|Year : 2020 | Volume
| Issue : 3 | Page : 179-188
Prevalence of Homoeopathic polar symptoms: A Follow-up pilot study conducted in Mumbai in 2017
Vaishali H Shinde1, Lex Rutten2, Anil Khurana3, Raj K Manchanda3, Ramesh S Bawaskar1
1 Regional Research Institute for Homoeopathy, Mumbai, Maharashtra, India
2 Independent Researcher, Breda, The Netherlands
3 Central Council for Research in Homoeopathy, New Delhi, India
|Date of Submission||05-Nov-2019|
|Date of Acceptance||28-Jul-2020|
|Date of Web Publication||28-Sep-2020|
Dr. Vaishali H Shinde
RRI (H), Sector-09, CBD Belapur, Navi Mumbai - 400 614, Maharashtra
Source of Support: None, Conflict of Interest: None
Background: Prospective assessment of homoeopathic symptoms is different from eliciting symptoms in daily practice. In prognostic factor research, we apply symptom questionnaires with Likert scales to assess symptoms in different intensities. In former research, we tested a 5-point Likert scale, which rendered a rather high prevalence for some symptoms even the strongest intensity, not useful in daily practice. A longer, 7-point Likert scale might render more useful outcome. Objective: To study if a longer Likert scale perform better in homoeopathic prognostic factor research. Methods: A 7-point Likert scale questionnaire with 30 polar symptoms was tested on 300 patients. Responses to various domains of temperature, climate, diurnal, influence of sleep, eating and desires/aversions were elicited. The outcome was compared with the former 5-point Likert scale. Results: The mean prevalence of all symptoms in the highest intensity with the 7-point Likert scale is (much) lower than in the 5-point scale, and for some symptoms, more useful. For a few symptoms, the prevalence remained high, even in the highest intensity. Conclusion: A longer Likert scale performs better in homoeopathic prognostic factor research, but not for all symptoms. The filling out of this questionnaire by patients should be guided by homoeopathic practitioners who are properly trained in prospective assessment of homoeopathic symptoms.
Keywords: Heuristics, Homoeopathic symptom, Likert scale, Prognostic factor research, Polar symptoms, Repertory
|How to cite this article:|
Shinde VH, Rutten L, Khurana A, Manchanda RK, Bawaskar RS. Prevalence of Homoeopathic polar symptoms: A Follow-up pilot study conducted in Mumbai in 2017. Indian J Res Homoeopathy 2020;14:179-88
|How to cite this URL:|
Shinde VH, Rutten L, Khurana A, Manchanda RK, Bawaskar RS. Prevalence of Homoeopathic polar symptoms: A Follow-up pilot study conducted in Mumbai in 2017. Indian J Res Homoeopathy [serial online] 2020 [cited 2020 Nov 26];14:179-88. Available from: https://www.ijrh.org/text.asp?2020/14/3/179/296244
| Introduction|| |
In the earlier paper published with the title, 'What is a homoeopathic symptom, in daily practice and research?' we discussed what makes symptoms useful homoeopathic symptoms. The most important property of a homoeopathic symptom is that it distinguishes one patient from others and this can be translated statistically that the prevalence of the symptom in the whole population is low. This low prevalence of a symptom is automatically achieved in peculiar symptoms (aphorism 153 of Hahnemann's Organon), but 'normal' symptoms also become peculiar if they are present in an abnormal intensity. In daily practice, doctors recognise by experience when a symptom is present in a peculiar degree. In prospective research, however, the symptom must be checked in every new patient. In that case, we also have to record the intensity of the symptom to be able to select the patients that have the symptom in a peculiar intensity. The intensity of the symptom is recorded in Likert scales that can have various lengths. In polar homoeopathic symptoms, symptoms with opposite values, a symptom like 'aversion or desire for open air' is expressed in a 5-point Likert scale as 'strong aversion', 'moderate aversion', 'neutral', 'moderate desire' and 'strong desire'. This gives us three intensities for each pole: neutral, moderate and strong.
Comparing the prevalence of a specific symptom in populations responding well to different medicines provides Homoeopathy with a suitable scientific identity, because this difference can be expressed as Likelihood Ratio (LR), the core of Bayes' theorem (posterior odds = LR × prior odds). Bayes' theorem is the scientific algorithm explaining how we learn from experience. Hitherto repertory entries were based on absolute occurrence of symptoms instead of prevalence. One of the consequences of this systematic mistake is that frequently prescribed medicines are over-represented in many repertory rubrics. If a symptom in the strongest degree has prevalence above 20%, it is not a good indication for specific medicines. The theoretical maximum LR of such symptoms is 5 (100/20), and in practice (considerably) lower because a prevalence of 100% in a population responding well to a specific medicine is rare. Symptoms with a prevalence below 10% are generally good homoeopathic symptoms, but if the prevalence is very low, <2%, there will be few cases unless we gather many cases, probably more than 8000. With a prevalence of 2%, a research sample of 8000 renders 160 patients with the symptom. These 160 patients with the symptom are divided over possibly more than 30 populations responding well to different medicines, rendering low numbers per medicine population. If the symptom prevalence is between 10% and 20%, the number of patients with the symptom is 5–10 times higher. In any case, the number of medicines that come up with higher prevalence of the symptom is variable. If that number of medicines is low, the symptom can be a good indication for those medicines.
In former research (2016), we tested a 5-point Likert scale for 70 polar homoeopathic symptoms at the CCRH Regional Research Institute (H), Mumbai, on 300 patients. It appeared that, for some symptoms, even the strongest intensity rendered a rather high prevalence of the symptom. Such a high prevalence leads us away from our implicit use of less common homoeopathic symptoms in daily practice: a common symptom is not a useful symptom in homoeopathic practice. For research of such symptoms, we must have more cut-off values, such as 'very strong', 'strong', 'moderate' and 'neutral'. This results in a 4-point Likert scale for non-polar symptoms and a 7-point Likert scale for polar symptoms. A second questionnaire with 30 polar symptoms in a 7-point Likert scale was tested in 2017.
| Methods|| |
After analysing the outcome of the 2016 questionnaire with 70 polar symptoms, the number of symptoms was reduced to 30 symptoms used in daily homoeopathic practice. There were some strong correlations (r ≥ |0.50|) found between symptoms related to weather and responses to weather and a few stronger correlations found between “cold aggravates” and “becoming cold aggravates” (r = 0.963) owing to which the reduction in number of polar symptoms was done. This was semantically obvious. There was also moderate correlation between many other symptoms. Few symptoms appeared to be unclear and so numbers of symptoms were also reduced to improve feasibility of the research. As a continuation of the previous study, this new questionnaire with a 7-point Likert scale was tested on the same lines on chronic cases attending the outpatient department from the period of 14th March 2017 to 31st March 2017 at the Regional Research Institute (H), Mumbai, under the Central Council for Research in Homoeopathy. With no human experimentation involved, the CTRI registration was not done. Ethical committee approval could not be sought. Verbal informed consent was obtained from the patients before the administration of the instrument. Responses to various domains of temperature, climate, diurnal, influence of sleep, eating and desires/aversions were elicited and incorporated by placing various intensities on a 7-point Likert scale, rendering a 4-point Likert scale for each pole such as 'neutral–worse–much worse–worse than in most people'. This questionnaire was tested on another 300 patients. The data were recorded in an Excel spreadsheet. The prevalence of symptoms at different cut-off values was analysed and compared with the outcome of the former 5-point Likert scale.
| Results|| |
The previously tested 2016 questionnaire with 5-point Likert scale rendered high prevalence of several symptoms even in a strong degree (degree 2 or − 2). The mean prevalence of all symptoms with different cut-off values is shown in [Figure 1].
|Figure 1: Mean prevalence of all symptoms on 5-point Likert Scale at different cut-off values. Cut-off value 1 refers to patients that have the symptom in a moderate or strong degree|
Click here to view
In the second test with the 7-point Likert scale, we see a low mean prevalence of symptoms in the strongest degree (3 or − 3) [Figure 2]. In [Figure 1], we see predominance of negative values and in [Figure 2] of positive values. This can be caused by statistical variation, but also by reducing the number of symptoms from 70 to 30.
|Figure 2: Mean prevalence of all symptoms on 7-point Likert Scale at different cut-off values. Cut-off value 1 refers to patients that have the symptom in moderate, strong or very strong degree|
Click here to view
We see that the mean prevalence of all symptoms in the highest degree is low. This offers us the possibility to select a small number of cases of a fairly common symptom (with high prevalence in lower cut-off values) in the highest intensity to discover what medicines are strongest related to that symptom.
In [Table 1], we show a comparison between the outcome of the questionnaire 2016 (5-point Likert scale) and the questionnaire 2017 (7-point Likert scale) for some symptoms. These symptoms would be useless or possibly useless (perspiration much) because of the high prevalence with the questionnaire 2016. The questionnaire 2017 offers us the possibility to select only patients with the symptom in very high intensity, where this is a good symptom.
|Table 1: A comparison of the outcome of the questionnaire 2016 (5-point Likert scale) and the questionnaire 2017 (7-point Likert scale)|
Click here to view
The longer Likert scale appeared not to work well for a few symptoms [Table 2]. The symptom 'desire vegetables' and 'desire fish' turned out to have very low prevalence in the strongest intensity, 0.3% for 'desire vegetables' and 0.0% for desire fish. The other cut-off values rendered too high prevalence. To overcome this problem, patients could be guided in filling in the questionnaire ['Discussion' section].
| Discussion|| |
In prospective research the symptom has to be checked in every new patient and in addition the symptoms in various intensities are also to be recorded. The Likert scales are taken as a tool to record a symptom in various intensities. This study has been a continuation of the previous study  which has concluded that with 5 point likert scale, a few symptoms even in moderate intensity precipitated higher prevalence in the general population. Consequently a need of longer Likert scales (more cut-off values) i.e., 7 point Likert scale was warranted [Table 3]. Moreover, some strong correlations (r ≥ |0.50|) between symptoms related to weather and responses to weather were observed. Moderate correlation (r between 0.30 and 0.50) between many other symptoms were also noted which was semantically obvious. Hence, in the next version, the number of questions in the questionnaire were narrowed down from 70 to 30, not only because of the confusion it caused in doctors and patients, but also because the principal component analysis showed that a few symptoms were related to various other questions in this questionnaire. This follow-up study validates the longer Likert scales (more cut-off values), i.e., 7 point Likert scale to ferret out the prevalence of symptoms in populations.
The second test with the 7-point Likert scale yielded a low mean prevalence of symptoms in the strongest degree (3 or -3). A few symptoms would be possibly useless (perspiration much) due of the high prevalence with the questionnaire 2016 where the questionnaire 2017 precipitated these to be more useful with low prevalence. A few symptoms such as 'desire vegetables' and 'desire fish' depicted very low prevalence in the strongest intensity, i.e., 0.3% and 0.00%, respectively, which possibly indicates a guidance in filling in the questionnaire.
Precipitation of some symptoms with very low prevalence in the highest cut-off value is not as big a problem. On encountering only high prevalence for a particular symptom, one can choose to select the cut-off value that comes closest to the optimal, provided such prevalence is available at one of the cut-off values.
Using a longer Likert scale can be a solution for improving the relevance of a symptom in prospective Prognostic Factor Research (PFR), but this is not the only factor that influences the outcome of a questionnaire. The formulation of the questions and the guidance in filling in the questionnaire also influence outcome. This requires creativity of the homoeopathic doctor assisting in filling in the questionnaire and also depends on his/her being familiar with cultural influences.
It appeared that symptoms such as 'desire vegetables' and 'desire fish' require attention of the group of doctors assisting in this research. How can we obtain a cut-off value for these symptoms that render prevalence between 2% and 10%? If you know how much vegetables the average person in a comparable group of people eats, you can ask for the amount of intake. Or ask the patient to place himself in a group of 10–50 people (search for example, such as class or work) and ask if his is the one with the strongest desire.
Homoeopathy is an art, interpreting symptoms in the context of every individual patient. However, the systematic mistake of the repertory, using absolute occurrence of symptoms instead of prevalence, should be corrected.
This is a pre-requisite step with a purpose to mend a serious systematic mistake of the repertory (absolute occurrence instead of prevalence) and to present Homoeopathy as a method with an underlying algorithm (Bayes' theorem) and to mend a serious systematic mistake of the repertory (absolute occurrence instead of prevalence) this is a pre-requisite step.
Indeed, symptoms in PFR should be collected with care and thorough knowledge and guidance in filling of the questionnaire. We can achieve a tremendous improvement of our repertory, but this is just a beginning of achieving our own scientific identity.
Looking at these two pilot studies in the same centre, we see that testing the questionnaires is essential for useful prognostic factor research. Assessing clinical symptoms in prognostic factor research is pre-requisite for correcting structural shortcomings of the repertory, and apart from the usual consultation, it requires new skills. As Bayesian methods can help in expressing the relationship between symptoms and expected results from medicines  and opens the possibility of investigating Homoeopathy in clinical practise, this is a baby step towards achieving it. Every doctor involved in this research should acquire experience and evaluation of this experience before the actual research starts.
| Conclusion|| |
The peculiarity of a homoeopathic symptom is indicated by its low prevalence which can be achieved using longer Likert scales and more cut-off values in our questionnaire. The longer Likert scale gives a choice to the patients/physicians to choose its intensity/gradation of symptoms revealing their true occurrences and unveiling their fallacious prevalence. The exercise also essentially has to be assisted with guidance in filling of the questionnaire.
Research always elicits new questions, but remaining ignorant by avoiding research is not an option. By testing the questionnaires being used in PFR, we can detect and improve problems that otherwise would have invalidated our research. PFR is new-fangled and we can foresee it to building a strong scientific footing for Homoeopathy, if we do it judiciously. The presented research shows how we can improve stepwise.
The assessment of the prevalence of symptoms provides Homoeopathy with a strong scientific rationale, but only after assessment of a considerable number of symptoms, we can make a new repertory that could be tested, e.g., in replications of old randomised controlled trials.
Financial support and sponsorship
Conflicts of interest
| References|| |
Rutten L, Muraleedharan KC, Shinde VH, Manchanda RK. What is a homoeopathic symptom, in daily practice and research? Indian J Res Homoeopathy 2017;11:11-20.
Hahnermann S. Organon of Medicine. 6th
ed. New Delhi: B. Jain Publishers (P) Ltd.; 2011.
Frei H. Polarity analysis, a new approach to increase the precision of homeopathic prescriptions. Homeopathy 2009;98:49-55.
Rutten L. Prognostic factor research in Homoeopathy. Indian J Res Homoeopathy 2016;10:59-65. [Full text]
Rutten AL, Stolper CF, Lugten RF, Barthels RW. Repertory and likelihood ratio: Time for structural changes. Homeopathy 2004;93:120-4.
Rutten AL, Stolper CF, Lugten RF, Barthels RW. A Bayesian perspective on the reliability of homeopathic repertories. Homeopathy 2006;95:88-93.
Rutten AL. Bayesian homeopathy: Talking normal again. Homeopathy 2007;96:120-4.
[Figure 1], [Figure 2]
[Table 1], [Table 2], [Table 3]