Monday, November 14, 2005

Homeopathy Cures ADHD

On the Second Opinion guestbook, a study conducted into the effect of homeopathy in the treatment of children with ADHD has been raised a number of times, particularly since it is a "randomised, double blind, placebo controlled crossover study" It is usually in response to mention of the recent Lancet metastudy showing no effect above placebo from the use of homeopathy. Unfortunately, this Swiss study is mentioned in a disengenuous way, and apparently by people who haven't read the study itself.

The study is available for purchase at European Journal of Pediatrics. It has only been published online.

Since the True Believers are wary of mainstream studies, arguing they are funded by Big Pharma and therefore fatally biased and flawed, it is interesting to note that, by this logic, the Swiss study must also be disregarded since, of the eleven authors, one is from the Swiss Association of Homeopathic Physicians and two from the Kollegiale Instanz für Komplementärmedizin (KIKOM)/Homeopathy. Oooh! Talk about conflict of interest! (EoR doesn't necessarily hold to this argument, but dismally hopes the alternatistas will stand by their unmoveable principles and condemn this study accordingly).

The publicly available abstract concludes

"The trial suggests scientific evidence of the effectiveness of homeopathy in the treatment of attention deficit hyperactivity disorder, particularly in the areas of behavioural and cognitive functions."

So what did the study involve? How was it actually conducted? And what are the real results?

In the first phase of the study, children (aged 6-16 years, both genders) openly received individually prescribed homeopathic remedies (remember that homeopaths argue that there is no specific prescription for any specific disease, but that the cure must be prescribed specifically to an individual based on that individual's symptoms - this of course ignores that the homeopath must still be using some system of principle(s) to determine the remedy - "according to the guidelines described by Hahnemann and Boenninghausen" according to the authors - if it is not assumed to be random choice). Only children who showed a positive response were allowed into the subsequent part of the study.

"Patients must reach an amelioration of 50% of the initial CGI value or at least 9 points during the screening phase."

The mean period to achieve this cutoff was 5.1 months (range 1-18 months). The CGI or Connors Global Index is a subjective assessment of ten items rated from 0 (never) to 3 (very often) ie 4 points. Therefore this does not appear to be a random sample but a specifically selected sample with a pre-existing inherent bias (more about how this can skew results later).

The sample was then split in two. One half received verum for six weeks, and then placebo for six weeks, while the other half received the placebo initially, and then the verum. This was performed as a randomised double blind trial. Both groups then continued receiving open label treatment for six weeks. This six week period was determined by a previous study involving only four(!) children.

An assumption made was

"A reduction of 5 points [in the CGI] from pre-treatment value due to verum was considered clinically relevant and a zero point reduction was assumed under placebo."

This seems to assume that the placebo effect is not measurable, something which would be a surprise to most researchers. It should also be noted that

"patients who dropped out after the first crossover period could also be included in the analysis by assuming missing at random"


"For other types of analyses, patients with missing values were excluded."

62 children participated in the study. 8 dropped out "due to insufficient response to homeopathy." A further 5 dropped out due to "compliance problems."

"Reasons for drop out were increasing tics, behavioural problems and a reactive depression."

Therefore, only 49 children completed the study. Statistics presented represent all 62 children though.

Some of the more interesting findings include:

"Within-patient comparison of treatment effect shows that the CGI, the primary endpoint, decreased under verum with a mean of 1.67 points."

Bear in mind this is based on a scale with 4 points for each parameter. A parent changing a result from a 3 to a 2 has scored a 25% reduction!

"Arm A: unexpected rise in CGI during verum phase, possibly due to expectation of receiving placebo in first phase. Persistent high CGI with placebo in second phase, normalisation of CGI with open label verum after crossover trial. Arm B: CGI rose with placebo in first phase and returned to CGI values within the normal range with verum treatment in second phase, as expected."

So both groups (placebo and verum) scored poorly to begin with due to "expectations"? Isn't that the same as saying the effect was a placebo? Didn't homeopathy actually fail here, or am I missing something?

"Comparison of the scores CPRS scores between start of treatment and 14 weeks after crossover trial still revealed highly significant improvements in all subscales, in both mothers' and fathers' ratings. Teachers ratings (CTRS) showed a significant improvement in behaviour, and a trend in improvement of the CGI. Impulsivity/hyperactivity and passivity were improved, but did not reach significance in the teachers' observations."

Here's another problem: the assessment scale used to produce the miraculous results of homeopathic power are subjective. Teachers (who presumably have less emotional investment in the child) observed less improvement than parents.

"The latest follow-up during open label treatment showed a maximum [my emphasis] amelioration of 12 points (63%)"

This reflects an improvement from an initial CGI of 19 (range 15-25), to 7 (range 2-15). This is the other major problem with this study, that small numbers of participants, and small scales of measurements can easily produce massive changes in percentage ratings. And notice that important word "maximum". In the discussion section, the authors state

"CGI decreased by only 17% (1.67 CGI points), somewhat less than that expected from the results of an earlier study."

So does this study in fact prove that homeopathy is less effective than expected?

Apart from the subjective parent and teacher assessments, a large number of cognitive tests were also performed on the children. The results of these were more mixed:

"positive: improved auditory short-term memory, a trend to increased stability of mood, better reactions to unexpected events; negative: decrease in alertness to visual details and the trend towards decreased visual spatial organisation"

and the researchers conclude that

"the CGI difference between placebo and verum was considerably smaller than expected."

Presumably, then, homeopathy is actually counter-productive to certain aspects of ADHD? Are the proponents of homeopathy publicising this? No.

The penultimate paragraph is significant:

"The CGI and the CPRS scores (parent ratings) decreased between 37% and 63% over the long-term observation period, most probably related to treatment induced adjustment of behaviour. The CTRS improvement ratings by teachers were smaller (between 28% and 37%) than those of the parents, reflecting the higher cognitive stress for patients in school situations. However, on the whole, the overall intensity of ADHD symptoms appears to be lower during treatment and results in an improvement in the childrens' social, emotional and scholastic behaviour. The question whether these long-term improvements are a treatment effect or merely due to a spontaneous change in development of the children cannot be definitely answered by our trial data."

Bearing in mind the effect of subjective assessment (higher by parents, lower by teachers) and the scaling effect of changing a low scale rating to a percentage, we get an improvement of 28% to 63%? A vast range indeed. Surely not an artefact of a small sample size? What would the placebo effect on its own be?

It should also be noted that the change in CGI touted by the authors as so large reflects the change before the children were selected for entry into the trial to the end of the actual trial. The children who actually entered the trial had a CGI of 8 (range 3-16) for one group, and 9 (range 4-20) for the second immediately prior to commencing the study. The difference over the period of the trial to a CGI of 7 (range 2-15) at the end can therefore be seen to be statistically insignificant.

So what this study actually showed is no significant change in CGI from the start of the trial to the end of the trial, though during the first six weeks there was a significant rise in CGI (for both groups on verum and placebo ie no difference whether the child was on a placebo or a homepathic treatment). The largest reduction in CGI occurred during the open label treatment (ie not randomised or blinded).

The authors conclude

"The results of this trial point to the effectiveness of homeopathy in the treatment of ADHD. To corroborate the findings presented here, the authors suggest a larger and independent multicentre study."

Well, I guess it points to the effectiveness of homeopathy as much as any small scale statistically doubtful trial could. Though it tends to point more to the desperate clutching at straws of the True Believers for any slight indication that homeopathy just might be something more than a placebo effect. Unfortunately, this study proves nothing else. I wonder what the larger scale trials have shown... Oh, hang on, we already know. Waves copy of Lancet threateningly at hordes of angry homeopaths...

1 comment:

  1. Thanks for this detailed analysis. I have seen the article in question held up by many homeopath people, and I have not had the time to look up the article yet. Your trenchant analysis has saved me quite a bit of time, though I will read the article eventually myself.

    The study design as you describe it certainly belies the conclusion of the authors. It is shameful that the editors of this the European J of Pediatrics allowed this flawed research into their publication.


Note: only a member of this blog may post a comment.