large-scale prevalence studies might also find more people who experience occasional reactions that do not lead to significant impairment. In the latter case, better diagnostics would not give us any more useful information on food intolerance as a public health problem.

Since self-reported prevalence rates vary more between countries than clinically-verified rates do (see ref 16 for examples), the inflation seen in self-reported rates is probably related more to cultural effects than to limitations of the DBPCFC. Based on self-reported rates, the real lifetime prevalence of food intolerance should never be more than 20-30%, which still supports our estimate of 5-10%. If we knew more about the mechanisms behind food intolerance, we could set upper limits on the prevalence rates for sensitivities to individual trigger chemicals; however, this will not necessarily be an easy task, because for each chemical sensitivity, there are still likely to be different mechanisms at work in different people.


A lifetime prevalence estimate of 5-10%, as given by Loblay and Swain, is probably a good, and perhaps conservative, estimate of the true prevalence of food intolerance. Better prevalence rates will not be available until researchers standardize the double blind placebo-controlled food challenge, define food intolerance consistently in terms of trigger chemicals and recognized symptoms, and make progress uncovering the mechanisms behind food intolerance.

Last updated March 29, 2015

© 2014 Anna (Laurie) Laforest. All rights reserved.
Photo © Can Stock Photo Inc. – Food intolerance resource with a scientific twist


1. Nwaru BI, Hickstein L, Panesar SS, Muraro A, Werfel T, Cardona V, et al. The epidemiology of food allergy in Europe: a systematic review and meta-analysis. Allergy. 2014 Jan;69(1):62–75. PubMed

2. Kuznar W. Food Allergies May Not Be as Common as Reported. American Journal of Nursing. 2010 Aug;110(8):18.

3. Madsen C. Prevalence of Food Additive intolerance. Human & Experimental Toxicology. 1994 Jan 1;13(6):393–9. PubMed

4. Fuglsang G, Madsen C, Saval P, Østerballe O. Prevalence of intolerance to food additives among Danish school children. Pediatric Allergy and Immunology. 1993;4(3):123–9. PubMed

5. Fuglsang G, Madsen C, Halken S, Jørgensen M, Østergaard PA, Østerballe O. Adverse reactions to food additives in children with atopic symptoms. Allergy. 1994;49(1):31–7. PubMed

6. Osterballe M, Mortz CG, Hansen TK, Andersen KE, Bindslev-Jensen C. The Prevalence of food hypersensitivity in young adults. Pediatric Allergy and Immunology. 2009;20(7):686–92. PubMed

7. Young E, Patel S, Stoneham M, Rona R, Wilkinson JD. The prevalence of reaction to food additives in a survey population. J R Coll Physicians Lond. 1987 Oct;21(4):241–7. PubMed

8. Zuberbier T, Edenharter G, Worm M, Ehlers I, Reimann S, Hantke T, et al. Prevalence of adverse reactions to food in Germany–a population study. Allergy. 2004;59(3):338–45. PubMed

9. Maintz L, Novak N. Histamine and histamine intolerance. Am J Clin Nutr. 2007 May 1;85(5):1185–96. PubMed

10. Headache: Hope Through Research: National Institute of Neurological Disorders and Stroke (NINDS) [Internet]. [cited 2013 Dec 1]. Available from: (Archived by WebCite® at

11. Smolinska S, Jutel M, Crameri R, O’Mahony L. Histamine and gut mucosal immune regulation. Allergy. 2013 Nov 29; PubMed

12. Loblay R, Swain AR. Food Intolerance. In: Wahlqvist ML, Truswell AS, editors. Recent Advances in Clinical Nutrition. London: Libbey; 1986. p. 169–77.

13. Pollock I, Warner JO. A follow-up study of childhood food additive intolerance. J R Coll Physicians Lond. 1987 Oct;21(4):248–50. PubMed

14. Hayder H, Mueller U, Bartholomaeus A. Review of Intolerance Reactions to Food and Food Additives. International Food Risk Analysis Journal. 2011;1.

15. Swain A. The role of natural salicylates in food intolerance [Internet] [PhD Dissertation]. University of Sydney; 1988.

16. Assumptions: The disappearance of symptoms follows a fixed probability and can be modelled by an exponential distribution; 81% still experience symptoms after 3 years. The rate of decline would be 0.07 events/year, which would lead to an expectation of 1/0.07 ≈ 14 years/event. For adults, assume a lifetime period of 40 years: 40/14 ≈ 3.

17. Skypala I. Other Causes of Food Hypersensitivity. In: Skypala I, Venter C, editors. Food Hypersensitivity: Diagnosing and Managing Food Allergies and Intolerance. John Wiley & Sons; 2009. p. 210–42.

18. Gellerstedt M, Bengtsson U, Niggemann B. Methodological issues in the diagnostic work-up of food allergy: a real challenge. Journal of Investigational Allergology and Clinical Immunology. 2007;17(6):350. PubMed

19. Niggemann B, Beyer K. Pitfalls in double-blind, placebo-controlled oral food challenges. Allergy. 2007 Jun 15;62(7):729–32. PubMed

20. Bindslev-Jensen C. Standardization of double-blind, placebo-controlled food challenges. Allergy. 2001;56(s67):75–7. PubMed

21. Niggemann B. When is an oral food challenge positive? Allergy. 2010 Jan;65(1):2–6. PubMed

22. Di Lorenzo G, Pacor ML, Mansueto P, Martinelli N, Esposito-Pellitteri M, Lo Bianco C, et al. Food-additive-induced urticaria: a survey of 838 patients with recurrent chronic idiopathic urticaria. Int Arch Allergy Immunol. 2005 Nov;138(3):235–42. PubMed

23. Allen DH, Van Nunen S, Loblay R, Clarke L, Swain A. Adverse reactions to foods. Med J Aust. 1984 Sep 1;141(5 Suppl):S37–42. PubMed

24. Henz BM, Zuberbier T. Most chronic urticaria is food-dependent, and not idiopathic. Exp Dermatol. 1998 Aug;7(4):139–42. PubMed

25. Zuberbier T, Pfrommer C, Specht K, Vieths S, Bastl-Borrmann R, Worm M, et al. Aromatic components of food as novel eliciting factors of pseudoallergic reactions in chronic urticaria. J Allergy Clin Immunol. 2002 Feb;109(2):343–8. PubMed

26. Lessof MH. Food intolerance. Proc Nutr Soc. 1985 Feb;44(1):121–5. PubMed

27. Stevenson J, Sonuga-Barke E, McCann D, Grimshaw K, Parker KM, Rose-Zerilli MJ, et al. The role of histamine degradation gene polymorphisms in moderating the effects of food additives on children’s ADHD symptoms. Am J Psychiatry. 2010 Sep;167(9):1108–15. PubMed