Probabilistic Reasoning in Homeopathy Practice and its Implications for Clinical Research.

Abstract

The principle of similitude put forward by Hahnemann has challenging practical consequences in the selection of the right homeopathic medicine for a patient. According to this principle, only the medicines that best fit the totality of the symptoms of a given patient are supposed to really cure: this greatly depends on the homeopath's clinical analysis. In addition, a patient's illness may be more or less curable, depending on the characteristics of the disease. In their daily practice, homeopaths typically apply Bayesian reasoning to deal with uncertainty associated with both medicine and disease. We suggest that clinical research on homeopathy would gain by integrating this kind of prior estimation of (1) the probability of a given medicine being effective for that particular patient and (2) the probability of the patient's disease (or symptoms) being curable. We therefore suggest that future trials of N-of-1 design may gain (1) by testing a small number of "best candidate" medicines (instead of one) for a given patient facing a given disease, and (2) by including careful prior estimations of the probabilities that (a) each selected medicine will be efficient for that patient and (b) the patient's disease will be reversible with the medicine.