Validity and Reliability of the Revised Convergence Insufficiency Symptom Survey in Children Aged 9 to 18 Years

I read with interest the article by Borsting et al. 1 regarding validity and reliability of the Convergence Insufficiency Symptom Survey (CISS). The study provides solid data and analysis on the test-retest reliability of the CISS, and I agree, as the authors conclude, that this would allow it to serve as a good testing instrument to measure the efficacy of various treatments to alleviate a patient's symptoms. However, I find the remaining presentation of the data potentially misleading, specifically, their analysis of the survey in terms of its specificity and sensitivity. These terms traditionally are used to assess the results of a diagnostic or screening test in properly identifying a condition. Sensitivity is a measure of the probability of correctly identifying a person without the condition, and specificity is the proportion of people with the condition who are so identified by the diagnostic or screening test. To assess a test properly in these terms, an unbiased population sample must be used. The authors of this article state: “To assess the ability of the CISS to correctly classify subjects as CI or NBV… a cutoff value of 16 (i.e., CISS ≥ 16) yielded a sensitivity (95.7) and specificity (87.5).” However, these high sensitivity and specificity values are only true when comparing the selected convergence insufficiency (CI) subjects in their study. That is, to be an eligible CI for this study, subjects had to score ≥9 on the original 13-item CISS. Going back to the results of the original 13-item CISS, 2 a score ≥9 gave a sensitivity of 86% and specificity of 86% in their sample of 14 CI and 14 normal binocular vision (NBV) subjects. By using this cutoff score on the original survey to preselect their CI subjects in this study, they have ensured that they are only including patients with symptomatic CI, at a symptomatic level that would differentiate them from normal subjects with high sensitivity and specificity (86%). The authors also report that the results of this study are similar to that of their earlier study. It is not surprising the results are similar. What they have shown is that children with CI who score high on the original survey will score high on this slightly revised survey. The new cutoff criteria score for highest sensitivity and specificity comes from adding two questions and increasing the number of response categories and point assignment for the responses from 0 to 3 to 0 to 4, allowing a higher potential total score for this new survey. Although there is ample support in the literature that, on average, children with clinically significant CI (three or more signs of CI) score higher on the CISS than children with NBV, this study does not make the case that the CISS could classify a subject as CI or NBV in the general population. 2-5 Hence, it may not serve as a useful screening tool. Lynn F.