Objective: To compare retinal nerve fiber layer (RNFL) thickness and diagnostic capability of spectral-domain (CirrusTM) versus time-domain (Stratus®) optical coherence tomography (OCT).

Methods: A total of 123 eyes of 123 subjects including 68 normal, 32 glaucoma suspect, and 23 patients with glaucoma were prospectively recruited for the study. All subjects were scanned by Stratus® and CirrusTM OCT in the same session. Average and quadrant peripapillary RNFL measurements by both machines were correlated using Spearman correlation coefficient, and agreement between testing methods was analyzed by Bland-Altman plots. The area under receiver operating characteristic curves (AUC) for glaucoma diagnosis was calculated.

Results: Average RNFL thickness were significantly thinner on the CirrusTM compared to the Stratus® OCT in normal subjects and glaucoma suspects (p<0.001), but thicker on the CirrusTM OCT in glaucoma patients, though the difference was not statistically significant (p = 0.53). There was good correlation between the measurements in all 3 groups. In normal controls, the average, superior, nasal, inferior, and temporal correlations were r = 0.668, 0.601, 0.508, 0.620, and 0.660, p<0.001, respectively. In glaucomatous eyes, the corresponding values were r = 0.560, p = 0.005; r = 0.423, p = 0.04; r = 0.117, p = 0.596; r = 0.742, p<0.001; r = 0.669, p<0.001, respectively. The 95% limits of agreement of average RNFL thickness were -30.2 to 13.8 µm. Area under receiver operating characteristic curves for diagnosing glaucoma were comparable (superior RNFL thickness by CirrusTM 0.925; average RNFL thickness by Stratus® 0.987). Highest correlated AUCs were for inferior and temporal quadrants.

Conclusions: Retinal nerve fiber layer measurements on the CirrusTM and Stratus® OCT correlate well but do not have clinically acceptable agreement between their measurements. The instruments may not be used interchangeably.