Glaucoma detection using optic disc images from the English national screening programme for diabetic retinopathy

We performed this study to determine the positive predictive value of a Diabetic Retinopathy Screening Program (DRSP) in detecting early glaucoma using diabetic retinopathy screening images.

The study was carried out on patients referred to the glaucoma clinic from the local DRSP. Retinal images considered by DRSP graders to show possible glaucomatous optic discs were referred to the final referral grader, a medical retina specialist (MR). The same images were independently graded by a glaucoma specialist (GS). GS-positive patients were referred to the glaucoma clinic. Possible outcomes at 1 year were true cases (patients found to have glaucoma or glaucoma suspects) or false cases (patients for whom a diagnosis of glaucoma was excluded).

Of 11,565 diabetic patients screened, 216 were suspected to have glaucoma after DRSP grading (1.87%). A total of 170 were graded glaucoma positive and referred to a clinic. At 1 year, 113 were true cases (98 per 10,000 screened with previously undiagnosed glaucoma or glaucoma suspect) and 22 were false cases and discharged from clinic (19 false positives per 10,000 screened). Of the 113 true cases, 89 were given a positive grade by the MR and 24 were given a negative grade, giving the DRSP a positive predictive value of 78.8% in detecting glaucoma in the population. Comparison of the MR and GS grading showed an agreement of 76.4%. The κ-coefficient was 0.404 (95% confidence interval, 0.272-0.536).

We conclude that optic discs imaging in DRSP can be useful as part of a glaucoma screening strategy to identify new disease within a diabetic population.