Abstract
Abstract: :
Purpose:To determine the short–term repeatability of the GDx VCC parameters and evaluate the influence of diurnal and daily variation, as well as machine–to–machine variation. Methods:Nine normal subjects were included in the study. Subjects were scanned 3 times on 5 different machines, twice a day (morning and afternoon), on 3 separate days. Independent variables were machine, time of day, day, and subject. Dependent variables were the GDx VCC output parameters. Analysis consisted of variance components estimation and repeated measures ANOVA. Results:Estimates of the standard deviation of repeated measurements using the GDx VCC are 1.25, 2.10, 2.30, and 1.67 µm for TSNIT average, superior average, inferior average, and TSNIT standard deviation, respectively. There were no substantial diurnal fluctuations in the parameters. TSNIT average was the parameter most significantly (p=0.03) influenced by diurnal effects and higher by only 0.34µm[0.06–0.63 95%CI] in the afternoon. Estimates for the variances components for the parameters were largest for subject effects and ranged from 4.79–123.82µm2. Day–to–day variation never accounted for more than 0.1% of the total variation (0.01–0.02µm2). Machine variance components ranged from 5.5–13% of the total variance and were similar in magnitude to the residual noise (0.88–8.93µm2). After accounting for the study design factors, the repeatability (residual measurement variation) accounted for 2–23% of the variation in the parameters (1.56–5.31µm2). Conclusions:Measurement variation inherent to the GDx VCC is small. Measurements are not influenced by diurnal and daily variation to a degree that is clinically relevant. Machine–to–machine variation is notable but only important in the detection of changes on the order of 6 or less microns.