All study examinations were performed by a single masked expert operator (SR). Before the study began, the reproducibility of data was tested by using the following procedure. The evaluator collected five scans of the plexus of a volunteer along the corneal axis, during the same day, at intervals of 60 minutes and at the end of each acquisition, he selected the best-focused image of the plexus. Masking was obtained by an independent investigator mixing these 5 images with 10 control images, and the operator was asked to grade all images considering number, tortuosity, and reflectivity of fibers. For each parameter, agreement of the five images was defined in the presence of values within ±1 from the mean value, this cutoff value being derived from previous studies on corneal CM.
14 Intraobserver agreement for image scores were calculated using the intraclass correlation coefficient (ICC): ICC = σ
a 2/(σ
2 + σ
a 2), where σ
2 = QM(e), or intraclass variability: σ
a 2 = [QM(a) − QM(e)]/
n or (interclass – intraclass variability)/number of classes.
20,21 Agreement was defined according to the guidelines proposed by Landis and Koch
21 : 0, chance agreement; 0.01–0.19, poor agreement; 0.20–0.39, fair agreement; 0.40–0.59, moderate agreement; 0.60–0.79, substantial agreement; and 0.80–1.00, almost perfect agreement. For our series of data, agreement was 0.70 for tortuosity, 0.75 for reflectivity, and 0.9 for number of fibers.