DIABETIC RETINOPATHY SCREENING THROUGH TELE-OPHTHALMOLOGY IN A PRIMARY CARE CENTER IN SOUTHERN BRAZIL

Diabetic retinopathy (DR) is a major cause of blindness and its prevalence varies between 15 and 20%. Brazilian people face limited
access to ophthalmological services, so strategies in telemedicine to monitor retinal status may help to early diagnose patients developing
DR. Objectives: To determine RD prevalence and clinical profile of diabetic patients from a primary care center. To evaluate
diagnostic skills of family physicians in interpreting retinal photographies (retinographies). To estimate the proportion of cases in
which retinography was enough to monitor retinal status. Methods: Two hundred and eighteen patients from Hospital de Clínicas
de Porto Alegre’s primary care service had retinographies bilaterally taken with a digital non-mydriatic camera (Canon CR-2). One
ophthalmologist and 3 family physicians independently interpreted and classified the retinographies as no DR, mild, moderate or severe
nonproliferative DR (NPDR), proliferative DR (PDR), and with or without macular edema. Family physicians have had previous
15 hours of training in the matter. All professionals were blind to patients’ clinical records. Sensitivity, specificity, overall accuracy and
kappa statistics were calculated. Results: Patients had mean age of 64.83 ± 11 years and had been diagnosed for diabetes for 9,3 ± 8,6
years. Glycated hemoglobin of 8% or above was present in 34%. Insulin-users accounted for 27.8%. The prevalence of DR at any stage
was 15.1%. From the total sample, 1.4% had severe NPDR or PDR and 4.9% required referral to a tertiary center for RD evaluation
and/or treatment. Results from family physicians in diagnosing DR showed 87.4% of overall accuracy, specificity 86% to 92% and
sensitivity 77% to 87%. Interobserver agreement between the ophthalmologist and each family physician for the DR stage ranged from
kappa = 0.359 to 0.522, while for macular edema ranged from kappa = 0.532 to 0.646. Conclusions: Family physicians’ evaluation
of retinographies had high sensibility and sensitivity for the detection of RD, although the agreement regarding the RD stage between
ophthalmologist and family physicians was rather poor. This means that family physicians were able to detect retinal signs of diabetes
but overall did not achieve to classify DR stages correctly. Only 4.9% had to be referred to a tertiary center, thus the remaining 95.1%
are amenable to subsequent follow through annual retinography at the primary care unit.