Point-of-Care Evaluation of Diabetic Retinopathy Using Ultrawide Field Imaging in a Teleophthalmology Program

Abstract

Purpose: To evaluate the ability of trained nonphysician retinal imagers to perform point-of-care diabetic retinopathy (DR) evaluation at the time of ultrawide field (UWF) retinal imaging in a teleophthalmology program.

Methods: Diabetic individuals received Joslin Vision Network protocol retinal imaging as part of their standard medical care in a specialty diabetes clinic. Imager training included a standardized 4 hours didactic and 12 hours of guided image review. Real time grading results were compared to results from masked standardized grading at a dedicated reading center.

Results: 3,978 eyes (1,989 patients) were studied. By reading center evaluation of UWF images, 3,769 eyes (94.7%) were gradable for DR, with 1,376 eyes (36.5%) having DR, and 580 eyes (15.3%) having referable DR (moderate nonproliferative DR or worse or diabetic macular edema). Compared to the reading center evaluation, real time point-of-care imager grading had a sensitivity of 95% (95%CI 0.94-0.97) and a specificity of 84% (95%CI 0.82-0.85) for identifying more than minimal DR. Sensitivity and specificity for detecting referable DR were 99% (95% 0.97-1.00) and 76% (95% 0.75-0.78), respectively. Only 3 (0.15%) patients with referable DR (3 patients with moderate NPDR) were not identified by imager evaluation. No cases of severe NPDR or worse or clinically significant macular edema were missed by the imagers.

Conclusions: Evaluation of UWF images by nonphysician imagers following established teleophthalmology program protocols had good sensitivity and specificity for detection of DR and identification of referable retinal disease. These data suggest that immediate evaluation by retinal imagers at the point-of-care might reduce reading center image grading burden by approximately 60% with the added benefit of rapid patient feedback.