Objective: To evaluate the efficacy of nonmydriatic digital retinal images for determining diabetic retinopathy.

Material and Method: Single field 45-degree digital retinal images of 225 eyes from 142 diabetic patients were obtained with a nonmydriatic camera. The images were diagnosed and graded by a general ophthalmologist. These results were compared with clinical diagnosis obtained by retinal specialists, after examination by using biomicroscope with plus lens and indirect ophthalmoscope of the patients. International clinical diabetic retinopathy disease severity scale was used for grading diabetic retinopathy in all cases.

Results: Presence of diabetic retinopathy was detected in 70 eyes (31.1%). The sensitivity and specificity for determining diabetic retinopathy was 68.57% (95% CI 57.00-78.20) and 92.25% (95% CI 87.00-95.50), respectively. The positive predictive value and negative predictive value was 80.00% (95% CI 68.20-88.20) and 86.67% (95% CI 80.60-91.00). Overall accuracy was 84.89%.

Conclusion: Single field 45-degree nonmydriatic digital retinal images were limited by fair sensitivity for determining diabetic retinopathy although overall accuracy from the present study was relatively high. Upcountry, this tool might facilitate increased access of diabetic patients for eye evaluation but cannot replace standard eye examination.

Keywords: Diabetic retinopathy screening, Digital retinal images, Nonmydriatic retinal images