There are two principle research applications of the Randomized Controlled Trial (RCT), in proving efficacy of homeopathic practice. Trials measuring effectiveness of homeopathic treatment of diseases are the most familiar and natural-seeming method, but are nevertheless the more difficult of realization. The second is the proving trial, which focuses on the narrower, and presumably easier to judge question whether the homeopathic remedy has any effect of any sort on the human organism. In short, the proving trial is not directly concerned with healing effects of homeopathy, but with the more foundational question, whether there is anything substantial, or “real,” in homeopathic remedies in the first place.
In Part I, following, I will briefly discuss the homeopathic treatment trial, primarily with an aim to identify a few of the key issues that, from the homeopathic point of view, have invalidated trial outcomes in the past. I will not pursue this thorny topic very far, but will be satisfied to identify a few problems that have characterized research trials conducted in this venue to date, in hope that work may begin, in appropriate quarters, to meeting these concerns and to developing more adequate research instruments.
In Part II, the main question of this paper will be addressed, namely, how to design an efficacious trial, that is, an RCT that can claim to measure accurately the effects of homeopathic remedies upon trial participants. Although it is normal to speak of a trial measuring the “efficacy” of a medication, I propose that we are justified, in view of the consistent failure of RCTs to measure any effect in homeopathic remedies, to question the “efficacy” of the RCT itself. This section will conclude with a recommendation – astonishing in its simplicity – that, frankly, should guarantee unmistakable evidence, in fully replicable experimental protocols, in proving trials of the future.
Finally, in Part III, a number of additional questions will be addressed, that should be kept in mind in designing an appropriate RCT. Together, these considerations, with the recommendations included in Part II, provide a range of parameters that should satisfy any critic, that the perspicacious researcher has fulfilled his obligations, to design and implement a credible scientific trial of homeopathy, by combining – at long last – academic and logical scruples, with his technical and statistical finesse.
Part I – Research in Homeopathic Treatment
The problem conducting research into homeopathic treatment is at once easy to define, difficult to understand, simple to rectify in principle, and intractable of resolution in practice. The problem can be stated with reference to the by now trite idea, that homeopathy operates according to a paradigm that is different from the paradigm of conventional (“allopathic”) medicine, and cannot be measured utilizing the same research instruments. This seems the most natural statement of fact to the homeopath, but is infuriating to the research scientist, who protests that, if a school of medicine claims to heal, by whatever means, then it ought to be possible to observe the fact that the patient has been healed – which is all the RCT wants to accomplish.
In fact, the researcher’s objections are perfectly reasonable. But so are the reservations expressed by the homeopath. And the two positions are not even difficult to reconcile, at least in principle, though once again, the hydra headed monster, practicality, laughs implacably at our efforts to find a direct route to a satisfactory conclusion.
Commonly, those skeptical of homeopathic claims will say something to the effect, “let’s design a trial to show that homeopathy can cure high blood pressure.” Of course, any other disease state can be substituted. Immediately, the rejoinder is heard, that homeopathy treats the whole person, that prescriptions in homeopathy are made upon the patient’s totality of symptoms, and not upon a discrete pathological entity. The homeopath concludes, therefore, that it is impossible to test homeopathy in reference to a discrete diagnosis.
However, if one considers choice of words more closely, it will be realized that the conventional, or “allopathic” diagnosis, whatever clinical entity is chosen, is but a different way of referencing what, from the homeopathic point of view, would (or could in some cases) be termed the patient’s “chief complaint.” And, in fact, in homeopathic treatment, therefore, one would expect that the chief complaint, or “presenting diagnosis,” would be resolved, or cured, by the end of a successful course of treatment. The problem arriving at that end, in a research trial, is simply that the trial is not individualized.
This is easy to overcome, as by now has often been observed: select a diagnostic entity to treat, gather a sample population, and allow the homeopaths to treat their patients according to their own methods, individualizing to their heart’s content. In fact, this should work, though it has not as yet been implemented in a way that would truly satisfy the requirements of individualization, in homeopathic practice. In short, to be successful, there are at least three main requirements, in designing such a trial that must be met, to establish efficacy of the trial, that is, to establish that the trial protocol is sufficient to the task of measuring homeopathic action:
First, the homeopaths participating in the study must be adherents of classical methods, of Hahnemannian homeopathy. Such practitioners would pass the ‘test’ of mainstream practice, in adhering to the four basic tenets of classical homeopathy: single remedy, single dose, minimum dose, wait and watch. The reason for insisting on this point, is that there are by now so many variants on classical methods, that we must focus clearly on our task: if we are testing “homeopathy,” then we are not testing combination remedies, preventative homeopathy, homeopathic remedies as specifics, or any number of other derivative practices.
This is not to suggest there is no value in such practices, only that, practically speaking, a research trial will show different treatment results, if different treatment methodologies are used. Now, granted, it may be worthwhile to do research into these other methods, and the variable uses to which homeopathic remedies may be put, but that is not the same as testing homeopathy. Simply put, our ideal research trial must define its terms rigorously, and test what it says it wants to test.
Second, there can be no limitation upon the homeopath’s choice of remedy. Often, trials such as these are published, with a view to demonstrating efficacy, for example, of Arnica in treating pain. But this commits the cardinal sin of treating a homeopathic remedy as a specific; in this circumstance, even though Arnica is, indeed, well known for its successful application to treatment of pain in some circumstances, it is far from the only remedy to consider and, furthermore, its ultimate selection must be made on the basis not only regarding the chief complaint, but the totality within which that complaint is embedded. Plainly, it must be understood, that Arnica is not as “good,” if by that we mean “reliable,” as aspirin, in resolving complaints of “pain;” but, homeopathy is certainly as good as “aspirin,” that is, if the prescription for pain is individualized, then we have a right to expect success in as “good” a percentage of cases as we get from aspirin, though that will mean we are not limited to prescribing Arnica, but can draw freely from the homeopathic pharmacopoeia.
Even here, however, hydra-headed individualization rears it’s ugly countenance: the individualized prescription is more likely to be wrong, than a prescription of an allopathic drug that was, after all, specifically designed to address (suppress) the precise symptoms in question. So, in this circumstance, an efficacious trial must permit the homeopathic prescriber additional time to follow and possibly re-evaluate the case, for a second or even third prescription. Needless to say, this creates interpretive problems, as with an increase in time, more patients are, in fact, likely to show improvement just in the nature of the disease.
Third, and finally, there can be no limitation upon the length of the homeopathic treatment. This is true in all circumstances, but more so when the trial is attempting to measure treatment of a chronic problem. The longer standing the illness, the more the patient has been exposed to traditional, suppressive medical interventions, the older the patient, the more complex the case, then very likely, the longer treatment may last, even into years. This is not a circumstance that would be looked at in kindly lights, by those wishing to either fund or implement a research trial. But the fact remains, in homeopathy, the presenting complaint may be the first thing mentioned – and the last thing cured. But individualized treatment means, simply, “individualized treatment,” and that means it lasts as long as it takes.
This introduces obvious problems: financial; organizational – holding the research team together for such a sustained effort; participatory – being able to count on the trial participants to “stay the course.” All of these considerations suggest that such a trial would be on shaky ground to begin with. They suggest, further, that to the extent such a trial might be considered, it would be advisable to begin with a study of treatment of acute conditions in a young and generally healthy population. This does not mean that all treatment of all participants can be expected to be completed within 30 days, or even 6 months, but at least could be counted on to lessen the average duration of treatment.
Even so, the research scientist, who may not be well versed in clinical practice, should understand at the beginning, that even acute disorders can take root in compromised soil, becoming part of a pre-existing chronic disorder, or susceptible constitution, and this can easily lead to lengthy treatment.
Such practical considerations cannot be avoided, even though the researcher complains that they raise potentially insurmountable obstacles. It is the nature of the beast. It can be measured, but not without considerable effort, insight, and patience. Of course, another implication is the fact that the homeopath, “treating” the control group, may after awhile conclude that his patient is in fact taking only placebo, from the lack of any clinical response, good or ill, over a sustained period of time and numerous prescriptions. In this circumstance, it is possible that the trial will come to an early end, by identification of the placebo group, and this, in fact, could form the basis for an interesting, testable hypothesis of its own!
In any case, it is fortunate that these complications do not represent the only avenue available, in the pursuit of statistical evidence for homeopathy.
Part II – The Homeopathic Proving Trial
Once properly understood, there is nothing complicated nor difficult standing in the way of a proving trial for a remedy, that will show favorably for effects of homeopathy. Not the only element, but the key to the matter is this: size of dose. Indirectly, Hahnemann points us in this direction in Aphorism 32: “Every real medicine … acts at all times, under all circumstances, on every living human being, and produces in him its peculiar symptoms (distinctly perceptible, if the dose be large enough), so that evidently every living human organism is liable to be affected…” (Emphasis added).
Of course, in the treatment situation, the dose is kept small in order to minimize symptomatic response to the medication (aggravation). Similarly, in provings, the dose starts out small, and is given, according to Hahnemann, “…on an empty stomach, daily from four to six very small globules of the thirtieth potency of such a substance, moistened with a little water or dissolved in more or less water and thoroughly mixed, and let him continue this for several days” (Aphorism 128). Note, however, that already, in the proving situation, Hahnemann has increased the size of the dose, which in treatment situations, as we see repeatedly, he recommends use of a single globule only.
Further, if there is no response to the initial dose, in a proving trial, Hahnemann recommends that “If the effects that result from such a dose are but slight, a few more globules may be taken daily, until they become more distinct and stronger and the alterations of the health more conspicuous…” (Aphorism 129). And when the purpose of a proving is simply to identify the symptoms associated with a substance, then Hahnemann is even more specific, in stating that “…in that case the preferable course to pursue is to give … [the remedy] for several successive days, increasing the dose every day” (Aphorism 132).
In short, the utilization of relatively large doses, repeated often, is the critical feature of homeopathic provings, as compared to treatment process, and forms the kernel around which credible research trials into homeopathic provings should be built. Note that Hahnemann is seeking more dramatic symptoms, “more distinct and stronger” symptoms, quite the contrary to practice in homeopathic treatment.
This, of course, is nothing new for the clinician, who realizes that the smallness of dose in treatment is designed to reduce aggravations, while the relatively large size of the dose in a proving is designed to increase them, that is, the proving is specifically designed to create a symptomatic response. In the professional proving, the reason for this is to discover the medicinal qualities of a substance. But in the research trial, the purpose is, just as obvious, to demonstrate that the remedy has a real, physical effect on the human organism. Practically speaking, increased size of dose to 2-3 teaspoons of liquid remedy, repeated 3-4 times a day for perhaps a month, should certainly suffice to overcome resistance of even the most non-responsive of provers. But that is an extreme, even a ‘worst case’ scenario, which I have created for emphasis; more than likely, much less will be required.
Safety. But this brings us around again to the question of individualization, to which I will turn momentarily. First, however, I should say that the preceding considerations reflect the likelihood, as far as I can tell, that by incorporating the idea of increasingly large doses into protocol design, we have done enough to guarantee that the outcome of a proving trial shall show favorably for homeopathy, that is, the remedy will outperform placebo in production of symptoms. The only caveat, is that the administrators of the trial, and of course the provers themselves, be aware of the possible need to increase the size of the dose very considerably beyond normal levels. This, in turn, introduces questions of potential severity of symptomatic response in provers, and therefore their health and safety. This consideration, therefore, suggests the need to introduce an extra measure of supervision to the trial proceedings, to monitor participant response in the event of untoward reactions, as well as ethical considerations, which must be addressed first.
Part III – Other Considerations
For a number of years, there was only one man in the world, who was subjected, or subjected himself, to medicinal provings, and that, naturally, was Hahnemann himself, in the early years of evolution of homeopathy. As time progressed, he collected around him a small group of gentlemen that assisted him throughout the subsequent years in proving new remedies. It was a small group, that Hahnemann found to be conscientious and skillful in the work, and who possessed qualities he considered characteristic of an effective prover: “…temperate in all respects, [and] of delicate feelings…” (Aphorism 137).
Proving Symptoms and the “homeopathic” prescription.
Critics of homeopathy often express confusion that a tiny dose of a homeopathic remedy is enough to cause an aggravation in a patient, and to cure disease, but that in a proving trial, dramatic symptomatic displays have never been observed. Simply put, the effects of homeopathic remedies appear to evaporate, as the remedy performs no better than placebo, in provoking proving symptoms.
Two simple observations suffice to clarify the apparent dilemma, which turns out to be based in a lack of understanding of homeopathic principles and practice. First, the responsiveness of the patient is to a remedy that has been selected specifically for him – in short, the prescription has been individualized. In the proving trial, on the other hand, needless to say the remedy has been “prescribed” (administered) to the entire group, and may simply not be a match for many of the participants.
Second, and perhaps more to the point: although the changes in condition of health of the patient, in treatment, may be dramatic, they are achieved by application of the famous minute homeopathic dose, so that aggravations are usually very mild, barely even perceptible. Remember, that a key in homeopathy is to produce a rapid, but “gentle” cure – which means that effects of medications may be subtle in their appearance, or even barely perceptible.
A result of this is that, in the proving trial, if there is any slack in efforts to harvest symptoms rigorously, some of the subtle signs of remedy action may easily be missed. I hasten to add, though, that as with most of the points I raise in this paper, this may not be a major hurdle to showing action of homeopathy; but failure to address the question will nevertheless, with certainty, introduce at least a small additional measure of doubt, regarding the satisfactoriness of the research methodology used to judge homeopathy.
The layman (including the research scientist) should keep in mind, in this regard, that homeopathic remedies operate on the principle of “like cures like,” which is to say, they are active primarily when applied to a patient (or, in fact, a prover) who happens to possess characteristics (symptoms) that are similar to the characteristic action of the remedy. This is markedly different an effect, than found in conventional, or “allopathic” medications, which are designed to act on a specific symptom or set of symptoms, regardless the characteristics of the patient himself. This creates a much more direct and satisfying access to an efficacy trial for conventional medication, for the researcher knows precisely what to look for in each and every case.
Harvesting symptoms. In the homeopathic research trial, harvesting symptoms becomes a key issue, for the two reasons alluded to above: variability of the individual response to the remedy, and mildness of effects. Note, for example, the lengths to which Hahnemann goes, in his provings, to ensure identifying symptomatic responses:
On experiencing any particular sensation from the medicine, it is useful, indeed necessary, in order to determine the exact character of the symptom, to assume various positions while it lasts, and to observe whether, by moving the part affected, by walking in the room or the open air, by standing, sitting or lying the symptom is increased, diminished or removed, and whether it returns on again assuming the position in which it was first observed – whether it is altered by eating or drinking, or by any other condition, or by speaking, coughing, sneezing or any other action of the body, and at the same time to note at what time of the day or night it usually occurs in the most marked manner, whereby what is peculiar to and characteristic of each symptom will become apparent (Aphorism 133).
Hahnemann specifies further that the prover “…must note down distinctly the sensations, sufferings, accidents and changes of health he experiences at the time of their occurrence, mentioning the time after the ingestion of the drug when each symptom arose and, if it lasts long, the period of its duration. The physician looks over the report in the presence of the experimenter immediately after the experiment is concluded, or if the trial lasts several days he does this every day, in order, while everything is still fresh in his memory, to question him about the exact nature of every one of these circumstances…” (Aphorism 139).
Compare this process, to the relatively casual approach to harvesting symptoms in one recent proving trial of belladonna:
The study medication was taken twice daily…. Subjects recorded, daily during the study, any new symptom that could not be attributed to any other cause, or any exacerbation of pre-existing symptoms…. Subjects were telephoned weekly to monitor …” progress. *
Aside from the infrequent professional monitoring (once a week), note that symptoms were to be excluded from the subject’s diary, if the subject felt the symptom could be accounted for by “any other cause.” One is forced to withdraw to sarcastic wonderment, and to inquire what diagnostic criteria these trial participants employed, in determining, differentially, whether a particular response was due to the remedy or to some other cause. Such distinctions, after all, are not necessarily easy to make even for the experienced professional; leaving them in the hands of untrained trial participants is not a “technique” well calculated to encourage confidence in the ‘scientist’s’ findings. In addition, subjects were substantially limited in their reporting to a selection from 5 symptoms on a provided checklist, drawn from the established proving record of Belladonna. But, as one example of that record, we have the Materia Medica Pura, in which Hahnemann lists 1,440 symptoms.
It seems to me, in all frankness, that findings presented in such a report, are hardly to be credited with even a sneeze in recognition.
In consideration of the sharp debate in the scientific community, as between research statisticians and homeopaths, regarding the credibility of findings from research trials such as this, it is prudent to insist on the most stringent adherence to the most conservative guidelines of clinical practice, as the standard against which methodology in trials are judged. True, I agree with the opinion that has it, that such absolute rigor will ultimately be found to be unnecessary; but until the time when the quality of these trials have improved sufficiently, as to enable them to measure responses to homeopathic remedies in the first place, as much should be done as is practically possible, to eliminate concern for any possible influence from as many potential confounding factors as we can.
Sensitivity. In Aphorism 281, Hahnemann notes “There are patients whose impressionability compared to that of the insusceptible ones is like the ratio as 1000 to 1.” In practice, this leads to a wide variability, in how much of a remedy constitutes an effective “minimum dose” for one patient as compared to another. The first patient may have, for example, a strong reaction to merely sniffing at the opening to the remedy bottle, and taking none of the physical remedy itself, while his neighbor might have but little reaction to 3 teaspoons of the liquid.
In the treatment situation, the homeopath is able to exercise his professional judgment, and modify the size of the dose according to the sensitivity of the individual patient. This helps to assure consistency of response, from one patient to the next. In the proving trial, on the other hand, all subjects are treated alike, and this means that everyone, the least sensitive to remedies as well as the most sensitive to them, will be given the same, relatively small dose. To one degree or another, this built-in limit imposed on size of dose, will necessarily have an impact on reducing possible numbers of symptomatic responses in the verum group, as, past a certain, conjectural point, the most unresponsive of subjects, as well as some of those toward the middle of the continuum, will never receive a dose large enough to cause a reaction.
Obviously, one way around this problem is, as I have already suggested, building in a progression in dosing routine, so that non-responders are given increasingly large doses. But another possibility is to select as trial participants, subjects whose profile suggests that they are at the more sensitive end of the spectrum. On the one hand, this provides better assurances that a larger percentage of subjects will show proving symptoms; on the other hand, it simply moves the risk, of untoward reactions, further down the continuum, as the non-responders are eliminated from the trial group. In any case, this strategy has the other advantage, of, presumably, permitting a larger percentage of responders while the dosing routine is still at a relatively small size of dose. In this way, any untoward reactions might at least be minimized in frequency and severity.
Other considerations. Other considerations may be mentioned briefly, including effects of diet and other antidoting factors, especially consumption of drugs and conventional medications, in reducing production of symptoms in the verum group. Of interest, is the possibility that these same factors could actually increase production of “placebo response” in the control group! After all, if a substance, such as even spicy foods much less a strong medication, effects an individual in the control group, any symptoms he produces will be interpreted as placebo, thus, effectively, improving performance of placebo, while its antidoting effects serve to reduce performance of verum! Calculation of this effect is complicated by the fact that, among verum subjects, some will produce both real and “false” (stimulated by “antidoting influences”) symptoms, but the ironic fact remains, that this consideration actually favors placebo statistically.
As with other points I have raised in this paper, I do not mean to over dramatize the importance of this effect, which I suspect would be of marginal significance, at best. Still, when there are numerous factors, each of which are suspected in compromising the integrity of research, to one degree or another, the cumulative impact of multiple confounders can be significant, even if the effects taken individually are negligible.
I have argued in this paper that the key element in constructing a successful research trial, proving efficacy of homeopathic remedies, is size of dose. I remain uncertain how far a variety of confounding factors might limit trial group reactions, but I operate under the general assumption that, if the size of dose is given a broad scope, it will show favorably for homeopathy under almost any conditions. However, the important caveat, attached to this idea, is that the more liberty one gives to the research team, to administer larger and more frequent doses, the more risk one runs of harmful effects on trial participants. This has obvious practical and ethical implications that will need to be addressed, in each new trial design.
The major problems identified above, in research trials carried out to date, are, in summary: failure to identify size of dose as the premier factor stimulating symptomatic response to remedy; failure to understand individual sensitivity of subjects as a limiting factor in verum response as a group; failure to adequately harvest proving symptoms; failure to control for antidoting influences; underestimating the difficulty of arranging a viable proving trial, due to failure to understand differences implicit between homeopathic treatment and proving methods – the minimum dose of treatment is not optimal for the purpose of stimulating symptomatic response in the proving trial; and minimizing the need to scrutinize protocol design in the first place, due to a misplaced confidence in the idea that, an instrument of “proven” value in measuring efficacy of conventional medicines could be considered, ipso facto, of equal value in measuring efficacy of homeopathic remedies. It has been an important objective in this essay, to clarify some of the problems with the last assumption, and I will consider this paper successful if it has contributed even a little to clarifying problems inherent in that point of view.
Needless to say, none of the preceding puts homeopathy outside the range of RCT. Yet, these observations are intended to highlight that statistical research into homeopathic effects is not so straightforward as might be expected, and requires attention to details of observation and measurement, of a type and to a degree, that are not required, for example, when examining effects of conventional medicines.
It is my strong belief, that substantial flexibility is possible in such trials, so long as a few basics are adhered to. In the present paper, however, I have tried to delineate a reasonably full range of the most demanding standards, so that, in reviewing future trial results, we may have a clear sense of the context in which decisions regarding protocol design are made. It has been too long that research scientists have built their observations upon incompetent experimental designs, and then complained that homeopaths are backpedaling when, after conclusion of the trial, they nitpick the results. It is time to build some Quality Assurance into the trial design process in the first place, so that some reasonable work may be accomplished in pursuit of greater understanding of homeopathic process.
The future evolution and progress in research technology itself, will also be a beneficiary of increasing the preparedness of researchers, who may not be well equipped, otherwise, to study corners of the scientific world with which they are not personally well acquainted. The issues are too important, to tolerate any further such lame excuses, if nevertheless popular in some corners of the research community, such as it being unnecessary to understand that which is being tested. Such a mockery of the tools of enlightenment will not be missed, when the errors introduced on its account are revealed in future trials, through the instrumentation of improved, more reasonable research practices.
*Brien, et. al. 2003. Ultramolecular homeopathy has no observable clinical effects. A randomized, double-blind, placebo-controlled proving trial of Belladonna 30C.” British Journal of Clinical Pharmacology, 56:562-568.