reports like this one that starts off with -
"A popular diet intended to curb the effects of autism received yet another blow today in the form of a new study that found that autistic children who adhered to a gluten-free, casein-free diet showed no improvement in their symptoms"
The problem that reports like this is that they overstate what the study shows. Or perhaps I should say what it appears to show as all I have to go by is media reports, the abstract, and the clinical trial registration. I would feel much more comfortable if the study had underwent the normal peer-review process and had been published in a journal, that way there would be something more concrete to talk about. Regardless, the study was presented and discussed widely in the media, so here we are.
Before I talk about the study, let me first ask a question. Do you think a gluten-free casein-free diet can rapidly "cure" the symptoms of autism in every child?
I think the answer is an obvious no. If the diet were an instant cure or made a huge difference for every child, there wouldn't be any question of whether the diet worked and demonstrating the effectiveness in a study would be trivial.
I think that people forget that this is a dietary change and not some sort of drug. People are used to taking a pill and seeing quick results. If you approach a treatment like the GFCF diet with that expectation, you are going to be disappointed. Think of it this way, if you alter your diet to lose weight or because you need to lower your cholesterol or because you need to manage a chronic condition like diabetes, would you expect to see changes overnight? Of course not, you would understand that these things take time. So while I am sure that there are some children who will respond very quickly and very strongly to the diet, I am equally sure that majority of children will not will do so.
The question then becomes whether there is a subset of children with autism who can benefit from the diet, what exactly the diet can do for them, and how quickly they can be expected to respond.
With this in mind, lets look at what what question the researchers were attempting to answer in this study. According to the abstract, the goal was "to evaluate the nutritional adequacy, physiological effects, and efficacy of the GFCF diet on symptoms of ASD using randomized double blind placebo controlled challenges in preschool children with ASD".
The researchers recruited 21 children who were determined to have autism using the ADI-R and/or ADOS and were receiving at least 10 hours a week of EIBI (ABA) therapy. I believe the researchers included the EIBI therapy because they wanted to attempt to control for improvements due to other treatments. These children were screened for milk/wheat allergies, celiac disease, and nutritional (iron) deficiencies. They found that 3 children had at least one of these problems (2 potential celiac, 1 iron deficiency) and these children were excluded from the study.
These 18 children were then placed on a strict GFCF diet for 4 weeks before being randomly challenged weekly with either wheat, dairy, or a placebo in some manner. I say in some manner because there seems to be some disagreement between the abstract and media reports as to how many of these challenges there were and how often they occurred. I think the challenges happened only once per week, but it isn't clear if there were challenges every week or only some weeks. Then there is the line that challenges only occurred if "measures were at baseline levels" - I have no idea what that meant in practical terms.
During the course of the study, 4 participants dropped out either because they couldn't adhere to the diet or because they withdrew from EIBI therapy, which left only 14 participants in the study.
Data was collected at the start of the study, 6 weeks, 18 weeks, 30 weeks, and before and after each challenge. I am not sure what was being happening at 18 weeks or 30 weeks as the study only seems to have lasted 16 weeks. It isn't clear what exact data was collected at each point but the following measures were used at some point during the study - Bristol Stool Scale, Sleep Diaries, Actigraphy, Conners Abbreviated Rating Scale, Target Symptoms Scale, Ritvo Freeman Real Life Rating Scales (RFRLRS), as well as some form of nutritional monitoring. Most of these measures have to do with behavioral or sleep issues.
Given the above, the results of the study aren't too surprising. The researchers found "no favorable effects of the GFCF diet on attention, sleep and stool patterns" but "such effects may occur for individuals or for subgroups of children (e.g. with significant GI disease), providing the basis for positive anecdotal reports."
This study was well designed and well executed, but there are a few problems.
The first problem is that the number of children in the study is very small AND the children who showed a potential intolerance to wheat/dairy or showed iron deficiency were excluded. As I talked about above and as the researchers acknowledged in their abstract, only some children seem to benefit from the diet. And yet the researchers not only failed to find the children who would benefit from the diet, they also kicked out the children who were most likely to benefit. This coupled with the fact that there are only a handful of children in the study means that the diet would have had to have a very large effect to be seen.
The second problem is the length of time of the strict portion of the diet before the challenges. While there is no established length of time to try a diet, I think it needs to be longer than 4 weeks. If you look at other conditions where a restricted diet is used as a treatment (such as celiac disease), you will see that while some people respond quickly while others take a good while longer. The only children I would expect to see major changes in four weeks are the children who would most need the diet - and they were excluded from the study.
The last problem is the tests used to detect changes. From what I can see, the tests were all having to do with behavior, sleep , or stool. None of these tests (except the last one) have to do with the biological changes that happen when a child responds to a diet. When we tried the diet with our older daughters, we did so because there were several abnormal results in their blood work and one of the ways that we could tell that the diet was working was that these measures improved. We also saw behavioral changes over time from the diet but I don't know whether any of the measures used in this study would have been sensitive enough to detect the changes - especially in only 4 weeks.
As for the challenges, if the tests aren't sensitive enough to see subtle changes over a 4 week period, how are they going to see the subtle changes from a single dietary infraction? We have had accidental "challenges" over the years and while we can (almost) always tell from short lived behavior changes although I don't know whether someone who didn't know them a well would see the change. But when these "challenges" happen close to the time when they have blood tests done, the result of tests show the "challenge".
The bottom line is that when testing whether the GFCF diet works, you need to find the children that would benefit from the diet, allow a long enough time to see the changes, and have a better way of seeing these changes. As a result, this study does not add anything new to our understanding of the diet.