Verification of homeopathic symptoms has so far been a neglected field of research. In our opinion this is a serious omission; what is the use of efficacy research if the instruments of the method have serious flaws? The most serious flaw of the homeopathic repertory is that entries of medicines in a symptom-rubric are based on absolute occurrence of the symptom in the cured population, not on prevalence. This way, frequently used medicines are over-rated, especially regarding frequently used symptoms. Bayesian theory shows that the prevalence of the symptom is the only correct criterion. A symptom is an indication for a specific medicine only if the prevalence of that symptom is higher in the population cured by that medicine than in the rest-population.
If we keep adding medicines to the repertory-rubrics and materia medica using the old rule the computer will become a serious threat to the quality of our instruments,. Sooner or later each symptom will turn up in the population that is cured by a specific medicine, by mere chance, even if it is seldom seen in that medicine-population. The computer makes it very easy to update the content of the repertory. It has become a sales-argument for electronic repertory manufacturers to have the largest repertory with the most entries and their income depends on updates.
If you are interested in Bayesian thinking in homeopathy, please let me know (lexrtn@concepts.nl) and I can send you new newsletters.
Newsletters already published:
- August 2006. About repertory corrections based on two years propective assessment of six homeopathic symptoms. Starting the discussion about when to add or delete repertory entries.
- May 2007. Our results differ from the original repertory in about 50% of the entries. The six symptoms that we selected for this assessment were not known as keynotes for the same medicine. We checked interrelatedness of the symptoms for a number of medicines and found hardly any correlation between symptoms.
- October 2007. In this newsletter we explore the possible interrelation between the six symptoms further with multivariate analysis. Another possibility of multivariate analysis is pattern recognition. With statistical techniques that are used for the recognition of satellite images we can explore the existence of homeopathic images. Although the LR database is not ideal for this purpose, we think that multivariate analysis, and discriminant analysis in particular, is an interesting technique to validate the materia medica of our most successful medicines.
- January 2008. The datacollection is concluded and first results calculated. We suggest some corrections for the repertory, but you can check them for yourself. Some repertory rubrics, like 'Grinding teeth during sleep', are seriously flawed.