|LETTER TO THE EDITOR
|Year : 2013 | Volume
| Issue : 11 | Page : 690-691
Parikshit Gogate1, Debapriya Mukhopadhyaya2, Ashok Mahadik3, Thomas John Naduvilath4, Amit Shinde5
1 Gogate's Eye Clinic, Pune; Lions NAB Eye Hospital, Miraj, Maharashtra, India
2 Gogate's Eye Clinic, Pune; Bharti Vidyapeeth Medical College, School of Optometry, Maharashtra, India
3 District Blindness Control Society; Lions NAB Eye Hospital, Miraj, Maharashtra, India
4 University of New South Wales, Australia
5 Bharti Vidyapeeth Medical College, School of Optometry, Maharashtra, India
|Date of Web Publication||16-Dec-2013|
Dr. Gogate’s Eye Clinic, K-102, Kumar Garima, Tadiwala Road, Pune - 411 001, Maharashtra
Source of Support: None, Conflict of Interest: None
|How to cite this article:|
Gogate P, Mukhopadhyaya D, Mahadik A, Naduvilath TJ, Shinde A. Author's reply. Indian J Ophthalmol 2013;61:690-1
We thank you for your interest in our study of spectacle compliance among rural secondary school children in India. In this study the cluster size could not be predetermined as the subjects were those who were given glasses the previous year. However, the fact that the first stage of the sampling covered 50% of the population (7/14 talukas) and the second stage included 100 schools, which approximated to 10 subjects per school (1035 subjects from 102 schools is a very small cluster size), makes the choice of 1.5 for design effect justifiable relative to the sampling frame.
The choice of 4% absolute precision was to obtain approximately 10% relative precision for the anticipated 35% compliance rate. Sample size was calculated with a compliance rate of 35%.  Previous studies give rates of compliance at 19.5% from rural Central India  to 37.7% from rural China.  Maybe the rural Central India figures would have been more appropriate, but we still sampled half the universe, 1028 out of the 2312 children who had been dispensed spectacles, much more than what would have been required.
[Table 1] presents results from the univariate analysis of the entire sample.  These are based on Chi-square tests and not a logistic regression. [Table 4] presents results from a logistic regression analysis and this includes only myopic subjects. The odds ratio presented in [Table 4] is equivalent to slope and is a measure of the strength of association.
We do acknowledge the possibility of collinearity between vision and spherical equivalent. At the outset it must be stated that the vision variable relates to vision in the better eye, while the spherical equivalent refers to the refractive error (RE) of the worse eye. The fact that it was not explicitly defined in the table was an oversight. The reason for using these variables was that visual impairment (VI) in the community is defined using visual acuity (VA) in the better eye, while myopia in the community is defined using spherical equivalent in the worse myopic eye. It is true that if both were better eye definitions, it would have resulted in highly correlated factors (Spearman's rank r= -0.66). When better eye VI and better eye RE were added into the model, RE remained significant, but VI lost all its significance (P = 0.288). This would be a typical case of multicollinearity. In contrast, we noticed that the correlation of the better VI and worse RE was only mild to moderate (Spearman's rank r= -0.46). When better eye VI and worse eye RE were added into the model, VI did not lose its entire significance. The table below presents both univariate and multivariate results for the two variables for the outcome of noncompliance. An observation of the odds ratio and P values will show that most of the changes occur at the higher levels of these variables (i.e., worse than -3D and worse than 6/60). The following observations can be made: (1) While compliance increases with worsening in both VA and RE, the increase in compliance is more linear with progression in RE that with worsening VA, which is very gradual at the early stages of vision loss. (2) Students feel the need for specs only when their vision is worse than 6/18 (visually impaired), even when RE is accounted for. (3) Greater than -3D of RE will result in a 3× odds of compliance, but strangely in terms of VA a 3× odds of compliance is achieved only at VA worse than 6/60, which is legally blind. Having both variables in the model has served the purpose to highlight areas of significance for spectacle compliance without overstepping statistical assumptions.
Children who had unaided visual acuity >6/18 were less likely to wear them as the odds of compliance was 1.7 (0.9-3.4) and not different to the reference category (P = 0.123). However when VA was between 6/60 and 3/60 the odds of compliance increased by 3.8 (1.1-12.7) which was different to the reference category (P = 0.034). Both [Table 3] and [Table 4] do not provide a category as "<6/60," which is a combination of "6/60-3/60," and "<3/60." When these two categories are combined, the odds of compliance was 4.0 (1.2-13.1) which was significantly higher than the reference category (P = 0.020). Therefore the statement in the article cannot be considered as a chance finding.
[Table 2]a and b of our original article accounts for 912 subjects.  The summarized mean and SD are from the worse and better eye data of these 912 subjects. Some children were unable to answer the questions. You are right in stating that organic and nonorganic causes of decreased vision could be confounding factors or effect modifiers in spectacle compliance. [Table 3] and [Table 4] include only the 499 myopic subjects, of whom VA without correction could be obtained only for 495 subjects.
We hope that this study shall spur more work on this important topic of spectacle compliance as noncompliance would render the efforts of school eye screening programs to naught.
| References|| |
Gogate P, Mukhopadhyaya D, Mahadik A, Naduvilath TJ, Sane S, Shinde A, et al
. Spectacle compliance amongst rural secondary school children in Pune district, India. Indian J Ophthalmol 2013;61:8-12.
Khandekar R, Sudhan A, Jain BK, Tripathy R, Singh V. Compliance with spectacle wear and determinants in school children in Central India. Asian J Ophthalmol 2008;10:174-7.
Congdon N, Zheng M, Sharma A, Choi K, Song Y, Zhang M, et al
. Prevalence and determinants of spectacles non-wear among rural Chinese school children. The Xichang pediatric refractive error study report 3. Arch Ophthalmol 2008;126:1717-23.