Abstract
Purpose :
To evaluate the agreement between optom, grader, and consultant assessments of UWF retinal images and clinical outcome of grading in a cohort of patients with established diabetic eye disease.
Methods :
Patients attending a medical retina clinic with a diagnosis of diabetic eye disease were included. A single UWF image (Optos, plc, Dunfermine, Scotland, UK) centered on the posterior pole was included for both eyes. Retinopathy and maculopathy severity was graded according to United Kingdom National Screening Committee defined criteria. The presence of other retinal pathology was graded as present or absent, and outcomes of the virtual assessment were determined as; virtual follow-up within 1, 3, 6, or 12 months, or referral for a routine/urgent face-to-face consultation.
Results :
80 eyes of 40 patients were included. Kappa values for agreement were presented in the order of (1) consultant vs optom, (2) consultant vs grader, and (3) optom vs grader. Poor agreement was observed with retinopathy severity grading (Kappa= 0.26; 0.20; 0.30) which improved when retinopathy grades were risk stratified into two groups; none to moderate non-proliferative diabetic retinopathy (NPDR) and severe NPDR to PDR (kappa=0.72; 0.29; 0.43). There was good agreement with maculopathy grades (kappa=0.63; 0.48; 0.62) and for the detection of other retinal pathology (kappa=0.43; 0.72; 0.58). Outcome of the virtual assessment showed poor agreement (kappa=0.32; 0.32; 0.42) when divided into six possible outcomes and improved when reduced to two (virtual or face-to-face follow-up) (kappa=0.44; 0.46; 0.40). In the 11 cases where there was a difference in outcome grading, 3/11 (27%) were due to a disagreement regarding the activity of previously treated PDR, 5/11 (45.5%) presence of diabetic macular oedema, 3/11 (27%) a possible diagnosis of wet AMD.
Conclusions :
This exploratory study examines the use of teleophthalmology for monitoring diabetic eye disease. We observed good agreement for the identification of treatable diabetic maculopathy, and simplified retinopathy grades. Outcome decisions showed fair agreement and differences were largely due a questionable presence of macular lesions. This may be improved by the addition of optical coherence tomography images.
This is an abstract that was submitted for the 2016 ARVO Annual Meeting, held in Seattle, Wash., May 1-5, 2016.