Abstract
Abstract: :
Purpose: To validate a simplified diabetic retinopathy (DR) grading system designed for a method of screening using three color fundus photographs, and the severity level requiring referral. Methods: The screening grading system consists of five severity levels for DR (no DR (level 0), mild non proliferative DR (level 1), moderate non proliferative DR (level 2), severe non proliferative DR (level 3), and proliferative DR (level 4)) and two levels for maculopathy. The level 2 of the DR grading system is defined as more than 20 microaneurisms and/or dot hemorrhages, or a number of blot hemorrhages ranging from 1 to 10, or less than 3 IRMA, or 2 or more cotton wool spots, on the whole three fundus fields. The screening test was three 45° fundus color photographs (posterior pole, nasal, and temporal) obtained with a non–mydriatic camera in 100 diabetic patients (200 eyes). The grading results according to the screening grading system were compared to the reference diagnosis, from either ETDRS stereoscopic photographs, or a mozaic of nine color fundus photographs. Results: Exact agreement between the grading of the three photographs according to the screening grading system and the reference diagnosis was 95.8 % (weighted ĸ statistics: 0.99) for DR, and 94.1 % (weighted ĸ statistics: 0.93) for maculopathy. When the retinopathy level requiring referral was ≥ 2, the sensitivity and the specificity for the detection of moderate non–proliferative DR or more severe scales were 100 % and 92.3 % respectively. When the maculopathy level requiring referral was 1 (hard exudates within one disc diameter from macular center), the sensitivity and the specificity for the detection of maculopathy were 92.3 % and 96 % respectively. Conclusions: This grading system allows to define sensitive and specific levels requiring referral, and seems to be suitable for a method of screening using three color fundus photographs.
Keywords: diabetic retinopathy