PURPOSE: To assess the effectiveness of optometrists as screeners for diabetic retinopathy using slit-lamp binocular indirect ophthalmoscopy through dilated pupils. METHODS: Prospective study of a screening scheme. Screening was performed by 27 locally accredited optometrists in their practice. The referral protocol used a new simple grading system of retinopathy, especially designed for use in an optometrist screening programme. All positive referrals and 10% of negative referrals were reexamined by an ophthalmologist. Sensitivity, specificity, likelihood ratios and technical failure rates were calculated. RESULTS: The optometrists screened 4904 people with diabetes in 18 months. 'Subthreshold' (screen negative) reports accounted for 4438 (90.5% of 4904); 429 (9.67%) of these were re-examined at secondary screening. There was disagreement regarding grading in 13 patients, of whom 5 (1.16% of 429) had sight-threatening retinopathy (STDR); this extrapolates to 52 patients if all the 4438 test-negatives had been examined. Of the 371 'threshold' patients, 112 (30.18%) were false positives; the commonest cause for false positive referral was drusen in patients with background diabetic retinopathy. The sensitivity for identification of STDR was 76% (95% CI 70% to 81%) and specificity 95% (95% CI 95% to 96%). The likelihood ratio of a positive test indicating STDR was 16.54 (95% CI 14.17 to 19.23) and that of a negative test 0.25 (95% CI 0.20 to 0.32). The technical failure rate was 0.2%. CONCLUSIONS: Suitably trained and accredited community optometrists performed well when screening for diabetic retinopathy using slit-lamp biomicroscopy through a dilated pupil. This was facilitated by the use of simple grading and referral criteria. The sensitivity, positive likelihood ratio and specificity were high.
PURPOSE: To assess the effectiveness of optometrists as screeners for diabetic retinopathy using slit-lamp binocular indirect ophthalmoscopy through dilated pupils. METHODS: Prospective study of a screening scheme. Screening was performed by 27 locally accredited optometrists in their practice. The referral protocol used a new simple grading system of retinopathy, especially designed for use in an optometrist screening programme. All positive referrals and 10% of negative referrals were reexamined by an ophthalmologist. Sensitivity, specificity, likelihood ratios and technical failure rates were calculated. RESULTS: The optometrists screened 4904 people with diabetes in 18 months. 'Subthreshold' (screen negative) reports accounted for 4438 (90.5% of 4904); 429 (9.67%) of these were re-examined at secondary screening. There was disagreement regarding grading in 13 patients, of whom 5 (1.16% of 429) had sight-threatening retinopathy (STDR); this extrapolates to 52 patients if all the 4438 test-negatives had been examined. Of the 371 'threshold' patients, 112 (30.18%) were false positives; the commonest cause for false positive referral was drusen in patients with background diabetic retinopathy. The sensitivity for identification of STDR was 76% (95% CI 70% to 81%) and specificity 95% (95% CI 95% to 96%). The likelihood ratio of a positive test indicating STDR was 16.54 (95% CI 14.17 to 19.23) and that of a negative test 0.25 (95% CI 0.20 to 0.32). The technical failure rate was 0.2%. CONCLUSIONS: Suitably trained and accredited community optometrists performed well when screening for diabetic retinopathy using slit-lamp biomicroscopy through a dilated pupil. This was facilitated by the use of simple grading and referral criteria. The sensitivity, positive likelihood ratio and specificity were high.
Authors: M Violato; H Dakin; U Chakravarthy; B C Reeves; T Peto; R E Hogg; S P Harding; L J Scott; J Taylor; H Cappel-Porter; N Mills; D O'Reilly; C A Rogers; S Wordsworth Journal: BMJ Open Date: 2016-10-24 Impact factor: 2.692