Abstract

Stereoscopic fundus photography of seven fields has become standard procedure for detecting diabetic retinopathy. However, the question has arisen as to whether adequate detection of retinopathy can be achieved with fewer fields. This question was investigated in a population-based study of 2694 diabetic persons. Retinopathy levels derived from the detailed grading of all seven fields were compared with those derived from combinations of two, three or four fields, the data for these being selected from the grading of the full seven fields. For eight retinopathy levels, the rate of agreement with seven fields ranges from 80% for two fields to 91% for four fields. For four retinopathy levels, agreement increases to 85 to 95%. The rate of agreement also varies with the specific retinopathy level, with level 65 being especially sensitive to the number of fields. The sensitivity of two to four fields compared to seven fields for detecting any retinopathy varies from 87 to 95%. For detecting proliferative retinopathy, sensitivity varies from 74 to 90%, and for Diabetic Retinopathy Study high-risk characteristics, it varies from 81 to 91%. Analysis of two groups of 50 eyes in which the combinations of fields were graded independently shows a small but nonsignificant effect of field overlap. These results indicate that in some situations, fewer fields may be adequate.