Purchase this article with an account.
I.E. Zimmer–Galler, R. Zeimer; Evaluation of a Reading Center for Detection of Diabetic Retinopathy With the DigiScope in Primary Care Offices . Invest. Ophthalmol. Vis. Sci. 2005;46(13):1478.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Purpose: To assess the performance of a supervised reading center in evaluation of DigiScope images obtained to detect diabetic retinopathy (DR) in the primary care environment. Methods: The DigiScope ia a semi–automated fundus camera designed to allow staff in primary care physicians' offices to acquire fundus images of diabetic patients. DigiScope images are transmitted to a reading center for evaluation by supervised readers trained to detect DR who determine the need for referral based on the presence of DR. Referral is recommended for any patient with greater than mild nonproliferative DR. Patients with ungradable images are also referred. Agreement between a reader and retina specialist was determined for referral and no referral, ungradable images, and referral for pathology other than DR. This retrospective study evaluated reading center performance over a 3–month period from 5/2004–8/2004. Results: DigiScope imaging was performed on 813 patients (1,626 eyes). Referral to an ophthalmologist for evaluation of DR was recommended by the retina specialist in 86 (11%) patients. The reader correctly identified all patients requiring referral for DR (sensitivity and specificity of 1.0). Images from 54 patients (7%) were labelled as not gradable by the retina specialist with 70% (38 patients) likewise graded by the reader. An additional 113 (14%) patients requiring referral for ocular pathology other than DR were identified by the retina specialist. Of these, the reader identified 53 (47%) patients. The majority of reader false–negatives for pathology other than DR were for disc findings suspicious for glaucoma. Overall the reader and retina specialist agreed on referral or no referral for 737 (91%) patients. Conclusions: Reading center evaluation of DigiScope images by a supervised reader trained to detect DR appears to be reliable for identifying patients with DR who require referral to an ophthalmologist. The prevalence of patients with unreadable images or pathology unrelated to diabetes but needing referral for further evaluation indicates the need for additional reader training and development of comprehensive criteria for referral for pathology other than DR.
This PDF is available to Subscribers Only