ABSTRACT
PURPOSE: To evaluate intraobserver reliability and interobserver reliability of a computer-based digital image measurement of eyelid position in Graves' ophthalmopathy and to compare digital image measurement with clinical measurement. DESIGN: Cross-sectional study. PARTICIPANTS: Eighty-four eyes of 42 patients with mild to moderate bilateral Graves' ophthalmopathy. METHODS: Digital images were created from 35-mm color slides of both eyes of participants and projected onto a 15-inch flat-screen computer monitor. Three observers (2 oculoplastic surgeons and 1 ophthalmology resident) independently recorded eyelid fissure height, margin-reflex distance, and inferior scleral show for each eye. MAIN OUTCOME MEASURES: Intraobserver reliability and interobserver reliability of eyelid parameter measurements, as described by the intraclass correlation coefficient (ICC) and Bland-Altman plots. Agreement between digital image measurements of the investigators and clinical measurements taken on the same day as the photographs also was assessed. RESULTS: Excellent intraobserver agreement was found for the measurement of all eyelid parameters for all 3 investigators (ICC range, 0.93-0.99). Interobserver agreement for all eyelid parameters was also excellent for all investigators (ICC, 0.86-0.97). Agreement between the photographic and clinical measurements for eyelid parameters was fair to moderate (ICC range, 0.38-0.62). CONCLUSION: Measurement of several eyelid parameters in Graves' ophthalmopathy patients from computer-based digital images is reliable. Associations between photographic and clinical measurements for all parameters are weaker. Relative to clinical measurements, the photographic technique offers the advantages of potential for masking and ease of transmission that might be useful in clinical trials.
