The issue of adverse reactions to artificial food additives is not a new one. The search for this relationship stems from various parental reports of food additives responsible for their child’s various symptoms.
Investigation of the incidence of food additive intolerance is challenging because of the large number of additives involved and the need for extensive patient compliance. The large number of additives available also makes it difficult to associate specific additives with symptoms.
The gold standard for investigating the association between food additives and adverse reactions in children is a double blinded, randomised placebo-controlled challenge. Despite improved methodology of more recent studies, the overall relationship between these two variables has failed to be clearly established.
A variety of symptoms have been reported in children consuming artificial food additives in their diet. Table 1 shows the range of symptoms reported.
Table 1 Symptoms reported to food additives. Taken from (1)
Eczema Behavioural/mood changes
Urticaria/angioedema Musculo-skeletal symptoms
Wilson et al (2) performed a double blind assessment of additive intolerance in 29 children with a clear history of symptoms induced by artificial yellow colourings. The symptoms included cough and wheeze (14 children), behavioural disturbance (7 children), eczema (4 children), urticaria (2 childen) and abdominal pain, pallor and lethargy (2 children). Following a period of additive free diet, the children were challenged daily with drinks containing tartrazine and sunset yellow for 12 days. Out of the 19 children who completed the study, only three children were found to have exacerbation of symptoms, which proved a causal relationship with substances in the challenge drinks.
In 1997, Young et al (1) conducted a population study in Wycombe to find out the prevalence of food additive intolerance. A total of 30000 people were surveyed, of which 7.4% of those who responded claimed a reaction to food additives. Of the 649 were interviewed for participation in the study, 132 were selected to take part. The subjects were given a low and high dose challenge of additives or placebo (lactose) concealed in opaque capsules. Statistical analysis of the results provided an estimate of 0.01-0.23% prevalence of food additive intolerance in the Wycombe population.
Fuglsang et al (3) performed a similar study in Denmark but looked at children who were referred to paediatric allergy clinics for symptoms of urticaria, asthma, eczema or rhinitis, and found the incidence of intolerance of food additives to be 2% (6/335) on double-blinded challenge of additives.
The mechanism of food additive intolerance is not clearly defined. Supramaniam and Warner (4) had disproved the then-held view that altered prostaglandin production was responsible as a reaction can be induced in aspirin-sensitive patients, because aspirin intolerance was uncommon in their study.
Food additives and urticaria
Artificial food additives, particularly tartrazine and benzoates (1) have been shown to trigger urticarial reactions in children in many studies conducted over the last three decades. One of the first few reliable studies to emerge was by Supramaniam and Warner (4). They evaluated 43 children who presented with angioedema and/or utricaria and had responded to an additive free diet in a double-blinded study. These children were challenged with various food additives including tartrazine (E102), sunset yellow (E110), amaranth (E123), indigo carmaine (E132), carmoisine (E122), sodium benzoate and sodium metabisulphite. 24 of the 43 children were found to have reacted to 1 or more of the additives. The authors noted that a very small percentage of these children had a history of atopy; only 9.3% had asthma and 11.6% had postitive skin prick tests and came to the conclusion that food additive intolerance causing urticaria was not an IgE-mediated phenomenon.
Food additives and atopic eczema
No well-controlled studies have been performed to investigate the claims that food additives can induce atopic eczema until Van Bever et al (5) performed a double blind placebo controlled challenge in 25 children with severe atopic eczema. In all the children, a thorough history was obtained linking food intake with exacerbations of eczema. Furthermore, their eczema was poorly controlled despite the use of topical moisturisers and steroids. All the children were admitted to hospital and were fed an elemental diet via a nasogastric tube (NG tube) and 1-2 weeks after this treatment, an improvement in the childrens’ skin was apparent, such that they were almost free from active eczema lesions. They were then given a placebo or the food additives (tartrazine, sodium benzoate, sodium glutamate, sodium metabisulphite, acetylsalicylic acid and tyramine) via the NG tube. The study (5) found that all children challenged with food additives showed positive reactions within 10 minutes after administration and the reactions consisted of pruritus and redness of the skin.
It is not clear whether food additives worsen atopic eczema by inducing erythema and urticaria or whether they exert a direct effect (5). Although the study managed to show adverse reactions to food additives, it failed to describe or measure the severity of the reactions (6) or pick up late reactions as the observation period lasted only four hours (6).
Devlin and David (6) attempted to address these drawbacks in a study of 13 patients with severe eczema (requiring regular attendance at outpatient clinics). The subjects were randomly assigned a regimen of three placebo and three active weeks where they were given capsules containing either tartrazine (50mg) or glucose placebo (50mg) to be dissolved in orange juice and drunk using a straw through an opaque cup. The severity of their eczema was assessed using a chart to record the percentage surface area affected. The results of the study failed to find evidence of a clear relationship between tartrazine and eczema as only one patient out of the 12 who completed the study had a significant correspondence between symptom or disease severity score with tartrazine challenges, and this relationship could have occurred by chance.
The above two studies have failed to demonstrate a significant link between food additives and eczema, although both had evaluated mainly tartrazine, so reactions to other types of food colouring may not have been tested. Furthermore, the mechanism in which food additives trigger exacerbation of eczema is now well known.
Food additives and behaviour
The earliest report of an association between food additives and behavioural problems such as hyperactivity was in 1973 when Dr Benjamin Feingold, Chief Emeritus of the Department of Allergy at the Kaiser-Permanente Foundation Hospital in San Fransisco claimed that much of the hyperactivity and learning difficulties seen in school-aged children was due to the ingestion of foods containing naturally occurring salicylates and artificial colourings and flavourings (7). Feingold devised a diet free from these foods and named it the “Kaiser-Permanente” (KP) diet (8) (see Table 1).
Over 5 separate programs, Feingold managed 260 children whose primary complaint was behaviour disturbance with the KP diet and using the Conners rating scale, discovered that there was marked behavioural change within 3-21 days in 30-50% of the children (8). The studies reached no statistical conclusion but proved that the KP diet influenced behaviour. Feingold also noted that an individual child’s behavioural deficit varies in terms of type and duration and the child’s age influences the speed and degree of response; in early infancy the response may take 24-48 hours, 2-5 year olds, more than 5 days, 5-12 year olds, 10-14 days and in older, post-pubertal adolescents, several months (8).
Feingold’s proposal attracted widespread attention from the media and public but was criticized by many in the medical community because his studies lacked a structured diagnosis for the subjects, control groups, an objective measure of outcome and was not double blinded.
Table 1. The Kaiser Permanente Diet. Taken from (8)
Avoid all artificial colours and flavours contained in foods, medications andcosmetics
Avoid preservatives BHA and BHT (butylated hydroxyanisole and butylated hydroxytoluene)
Avoid the following foods containing natural salicylates:
AlmondsCurrantsPlums, prunes Cloves
Apples Grapes, raisins TangerinesCoffee
Apricots Nectarines Cucumber, pickles Teas
Berries OrangesGreen peppers Oil of wintergreen
Over the subsequent period of 35 years, many studies have been conducted to evaluate Feingold’s hypothesis and these were performed in children who were diagnosed with hyperactivity, ADHD or other behaviour problems. One of the earliest studies to be reported was by Conners et al. (9). The study looked at 15 hyperkinetic children using a double blind crossover design. These children were randomly allocated to 4 weeks of the KP diet followed by 4 weeks of a control diet or they were assigned to the control diet followed by the KP diet. Relative to a 4-week baseline period, parents and teachers were asked to rate the children based on a standardised rating scale of ADHD symptoms. It was found that from teacher ratings, the KP diet was significantly more effective than the control diet with approximately 15% reduction in symptoms (p<0.005) but not on parent ratings. However, when compared with the baseline period, both parents and teachers reported fewer hyperkinetic symptoms- 2.53% reduction in symptoms (parent) and 2.55% (teacher), on the KP diet (p<0.05).
Although this study had an improved methodology compared to Feingold’s, it was limited by a small sample size (n=15), inconsistent results, uncertain control of information and expectation held by parents and findings that the behavioural effect of the diets were related to the order in which they were administered.
Gross et al (10) performed a study of 39 children ranging in age from 11 to 17 with learning problems attending a private summer camp, 18 who had been diagnosed with Attention Deficit Hyperactivity Disorder (ADHD), amongst which, 17 were taking stimulant medication. All children were given the KP diet for 1 week, and then allowed to eat a typical American diet rich in cookies, cakes, candy, soft drinks and snacks from home during the second week. Each week, the childrens’ behaviour was monitored by videotape placed in the dining hall at 4-minute intervals by three blinded observers (one which included the study author) for motor restlessness, disorganized behaviour and misbehaviour. The authors found no difference in the behaviour of the children while on both diets and also commented that the children disliked the KP diet.
Despite being blinded and the investigators having complete control over the children’s diets, there were several weaknesses in the study. Firstly, there was no specification of the components of the diet rich in food additives. Secondly, the children were aware for the first week that they were eating a different diet (11). Thirdly, all the subjects except for one were taking stimulant medication (10), which arguably might have influenced the outcome of the study. In addition, the sensitivity to changes in children’s behaviour of the outcome measure (coding of videotapes) had not been evaluated (11).
Based on these initial investigations, there has not been a clear and consistent association between the KP diet and behaviour in children with symptoms of hyperactivity. Only a small proportion (11-13%) of hyperactive children respond to the KP diet such that there is an improvement in their functioning at home and in school (11).
Double blind placebo controlled crossover studies
Various studies have been done to investigate the effects of artificial food colourings (AFC) on hyperactive behaviour in children and adolescents by following a methodological pattern of a baseline diet free of AFCs, followed by a double blind, placebo controlled crossover challenge of AFCs (11). However, these studies vary in length, sample size, outcome measure and AFC challenges, which varied in the amounts and selected dyes. Most studies used a mixture of AFCs as a challenge-most commonly allura red, erythrosine, brilliant blue, indogotine, tartrazine and sunset yellow, and a similar amount of AFCs (26 mg) (11).
Swanson and Kinsbourne (12) performed a short-term trial with 20 hyperactive and 20 non-hyperactive children in a hospital setting. The children were given a diet free of food dyes, artificial flavours and preservatives for 5 days- 3 days of baseline and 2 days of placebo-controlled challenge of 100 to 150 mg of AFCs. The high doses of AFCs were given because the authors concluded from an earlier study (8) that a high dose of 100mg produced significant effect (p<0.001) on the children’s ability to perform a laboratory learning task compared to the 26mg dose. They found that the performances of the hyperactive children on learning tasks were significantly impaired (p<0.05) after the AFC challenge but the performances of the non-hyperactive children were not affected.
This study used an objective measure of outcome (laboratory learning test) and compared the effects of AFCs in hyperactive and non-hyperactive children, as compared to previous studies that only looked into children with diagnosis of hyperactivity or children with suspected behavioural problems associated with AFCs. However, the need to use a high dose of AFC (100mg) to provoke behavioural reactions in hyperactive children suggests that average doses of AFC found in the daily diets of these children might not do so.
Pollock and Warner (13) performed a 7-week double blind AFC crossover challenge, also using a high dose of AFCs (150mg) on 19 children between the ages of 2-8 years (mean 8.9), whose parents had observed that various behavioural problems in these children had improved on a diet free of food additives. The children studied were all normal except for one who had idiopathic global retardation and another who had been diagnosed with hyperkinesis. The food colours used in the challenge were tartrazine (E102) 50mg, sunset yellow (E110) 25mg, carmoisine (E122) 25 mg and amaranth (E123) 25mg, as they were often the blamed food additives causing adverse reactions. These were given in opaque capsules daily for two weeks while placebo capsules given the remaining five weeks, in random sequence. Parents were asked to complete a daily questionnaire of the child’s behaviour and somatic symptoms throughout the seven weeks. Results of the study showed that parents reported more behavioural problems (p<0.01) on the AFC challenge compared to placebo. However, only 2 children demonstrated clinical hyperactivity on their Conners’ score.
They (13) also suggested that food additives given in large doses act as a pharmacological trigger in a small percentage of children with behaviour problems, although their study showed that the effect was small.
Rowe and Rowe (14) investigated the effect of 6 doses of tartrazine (dose range 1-50mg) in 34 hyperactive children and 20 non-hyperactive children in a double blind placebo controlled study for 6 weeks. The parents of the children were asked to complete two rating scales (a behaviour rating inventory devised by the authors and Conners 10 item Abbreviated Parent-Teacher Questionnaire). In total, 24 children (22 hyperactive and 2 comparison children) reacted to the tartrazine challenge. These children demonstrated consistent variations in behaviour for at least 5 of the 6 challenges. The study also found that pre-schooled and school aged children responded differently to the AFC challenge; severe sleep disturbance was the main complaint in younger children (aged 2-6 years), while older (aged 7-14 years) children exhibited negative mood, impulsivity and whining.
This study was able to address the drawbacks in an earlier study by David (15) where tartrazine could not be disguised in capsules due to its bright and early recognizable colour, thus could not be performed in the home environment. In Rowe’s study (14), the capsules were colourless and tartrazine was planted in an inner capsule surrounded by the placebo (lactose).
Based on the above studies (13,14), it can be concluded that in non-hyperactive subjects, there exists a relationship between food additives and behavior but to a much smaller extent compared to hyperactive subjects.
In an attempt to address the problem with generalization of findings from previous studies limited by small samples, dependant on a diagnosis of hyperactivity or in children thought to show adverse behaviour triggered by food additives (11,13,14), Bateman et al (16) devised a population based study to test whether food additives have a pharmacological effect on behaviour. They looked at 277 children aged 3 years, registered with general practitioners in the Isle of Wight, who were given 20 mg in total of AFCs (sunset yellow, tartrazine, carmoisine and ponceau 4R; 5mg of each) and 45 mg of sodium benzoate during the second and fourth week of the 4-week study. As an objective measure of outcome, research psychologists using validated tests assessed the children’s behaviour weekly. In addition, parents were asked to rate changes in their child’s behaviour. The study found that parental ratings showed a significantly greater increase in the hyperactive behaviour during the active period (p<0.007).
The authors suggested that the reason parental ratings have a higher sensitivity to changes in behaviour is because parents experience their child’s behaviour over more prolonged periods of time and in varied environmental settings (16).
The most recent population study by Mc Cann et al (17) was performed on 2 groups of schoolchildren: 137 preschoolers (age 3) and 130 school-aged children (ages 8 and 9) from the general population. Each group was challenged with sodium benzoate combined with 2 different mixtures of dyes. Mixture A had the same content as the Bateman et al study while mixture B contained sunset yellow, carmosine, quinolone yellow and allura red. The doses of dyes were different in the two mixtures and also according to age group- mixture A contained 20mg (preschool) and 24.98mg (school-aged) while mixture B contained 30mg (preschool) and 62.4mg (school-aged). It was found that both age groups had significantly increased Global Hyperactivity Aggregrate scores when challenged with one or both dyes compared with placebo. The younger children significantly reacted to mixture A (p=0.044) but not mixture B while the older children reacted significantly to both mixture A (p=0.023) and mixture B (p<0.001).
Despite much-generated interest from parents and the public over the effect of food additives to children’s behaviour, evidence for this association is generally weak and as described above, some findings can be conflicting. However, recent population studies have managed to show a significant association between food additives and childhood behaviour, particularly in the older age group (8-9 years). Table 2 summarises the studies discussed above.
Table 2: Double blind placebo controlled studies on effect of artificial food colouring on behaviour in children
Studies Number ofRestricted Diet AFCs Medium Amount of Outcome Challenge
Subjects AFC (mg) Measure Effect
Swanson 20KP dietMixCapsules 100-150 Learningp<0.05
Pollock 19 Additive freeMixCapsules 125 PRS p<0.01
Rowe 34 Additive freeTartra-Capsules 1-50 PRS p<0.001
and Rowe (10) zine
Bateman 277 Mix Drinks20 PRS p<0.007
et al. (13)
Mc Cann 267 Mix Drinks GHAA-p=0.044
et al (14) A- 20(preschool) (preschool)
– 24.98(school-aged) B-p=0.023
-62.4 (school-aged) B-p<0.001)
AFC:Artificial food colouring, PRS: Parental rating scale , GHA: Global Hyperactivity Aggregrate, A:Mixture A, B:Mixture B
1) Young E. Prevalence of intolerance to food additives. Environmental Toxicology and Pharmacology 1997;4:111-114
2) Wilson N, Scott A. A double blind assessment of additive intolerance in children using a 12 day challenge period at home. Clinical and Experimental Allerfy 1989;19L267-272
3) Fuglsand G, Madsen C, Halken S, Jorgensen M, Ostergaard OA, Osterballe O. Adverse reactions to food additives in children with atopic symptoms. Allergy 1994;49:31-37
4) Supramaniam G, Warner JO. Artificial food additive intolerance in patients with angioedema and urticaria. Lancet 1986;2;907-909
5) Van Bever HP, Docx M, Stevens WJ. Food and food additives in severe atopic dermatitis. Allergy 1989;44:588-594
6) Devlin J, David TJ. Tartrazine in atopic eczema. Archives of disease in childhood 1992;67:709-711
7) Feingold BF. Adverse reactions to food additives. Paper presented at American Medical Association annual meeting 1973, Chicago, IL.
8) Feingold BF. Hyperkinesis and learning disabilities linked to artificial food flavours and colours. American Journal of Nursing. 1975;75:797-803
9) Conners CK, Goyette CH, Southwick DA, Lees JM, Andrulonis PA. Food additives and hyperkinesis: a controlled double blind experiment. Pediatrics 1976;58:154-166
10) Gross MD, Tofanelli RA, Butzirus SM, Snodgrass EW. The effect of diets rich in and free from additives on the behaviour of children with hyperkinetic and learning disorders. Journal of American Academic Child and Adolescent Psychiatry. 1987;26:53-55
11) Stevens LJ, Kuczek T, Burgess JR, Hurt E, Arnold LE. Dietary sensitivities and ADHD symptoms: Thirty five years of research. Clinical Pediatrics 2011;50(4):279-293
12) Swanson JM, Kinsbourne M. Food dyes impair performance of hyperactive children on a laboratory learning test. Science 1980;207:1485-1487
13) Pollock I, Warner JO. Effect of artificial food colours on childhood behaviour. Archives of Diseases in Childhood 1990;65:74-77
14) Rowe KS, Rowe KJ. Synthetic food coloring and behavior: a dose response effect in a double blind, placebo controlled, repeated measures study. Journal of Pediatrics. 1994;125:691-698
15) David TJ. Reactions to dietary tartrazine. Archives of Disease in Childhood 1987;62:119-122
16) Bateman B, Warner JO, Hutchinson E, et al. The effects of a double blind, placebo controlled, artificial food colourings and benzoate preservative challenge on hyperactivity in a general population sample of preschool children. Archives of Disease of Childhood 2004;89:506-511
17) McCann D, Barrett A, Cooper A, et al.Food additives and hyperactive behaviour in 3 year-old and 8/9 year-old children in the community: a randomised, double blinded placebo controlled trial. Lancet 2007;370:1560-1567