Testing Wheat- and Dairy-Free Diets for Autism

Image Credit: Hans Braxmeier / Pixabay.

What happens when children with autism on gluten- and casein-free diets are covertly challenged with wheat and dairy?

The first randomized, control study of a gluten-free and casein-free diet for autism found that parents reported their kids did significantly better in the diet group than in the control group, but that could just be the placebo effect, where “parents attribute changes to the diet, in part due to the great effort they are putting into it…[so] they will be biased toward seeing evidence of success” that may not actually be there. 

What if you don’t rely only on parental report? What if you do a blinded study of a gluten- and casein-free diet, as I discuss in my video Double Blind Clinical Trial of Diet for Autism? In such a study, the parents know what the kids are eating, but you don’t ask only the parents how the kids are doing; you have investigators objectively assess all the children without knowing who is in which group—the diet group or the control group. The result? The investigators found “a significant beneficial group effect at 8, 12 and 24 months of intervention on core autistic and related behaviours” with a gluten- and casein-free diet. This was one of the largest such studies ever done. It began with 73 children, but about a fifth of the subjects dropped out, mostly from the diet group.

“If a family did not feel that their child was making strides on the diet, they may have been more likely to drop out of the study, thereby skewing the analysis toward those who believed their children were making progress”—for those kids for whom the diet appeared to work better. So, the remarkable results they got in terms of improved social interaction and fewer ADHD-type symptoms may have ended up exaggerating the effects of the diet, since the kids for whom it didn’t help may have gotten disproportionately weeded out. Also, the parents were very much aware whether their kids were in the diet group or the control group because they were the ones cooking the meals, so they may have changed their own behavior towards their children and that “may have influenced some of the observed patient responses.”

This is similar to what happened in a famous sugar study where researchers fibbed to mothers and falsely told them their children had just received a whopping dose of sugar. Not only did the mothers rate their children as “significantly more hyperactive,” they inadvertently changed their own behavior, too. The mothers who falsely believed their kids had just gotten a load of sugar were observed to have “exercised more control” over their children and were more critical of them than were control mothers. In this way, their expectation of an effect may have actually ended up having an actual effect in changing their children’s behavior. 

In these unblinded autism studies, the parents are upending their family’s diet, hoping and possibly even expecting that their children will get better. Perhaps the parents are even unconsciously treating them differently, such that the kids end up behaving differently when assessed later by the blinded investigators. This is why we need double-blind studies where no one—not the parents or the kids—knows who’s in the diet group or the control group. Why didn’t the researchers conduct that type of study? Why didn’t they secretly sneak some gluten or casein into the children’s diets to see if they got worse again? Because it wouldn’t be ethical, just as the researchers in a study I discussed in detail in my previous video Are Autism Diet Benefits Just a Placebo Effect? had determined. They just couldn’t bring themselves to slip any gluten or casein to the kids because they were so convinced those proteins may be harmful. That’s pre-deciding the outcome, though. That’s circular logic. We can’t test to see if it really works, because it may really work—but we can’t test it. What?!

Finally, researchers at the University of Florida broke through the impasse by performing a double-blind study, which is not an easy thing to do. All the meals and snacks had to be provided so the families remained clueless as to whether they had been randomized into the gluten- and casein-free diet group or were actually in the control group, getting the same foods, but with gluten and casein. Then, after six weeks, they switched, so the gluten- and casein-free group started getting wheat and dairy, and the control group were covertly switched to a gluten- and casein-free diet. 

Before “unblinding,” that is, before the codes were broken to see who was in which group, the parents were asked whether they thought their child was on the special diet during the first or second six-week period. Five got it right, two had ‘‘no idea,” and six were wrong. In other words, it was no better than a flip-of-a-coin chance. In fact, about half essentially thought their kids got better on the casein and gluten. So, there were “no statistically significant findings even though several parents reported improvement in their children,” claiming “marked improvements in child language, decreased hyperactivity and decreased tantrums”— so much so that a number of the parents decided to keep their kids on the gluten- and casein-free diet even though the researchers had just told them that it didn’t work.

Was anything missed? Some parents had claimed significant improvement, after all. The researchers re-examined the videotapes they had made of the kids before and after the diet intervention and showed them to blinded examiners. Did the children’s language really get better? Apparently not. The videotapes showed no such improvement, so, again, the study results did not support the efficacy of a gluten- and casein-free diet for improving some of the core symptoms of autism—“at least for a dietary intervention lasting for 6 weeks.” The non-double-blind studies that showed an effect had kids on the gluten- and casein-free diet for a year or even two. What does this mean? “The failure to obtain statistically significant difference between the diet conditions in the current [double-blind] study should not necessarily be interpreted to mean that the intervention method does not work for children with autism,” given “the relatively shorter duration of the diet intervention period.”

The same issue materialized years later in a 2014 study in Texas. The study design was simple: Put everyone on a gluten- and casein-free diet, and then randomize parents to get weekly baggies filled with either gluten- and casein-free brown rice flour (thereby sticking to the diet) or an identical-looking powder with gluten and milk mixed in. So, no one knew until the end who truly remained gluten- and casein-free. The result? No meaningful changes were found in either diet group. Okay, but this study only lasted four weeks, and diet proponents suggest it may take months off gluten and casein to properly assess a response.

The problem is there hadn’t been any double-blind studies that lasted that long…until recently.

This article discusses the fifth video in a six-part series on the role of gluten- and dairy-free diets in the treatment of autism. For the first four installments, see:

And, for the finale, check out Pros and Cons of Gluten-Free, Casein-Free Diets for Autism.

Keep abreast of all of my videos on autism here.


You may be curious why I didn’t just skip to the chase and make a single video on the topic instead of produce this whole series. Well, it’s such a controversial and contentious issue I wanted to do a deep dive to offer a comprehensive overview. 

If you’re ever uninterested in a subject I’m discussing, please feel free to dip into the videos on the more than 2,000 other topics I’ve already covered on NutritionFacts.org.

In health, 

Michael Greger, M.D.

PS: If you haven’t yet, you can subscribe to my free videos here and watch my live presentations:

Pin It on Pinterest

Share This