Screening for diabetic retinopathy in a clinical setting: a comparison of direct ophthalmoscopy by primary care physicians with fundus photography.

BACKGROUND: Type II diabetes mellitus is a major health problem among Native Americans, and diabetic retinopathy is a frequent complication of this disease. Screening for retinopathy can identify early disease and prevent major vision loss, but the most cost-effective screening method has not yet been determined.
METHODS: In a rural clinic that served more than 400 Native Americans with diabetes, we compared the accuracy of referrals made based on two screening methods: ophthalmoscopy by trained primary care physicians and seven-view nonstereoscopic, mydriatic fundal photography read by two general ophthalmologists and a retinal specialist. Patients in whom abnormal findings were detected by either screening method were then referred to a general ophthalmologist for further evaluation.
RESULTS: Two hundred forty-three examinations were performed and 83 referrals made. Both screening methods had high sensitivity for referring patients with retinopathy that required treatment or follow-up sooner than 1 year (100% for direct ophthalmoscopy by primary care physicians, 94% for the general ophthalmologist photography readers, and 100% for the retinal specialist reader). The calculated costs of screening by direct ophthalmoscopy and by retinal photography were 64% less and 44% to 35% less, respectively, than the cost of yearly ophthalmological examinations by ophthalmologists.
CONCLUSIONS: Careful screening for treatable diabetic eye disease by trained primary care physicians proved to be a clinically acceptable, cost-effective strategy. Screening methods for diabetic retinopathy should be evaluated based on the absolute sensitivity, specificity, and predictive values of their ability to correctly refer patients rather than their diagnostic accuracy.