Abstract
Purpose :
Cataracts remain the leading cause of blindness (51%) worldwide with 20 million people affected. Improvements in detection and remote gradings in tele-ocular eye screenings are essential for timely intervention and reducing false positive referrals to clinics. In this study, we investigated the use of a modified technique with an infrared flash-free near co-axial illumination system using a Canon RK-M3TM auto-refractor to standardize retinal reflex to improve cataract gradings.
Methods :
Thirty-eight eyes were imaged during community screenings. Anterior segment images of both eyes of all patients were obtained using a non-mydriatic retinal camera and a Canon RK-M3TM autorefractor in retro illumination mode at three near fixation angles for each eye (-15°, 0°, +15°). Images from each device were graded independently using the WHO Simplified Cataract Grading System. Cataract grades of 2+ or greater were referred to a specialist. Cohen’s kappa was used to evaluate statistical associations using Statistical Package for Social Sciences.
Results :
Interrater agreement between cataract-detection techniques was evaluated. Qualitative analysis of ocular images using the free-marginal multi-rater kappa test revealed moderate interrater agreement based on cataract grading (κ = 0.52) and substantial agreement based on decision to refer for treatment (κ = 0.74). Nine eyes were referred using the RK-M3, while 14 eyes would have been referred using the retinal camera. There was a strong positive correlation between cataract grades from the autorefractor and retinal camera (r = 0.90, p < 0.001).
Conclusions :
Current gold standard imaging using a retinal camera produces multiple Hirschberg central reflexes at the corneal apex, anterior apex of the lens, and the posterior part of the biological lens, interfering in clear view of cataracts. High referral agreeability supports the use of infrared retro-illumination, which allows for unobstructed view and grading of cataracts. Cataract grading by this method is more conservative and accurate than grading using a non-mydriatic retinal camera in the anterior mode.
This abstract was presented at the 2023 ARVO Annual Meeting, held in New Orleans, LA, April 23-27, 2023.