Purpose: Multivariate statistical and graphical methods were used to compare high, medium, and low contrast visual acuities in right and left diabetic eyes with and without diabetic retinopathy (DR) and with and without diabetic macular edema (DME).
Methods: Two hundred and two diabetic subjects were randomly recruited from a diabetic hospital clinic in Johannesburg. Contrast visual acuities were measured using the new ETDRS visual acuity chart, at 100% (high contrast), 10% (medium), and 2.5% (low). For all eyes, retinal photographs were evaluated by two independent ophthalmologists and individuals were classified by the presence or absence of DR and DME. LogMAR acuities for the different contrast levels were compared using box and whisker plots and multivariate stereo-pair scatter plots, distribution and confidence ellipsoids, and variance-covariance matrices.
Results: Analyses indicated larger volumes of 95% distribution ellipsoids for the right and left eyes with DR (2.5 logMAR(3) and 3.8 logMAR(3)) and DME (2.5 logMAR(3) and 4 logMAR(3)). Similar results were found for the volumes of the 95% confidence ellipsoids for the mean contrast visual acuities. A significant difference in means was found in the C2.5% level for right (p = 0.042) and left eyes (p = 0.035) of the DR subjects. In subjects with DME, significant differences were found for C10% (p = 0.021 for right eyes and p = 0.046 for left eyes) and C2.5% (p = 0.033 for right eyes and p = 0.045 for left eyes) levels. The variance-covariance matrices indicated that variances for the C100% and C10% measurements were small suggesting minimal variation whereas greater variation occurred in the C2.5% contrast acuity measurements.
Conclusions: These results suggest that diabetic eyes with DR or macular edema have a greater spread or variation of measurements and display poorer acuities at the lower contrast levels of 10% and 2.5% as compared with eyes without such disorders. By representing the contrast visual acuities in a 3-dimensional manner, the relationship between different levels of contrast can be assessed and may provide a clearer understanding of the nature of abnormal contrast sensitivity. Such methods may assist with earlier identification of DR and DME and may also become important in monitoring variation with either progression or treatment of diabetic retinal disease.