

ORIGINAL ARTICLE 

Year : 1988  Volume
: 36
 Issue : 1  Page : 1721 

Patterns of responses to alternative medicines in controlling allergic conjunctivitis
Deepak Batra, Madan Mohan, Pradeep Sharma, SK Gupta
Dr. RP. Centre for Ophthalmic Sciences, AIIMS, Ansari Nagar, New Delhi110 029, India
Correspondence Address: Deepak Batra Dr. RP. Centre for Ophthalmic Sciences, AIIMS, Ansari Nagar, New Delhi110 029 India
Source of Support: None, Conflict of Interest: None  Check 
PMID: 3150968
This analysis identifies the various patterns of the responses of the medicines in controlling allergic conjunctivitis. The mean S Deviation coefficient of variation, correlation matrix and loading of factors are worked out as stated in the tables. In the present case the Eigen values greater than 1.50 are retained. The four factors retained explain 68% of the total variations of the 16 responses. The first factor shows 23.38% of variations in total responses while first two and first three factors show 42.39% and 58.64% respectively. Thus medicine affective in controlling the symptoms are given In the descending order: oxymetazoline and sodium salicylate and Sodium cromoglycate, Oxymetazoline and Disodium CCromoglycate & Sodium Salicylate and disodium Cromoglycate.
How to cite this article: Batra D, Mohan M, Sharma P, Gupta S K. Patterns of responses to alternative medicines in controlling allergic conjunctivitis. Indian J Ophthalmol 1988;36:1721 
How to cite this URL: Batra D, Mohan M, Sharma P, Gupta S K. Patterns of responses to alternative medicines in controlling allergic conjunctivitis. Indian J Ophthalmol [serial online] 1988 [cited 2019 Dec 13];36:1721. Available from: http://www.ijo.in/text.asp?1988/36/1/17/26175 
Introduction   
External symptoms of various eye diseases, in general, are (a) watering (b) itching (c) redness and (d) heaviness of eye lids. There are certain medicines prescribed to control these symptoms. Different medicines are found to control different symptoms with varying intensity. These responses, however, also vary from patient to patient It will, therefore, be useful to identify the various patterns of the responses to these medicines. The analysis of these patterns would help the medical practitioners in choosing alternative medicines for different patients in controlling some of these symptoms. In the present paper an attempt has, therefore, been made to identify the dominant patterns of responses of four treatments. These treatment are:
(1) Control'X'
(2) Oxymetazoline and Disodium Cromoglycate 'A
(3) Sodium Salicylate and Disodium Cromoglycate 'B'
(4) Oxymetazoline and Sodium Salicylate 'C
Materials & Methods   
The data used in this analysis was obtained from a randomly selected sample of 30 patients from the out patient department of All India Institute of Medical Sciences. The patients were interviewed The medicines given to them were noted from the records, and responses of these medicines were noted through interviews. The effectiveness of each medicine was evaluated in graded form on a Six point Scale ranging from Zero to Five. The effect of the medicine which was most effective was graded as 'O' and the least effective medicine was graded as '5'. All responses in between were graded in the range 0 to 5. As the effect of four different medicines on four symptoms were observed, it gave 16 different responses The sample, thus generalised a 30 x 16 data matrix Each row corresponded to a patient and each column corresponded to a response. The data generated from the sample survey is given in Appendix  I [Figure  1]
To evaluate the effectiveness of these treatments in reducing the four symptoms the mean scores standard deviation and coefficient of variations are worked out Inter correlation matrix has been worked out to see the patterns of responses. This correlation matrix is further orthogonalised to generate dominant independent patterns underlying these responses.
Conventional tool of such analysis in most of the biometric research is analysis of variance, is a departure from this convention The main reasons of this deviation is the fact that analysis of variance gives us the significance of all the types of treatment taken together. It gives us very limited information about the patterns of responses related to individual treatment The orthogonalisation of correlation matrix is part of a factor analysis approach. Some of the work giving the details of it is given at the end.
Observations   
In order to evaluate the effectiveness of these medicines in controlling these symptoms, the means, Standard Deviation (S.D.) and Coefficient of Variation (C.V.) of each response was worked out and is given in [Table  1]
A close examination of [Table  1] shows that the mean scores are fairly high and the coefficients of variations are fairly low for the control position in all the symptoms. In other words we can say that the three medicines do have a better effect on reducing the symptoms than the control Other treatments did show lower values of means, but with varying levels of consistencies as reflected by the different values of coefficients of variations. Some of these treatments, however, are expected to have similar effect on some patients and different effect on some other patients. Thus, in order to investigate the patterns of relationship in the responses a correlation matrix is also prepared between the scores of 16 treatments The 16 x 16 matrix is zero order correlations is given in [Table  2].
A close look of the correlation matric reveals that a good number of coefficients of correlation are found to be statistically significant To synthesise these correlations, the correlation matrix was further orthogonalised to discern out the different underlying patterns of relationship clearly.
The Eigen values and the Eigen vectors of the above correlation matrix were worked out Since a correlation matrix is always a semipositive definite, all its Eigen values will be positive. In the case of the present exercise only those Eigen values are retained for further analysis which are greater than 1.50. Thus, only the first four factors were retained, these explained about 68% of the total variations of the 16 responses. There is no definite rule relating to the number of Eigen vectors choosen. The four Eigen vectors chosen here are on the basis of the convenience with which they are explained
The loadings of different factors, corresponding to Eigen values and the cumulative proportion of variation explained by these factors are given in [Table  3]
The factor loadings are in fact the coefficients of correlation of each of 16 variables with the composite scores of each of these factors. As the higher value of a score shows the lower effectiveness of the treatment the negative factor loadings will show the improvement effect of the medicine on a particular symptom.
A close examination of [Table  3], thus, shows, four different patterns in which the effect of these treatments in controlling the four symptoms can be summarised The first factor which shows the most dominant pattern, explains 23.38 percent of the variations in total responses. This factor very consistently showed negative loadings of medicine 'C for all the 4 symptoms All other treatment showed positive loadings. Thus we can say that predominantly the medicine Oxymetazoline and Sodium Salicylate help in curing all these four major symptoms of the eye diseases.
The second dominant factor showed negative relationship of the effect of medicine A with all the four symptoms. All these values are, however, quite small in the case of the redness and heaviness of eye lids, these values are very small All other loadings are found to be positive. Thus we can say that when medicine including 'C' Oxymetazoline and Sodium Salicylate does not have effect the second dominant pattern is found by the controlling effect of medicine Oxymetazoline and Disodium Cromoglycate. This medicine however has better value for controlling the watering and itching.
The third dominant factor shows positive loadings on medicine A and negative on medicine B for all the symptoms. In other words in those cases where medicine A is found to be ineffective medicine B is likely to be effective in controlling these symptoms.
Fourth dominant pattern is marked by the negative loading of the effect of four treatments on watering and itching The loadings related to redness and heaviness of eye lids are all positive. Thus the fourth pattern of relationship may be characterised as the controlling effect of the four treatments including the control treatment in watering and itching.
Discussion   
The findings of the above analysis lead us to the following.
The medicine Oxymetazoline and Disodium Cromoglycate, Sodicum Salicylate and Disodium Cromoglycate and Oxymetazoline and Sodium Salicylate when prescribed to different patients for controlling watering, itching redness and heaviness of lids in eyes, give different responses to different patients.
Medicine Oxymetazoline and Sodium Salicylate is found to be most effective in controlling these symptoms in general followed by the combination of Oxymetazoline and Disodium Cromoglycate. In some other cases the effect of medicines Oxymetazoline and Disodium Cromoglycate and Sodium Salicylate and Disodium Cromoglycate are found to be complementary. Medicine Sodium Salicylate and Disodium Cromoglycate have controlling effect and medicine Oxymetazoline and Disodium Cromoglycate have no controlling effect[5].
References   
1.  Anderson T.W. Introduction to Multivariate Statistical methods, John Wiley & Sons New York, p 272, 1958. 
2.  Cooly W. W. & Lohnes RP. Multivariate Data Analysis John Wiley & Sons New York 1971. 
3.  Kendall M.G. A course in Multivariate Analysis Charles and Griffin London 1957. 
4.  Rao C.R Advanced Statistical Method in Biometric Research, John Wiley New York p 236, 1965. 
5.  Hadley G. Linear Algebra Weseley Publishing Company. 
[Figure  1]
[Table  1], [Table  2], [Table  3]
