An Alternative Method of Grading Diabetic Retinopathy

Abstract

The purpose of this report is to (1) present a system for grading the severity of diabetic retinopathy that is a rapid, relatively inexpensive, and standardized alternative to the more detailed Early Treatment Diabetic Retinopathy Study (ETDRS) system; (2) present data on its reproducibility; and (3) compare it to the detailed ETDRS grading system. The alternative system was used to grade fundus photographs obtained during a large prevalence study, the Wisconsin Epidemiologic Study of Diabetic Retinopathy (WESDR). The alternative method involved grading seven stereoscopic standard fields as a whole, and assigning a level of severity for the eye according to the greatest degree of retinopathy using a modified Airlie House Classification scheme.Using an eight-level classification system of increasing severity of retinopathy, there was 78.3% exact agreement between the alternative and ETDRS systems. A grader regraded all 503 disagreements, and was in exact agreement 49.3% of the time with the alternative system, 35.7% of the time with the detailed system, and 15.0% with neither the alternative or detailed systems. Interobserver agreement for the alternative system was 78.5%; intraobserver agreement over a 9 month to 1 year period was 90.0% for grader 1 and 84.0% for grader 2. The alternative system of grading, when used by experienced graders, is a reproducible method for objectively determining retinopathy status in epidemiologic studies.