Reproducibility of measurements with the nerve fiber analyzer (NfA/GDx).

PURPOSE To determine the reproducibility of measurements with the Nerve Fiber Analyzer, a scanning laser polarimeter designed for quantifying glaucoma in healthy patients and patients with glaucoma. The authors also assessed the variance of measurements between instruments. METHODS Measurements were made with the third generation Nerve Fiber Analyzer, the GDx. The study consisted of three parts. In the first part, the authors measured the right eyes of 10 healthy volunteers on 5 consecutive days. In the second part, 45 patients with glaucoma underwent Nerve Fiber Analyzer measurements of one randomly selected eye on two separate days in a 5-week period. For all 14 available parameters, reproducibility of measurements was expressed in terms of 95% limits of agreement and as the intraclass correlation coefficient. The Nerve Fiber Analyzer software has an option of creating a mean image from a selection of single images; for both parts of the study, the reproducibility of measurements was calculated for a "single image," and a "mean-of-three" image. In the third part of the study, 17 volunteers underwent repeated Nerve Fiber Analyzer measurement sessions on each of three different instruments. Using multivariate analysis of variance, the authors determined the variance of measurements between instruments. RESULTS The reproducibility of measurements varied considerably across parameters. Limits of agreement in mean images for superior maximum and inferior maximum were 7.2 microm and 7.7 microm, respectively in the healthy volunteers, and 8.7 microm and 7.9 microm, respectively in the patients with glaucoma. For healthy patients, the intraclass correlation coefficient was greater than 90% in 10 of 14 parameters. In patients with glaucoma, the intraclass correlation coefficient was greater than 90% in 13 of 14 parameters. Some parameters reproduced better in a mean than in a single image; these differences, however, were small and generally not statistically significant. The between-instruments component also varied across parameters and was highest in ratio-based parameters. CONCLUSIONS The reproducibility of measurements varied across parameters. In general, the reproducibility of measurements with the Nerve Fiber Analyzer was high. The reproducibility of measurements was similar between healthy patients and patients with glaucoma. Any measured change in nerve fiber layer thickness would be statistically significant if it exceeded approximately 7 or 8 microm in the superior maximum or inferior maximum parameter in healthy patients. Reproducibility of measurements hardly differed between single images and mean images. The reproducibility of measurements among the three instruments we used was highest for straight parameters.