I’ve been thinking a lot about the MAHA movement’s rhetoric around food and neurodivergence, where food additives—and processed foods in general—are singled out for “causing” conditions like autism and ADHD, despite a lack of sound scientific evidence. Since I’m still getting back into the swing of things after maternity leave, I’m re-sharing this piece I wrote a few years ago about food additives and ADHD. I found this research fascinating, and I hope you will, too.
I’ve unlocked the paywall to make this info available to everyone, but this is the kind of post normally reserved for paid subscribers because it’s so labor intensive. Upgrade to paid for a whole library of Q&As like this!
Can you provide insight to any impact food colors/additives have on ADHD? Is there a correlation or is it just BS?
FYI: my answers here are for educational and informational purposes only, aren’t a substitute for medical or mental-health advice, and don’t constitute a provider-patient relationship. Also: I’ve never taken money from “Big Food/Beverage” or worked with them in any way, and I never plan to.
There’s a longstanding myth in wellness culture that artificial food colorings and other additives cause ADHD, but the science doesn’t bear it out. Still, although food dyes aren’t responsible for giving people ADHD in the first place, there’s some evidence from randomized controlled trials that consuming artificial colorings may lead to a small increase in behavioral issues in a small subset of kids with ADHD. The research on other additives is much sparser and includes a number of dubious studies, though again, a small subset of kids with ADHD may respond to a more extensive elimination diet (or this finding may be simply the product of biased research). There’s very little data on these questions in adults, but the same issues likely apply.
Paywall Unlocked! Become a Paid Subscriber for More Like This
The misleading idea that diet “causes” ADHD first took hold in 1975, when an allergist named Benjamin Feingold published a book alleging that artificial colors, flavors, and preservatives, along with a few naturally occurring compounds, were the main cause of hyperactivity (ADHD didn’t get its current name until 1987). These ideas—and his eponymous Feingold diet, which removed the offending additives—weren’t based on any scientific evidence but simply on the doctor’s own experience and speculation, which are subject to many forms of uncontrolled bias.
Even though there was no proof that his diet worked, it became popular among laypeople and health professionals alike. Soon scientists did start testing the diet, and by the mid-1980s there were enough small “challenge studies” (where food additives were removed and then brought back) to warrant a review in a respected pediatrics journal. The 1986 review found that just 1 percent of children had consistent improvements in behavior (based on parent/teacher reports) on the Feingold diet, and only 10 percent had an increase in symptoms when standardized food dyes were reintroduced. The review concluded there was no evidence to support recommending dietary changes for kids with hyperactivity, though more recent evaluations suggest that a small subset of kids with ADHD may be sensitive to food colorings, as we’ll discuss.
But the 1986 review seems to have had limited impact. Feingold advocates were numerous by that point, and the notion that food additives were responsible for hyperactivity had become something like conventional wisdom. A 1994 study, for example, used causal language emblematic of some of the more dubious research in this area: “Foods and additives are common causes of the attention deficit hyperactive disorder in children.” It based this conclusion on results from just 26 kids with ADHD, only 16 of whom underwent a double-blind, placebo-controlled food challenge—and children with preexisting allergies had a significantly higher response to the challenge than those without, meaning these results may not be generalizable beyond a subset of kids with ADHD. But that didn’t stop the researchers from making a big leap to say that the condition itself is caused by foods and additives.
A decade later, a 2004 meta-analysis shed some light on the issues with research in this area. The study evaluated 15 trials with a total of only 219 participants—including 2 studies with just 1 or 2 participants. They were all double-blind, placebo-controlled trials (though not all randomized), but the researchers only gave two of the studies “A” ratings for quality.
The analysis found that health professionals’ and teachers’ ratings of kids’ behavior in response to artificial sweeteners were NOT statistically significant, but parents’ ratings were—which may be because the outcome scales used to measure symptoms were non-standardized and skewed toward things parents would be more likely to observe (e.g. sleeplessness, irritability). What’s more, only trials that screened for responsiveness to dietary changes (according to parental reports or unblinded tests) before enrollment found a statistically significant effect of artificial colors—but an additional trial that screened participants prior to enrollment with a blinded trial of the Feingold Diet did NOT show any statistically significant effects. This raises a couple possibilities: 1) that the pre-screened kids were already more likely to be sensitive to dietary changes than the average ADHD population (or non-ADHD population); and 2) that there were placebo/nocebo effects at play in the study populations (who may have already had placebo responses to food additives in the pre-screening process) or their parents (who were more likely to enroll their kids in a diet study in the first place because they believed diets worked), and those effects weren’t controlled for in the trials.
It’s likely that possibility #2 has biased at least some of the results in this area, as several researchers have pointed out. But if possibility #1 is also true, there may be a small subset of kids with ADHD who are sensitive to artificial food colorings. Indeed, a 2012 meta-analysis (which had similar discrepancies between parent and teacher reports) found that about 8 percent of kids with ADHD may have symptoms linked to artificial food colorings.
However, the 2004 meta-analysis found that the effect size of food-coloring exposure among these kids is relatively small—only about a third to half the likely effect size of taking kids off stimulant medications for ADHD. That is, medication still has a much stronger effect on ADHD symptoms, even in the (possible) subset of kids who are sensitive to food colorings. And the effect size for artificial food colorings was even lower in the 2012 meta-analysis (g = 0.12, vs. SMD = 0.21 in the 2004 meta-analysis). If there is a small percentage of kids who react to artificial food colorings, their reactions are likely small as well.
Given the limitations of the existing evidence and the ongoing need for more research, I don’t think it’s wise to recommend as a general rule that kids with ADHD avoid artificial food colorings. Scientists still don’t know the mechanism (if any) by which artificial food colorings might affect a small subset of kids with the condition. And as the 2004 meta-analysis pointed out, “the restrictiveness of an AFC [artificial food coloring]-free diet may burden hyperactive children, who are already at risk for poor psychosocial outcomes.” In fact, kids (and adults) with ADHD may have a higher risk of eating disorders, which makes any dietary restrictions especially risky.
The only evidence I could find in adults is this small pilot study from 2022, which involved just 18 ADHD participants and 11 controls who were randomized to a double-blind, placebo-controlled crossover challenge with artificial food colorings. It found that there were no significant differences in inattentiveness, scores on the Adult ADHD Self-Report Scale, or other attentional measures for either group in the AFC challenge periods. There were some small but statistically significant differences in two types of brainwave activity in the ADHD students after reintroducing AFCs, but the researchers acknowledge that it’s unclear how to interpret these results—they’re simply “exploratory.”
Other Additives
Artificial colorings are the most-studied food additive in ADHD, but some researchers have focused on other additives or overall restricted diets as well. Unfortunately, much of that research is shaky—and some is incredibly shoddy.
Take this retracted paper from 2022, which purported to review the harmful effects of various food additives. The dubious article (not an RCT) made many leaps from correlation to causation, and said activist-y things like, “we must take specific precaution to avoid consuming dangerous compounds before they begin to wreak havoc on our health.” I would’ve been suspicious of this article even without the big, red retraction notice at the top, but that made it all the more clear that this was not real science. The publisher’s investigation found evidence of “systematic manipulation of the publication process,” which “undermines our confidence in the integrity of the article’s content and we cannot, therefore, vouch for its reliability.”
This recent (2024) paper is different in that it hasn’t been retracted, and its tone is less egregious, but it still uses advocacy language that isn’t justified by its design (a literature review, not an RCT). For example, from the abstract:
Little research has focused on children who are the largest consumers of UPFs [ultra-processed foods]. Potentially, they are a ticking time bomb for adult obesity, metabolic syndrome, cardiovascular diseases, mental health disorders and cancers. Based on risk/benefit analysis, azo dye AFCs should be banned. Benzoates, emulsifiers and sweeteners require assessment of quantitative limits and cumulative effects of combinations. Consumers purchasing UPFs require information through ingredient health warnings and recommendations to use natural unprocessed foods which have well-described health-promoting properties.
The journal bills this paper as a review, but it reads more as an opinion piece, with phrases like “ticking time bomb” and “should be banned” and a recommendation to put health warnings on UPFs and push consumers toward “natural unprocessed foods.” I don’t have anything against opinion pieces—I sometimes write them myself! But this is not an unbiased scientific study. Instead, it seems like more hype about ultra-processed foods that’s unwarranted by the science we have to date, as Laura Thomas, PhD, RNutr points out in the above-linked episode.
That said, there is some limited evidence that a “few-foods diet”—a descendant of the Feingold diet that eliminates many foods and additives, including artificial colors and flavors as well as lots of other food groups—may have some effect on behavior in kids with ADHD, though the data is mixed and should be interpreted with caution. Previous systematic reviews have found that there was no significant effect of a few-foods diet, but a 2017 systematic review reached a different conclusion. It analyzed the results of two previous meta-analyses of double-blind placebo-controlled trials of the diet and found an apparently large effect.
But that review has serious conflicts of interest: the lead author, Lidy M. Pelsser, is “franchiser of the restricted elimination diet (RED) protocol,” meaning she benefits financially from recommending this diet—which is another name for the few-foods diet in her home country of the Netherlands 🤯 Her Pelsser RED Centers website (translated from Dutch) uses causal language that goes far beyond what the research has found, such as: “Diet removes a cause of ADHD. Medication only suppresses the symptoms.” This conflict in Pelsser’s systematic review is akin to a pharmaceutical company finding that its weight-loss drugs work, or WeightWatchers concluding that its diet program works: a major source of potential bias that means the results should be taken with a grain of salt.
It’s possible that this bias explains something odd about the way statistics are handled in this study. When Pelsser and her colleagues analyzed the results of previous meta-analyses, almost none were statistically significant (p > .05), meaning that any differences between diet and placebo groups could simply be due to chance. But they gloss over that in the results section of the systematic review, and in the discussion section they make a big deal about how statistical significance can’t tell the whole story anyway; instead they harp on effect sizes and confidence intervals. While there have been valid critiques of the overemphasis on statistical significance in social-science and biomedical research (and some thoughtful responses to the critiques as well), I’m not sure if that’s really why Pelsser et al. took this stand. In fact, this same team of researchers went right back to reporting p-values and emphasizing statistical significance in this 2020 study on the few-foods diet, which had results that were more favorable to the diet (and therefore to Pelsser’s business). If their position on statistical significance in 2017 was genuine, it was also apparently short-lived.
Beyond the conflicts of interest, the 2017 systematic review’s supplemental materials make it clear that any potential effect a few-foods diet (FFD) only exist for a small subset of children with ADHD. As in the food-coloring studies, the kids in the FFD trials were openly given the diet before randomization, after which only the kids who had a change in symptoms moved on to the next phase (an open challenge where one food a week was reintroduced to pinpoint any reactions). Only then did the responders move into the double-blind, placebo-controlled phase of the studies—so the response rate really reflects kids with ADHD who were pre-selected for either a genuine response to foods, a placebo effect, or perhaps both.
Indeed, Pelsser and colleagues acknowledge the limitation that these studies may only have included parents interested in dietary treatment for ADHD, since that’s who signs up for a diet trial. The seeming effectiveness of a few-foods diet in these kids could be largely a matter of parental placebo effect, where the parents are expecting to see benefits from the diet and so they do.
Finally, even with the significant pro-diet bias of this systematic review due to its conflicts of interest, the authors still caution that a few-foods diet “is not a long-term treatment, but a short-term diagnostic procedure” meant to identify any food sensitivities, at which point children can be given personalized diet advice. They also acknowledge that this ultimate goal could take at least a year, and that it would be burdensome to families and might not be achievable except by the most motivated parents with younger kids (who’d presumably be less likely to protest the restrictions). They conclude that “large-scale implementation of the few-foods approach would not be a realistic recommendation.”
The Bottom Line
Like so many areas of nutrition research, this one is fraught with limitations and conflicts of interest. Based on the available evidence, it’s possible that a small subset of kids with ADHD would have a small reduction in certain symptoms if they avoided artificial food colorings—but we don’t know who or why, since the mechanisms for this potential link are still unknown. And the potential drawbacks include increasing the burden on kids and families and putting an already high-risk group of kids at even greater risk of disordered eating.
*
Thanks so much to the person who asked this question! Ask one of your own for a chance to have it answered in an upcoming Q&A.
