Reply to D Riley

I am grateful to Dr Riley giving me the opportunity to explain some elements of our study more clearly. I agree with him: If the “typical” Cantharis symptoms, as determined by the homeopathic expert, were symptoms commonly associated with non-specific or placebo responses then the increase in symptoms could be explained by other causes. The problem here is the if. There is no way to determine which symptoms are specific for placebo, because there is no standard. Placebo can generate virtually any symptom. Hence, we have no external standard to answer this question, only the internal comparison within a given study. In our study, more symptoms typical for Cantharis were observed in both groups, compared to atypical symptoms.

Bear in mind the following: the potential number of symptoms not typical for Cantharis, i.e. placebo symptoms, is unlimited, while the number of Cantharis symptoms is limited. Thus, baseline probability for a typical Cantharis symptom is very low compared to all other symptoms that could potentially be experienced. It is true that the symptoms could also be commonly associated with placebo responses, but in fact they were not. Symptoms were generally quite specific, and a lot of them were in the area of mind and dreams (possibly due to attention bias, or because this proving happened to be more effective in this area; incidentally one of the reasons Cantharis was not identified by the homeopathic expert was that it does not have an extensive mind-symptoms proving picture). But the symptoms were not of the sort that one would commonly associate with placebo like feeling bad, nausea, pressing headache all day long, could not sleep well.

Considering that, by pure chance, there may be many thousand more symptoms which are not-typical for Cantharis, we thought that the pattern of the data, not the absolute figures, convey the important information. We see more atypical and typical symptoms alike during proving compared to baseline—which admittedly was methodologically unsatisfactory—an effect which can be attributed to more attention. But it would be very difficult to explain the fact that, compared with non-typical symptoms, typical symptoms are relatively more prominent in both groups.

One other explanation is possible: if the materia medica expert rated more symptoms as typical for Cantharis because of her lack of knowledge or bias, this would generate the same pattern of data. This is a limitation of the method and could only be checked by a secondary analysis of our data by another expert, starting afresh. While this would have been desirable from the outset and would have taught us something about the replicability of homeopathic practice, it was not feasible with the funds available. However, we are happy to offer both our data and collaboration, should anyone wish to delve into this issue.