Purpose: Diabetic retinopathy (DR) screening methods are costly, requiring specialized personnel and patient mydriasis. The Optos apparatus can be operated by nonspecialists and provides ultra-wide-field imaging, with 200° views of the retina in a single image. We compared DR grading obtained from Optos imaging with DR grading from conventional Topcon imaging.
Methods: We conducted a cross-sectional study of 101 persons with diabetes who participated in the Addition-DK 10-year follow-up study. Retina fundus photos using both Optos and Topcon imaging were taken for both eyes. All photos were graded by specialist ophthalmologic nurses, using a modified version of the 'Proposed International Clinical DR severity scale'. We constructed frequency tables and assessed levels of agreement between Optos and Topcon gradings using weighted Kappa statistics, separately for peripheral and macular grading. We tested if the agreements between the two cameras were significantly different using a stratified Wilcoxon test.
Results: Significantly, more lesions in the periphery were identified by Optos compared with Topcon (p < 0.01), resulting in a fair Kappa agreement of 0.21. We saw no significant differences in the grading of the macula region (p = 0.97) between the two cameras. Although only 7% of the macula image gradings differed between the cameras, the Kappa agreement of macula grading was only moderate (Kappa = 0.52).
Conclusion: The inter-camera agreement was acceptable for macula image grading, but only fair for peripheral grading, with Optos identifying more microvascular changes in the periphery. The clinical significance and impact on screening modality and frequency remain to be explored.
Keywords: diabetic retinopathy; screening; ultra-wide-field; validation.
© 2019 Acta Ophthalmologica Scandinavica Foundation. Published by John Wiley & Sons Ltd.