Abstract
Purpose: :
Glaucoma is the leading cause of irreversible blindness in the world. It is estimated that £1 billion in non-treatment costs would be saved annually if just 10% of unrecognised glaucoma were detected. In contrast to general population screening, targeting high risk groups such as diabetic patients may be cost-effective. In England, annual population screening for diabetic retinopathy (DR) uses digital retinal imaging. Optic disc changes suggestive of glaucoma are frequently seen and referred for further evaluation. This study aims to determine the ability of the National DR Screening Service (DRSS) to detect glaucoma in referred patients and predict the value of DRSS images for glaucoma screening.
Methods: :
We looked at outcomes of DRSS images referred to ophthalmology with suspected glaucoma from the screened diabetic population of one London Borough from November 2006 to February 2008. Image sets (two 45o retinal images per eye taken with Topcon TRC NW6S and Nikon D70 camera using OptoMize iP® software) were examined for final DRSS grading by a medical retina consultant and classified into 2 groups: DRSS glaucoma / DRSS not glaucoma. The same images were independently graded by a glaucoma specialist. Those with positive findings from glaucoma grading were referred to clinic for diagnosis and management. Referred patients had 4 possible outcomes at 1 year: (a) glaucoma: treatment started (b) glaucoma suspect: no treatment but review in clinic (c) no glaucoma: discharged (d) no data.
Results: :
Of 11565 patients screened, 216 were suspected glaucoma after initial DRSS grading (prevalence 1.9%). After final DRSS grading, 147 were classified DRSS glaucoma and 69 DRSS not glaucoma. After glaucoma grading, 170(79%) were referred to clinic. Comparing results of final DRSS grading with glaucoma specialist grading showed 78% sensitivity, 70% specificity and 90% positive predictive value.Of 170 clinic referrals at 1 year: 42(25%) have glaucoma, 50(29%) are glaucoma suspects, 19(11%) were discharged and 59(35%) have no data (30 failed to attend follow up, 27 have incomplete data and 2 patients died). Where diagnosis is known, detection of definite glaucoma after final DRSS grading had 83% sensitivity, 26% specificity and 71% positive predictive value.
Keywords: clinical (human) or epidemiologic studies: treatment/prevention assessment/controlled clinical trials • optic disc • diabetic retinopathy