Screening for diabetic retinopathy in general practice using 35mm colour transparency fundal photographs

The objective of the study was to determine whether a general practitioner could effectively interpret fundal photographs obtained from clinics organised in a primary care setting ‐ in this case, a rural group general practice in Mid Wales. All patients registered with the practice and indentified as having diabetes were included into the study. The study was designed as a population‐based study and the GP diabetes register used to arrange photographic clinics in general practice. Comparison was made between the GP's and the ophthalmologist's (reference standard) interpretation of fundal photographs. Observers were unaware of the identity of the subjects and of each others results. Ascertainment rate for diabetes was 2.2% (166 patients). Photographs were obtained in 143 (86%); quality was sufficient for analysis in 137 (82.5%). Twenty‐three patients were identified by ophthalmologists as having retinopathy, eight (6% of whole sample) with sight‐threatening lesions. The GP's sensitivity for the presence of retinopathy was 87% (95% CI 66%‐97%), and specificity 77% (95% CI 70%‐85%). No sight‐threatening lesions were missed by the GP. Of 18 false positives with good picture quality, 10 (56%) had drusen; eight other false positives had photographic artefacts or poor picture quality. It was concluded that, given adequate training, GPs could act as first screeners for diabetic retinopathy by utilising fundal photographs taken in general practice. Analysis of the effectiveness of screening methods for diabetes should take into account ascertainment rate of diabetes in the study population.