RESUMEN
PURPOSE: To assess the sensitivity and specificity of lateral midpupil lid distances for the detection of upper lid lateral flare. METHODS: Lateral lid flare was determined by unanimous agreement among six experienced oculoplastic surgeons in the grading of photographs obtained for patients with Graves orbitopathy (GO). Bézier lines were employed to extract the upper eyelid contours of the patients and a control group of age and sex matched subjects. Custom software was employed to determine 5 lateral midpupil eyelid distances. The sensitivity and specificity of each measurement in detecting lateral flare were estimated from receiver operating characteristic curves. The non-parametric Kruskal-Wallis one-way analysis of variance (ANOVA) with Dunn's posthoc test was used to compare the median values of the contour parameters between groups. RESULTS: The degree of agreement between judges evaluated with the Fleiss' Kappa test was relatively high (K = 0.69, z = 16.6, p < .0001). The raters classified 12 lids with lateral lid flare (LLF) and 7 without LLF in patients with GO. There was no agreement on the presence or absence of LLF in 11 lids. In all eyes, lateral midpupil lid distances diminished from the center of the eyelid towards the lateral canthus. Receiver operating characteristic analysis for the midpupil distances revealed that the fourth distance from the center demonstrated high sensitivity and specificity in detecting flare. At this location (2.5 mm medial to the lateral canthus) a midpupil distance equal to or greater than 60% of the margin reflex distance (MRD1) indicated the presence of flare. CONCLUSIONS: - A single measurement of a lateral midpupil eyelid distance 2.5 mm medial to the lateral canthus is a sensitive and specific measurement for the diagnosis of the LLF.