Eye‐screening models using digital imaging for the Diabetes NSF

The typical eye-screening model for the Diabetes National Service Framework (NSF) involves the collection of digital images for grading on a separate occasion at a Grading Centre. In 150 consecutive screenings, we assessed an alternative digital image model involving full clinical grading at the point of screening by an optometrist who may also use, according to need, direct ophthalmoscopy and/or slit lamp. Using this model, important additional information was gained in 46/150 (31%). This broke down into: 21/150 (14%) poor quality pictures usually due to cataract; 14/150 (9%) confirmation of presence/absence of suspected lesions and their nature; 10/150 (7%) lesions not seen on digital images; and 1/150 (0.7%) diagnosis of unexplained poor vision. Diagnoses made by the addition of direct ophthalmoscopy and/or slit lamp included macula red dots, macula hard exudates hidden by lens opacities, peripheral neovascularisation and suspicious peripheral pigmented lesion. If the Grading Centre model had assessed this group of patients, 22/150 (15%) would have needed to be recalled for further assessment on another occasion. Grading at the point of screening by a screener who can also use direct ophthalmoscopy and/or slit lamp improves diagnostic accuracy and significantly reduces the number of patient recalls with their associated costs. When different models are being considered to achieve the ambitious eye-screening targets of the Diabetes NSF, these factors should be taken into account. Copyright © 2005 John Wiley & Sons, Ltd.