Is This Good Evidence For Homeopathy?
Pigs, a spray with nothing much in it, some statistics, and a headache for skeptics.
An Interesting Study
Recently, a paper was published in Homeopathy magazine (Volume 99, Issue 1), entitled Homeopathy as replacement to antibiotics in the case of Escherichia coli diarrhoea in neonatal piglets, by I. Camerlink, L. Ellinger, E.J. Bakker and E.A. Lantinga.
I know, sounds fascinating, but bear with me.
I discovered a link to this by accident whilst browsing a homeopathy website. Usually, links like this contain the typical evidence presented for homeopathy - a litany of poor blinding, placebo effect, rubbish experimental design, and statistics abuse.
But this appears, in fact, an interesting study.
And that presents both a problem and an opportunity for a proper skeptic. So let's take a look at the paper in detail.
Farm pigs, it will surprise no-one to discover, do not live in a sterile environment. Escherichia coli is a constant presence in the farm environment, and it can lead to serious diarrhoea in newborn piglets.
This is a problem since it's bad for the piglets, killing more than 1 in 10, and this in turn is bad for the farmers. The usual solution to the problem is antibiotics, which are expensive and have side effects.
Homeopathic remedies may not be cheaper than antibiotics, but homeopathic practitioners often claim that there are no side effects to homeopathic medicines, so it is attractive to consider homeopathic alternatives to antibiotics.
The homeopathic preparation used in the paper was designed on the homeopathic principles of "like cures like", and thing most like the Escherichia coli bacterium is, well, Escherichia coli itself.
The preparation is shown as Coli 30K, the K of which means that the initial culture of E. coli bacteria has been diluted using the Korsakoff method. A vial of the mother liquor is prepared, with a capacity of 100 drops (about 10ml), which is then poured away, assuming that about 1 drop remains clinging to the inside of the vial. The vial is then topped back up to the 100 drop level, and shaken. This process is then repeated 30 times.
This differs from the 30C dilution, which uses the Hahnemannian method, where a new vial is used for each stage in the process.
A drop of this was then added to a dilute sugar solution. The unadulterated solution was used for the placebo.
Assuming that the mixing at each stage is more or less complete, the dilution over the whole process is 1 part in 1060. This is far larger than the number of bacteria in the original sample. It is, in fact, far larger than the number of molecules in the bacteria in the original sample. In fact, it is far larger than the number of molecules in the Earth. And Jupiter. And the Sun.
The principles of homeopathy, however, dismiss this as mathematical sleight-of-hand since it's the added water molecules that are supposed to remember the original E. coli and produce the same effects. Under this assumption, the actual dilutions are irrelevant.
So, if we ignore the implausibility of the proposed mechanism, the proper question for a skeptic is "does it work?"
So, Does It Work?
In this study, the paper says yes.
52 pregnant sows were selected and randomly assigned to either the control or experimental groups. The groups were similar in number of piglets born, and over a similar period. They were housed together, presumably to ensure that the environment they were raised in was as similar as possible.
Either the placebo or the experimental preparation was administered to the sows by means of a spray to the vulva - this being a soft tissue and, according to the experimenters, a good way to introduce homeopathic treatments to the body.
The general condition of the sows, and the faecal consistency from the piglets, was assessed daily by observers who were blind to the group assignments.
At the end of the study, the experimental group showed a significant reduction in both mild and severe diarrhoea (also referred to in the paper as scours).
I will skip the statistics - it's all in the original paper, and the difference between the groups is quite striking.
So, is this the study that turns all the skeptics into believers? Mmmmmaybe.
But, and it's a big but. Mmmmmaybe not. Although the design seems sounder than any of the other homeopathic studies I have seen, there are still a few confounding factors I would like to see controlled.
The first oddity, in my opinion, was the application of homeopathic remedies to the vulva. This strikes me as unusual. Why not add it the feed of the sows? Homeopathy is described in the paper as activating the healing mechanisms of the body. If this is, as implied, a systemic effect, then the means of application would have little effect. Admittedly, I have little direct experience with pigs, but it would seem to me that from a purely practical perspective it should be easier to administer the dose orally rather than vaginally.
The blinding on the study initially seems adequate - the assessment of the sows and piglets was done observer-blind. This is a single-blind test. The administration of the remedies is not reported in the trial as being blinded, so we have to assume that it was not.
Although the percentage composition of the dose was stated in the paper, the total volume administered was not. Was this a quick spritz with a few milliliters, or a whoosh with a garden sprayer? (Please forgive me if I am getting too technical here.) A large dose is basically washing the pudenda of the sows; surely a procedure that would plausibly have an effect on the bacterial environment of the reproductive tract.
Now that I come to think of it, were the pigs cleaned in any way before administration of the dose? This is omitted.
The combination of these omissions suggests a possible mechanism for unconscious bias to creep in. If the administration was not blinded, differences in dosages or pre-preparation might have been enough to skew the results.
I would like to point out that this is absolutely not meant to imply any conscious bias by the experimenters - Justice Eady take note.
There is another possible counfounding factor. I would like to see if there is any difference between 30K and 30C potencies.
Why do I ask this? Because of another paper, published in Colloids and Surfaces B: Biointerfaces (Volume 50, Issue 1) with the fascinating title of Laser trap studies of end-on E. coli adhesion to glass by Joseph F. Jonesa and Darrell Velegol.
The abstract states that "15.9 ± 3.4% of the bacteria adhered when presented end-on for 15 s to a cleaned glass surface that was not treated for specific interactions. These bacteria were found to adhere either instantaneously (approximately <1 s) or not at all." So, the Korsakoff dilution process for E. coli bacteria may not remove the bacteria from the walls of the initial vial. This may act as a traditional inoculant.
So I say - interesting effect. A reduction of 10% infant mortality rates in farmed pigs is well worth having, but it was not compared to the success rate for the much simpler treatment using preventative vaccination.
The thing about this kind of study is that any experimental bias, however unlikely, is still more likely that the proposed mechanisms behind homeopathy, which have been adequately deconstructed elsewhere. Studies in the field of psychic phenomena suffer from the same kind of problem.
As a result, experimental design has to be very, very careful.
I will admit to not being an experimental biologist, so it's perfectly possible that the safeguards I have identified were in fact either (a) already controlled for in the experiment, or (b) not significant enough to explain the effect.
Anyway, it will be interesting to see if followup studies show that it really is the homeopathy that's causing it, or whether it's all down to the way you pamper a pig's pudenda.
Epilogue - Non Sequitur Alert!
There are a number of interesting statements made in the introduction to the paper, which I think are non-sequiturs.
First, "Homeopathic remedies have significant benefits since there are no residues in animal products". Assuming that homeopathy has an effect, the physiology of the animal has been modified. To state that this has no "residue" - presumably referring to anything that causes effects in humans consuming the pigs - is unsupported. What happens if the homeopathic changes are transmissable to humans? They seem to be transmissable through water, sugar pills, absorption through soft tissue, so why not consumption?
Second "nor does homeopathy generates (sic) resistant micro-organisms." Why not? Assuming that there is an effect, is this 100% fatal to the target micro-organisms? If not, there will be a selective effect, with the surviving bacteria being more resistant the the effect. Even changes to the immune system in the animal will still cause some changes in the population of target organisms.