Comparing Repeatability and Agreement between Commonly Used Corneal Imaging Devices in Keratoconus

SIGNIFICANCE Keratometry and pachymetry are important for diagnosis and monitoring of keratoconus. Many corneal imaging units exist, but comparison to determine repeatability and agreement is paramount for adequate keratoconus management in primary and secondary care. PURPOSE This study aimed to evaluate the repeatability and agreement of a Placido disc–based videokeratoscope (Melbourne, Victoria, Australia) used in primary care with spectral-domain optical coherence tomography (Optopol Technology, Zawiercie, Poland) and Scheimpflug corneal tomography (Oculus, Wetzlar, Germany) in secondary care. METHODS This was a prospective single-center study where one eye was randomized to have central corneal thickness;thinnest corneal thickness (TCT); and maximum, mean, steep, and flat keratometry (Kmax, Kmean, Ksteep, and Kflat, respectively), measured with all three devices. Three measurements were completed per device to assess intraobserver repeatability. RESULTS A total of 110 eyes from 110 patients with keratoconus were analyzed. Repeatability was best with the Pentacam for central corneal thickness, Kmax, Kmean, Ksteep, and Kflat parameters (precision, 9.21, 0.8, 0.38, 0.52, and 0.58). The Medmont had better repeatability than the Revo with Kmax, Kmean, Ksteep, and Kflat (precision, 1.41,1.35,1.43, and 1.59). Revo had the best repeatability with TCT (precision, 3.81). The intraclass correlation coefficient was >0.94 for all parameters in all devices. Agreement was generally poor between devices. However, there was good agreement between Pentacam and Medmont Kflat measurements ( P > .05). CONCLUSIONS Repeatability of keratometry parameters with the Pentacam and Medmont was greater than the Revo, suggesting a lower threshold for detecting change in anterior corneal parameters. The Revo had the greatest repeatability for TCT, suggesting a lower threshold for detecting thinning in disease progression and corneal-crosslinking safety. There was poor agreement between devices, and it is not recommended for them to be used interchangably.