May 20 2010

Special Diets Do Not Work for Autism

One of the burning controversies in the autism community is whether or not special diets, specifically gluten-free or casein-free, are of benefit to the symptoms of autism. About 1 in 5 children with autism are on a special diet, and many parents strongly believe it is of benefit, but there has been no scientific evidence to back up their anecdotal observations.

A recent study adds to the evidence for a lack of benefit from such diets. The study is not yet published, but is summarized on the University of Rochester website. The main weakness of the study is that it is small – the study enrolled 22 children and 14 completed the study. It’s strength is that it was tightly controlled. Children were placed on a gluten-free (no wheat and barley) and casein-free (no dairy products) diet for 4 weeks. They were then challenged in a double-blind manner with snacks containing gluten, casein, both, or neither (placebo) and observed for behavior and GI symptoms. The study found no benefit for any outcome measure.

By itself this study is far from definitive and I am sure it will not be the last word. It is a small addition to the growing evidence showing lack of benefit from such special diets in autism.

It should further be noted that the evidence does not support the notion that children with autism have a specific gastrointestinal disorder. A recent consensus statement based upon a systematic review of the literature concluded that no such autism-associated GI syndrome exists and that children with autism should be treated as any child would when they have GI symptoms. It is probable that children with autism may respond to GI symptoms when they occur with a worsening of their behavior. This may then cause parents to associate the behavior with the GI symptoms and falsely conclude that the GI symptoms cause the behavior and the underlying autism. This concept comes up frequently in medicine – differentiating an exacerbating trigger from an underlying cause.

It is also interesting to consider why the general notion that diet can dramatically affect behavior in children (whether typical, autistic, or with other disorders like ADHD) is so widespread despite lack of scientific evidence. Many parents believe that sugar makes children hyperactive, but the evidence does not support this believe – it’s basically a myth. What appears to be happening is simple confirmation bias – which can have a powerful psychological effect on belief. When we think there is a pattern we will tend to seek out, notice, and remember events that seem to confirm that pattern. Meanwhile we ignore or explain away disconfirming evidence. This can leave us with a powerful, even if completely false, sense of confirmation.

A common scenario, which parents have reported to me many times, is that parents will first observe that their child’s behavior has suddenly worsened – they are hyper and out of control. They then ask themselves – why is this happening now – and search for a cause. So they inquire about recent eating and may find that they just had a slice of cheese or bread or a piece of candy. “Aha,” they declare, “it must have been the cheese,” or whatever they found. But they do not realize that they don’t investigate what their child recently ate when they are not acting up, and so they have no baseline for comparison.

Unless someone is familiar with the concept of confirmation bias, it is difficult to impossible to convince them that what they believe they have directly observed is not true – just an illusion or trick of memory. When the topic is emotionally charged, such as a parent’s desire to understand, control, and help their child’s condition, it is even more difficult.

Again – we get back to the need for controlled observation (i.e. science). For individual patients I will often have them keep a written diary of symptoms and possible triggers, and use that as a basis for looking for patterns. This is still not perfect, but much better than raw memory. For the underlying question itself, blinded tests where variables are controlled as much as possible are necessary. We frequently find that effects noticed anecdotaly vanish under controlled observation.This happens so often in the history of science that we can take it as a proven principle that anecdotes are unreliable. They are useful for generating hypotheses to test, but they are worthless for arriving at reliable conclusions.

In a perfect world the scientific data would be the final word on such questions. We all just want to know the truth so that we give optimal treatment, whether to our children or our patients. But we live in the modern world of the democratization of information and expertise. When the science conflicts with anecdote some people choose to believe their anecdotal experience and distrust the science. They form grassroots organizations that take on a life of their own, and increasingly become isolated from and distrustful of the scientific establishment with which they disagree. They start to accuse scientists of “attacking mothers” and not listening to parents, and worse – of being part of a nefarious conspiracy.

That response, of course, is incredibly counterproductive. Some researchers may then overreact and discount parents completely or view them as hostile. The very people who can most benefit from the research become alienated from it. They may then form their own subculture, with their own experts and practitioners to serve their beliefs. They will attract more parents, as they are claiming to have a cure or at least a treatment that is effective. The lay media will generally be worthless to counterproductive in sorting out the situation for the public.

We have seen this very scenario play itself out with the Defeat Autism Now (DAN) community. We have also seen it with the chronic lyme disease community, with patterning, bee venom therapy, and many other diseases and treatments.

I don’t have the magic solution to this problem. I think part of the answer is constant education. We need a more scientifically literate public, and a more skeptically literate public. Scientists also need to be better and more aggressive at dealing with the public and the media, and the media needs to be more responsible and have higher standards when it comes to serious scientific reporting. And I think we need more effective regulation, and we need to specifically eliminate regulation which is designed to facilitate the existence of these medical subcultures that exist outside of the standard of care. And we also need to be vigilant toward the transparency and validity of our scientific institutions to minimize fraud and conflict of interest.

We will see how the autism community responds to this new study. It is small and should not be the final word – but it is soundly negative and does add to the body of research which is pointing away from the conclusion that gluten-free or casein-free diets are helpful in autism. Parents need to understand that the very fact such research is being done means that researchers are listening to parents – that was the only reason for doing such a study in the first place. But while scientists listen to parents to generate hypothesis, in the end they listen to the science, and so should the parents.

24 responses so far

24 thoughts on “Special Diets Do Not Work for Autism”

  1. mindme says:

    Canada’s CBC recently ran this report on the use of Hyperbaric Oxygen Therapy and diet. They did not seem to interview one single skeptical voice. They sort of hand waved towards the skepticism but played the emotional card to the max. Parents need to do something. Parents see amazing results. They repeated the whole notion of toxins causing autism without any real counter. Skeptical doctors call progress a mere fluke.

    There was a q/a between the segment producer and the anchor after the video that brought a bit more skepticism to the claims but it certainly won’t live on and I doubt would have changed the mind of any fence sitter swayed by the appeal to emotion and the kindly doctor charging $200 a session (with a min of 40 sessions required … which resulted in at least one parent remortgaging her home to pay for) with some science-y sounding talk about autism caused by not enough oxygen to areas of the brain.

    In a slightly unrelated story, people have been lobbying the Ontario government to allow doctors to peform some kind of surgery to increase blood flow, based on the idea blocked neck veins cause MS. I think the doctor who came up with the idea is urging people to wait but at least one hospital in Ontario was performing the procedure and billing OHIP under some paper subterfuge.

  2. XYZ says:


    I have MS and live in Ontario and have been (somewhat) following the idea about blocked neck veins causing MS. The procedure uses balloon angioplasty to open up the neck veins. There was a doctor in California using stents, but he has since stopped, I believe.

    In November 2009 a Canadian news magazine (W5) ran a story on Dr. Zamboni, CCSVI and the “liberation procedure”. It was an emotionally charged piece and got people in the MS community up in arms. Predictably as doctors are urging people to wait until more studies are done, it has turned into a they’re-all-against-us, doctors-and-the-MS-Society-want-us-to-stay-sick-so-they-don’t-lose-their-jobs, everyone-is-in-bed-with-Big-Pharma type thing. It’s a bit of a mess, I think.

    I would love to know Steven’s take on all of this.

    Sorry, very off topic.

  3. I wrote about CCSVI here:

    Basically – I think the plausibility is low and the evidence is preliminary and now conflicting. Zamboni may have hit upon a side consequence of MS, but it is highly unlikely given what we already know that he has hit upon an underlying cause. It is also possible his observations are nothing but wishful thinking. We need some definitive science to sort it out.

    I also think it would be an awful coincidence that a neurosurgeon who’s wife has MS set out to find a treatment for MS and found a neurosurgical treatment, completely out of the blue. This does not make it wrong – it would just be a huge coincidence, so it raises a red flag.

    Bottom line – we need time for the science to sort itself out, and I do not think there is sufficient evidence or plausibility to give this treatment outside of a clinical trial.

  4. xyz,

    Steven Novella runs, and he and another contributor have each posted articles about the Zamboni procedure. Check them out:

  5. enjay111 says:

    If it was true that carotid stenosis causes/’contributes to MS, then would we not see an awful lot more older people diagnosed with and suffering from MS? As far as I remember the biggest epidemiological factor in a daignosis of MS is visiting a neurologist – humorous but apprently true…..

  6. XYZ says:

    Steven and Alison,

    Thanks for the links. I will definitely check them out.

  7. I was saddened to note that my local library has many books of the archtype “Cooking to avoid autism” or “the anti-autism diet” which seem to just be gluten-free or caisen-free recipes. I think these titles are harmful to the public but don’t wish to be fulfill the stereotype of “covering up what the man doesn’t want you to know”. Would it be unreasonable to challenge these books being available or is there a precedent for the skeptic movement to challenge library books?

  8. Calli Arcale says:

    Knowing people with actual gluten intolerance (celiac sprue), actual lactose intolerance, actual casein allergies, and other actual allergies, I find the whole idea of treating autism with a GF/CF diet to be absurd. If the kid has an actual intolerance or allergy, absolutely adjust the diet. But otherwise, you’re setting yourself up for some major headaches, because GF/CF is a very hard diet to follow, especially if you’re REALLY following it. It’s going to be expensive, it’s going to be socially awkward, and it has a very real risk of impeding nutrition unless great care is used. I’m fine with reducing intake of refined grains (whole grains are better, now that we have preservatives to keep them from spoiling) and dairy products (I love cheese passionately, but there is definitely such a thing as too much of a good thing). But completely eliminating them is silly — and makes it harder for people to take seriously the real allergies and intolerances.

  9. Calli Arcale says:

    BTW, autistic children are more likely to be picky eaters, as they tend to fall into habits very easily and are comforted by familiarity. So I could see value in a book about cooking for autistics (or even just cooking for picky eaters). But it wouldn’t be about treating autism by eliminating gluten or casein. There are already cookbooks for those, thanks to the real allergies and intolerances. (“The Gluten-Free Gourmet” is a good one.) It would be about clever ways of hiding nutrition in the foods the child is willing to eat, ways of building a diet plan that will vary the diet sufficiently within the child’s chosen foods, and strategies for successfully introducing new foods. That’s where I’d like help, anyway. That would actually be *useful*. My girl can safely eat any food; she just refuses to *try* them (or insists they are delicious but can only be eaten at specific family occasions, like Easter or Thanksgiving). I’m finding tortellini and lasagna to be a great help there; it’s amazing what you can hide in that if you puree it finely enough, and mix it with tomato sauce to hide the color. 😉

  10. tmac57 says:

    Steven, I’m a bit surprised that you titled your article the way you did.The article itself seems more circumspect.

  11. tmac – I am basing my conclusions not just on this article, but my reading of the entire literature plus lack of plausibility. The authors were limiting their conclusions to their study alone, and were careful to point out what was not covered by their study.

  12. tmac57 says:

    That ‘hyperactivity caused by sugar’ myth seems to be one of those unsinkable rubber ducks.I think that it has been at least 15 years since that was 1st shot down, and every time I hear it repeated,and I try to tell someone (usually a mother )why it has been disproven,they will say something like “Yeah I heard that,but I know for a fact that anytime my kid has cokes and candy,they just turn into a little crazy person”.(Sigh)

  13. provaxmom says:

    Several years ago at a conference for families dealing with the condition my son has, they had a guest speaker. I forget precisely how she billed herself, but it was something to the effect of “autism nutritionist” or “autism dietian” or some such thing. Upon researching her, I found that she has discovered several solutions for your child’s “gut problems.” Insert huge eye roll here. And of course she offers many gf/cf cookbooks.

    Of course whenever I express my ahem, concerns or skepticism to us supporting such things, I am always found to be the ornery one. All the while our fundraising money, which families work hard to do, go to pay for it.

  14. Calli Arcale says:

    tmac57 — it’s a wonder they don’t think about the other things that go along with the cokes and candy, like caffeine (Coca-Cola being, of course, a caffeinated beverage) and FUN!!! There is a strong association in children’s minds between sweets and excitement. Halloween, parties, birthdays . . . even most religious holidays have an element of giving candy to children. They are accustomed to associating candy with excitement and with rewards, so of course they’re going to get silly when they eat candy. It’s FUN!

  15. ccbowers says:

    If candy is merely mentioned, my daughter (she is almost 4) will sometimes start jumping up and down on her own. The myth about sugar is a primary example of how people are often terrible at being able to draw accurate conclusions from observations and experience when there is an expectation.

  16. Matt P says:

    The only part of this post that I don’t like is the reference to “the autism community.” If that means “parents of children with autism,” then I can assure you that many of us readily accept, and even seek out, the scientific evidence on autism treatments.

    I think the phrase is supposed to refer only to those people who use “alternative” treatments for autism.

    There is no autism “community.” Those of us intimately involved with the disorder are as diverse as any other group of randomly-selected people. The anti-conventional-medicine crowd hijacks the phrase in an attempt to show that everyone with autism in the family is on their side. Skeptics should be careful not to fall into that trap.

  17. tmac57 says:

    Good to hear your perspective Matt.The anti-vax/alt med. element of parents with autistic children will naturally attract the attention of the Skeptical groups more,by their very nature.It is always an important reality check to put them in the broader context of the overall numbers of those affected by autism.They just seem to have a louder voice.

  18. s says:

    I think you doubt parents too much, and overstate the results from the study. The group was just too small to draw any reasonable conclusion however “well controlled” the diet was. Hypothetically it could also have consisted only of children not affected by diet. Four weeks is also a very short intervention time for uncovering subtle dietary causes or other environmental causes. Whatever cause autism has, the MS study by Baranzini et al (Genome, epigenome and RNA sequences of monozygotic twins discordant for multiple sclerosis) finding no genetical cause of MS (!!), does sure make me stop and think before I off-handedly dismiss environmental factors like diet.

  19. ccbowers says:

    “I think you doubt parents too much”

    I’m not sure what this sentence means. Do you? THis is not the only study on this topic, and so far there is no compelling evidence of a relationship between diet and autism. It is to the point that perhaps we should stop looking at this since there are other potentially more fruitful avenues to pursue.

    We don’t find out about the truth about the world by just asking people what they think. We control varables and testing conditions to ensure that our results are not due to biases or other variables. Just like we don’t test the efficacy of medications by asking people if it worked for them, we shouldnt do the same for somehting like diet. Popular opinion is not how science works.

    “Hypothetically it could also have consisted only of children not affected by diet. ”

    So somehow there are children who would benefit from diet yet were sytematically excluded from the study? What is your evidence for this? If none, then appear that you want it to be true, and are coming up with a rationale that is unrelated to any evidence. If there is a subset of the population that could benefit we should be able to see this, and have a rationale for why this subset is different. But we have no evidence for any of this, and just speculating based upon nothing.

    Why do people have such attachements to certain theories? Is it wishful thinking or something more?

  20. s says:

    Reply to ccbowers

    I am always very fascinated at the acrimony risen by certain subjects ventilated on this forum. CCbowers response to my reflection is charcteristic of it. Instead of trying to out forth a scientific rebuttal it assumes a very minor study is the last and almost godly word on the subject. A very dogmatic and unscientific response that is just the mirror image of how dogmatic believers in the paranormal etc etc react.

    Science does not advance due to preconceived notion of what is right and wrong, but by questioning everything. Even purported notions of scientific truth. Especially questioning results in such a complex field as environmental effects on human health. A questioning that goes both ways instead of just faithfully following party lines. And reading this forum always reminds me of how easily one could fall in the trap of superbia.

    Anyhow, dietary hypotheses are anything but disproven. There are myriads of studies going proving this and that (see PubMed) and the twin study i refer too was the main point, not whatever ccbowers chose to cherrypick and create a response to. Main point because the study did not show any genetic or epigenetic causality. Which is extremely interesting as MS research is then back to square one: the environment. And as the MS debate around diet is similar to that about ASDs it is relevant to make that point.

    Follows some randomly chosen recent diet – autism studies coming indicating dietary effects: (24 month study of 72 children) (expert consensus on gastrointestinal disorders in ASDs) (review of dietary intervention in autism, concluding we still do not know enough) (an extremely interesting article on the metabolic regulation of adenosine and ketogenic diets and the “therapeutic implications … to acute and chronic neurological disorders as diverse as brain injury, inflammatory and neuropathic pain, autism and hyperdopaminergic disorders.”)

  21. Cay says:

    Calli Arcale – One of my sons has some autistic behaviors, hand-flapping, toe-walking, really picky eater. Years ago, he adopted Pizza Squeeze as the only acceptable tomato-based topping (besides ketchup.) These kids are dastardly clever!

  22. ccbowers says:

    S –

    Questioning everything is fine. I’m not trying to be dogmatic… we just have to look as plausibility in science. We don’t have unlimited resources so we have to look at most plausible and helpful avenues with a greater priority. Diet has been looked at, and there just doesnt seem to be anything there. I’m not sure what the rationale for removing casein or gluten anyways, it appears to be grasping at straws.

    The “studies” you point to are very weak. They are either basic research only and/or not designed to address the question. Not one is double-blind which is very important to determine a real effect for a study addressing dietary interventions. A review saying ‘more research is needed’ is not evidence… they use that type of statement in accupuncture studies. Just like a lot of research into things that don’t work, the better studies show no effect.

    I’m not philosophically opposed to diet affecting autism, its just not very plausible and the evidence is just not there. If better evidence comes along I’ll be all for it. It would be a relatively easy and inexpensive intervention. I just don’t see it happening.

  23. Zanduby says:

    Not sure you will see this Steven, but if you do see this a response would be awesome.

    My question is; Is there any updates that you’ve seen to this? My son has been recently diagnosed with HFA and my wife has been reading all the woo out there. I’m trying to sway here away from all of it. I’m glad to have found this, but it is a couple years old. Any way you could revisit the information or help point me in the right direction?

    I have googled and searched pubmed for articles and found two ( and

    Thanks for all you and your cohorts do.

Leave a Reply