A quick note on another homeopathy paper doing the rounds.
Before you read this, take a look at the paper, published in pro-homeo journal, Homeopathy, and then read this blog post at moteprime.org.
As noted in the moteprime piece, this paper is one of the better attempts at conducting a homeopathic RCT. The basic methodology seems pretty sound, and the conclusions are largely borne out by the reported results.
In addition to what Sean at moteprime mentions, I have a couple of remarks about this paper.
EDIT 22.09.10 – two things have been brought to my attention recently:
1) The significance of antibodies in colostrum is not widely appreciated. See 6).
2) Adam (comment 1) has pointed out that treating the piglets as individual, independent data points is not valid, as teat sharing and close proximity within a litter would accentuate the spread of scours within said litter. As a result of this the reported p-value of 0.0001 is extremely optimistic. The N number of 525 piglets is also invalid, as the data should really be taken on a sow-by-sow basis – it was the sows that were treated, after all. So in reality, N=52. The statistical power of this study is not nearly as impressive as what the authors lead the reader into believing.
(Yes – I am kicking myself that I got sucked in and missed this – If I get a chance I might do the numbers on this…)
3) Despite hypothesising that E.coli was to blame, and an E.coli nosode would cure them, no E.coli was detected in the faeces of the pigs – suggesting that maybe another agent was to blame (Clostridia, coliform, viral infection…). Therefore the logic of using an E.coli nosode would appear to break down – but then, I suppose if one only considers symptoms and not causes…
4) Having determined that E.coli were not present in the faeces, the authors state:
Because treatment with Coli30C had worked before, and E. coli diarrhoea generally can be distinguished based on day of appearance and colour, this was not further investigated.
Coli30C (a similar homeopathy nosode remedy of E.coli) working before is not referenced at that point. It is however mentioned in the introduction: (my emphasis)
Homeopathic treatment of E. coli has been studied by Velkers and others (10) in commercial broilers. In that study, broilers were infected with E. coli and treated afterwards with an antibiotic or with different combinations of homeopathic remedies, including a nosode of Coli30C. None of the homeopathically treated groups differed significantly from the controls. In another experiment E. coli nosodes were administrated to calves suffering from scours(11). Here, the nosode treated group did not differ significantly from the control group, but the study was underpowered, due to small numbers of diseased animals in the treatment and control group. Many experiments in the homeopathic field have failed to prove an effect of the treatment. Reasons for that could lie in the methodology of medicine testing as applied in regular medical science, which partly contradicts with the homeopathic philosophy (12).
So previous studies have clearly not shown that Coli30C works. But this must be because regular medical science “contradicts with the homeopathic philosophy” – right?
So having established that Coli30C doesn’t work, and regular medical science contradicts homeopathic philosophy, the authors go on to test homeopathic Coli30K using regular medical science.
5) Another potential source problem is that the exposure of the sows to the diarrhoea causing agent prior to or during is not measured or controlled for. It’s almost as if no consideration was made for variation in the natural immunity of the pigs.
But this isn’t tested, checked, acknowledged or controlled for… there will also genetic factors which may make pigs more/less susceptible to E.coli, but these aren’t checked for either.
6) Amounts of colostrum ingested by individual pigs also has an effect – this is mentioned in the conclusions, but not randomised/controlled for.
Antibodies raised against local pathogens (i.e. those to which the piglets are most likely to get exposed to) are passed from sow to piglet. The amount of colostrum produced by sows and imbibed by piglets is correlated with scours.
7) Birth weight of piglets is also a major risk factor, and this is not randomised/controlled in the trial.
NB) With points 5, 6 & 7 – one might expect these to have a normal distribution through-out the herd of pigs, but unless this is tested for, acknowledged or controlled, they exist as caveats for the conclusions of the paper.
Note also that the authors state – “Independent repetition on different farms with standard preventive treatment against E. coli, is required.” – a caveat which is curiously missing from many pro-homeopathy blogposts about this study.
So, in conclusion, better than many efforts, but still important questions remain unanswered.