Interexaminer Reliability of Meibography: Upper or Lower Eyelid?

Abstract

OBJECTIVES: To investigate the interexaminer reliability of meibography evaluation and the impact of eyelid selection for the procedure. METHODS: Meibography images of 30 subjects' right eye, including both lids, were evaluated by three clinicians. Meibomian gland (MG) dropout was noted with respect to both percentage and grading (0-4). The examiners' (E1, E2, and E3) consistency was assessed by the intraclass correlation coefficient for MG loss percent and weighted kappa for grading. Correlations between the meibography findings and the fluorescein tear break-up time (FBUT), ocular surface disease index (OSDI) questionnaire, and noninvasive tear break-up time (NIBUT) data were also analyzed. RESULTS: The patients' mean age was 49.2±13.4 years, and the male/female ratio distribution was 15/15. Mean FBUT was 9.6±3.1 sec. OSDI and NIBUT were not correlated with meibography findings. FBUT measurements were correlated with the MG loss assessments of all examiners for the upper lid (Spearman correlation test, E1: r=-0.617, P<0.001; E2: r=-0.428, P<0.018; E3: r=-0.429, P=0.018) but not for the lower lid. Weighted kappa statistics based on MG dropout grading revealed moderate-to-good agreement for the upper lid (E1-E2: κ=0.516, P<0.001; E1-E3: κ=0.621, P<0.001; E2-E3: κ=0.650, P<0.001) and fair-to-moderate agreement for the lower lid (E1-E2: κ=0.212; E1-E3: κ=0.272; E2-E3: κ=0.530). CONCLUSIONS: Meibography is an objective tool for the detection of MG loss. Because of the correlation with FBUT and better interexaminer agreement on dropout by grading, the upper lid might be the preferred lid to make an evaluation.