As Dr Crippen notes, Patrick Holford has taken it upon himself to educate the British Dietetic Association (BDA) on the benefits of dietary interventions for autism. On this blog, Shinga has also analysed Holford’s wisdom on this issue. I’m also going to look over some of Patrick Holford‘s ‘evidence’ base on this.
If you’re going to take it on yourself to lecture a learned body like the BDA, you had better make sure your research stands up to scrutiny. Sadly, the ‘evidence’ that Holford provides for a gluten free casein free (GFCF) diet to treat autism doesn’t stand up to even cursory scrutiny.
Holford’s first piece of autism-specific evidence is a link to Robert Cade’s work. Unlike in other Holford work, the link works this time – I suppose one should give Holford some credit for this. However, he doesn’t get any credit for the quality of this ‘evidence’.
The Robert Cade piece that Holford links to is a short letter on what reportedly happened to some children on the autistic spectrum when they followed a GFCF diet. This isn’t a journal article (or even a letter published in a reputable journal – the letter is reproduced on the paleodiet website). The letter hasn’t been peer-reviewed, and there’s relatively little detail on the methodology. This lack of detail on methodology – along with the lack of peer review – means it’s pretty much impossible to know whether the study referred to was any good. It is hard to see why Holford would want to reference this letter anyway – there have been journal articles on the GFCF diet, so why not reference the academic literature instead of a letter on the paleodiet website?
Holford then cites Jynouchi et al’s paper on inflammatory reactions in people on the autistic spectrum. For Jyonouchi et al, “Immune reactivity to [dietary proteins] may be associated with apparent [dietary protein intolerance] and GI [gastrointestinal] inflammation in ASD children that may be partly associated with aberrant innate immune response against endotoxin, a product of the gut bacteria.” These are interesting ideas and, if people on the autistic spectrum do suffer GI problems then no-one – least of all dietitians – would deny that they should be provided with good quality medical treatment. However, without even starting on the problems with Holford’s conception of IgG reactions, Jyonouchi et al’s paper has nothing to do with Cade and Holford’s apparent belief that GFCF diets can help to reduce “autistic traits” (unless you define GI problems as an ‘autistic trait’).
Holford then cites the Cochrane review of GFCF diets for autism. Holford states that:
A review of wheat and dairy-free diets in relation to reducing autistic traits reported “a significant beneficial treatment effect for the combined gluten- and casein-free diet”.
The Cochrane review is an excellent source – but Holford has chosen an odd way of representing its conclusions.
As part of a commendable attempt to make science accessible to non-experts, the Cochrane review has a nice, clear plain text summary:
Extensive literature searches identified only one randomised control trial of gluten and/or casein free diet as an intervention to improve behaviour, cognitive and social functioning in individuals with autism. The trial was small scale, with only 10 participants in the treatment group and 10 participants in the control group. Results indicate that a combined gluten and casein free diet may reduce some autistic traits. This is an important area of investigation and large scale, good quality randomised control trials are needed.
This gives a very different impression to – and contains far more qualifiers than – the section of the Cochrane review that Holford quotes. Holford chooses to quote from the review’s abstract, instead of the plain text summary. The results section of the abstract states that
The one trial included (n= 20) reported results on four outcomes. Unsurprisingly in such a small scale study, the results for three of these outcomes (cognitive skills, linguistic ability and motor ability) had wide confidence intervals that spanned the line of nil effect. However, the fourth outcome, reduction in autistic traits, reported a significant beneficial treatment effect for the combined gluten and casein free diet.
In other words, there is some, small-scale, evidence of a statistically significant effect on one out of four measured outcomes. It would be easy to get a different impression from Holford’s statement that the Cochrane review finds “a significant beneficial treatment effect for the combined gluten- and casein-free diet”.
The other autism-specific journal article that Holford provides is Green et al’s “Internet survey of treatments used by parents of children with autism“. I’d imagine that – to Holford Watch’s intelligent, sensible readers – many likely problems with this survey will be apparent from the title, but I’ll give a few details here.
As article points out, “the data on most commonly used treatment must be interpreted with caution. Common use may not necessarily indicate preference for the treatment.” And, of course, treatment preference doesn’t necessarily demonstrate efficacy – ineffective and dangerous treatments can sometimes become popular because of, for example, skillful marketing.
Talking about ineffective and dangerous treatments for autism – some of the statistics in Green et al’s article are alarming. They find that 7.4% of children were being chelated and 1.6% were having secretin used on them (down from an alarming 11.3% who had had this drug used on them in the past). It’s odd that Holford chose not to take this opportunity to lecture the BDA on how “valuable [and] naturopathic in concept” chelation therapy is (or its occasional fatalities), or to argue again that the off-label use of the drug secretin is “Worth considering“.
At any rate, I hope that the 20-odd percent of parent who Green et al found had put their kids on a GFCF diet had accessed a better evidence-base than that provided by Holford before they did so. Bear with me – we’re almost at the end of Holford’s attempt to educate the BDA. His final piece of ‘evidence’ is a link – again, not broken, so he gets some brownie points for that – to what looks like a series of e-mails about casein and gluten. Surely Holford doesn’t expect dietitians to base their decisions on how to treat vulnerable kids on a handful of e-mails reproduced online – does he?
To summarise Holford’s evidence on GFCF diets for autism, then, he uses this article to tell the BDA about:
- One letter describing a small study (not published in a journal, and without enough information on the method for one to assess the quality of the research).
- An article which suggests that there may be an increased incidence of GI problems and food intolerances in autistic children – which, even if true, does not show that a GFCF could ‘cure’ any children on the autistic spectrum, or that those children without gluten/casein intolerances should follow a GFCF diet.
- A somewhat misleading quote from the Cochrane review of GFCF diets and autism.
- An article on an Internet survey on the treatments that parents use on their autistic spectrum kids.
- A link to what looks like a collection of e-mails, some of which relate to GFCF diets and autism.
I find the idea that anyone would impose dietary restrictions on vulnerable children, basing their treatment decisions on such flimsy evidence, rather worrying. Dietary interventions are certainly not risk-free – or particularly pleasant for the kids involved – so you don’t want to impose them without good reason. As the BDA notes:
Although the perception is that dietary change is much safer than the use of medications,
excluding foods without making sure the diet stays balanced can cause dietary deficiencies,
weight loss and poor growth. The risks increase as more foods are excluded, and children who
are extreme faddy eaters can be at high risk of having an inadequate diet.
There is some interesting evidence that GFCF diets may benefit some people on the autistic spectrum: as the BDA puts it, the evidence on this topic is “inconclusive”; I would therefore support the Cochrane review’s recommendation of further good-quality research. Holford’s analysis of the literature on GFCF diets and autism is, however, sufficiently poor that it is not a useful contribution to the debate. I very much doubt that the BDA will be impressed by Holford’s attempts to enlighten them. I very much hope that healthcare practitioners will not use this ‘evidence’ as a basis for treatment decisions.
Towards the end of his article, Holford quotes from (though, oddly, chooses not to link to) Simon Crompton’s excellent critique of Food is Better Medicine than Drugs. In this piece, the dietitian Catherine Collins argues that “It’s unfair on people to say that dietary treatment can be anything more than an adjunct to medicine when it comes to serious conditions”.
Holford thinks that Collins is being too dismissive of his work. However, having waded through way too much of Holford’s writing on ‘nutritional’ and other ‘alternative’ treatments of autism, I am not convinced that Holford’s approach to autism is even a useful adjunct to more ‘conventional’ support and treatment.
Dietary treatments for conditions such as autism may be valuable when carried out by qualified healthcare practitioners. However, the thought of children on the autistic spectrum being treated by ‘nutritionists’ drawing on the type of ‘research’ analysed above, on Holford Watch ,and elsewhere in the wider Internet frankly fills me with horror.
Holford asks, rhetorically, whether “showing this kind of evidence will shift…thinking about optimum nutrition”. I very much hope that it doesn’t, and that qualified dietitians continue to draw on a much superior research base.
Update: Feb 2009. Chelation is not a risk-free procedure. There are clear indications that popular chelation interventions for children with autism may be contributing to neurological damage. It is long past time for Holford to change his advice and cease to recommend such dubious procedures.