Screening of diabetic retinopathy: effect of field number and mydriasis on sensitivity and specificity of digital fundus photography.

Purpose: To evaluate the sensitivity and specificity of one- and three-field, nonmydriatic and mydriatic, and 45 degrees digital colour photography compared with mydriatic indirect ophthalmoscopy for diabetic retinopathy (DR) screening.

Methods: A group of 79 patients (158 eyes) were included in this prospective study. Colour fundus photographs were taken with a Topcon TRC-NW6S digital camera, using four different techniques--single-field nonmydriatic; three-field nonmydriatic; single-field mydriatic; and three-field mydriatic--followed by dilated ophthalmoscopy. Two independent ophthalmologists classified blinded photographs according to the presence or absence of specific diabetic retinal findings. The sensitivity, specificity and agreement (kappa analyses) of the four methods were calculated for the presence or absence of DR and for all diabetic retinal findings.

Results: The sensitivity and specificity of digital photography compared with ophthalmoscopy for detection of DR were, respectively: 77 and 99% using single-field nonmydriatic; 92 and 97% using three-field nonmydriatic; 90 and 98% using single-field mydriatic; 97 and 98% using three-field mydriatic. The degrees of agreement for the four methods were 0.82, 0.90, 0.90 and 0.95, respectively. For specific retinal findings, sensitivity was greater for detection of hard exudates, nerve fibre layer haemorrhage and venous beading, and lower for detection of microaneurysms, dot-blot haemorrhage, cotton wool spots and intraretinal microvascular anomalies.

Conclusion: The three-field strategy without pupil dilation represents a good compromise, with reasonable sensitivity and good comfort (short examination duration, able to drive after photography) favouring patient compliance with the screening programme.