Chronic and acute effects of artificial colours and preservatives on children's behaviour

Last updated:
15 January 2015
This research project examines the impact of artificial food colourings and preservatives on children's behaviour.
Study duration: September 2004 to March 2007
Project code: T07040
Contractor: University of Southampton

Sarah Hardy
Tel: 020 7276 8532


Attention deficit hyperactivity disorder (ADHD) is one of the most common neurological conditions found in children. There has been a longstanding suggestion that some artificial food colourings and preservatives (AFCPs) may affect children's behaviour, particularly in relation to ADHD. This project will build upon a previous study commissioned by the Government that examined the effects of AFCPs on the behaviour of a cohort of three-year old children. Although that study found that AFCPs may affect children’s behaviour, the results were considered inconclusive as the effects were only observed by parents and were not confirmed independently by clinicians. A further limitation of the earlier study design was the observed 'placebo effect'. A large proportion of children in the first study were unaffected (or, in some cases, showed an improvement in behaviour) when on the additives mix, but showed worsening when on the placebo. The present study addresses these limitations and includes further experimental design improvements by taking into account guidance from the Food Additives and Behaviour in Children (FABIC) working group and the Committee on the Toxicity of Chemicals in Food, Consumer Products and the Environment (COT).

Research Approach

The project was divided into two stages, each involving double blind placebo controlled food challenges (DBPCFC). Stage one investigated the chronic effects of AFCPs on behaviour, acute affects were investigated in stage two.

Stage one: chronic effects of AFCPs
Stage one involved a sample of 120 three-year old children and 120 eight/nine-year olds selected to represent the full range of ADHD symptoms. Two AFCP mixtures were used. Mix A was identical to that used in the previous study (i.e. 20mg in total of sunset yellow, tartrazine, carmoisine and ponceau 4R in equal amounts plus 45mg of sodium benzoate). A second mixture that is more typical of the AFCPs currently found in children’s foods (mix B) was also developed. Mix B contained quinoline yellow, sunset yellow, carmoisine, allura red and sodium benzoate in amounts representative of the average daily intake of these additives by young children in the UK. With the exception of sodium benzoate, these were the most frequently found additives in recent Agency surveys of children’s sweets and soft drinks, which together are responsible for a major part of the intake of food additives in the diets of children. Sodium benzoate was included as this additive is also widely used in soft drinks currently sold to children, frequently in combination with some of the above colours. Also, the inclusion of sodium benzoate made Mix B more comparable to Mix A. It is important to note that neither of the challenge drinks equates to any drinks that might be commercially available, as they will be used as vehicles via which to administer the total daily dose of additives. The active and placebo drinks were provided as daily doses, delivered at irregular intervals and presented in identical, sealed bottles and with the same appearance. The drink used as the basis for both test drinks and the placebo was a mixture of natural fruit juices.

The design of the DBPCFC used in stage 1 of this project is represented in Figure 1. The challenge followed the format of that previously used with pre-school children but with the addition of a second active challenge mix (mix B) and the consumption of the placebo drink during the initial withdrawal period and the washout periods i.e. the child received a drink of some kind throughout. This latter modification removed any unblinded placebo effects influencing parental reports during the withdrawal period. Over a six-week period the children were assessed at baseline (T1). From here onwards the children received continuously for the next 6 weeks either the placebo or the test drink. They were then assessed at T2 before a period of either active (mix A or mix B) or placebo challenge (W2), at T3 before a wash out period of one week (W3), at T4 before the second placebo or active challenge period (W4), at 5 before the second washout period (W5), at T6 and T7 before and after the final placebo or active challenge period (W6). During the challenge periods, each child received either mix A, mix B or placebo drinks for one week. All children consumed all three drinks over the duration of the study. Throughout the study, other sources of AFCPs were removed from the children's diet and the parents were asked to keep a diary of dietary violations during the course of the DBPCFC.

Figure 1: Design of the double blind placebo controlled food challenge
Figure 1: Design of the double blind placebo controlled food challenge

Assessment was carried out in the children’s pre-school or school and included a daily observation by a research assistant using validated methods. A standardised measure of attention was also used to assess the older group. In addition, ratings of the children’s behaviour were made by parents and by teachers throughout the six-week study period. The parents and all members of the study team were blind, apart from the study administrator.

Several factors that may moderate responses to AFCPs were investigated. Firstly, the effect of the children’s pre-trial level of ADHD behaviour was investigated by stratifying the group on the basis of initial ADHD status (measured at T1). The inclusion of the older children also allowed any age-specific effects of AFCPs to be explored. Furthermore, cheek cells provided by all of the children involved in stage one were genotyped in order to determine whether allelic variation in genes implicated in ADHD (e.g. DAT1, DRD4) influenced the effects of AFCPs.

Stage two: acute effect of AFCPs
Stage two investigated the acute effects of AFCPs in responders and non-responders selected from the older group in stage one. Each child was assessed during two half-day sessions a week apart. At each visit the child was given either a placebo or active drink similar to those used in stage one. However, the amount of AFCPs in mix A and mix B was equivalent to the total daily dose used in stage one but administered in one ingestion. Behaviour and psychological factors were assessed under carefully controlled laboratory conditions at multiple timepoints over a three hour period. Metabolic factors that could mediate a response to AFCPs was investigated by assaying salivary samples for levels of tryptase and by testing urine samples for histamine.

Overall, the questions to be addressed by the project were as follows:

  • Do children respond to the withdrawal and introduction of AFCPs in their diets?
  • Is any effect found with pre-school children also found in older children?
  • Is any response seen in teacher ratings, direct observations of behaviour and test performance as well as in parent ratings?
  • Is any response to AFCPs related to initial levels of ADHD i.e. is the response greater in children at the extreme end of the continuum?
  • Do genetic differences moderate any effect?
  • What are the metabolic and psychological mediators of any effect?


The study recruited over 130 children in each of the two age groups (3 year olds and 8 to 9 year olds) from schools and nurseries and pre-schools in the Southampton area. From the families who initially volunteered to take part in the study, children were recruited to represent the full range of behaviours from normal through to high level hyperactivity.

The results showed that in the sample of 3 year old children Mix A showed a significant adverse effect on the average level of hyperactivity within this group, compared with the placebo, but Mix B did not. For the 8-year old children, Mix B but not Mix A showed a significant adverse effect on the average level of hyperactivity compared with the placebo.

There were marked individual differences between children in their response to the artificial food colour and preservative mixtures with some responding substantially and others not at all. The researchers reported that one factor influencing this variable response to additives was the child’s genetic make-up. For both 3 year old and 8 year old children the researchers found that some specific types of genetic make-up, specifically to do with variations in certain genes involved in the release of histamine from cells, were associated with greater responses to either or both of the test mixtures.

The researchers consider that the results from this study add to the previous work on the Isle of Wight. The study demonstrates that using an aggregated measure of hyperactivity based on parent ratings, teacher ratings, direct observation of behaviour and, in the case of older children, direct testing of attention, certain mixtures of additives based on artificial colours and benzoate preservative, had adverse effects on the hyperactive behaviour of some children. The significant effects that were observed in this new study were shown in terms of an increase in the mean level of hyperactivity for the group. Not all children show more hyperactive behaviour on the active mix than on the placebo challenge. Importantly this work is based on a representative sample of children selected from the general population; the results are not limited to an extreme group of children showing Attention Deficit Hyperactivity Disorder.

The results of this study were presented to the Agency’s independent advisory committee, the Committee on Toxicity, for their consideration of the work, during March to August 2007. A statement of their views has been published on the COT website. See link below.

Published Papers

  1. Stephenson, J., Sonuga-Barke, E., McCann D., Grimshaw, K., Parker, K.M., Rose-Zerilli, M.J., Holloway, J.W. and Warner, J.O. (2010) American Journal of Psychiatry 167 1108-1115. DOI: 10.1176/appi.ajp.21010.09101529