Dependability of posterior-segment spectral domain optical coherence tomography for measuring central corneal...

The aim of this study was to assess the intersession and interobserver reliability of central corneal thickness (CCT) measurements using a posterior-segment spectral domain optical coherence tomography (SD-OCT).

Sixty patients were examined with a posterior-segment SD-OCT system (Topcon 3D-OCT). To analyze intersession reproducibility, 1 examiner measured CCT in 60 eyes 2 times 1 week apart, and to study interobserver reproducibility, a second examiner obtained another measurement during the first session. Within-subject reproducibility, coefficient of reproducibility (CR), and intraclass correlation coefficients (ICC) were calculated. Agreement between Topcon 3D OCT and ultrasound pachymetry (USP) was also assessed.

Posterior-segment SD-OCT showed an intersession reproducibility of 12.7 μm (CR = 2.4%) and an excellent ICC (0.96). No systematic difference between sessions was found (average difference, 1.7 μm; P = 0.15). We found no significant bias between observers (mean difference, -1.5 μm; P = 0.29). The interobserver reproducibility was 16.4 μm (CR = 3.1%), and the ICC was 0.94. Topcon 3D OCT CCT values were lower than USP readings (mean difference, 7.6 ± 8.1 μm; 95% confidence interval, 9.9-5.3 μm; P < 0.001).

With a commercially available retinal SD-OCT system, reliable intersession and interobserver CCT measurements can be feasibly obtained. The criteria for a significant change, which would be the one exceeding the reproducibility, indicate that intersession and interobserver variations in CCT of more than 13 and 16 μm, respectively, may reflect true corneal change with this particular device. These estimates should help investigators and clinicians differentiate actual CCT modification from measurement random error. The slight CCT underestimation with respect to USP is within the range of other OCT devices.