Tuesday, 13 December 2011

Vitamins and autism: double-blind, placebo-controlled RCT

It is the gold-standard of experimental trial designs, second only to the meta-analysis on the hierarchy of evidence-based practice: the double-blind, placebo-controlled, randomised controlled trial. To you and me it means that neither participants nor investigators knew who was taking what during the trial, the 'what' experimental treatment was compared against a placebo or sham treatment option, and when it came to deciding who got 'what' or placebo, it was a very random affair.

So it was with a new study by Jim Adams and colleagues* looking at the effects of an individually titrated vitamin and mineral supplement for children and adults with autism spectrum conditions. Regular readers will know that Jim Adams has cropped up more than once on this blog with his various research into the more nutritional aspects to autism, or at least linked to some cases of autism. Indeed the current paper rounds off quite a research publication intensive year for him.

The study. Well it is the best kind of study (open-access) with results which are difficult to ignore or put down purely to methodological issues given the study design and participant group size. To summarise:

  • The trial was registered with the NIH Clinical Trials directory (here). Recruited participants diagnosed with an autism spectrum condition formed either an Arizona or National group reflecting geographic status but also whether or not a series of medical exams looking at nutrition and metabolism were carried out. 72 participants split between the Arizona and National groups were allocated to a vitamin-mineral supplement group and compared with 69 participants receiving a placebo designed to look and taste the same as the active supplement.
  • The supplement itself contained quite a bit including things like MSM. Most vitamins and minerals were within current RDA guidelines where data on the typical and upper-limit values were available. A few exceptions were things like vitamin D[3], biotin, chromium, iodine, molybdenum, manganese and zinc which were all at the upper ranges in terms of concentration.
  • The study analysed data from baseline upto 12 weeks on supplement (or placebo) based on a variety of psychometric and biological parameters. The ATEC is mentioned (and its growing popularity). I've talked about the relative lack of good measures to analyse behavioural 'change' in autism before and it looks like the authors used more than one source schedule for their measurements to perhaps compensate for this void in the research literature.
  • The findings. Well there are quite a few of them and it is slightly difficult to say what individual component did what, but here goes: (a) quite a few of the levels of vitamins, the 'b vitamins' in particular, were increased in the experimental group who underwent nutritional testing when assayed for in blood or urine, (b) a general 'normalisation' of minerals compared to control values was observed; most interesting that physiological levels of lithium (+485%) and calcium (+43%) went up alongside levels of iodine (+54%) and molybdenum (+48%), whereas iron levels seemed to drop (-9%), (c) levels of sulphate (total and free), that dusty compound under the bed compound, increased (+83% & +17% respectively) as did levels of glutathione (reduced) (+17%), (d) levels of ATP, NADH and coenzyme Q10 all increased, (e) levels of methylmalonic acid seemed to show improvement (-41%), and homocysteine levels also dropped (-34%), (f) behaviourally, positive effects were noted in areas of hyperactivity, number of tantrums, receptive language and play compared to the placebo.
  • The attrition rates were slightly unexpected in that more participants dropped out of the placebo group (n=11) than the experimental group (n=8) even when comparing drop-outs due to adverse effects (n=5, n=3 respectively). Temporary adverse effects were also reported during the early days of getting the dosages right but these seemed to be short-lived for the majority of participants.

I have already made a few active links in the above summary to posts previously made on this blog in relation to the current findings. Without trying to repeat myself, the interesting things from my point of view were that glutathione and sulphate (sulfate) levels went up, homocysteine and methylmalonic acid levels went down; that and an overall 'normalisation' of various vitamin and mineral levels so important for lots of things including being co-factors for various enzymes. The accompanying behavioural changes also reported for supplemented participants reiterates the fact that participants physical health might just tie into behavioural presentation.

This is a big paper both in terms of content and results and also in terms of potential implications to autism. I know some people might look at the competing interests statement at the end of the paper and see that there are various connections between the supplement suppliers, testing companies and some of the authors. I have to say that this does not worry me so much knowing what it means to register a trial with the NIH and the requirement to publish your results whatever they show. That and the gold-standard trial design instills quite a bit of faith that the results are accurate, at least in the group examined. The next questions: finding out which people with autism might be 'best' responders to such a supplement and looking at the longer term profile of supplementation with regards to safety and outcome. A definite watch this space.

* Adams JB. et al. Effect of a vitamin/mineral supplement on children and adults with autism. BMC Pediatrics. December 2011.


  1. Genuine question: how do they maintain double blindness at the same time as adjusting the dose?

    "Temporary adverse effects were also reported during the early days of getting the dosages right but these seemed to be short-lived for the majority of participants."

  2. Good question Jon and I need to perhaps look through the paper with greater assiduity to see if there is an answer. I suspect (and it is only that) that some kind of study marshall might have been involved as an independent observer.

  3. Interesting article and discussion. I look forward to any follow up on the reply to Jon Brock's question.

    Forgive my ignorance but do study marshall's exist in other studies and can such a study marshall's participation save the double blind status of the test?

  4. Thanks Harold. Another look at the paper seems to offer some guidance in this area. The dosage of both active & placebo supplement was gradually built up for all participants over the course of the first 21 days of study as a function of body weight. So days 1-4 included a sixth of the final dose, etc. up to 100% dose (for those that could tolerate their assigned med). This was done irrespective of whether experimental or placebo was assigned. Given that more children in the placebo group seemed to show adverse effects to their allocated intervention over the active intervention group gives a little weight to the proper way this was managed.
    Protecting the double blind status is always an issue in these kinds of study where dose is not just given as a maximum from the start. The authors however seem to have gone to quite a lot of trouble to minimise the risk of the blindness being broken with regards to doing the same for both groups and other things like same flavouring, same packaging, etc for active and placebo.
    As to the issue of study marshalls, this is not uncommon in such experimental design particularly in the newer 'adaptive' designs where for example, some way into a double-blind trial, you need to take a snapshot of where a trial stands.