There has been a recent increase in attention paid to the old question about food dyes and behavior in children. The idea that food coloring causes hyperactivity in children started with Ben Feingold in the 1970s. He popularized his “Feingold diet” for ADHD, which is still being promoted by some today.

Initial research showed a possible connection between certain food dyes, especially synthetic dyes, and hyperactive behavior in children. However, the next 20 years produced better controlled studies that did not show the alleged effect. It seemed like just another case of preliminary positive evidence that did not hold up to later more rigorous replication. Serious scientific interest in the question waned with this negative data.

However, recent popular interest in such issues has caused another wave of research. Dr. Oz’s website, for example, discusses the issue, giving it credence. Unfortunately, while it has renewed interest in the food dye question, the more recent research has not definitively answered the core question.

This is a complex research area because hyperactivity is often a subjective judgment, and behavior in children is subject to observer effects and placebo effects. Strict blinding and controls are therefore necessary.

Several recent reviews come to similar conclusions, which are more than a bit wishy washy. The Dr. Oz article cited one meta-analysis, but the conclusion of the analysis hedges:

Despite indications of publication bias and other limitations, this study is consistent with accumulating evidence that neurobehavioral toxicity may characterize a variety of widely distributed chemicals. Improvement in the identification of responders is required before strong clinical recommendations can be made.

Like other reviews and meta-analyses, they find a small effect, but also find serious problems in the data such as publication bias. They also find that the effect is most significant in a subset of children (called responders). While it is plausible that a subset of children, due to genetic differences, may have an adverse effect to some food dyes, while others do not, this makes interpreting the research more tricky.

A 2012 meta-analysis showed:

A restriction diet benefits some children with ADHD. Effects of food colors were notable were but susceptible to publication bias or were derived from small, nongeneralizable samples. Renewed investigation of diet and ADHD is warranted.

Again we see there are problems with publication bias (a tendency to publish positive studies more than negative studies), but also point out problems with the sample sizes. They also found an effect in some children but not others.

Conclusion

At present there does not seem to be a definitive answer to the question of whether or not any specific food coloring or combination of food dyes worsens the symptoms of ADHD or contributes to hyperactivity in children with or without ADHD. To clarify, there is no research showing that food dyes cause ADHD, the question is restricted to short term effects on behavior.

There are some common themes in the reviews and meta-analyses. The existing research is currently inadequate to definitively answer the question, and further research is warranted. There appears to be publication bias affecting such reviews. Many studies are small, have problems with selecting subjects, and have small and inconsistent effects.

It is possible there is an adverse effect on behavior in a subset of children.

I agree with one reviewer who concluded:

While these strictures could have positive effects on behavior, the removal of food dyes is not a panacea for ADHD, which is a multifaceted disorder with both biological and environmental underpinnings.

An effect from food dyes is possible, but its overall effect on ADHD is likely minor and only one piece to a larger puzzle. Unfortunately this leaves us without a clear recommendation for consumers.

It may seem obvious to recommend to parents of children with ADHD to try a dye-restricted diet in their children to see if it works. However, such subjective evaluation by parents is likely to be overwhelmed with confirmation and observer bias. It still may not be unreasonable to try it. At least there is no health risk to such a trial. But I would advice caution in interpreting the results, and weigh any perceived benefit against any inconvenience or added cost. I would not recommend draconian measures in the hopes of a dramatic effect.

Mostly I hope to see some further rigorous trials to more definitively put this question to rest.