Attention Research Update

April 2005

"Helping parents, professionals and educators stay informed about new research on ADHD"

David Rabiner, Ph.D.  Senior Research Scientist, Duke University

Dietary Intervention for ADHD: A Meta-Analysis

An ongoing debate among ADHD researchers, healthcare professionals, and parents concerns the extent to which dietary factors contribute to children's ADHD symptoms.  On the one hand, researchers and physicians have generally minimized the importance of diet for most children with ADHD, while recognizing that it may be an important influence for some. Many parents, on the other hand, report that diet makes an important difference in their child's behavior, and often initiate dietary interventions in the absence of physician support for, or even awareness of, this practice.

The possibility that diet exerts a significant influence on ADHD symptoms was proposed over 3 decades ago by Dr. Ben Feingold, a pediatrician who suggested that eliminating a variety of artificial food colors (AFCs), naturally occurring salicylates (salicylates are chemicals that occur naturally in many fruits and vegetables), artificial flavors, and particular preservatives could substantially reduce ADHD symptoms in many children. Controlled trials of the Feingold diet (FD) first appeared in the literature during the 1970s, and by now, a number of qualitative reviews and one statistical summary of the effects of this diet on ADHD symptoms have been published.  The qualitative reviews have reached mixed conclusions on the benefits of the Feingold Diet, and the debate between different authors has frequently been highly contentious. The only statistical summary of the different studies was published over 20 years ago and concluded that placing children on the FD had only a small and non-significant affect on ADHD symptoms.  (If you would like to learn more about the Feingold Diet, yu can do so at ).

A paper published in a recent issue of the Journal of Developmental and Behavioral Pediatrics provides an updated analysis on the role of AFCs in the behavior difficulties of children with ADHD (Schab & Trinh (2004). Do artificial food colors promote hyperactivity in children with hyperactive syndromes? A meta-analysis of double-blind placebo controlled trials. JDBP, 25, 423-434.)  The authors of this paper argue that an updated analysis of this issue is needed for several reasons, noting that the initial statistical summary was overly broad in the studies it included, and that this could have reduced the association between diet and ADHD symptoms that was reported.  In addition, several well-conducted studies have been published since that initial summary.

In the current review, the authors began by identifying all published studies that reported the results of placebo-controlled double blind trials on the relationship between the consumption of AFCs (artificial food colors) and behavioral change in children with ADHD.  A total of 15 trials that included a total of 219 participants were identified.

All trials were double-blind cross over trials which means that during the trial, children received two different diets, one that included AFCs and one that did not. Children's behavior was rated by parents, and/or teachers, and/or clinicians while on both diets.  In all cases, those rating children's behavior were unaware of which diet the child was receiving at the time.  This procedure is necessary to insure that any behavior differences found when children were on the different diets did not reflect the pre-existing biases of the raters, and is comparable to what is done in placebo-controlled trials of medication treatment for ADHD.

In a meta-analysis such as this one, rather than considering the results of each study separately, the researchers pool the findings across studies and conduct an analysis that that examines the overall pattern of results. The benefit of this approach is that it minimizes idiosyncratic findings of any individual study, and provides a more reliable basis for drawing conclusions about the overall pattern of findings in a particular research area.


The primary finding from the meta-analysis was that children's behavior showed a statistically significant improvement when AFCs were eliminated from their diet.  The size of the improvement was relatively modest- about a third to a half as large as the improvement typically associated with medication treatment for ADHD. 

The authors next examined whether eliminating AFCs resulted in greater behavioral improvement for children who were previously screened for responsiveness to AFCs.  This, in fact, was the case.  Children who had demonstrated responsiveness to AFCs through a non-blinded trial, or whose parents believed they were responsive to dietary factors, showed a more substantial improvement when AFCs were removed during the double-blind study.  This results suggests that parents are sensitive to whether their child's behavior is adversely affected by diet, and that if parents believe this to be true, eliminating AFCs from their child's diet is likely to result in behavioral improvement.

One important caveat to these findings is that the impact of eliminating AFCs on children's behavior was only evident in parents' ratings, and not in the ratings of teachers or clinicians.  Because parents, teachers, and clinicians were all unaware of which diet children were on when they completed their ratings, this difference cannot be attributed to preexisting biases on parents' part.  However, it does suggest that although removing AFCs may yield behavior improvements in children that parents are sensitive to, it is less likely to result in improvements that teachers observe.  This is quite different from what is generally found in studies of medication treatment or behavior therapy, where teachers generally report significant improvement in children's functioning.


Results from this meta-analysis provide strong evidence that the behavior of children with ADHD can be made worse by dietary factors, and that eliminating AFCs from their diets will, on average, result in behavioral improvements. This result is consistent with with accumulating evidence that neurobehavioral toxicity may result from a wide variety of distributed chemicals. 

The authors note that the mechanism by which AFCs may adversely affect children's behavior is not known, and suggest that it may occur because of allergic reactions, or because of actual pharmacological effects that AFCs induce.  They also suggest that results from the trials they analyzed may potentially underestimate the actual effect of AFCs because several trial employed doses that were well below children's true likely daily exposure.  In addition, because the impact of AFCs on behavior and learning may occur within several hours of ingestion, too much time may have elapsed between the administrati0n of AFCs and the measurement of outcomes to fully capture how participants were affected by ingesting AFCs.  Finally, although not discussed by the authors, the studies included in their meta-analysis only examined the impact of eliminating AFCs from children's diet, and it is possible that greater behavioral changes would result from eliminating other items, e.g., artificial flavors and preservatives, as well.

On the one hand, these results supports the belief held by many parents that their child's behavior is adversely affected by dietary factors.  As discussed above, this was particularly true for children whose parents held a prior belief that their child was sensitive to AFCs.  For such children, dietary interventions may thus play a meaningful role in helping to manage at least some of the behavioral difficulties associated with ADHD. What percentage of children with ADHD this subgroup represents is not discussed in the paper, however, and I am not aware of data on this important question.

Despite these basically positive findings, the authors also point out several important factors that qualify their results.  First, as noted above, ratings from parents - but not teachers - showed improvement when AFCs were eliminated from children's diets.  This is in marked contrast to what is typically found with medication treatment. Because improving children's behavior and learning at school is such a critical issue in ADHD treatment, this is a very important consideration.

Second, the degree of improvement parents reported when AFCs were eliminated were smaller than improvements typically associated with medication.  This was true, although to a less extent, even for those children who were screened for initial responsiveness to AFCs. 

Finally, it is important to note that several trials included in the meta-analysis employed ratings that were not specific to ADHD symptoms as the indicator of change.  Instead, outcomes were sometimes measured using reports of symptoms that are believed to be specific to AFC ingestion such as sleeplessness and irritability, and which de-emphasized restlessness and inattention.  Other studies measured child outcome using scales that were customized for each child based on parents' report of how their child was affected by diet.  Because these were blinded trials, the use of these alternative outcome measures does not invalidate the finding that parents, on average, observed improvement in children's behavior when AFCs were eliminated.  It does raise important questions, however, about whether such improvement will necessarily occur in the core symptoms of ADHD.

What are the clinical implications of these findings?  First, it is important to note that treatment guidlines from the American Academy of Pediatrics and the American Academy of Child and Adolescent Psychiatry both recommend recommend medication treatment and/or behavior therapy as the current treatments of choice for ADHD.  The authors themselves emphasize the need for additional research and are cautious about making any clinical recommendations.  They note that the restrictive nature of AFC free diets may place a burden on children and families, and suggest that until more certain methods have been developed to identify AFC-responsive children, "imposition of the diet should be done reluctantly". 

If parents would like to try an AFC elimination diet for their child, however, the findings reported in this study suggest that it may be useful in cases where parents have reason to believe that their child is sensitive to AFCs.  In addition, parents and children would need to feel comfortable with the restrictions that an AFC free diet will entail. If this approach is tried, it will be especially important to carefully monitor whether improvements that parents may observe are also reported by the child's teacher, as evidence presented above suggests that behavioral improvements in the classroom are less likely to occur.  Thus, parents cannot assume that their child is doing any better at school even if they observe improvements in the child's behavior at home.

In situations where children are struggling in these areas, even if their behavior has improved at home, additional interventions to address these difficulties will be important to implement.  This may include medication treatment and/or behavioral therapy and/or specific academic assistance, intervention components that currently enjoy the strongest empirical support for helping children with ADHD, and which be necessary treatment components even for those children who derive some benefit from dietary interventions.

Information presented in Attention Research Update is for informational purposes only, and is not a substitute for professional medical advice.  Although newsletter sponsors offer products and services that I believe will be of interest to subscribers, sponsorship of Attention Research Update does not constitute a specific endorsement or guarantee of any company's product or services.