Abstract
BACKGROUND: Diabetic retinopathy is the main cause of preventable blindness in the economically active population in western countries. Diabetic retinopathy screening is effective in preventing blindness and can be performed through various diagnostic methods. Our objective is to compare binocular indirect ophthalmoscopy (BIO) to telemedicine protocols of digital retinography for diabetic retinopathy screening in a large and heterogenous type 1 diabetes population in a developing country. METHODS: Data from 1266 Type 1 Diabetes Mellitus patients from a Brazilian multicenter study were analyzed. Patients underwent BIO and digital retinography, non-mydriatic and mydriatic. Images were sent to a reading center in a telemedicine protocol. Agreement between the different methods was calculated with kappa statistic for diabetic retinopathy and maculopathy classification. Clinical outcome was either observation or referral to specialist. RESULTS: Agreement between BIO and mydriatic retinography was substantial (kappa 0.67-0.74) for diabetic retinopathy observation vs referral classification. Agreement was fair to moderate (kappa 0.24-0.45) between retinography and BIO for maculopathy. Poor mydriasis was the main obstacle to image reading and classification, especially on the non-mydriatic strategy, occurring in 11.9 % of right eyes and 16.9 % of left eyes. CONCLUSION: Mydriatic retinography showed a substantial agreement to BIO for diabetic retinopathy observation vs referral classification. A significant amount of information was lost on the non-mydriatic technique because of poor mydriasis. We recommend a telemedicine-based diabetic retinopathy screening strategy with digital mydriatic retinography, preferably with 2 fields, and advise against non-mydriatic retinography in developing countries.