Comparison of subjective grading of lid wiper epitheliopathy with a semi-objective method

To validate a semi-objective method of grading lid wiper epitheliopathy (LWE) compared to subjective assessment.

Twenty upper and 20 lower eyelid margins of patients with LWE were photographed after instillation of fluorescein and lissamine green. The images were graded by two observers using a 0-3 grading scale for height (%) and width (mm) of the lid staining. The images were also processed using custom designed software in MATLAB. After manual delineation of the staining area, width and perpendicular height were automatically measured throughout the selected area. The height as a proportion of the lid margin width and width measures were then categorized into the same bins as in the grading scale.

Repeatability of the image analysis system showed a mean difference (95% limits of agreement) between repeats of -0.01mm (0.03 and -0.05mm) for LWE height, 0.04mm (1.16 and -1.08mm) for LWE width, and -0.11mm² (0.32 and -0.53mm²) for LWE area. The mean difference (95% limits of agreement) between image analysis and human grading for LWE height was -0.84 grades (0.54 and -2.21 grades), for LWE width was 0.31 grades (1.22 and -0.59 grades), and for the final grade (mean height and width) was -0.26 (0.44 and -0.96 grades) (all p<0.001).

Human observers tend to overestimate the height and underestimate the width of LWE staining. Lid wiper region is not well defined, thus, it might be a difficult process for human observers to judge the stained region as a proportion of the lid wiper total region.