Validity of keratoconus screening tests

Authors:
Open Access
Publisher website
Google Scholar
Abstract
Knowledge of reliable cut-off values for screening tests can facilitate early disease identification and referral. This study compared the validity of clinical tests used to screen for keratoconus in high school students. In this population-based, prospective, cross-sectional study, multistage cluster sampling was used to select participants who underwent visual acuity assessment, auto-refraction, retinoscopy and corneal topography. Students were classified into three groups: non-keratoconus, keratoconus suspect and keratoconus. A total 3051 students, mean age of 17.4 ± 1.6 years ranged from 13 to 25 years participated in this study. Of the 51 (1.7%) students diagnosed with keratoconus, one-third (31.4%) had a binocular visual acuity of ≤0.2 logMAR. The median visual acuity was worse, the spherical equivalent refractive error was more myopic and the astigmatism was higher in the keratoconus group compared to the non-keratoconus group (all p values < 0.05). However, there was a considerable overlap when comparing these parameters between the three groups, making it difficult to establish a reliable cut-off. Asymmetry was a feature of a large proportion of the students with keratoconus. When using scissor reflex on retinoscopy, maximum anterior corneal curvature or astigmatism individually as referral criteria, retinoscopy yielded the highest Youden Index (0.44). However, combining the use of all three parameters where either of them is positive: a scissors reflex or maximum anterior corneal curvature > 46.5 D or astigmatism >2.50 D was the most effective (Youden index 0.46). Monocular visual acuity, retinoscopy, keratometry and refraction should be included in keratoconus screening protocols with the inclusion of corneal topography to increase effectiveness. A combined referral criterion of a scissors reflex on retinoscopy or maximum anterior corneal curvature > 46.5 D or astigmatism > 2.50 D is recommended.

This publication has 32 references indexed in Scilit: