ObjectiveTo evaluate the ability of contrast sensitivity (CS) to discriminate loss of visual function in diabetic subjects with no clinical signs of retinopathy relative to that of normal subjects.Research design and methodsIn this prospective cross-sectional study, we measured CS in 46 diabetic subjects with a mean age of 48±6 years, a best-corrected visual acuity of 20/20 and no signs of diabetic retinopathy. The CS in these subjects was compared with CS measurements in 46 normal control subjects at four spatial frequencies (3, 6, 12, 18 cycles per degree) under moderate (500 lux) and dim (less than 2 lux) background light conditions.ResultsCS was approximately 0.16 log units lower in patients with diabetes relative to controls both in moderate and in dim background light conditions. Logistic regression classification and receiver operating characteristic curve analysis indicated that CS analysis using two light conditions was more accurate (0.78) overall compared with CS analysis using only a single illumination condition (accuracy values were 0.67 and 0.70 in moderate and dim light conditions, respectively).ConclusionsOur results showed that patients with diabetes without clinical signs of retinopathy exhibit a uniform loss in CS at all spatial frequencies tested. Measuring the loss in CS at two spatial frequencies (3 and 6 cycles per degree) and two light conditions (moderate and dim) is sufficiently robust to classify diabetic subjects with no retinopathy versus control subjects.