Agreement of corneal thickness measured by Scheimpflug, SS-OCT, and specular microscopy in Chinese myopia

To evaluate the clinical agreement, diagnostic consistency, and measurement discrepancies of central corneal thickness (CCT) obtained via Scheimpflug imaging (Pentacam HR), specular microscopy (NIDEK CEM-530), and swept-source optical coherence tomography (IOLMaster 700) in Chinese myopic patients. This cross-sectional study included 400 right eyes of candidates undergoing preoperative refractive evaluations. CCT was measured using all three devices under standardized conditions. The Pentacam HR additionally recorded the thinnest corneal thickness (TCT). Statistical evaluations included the intraclass correlation coefficient (ICC) for measurement reliability, repeated-measures ANOVA for inter-device differences, Bland-Altman analysis for limits of agreement (LoA), simple linear regression for proportional bias, and Cohen’s kappa ( κ ) for diagnostic agreement at the high-risk clinical threshold (CCT < 500 μm). Mean CCT values were 547.97 ± 29.01 μm (Pentacam HR), 546.67 ± 29.64 μm (NIDEK CEM-530), and 544.98 ± 29.98 μm (IOLMaster 700), revealing statistically significant differences across all platforms ( P < 0.001). Despite excellent reliability (ICC = 0.979; 95% CI: 0.972–0.984) and strong linear associations ( r > 0.96), Bland-Altman analysis demonstrated mean biases of − 3.00 μm (95% CI: −3.60 to − 2.40) between IOLMaster 700 and Pentacam HR, − 1.30 μm (95% CI: −1.92 to − 0.69) between NIDEK CEM-530 and Pentacam HR, and − 1.69 μm (95% CI: −2.13 to − 1.26) between IOLMaster 700 and NIDEK CEM-530, with relatively broad 95% LoA and significant proportional biases. Notably, CCT values from the NIDEK CEM-530 and IOLMaster 700 were numerically lower than the Pentacam-derived TCT in 26.0% and 33.25% of eyes, respectively. However, all three devices exhibited substantial to almost perfect diagnostic agreement ( κ = 0.735–0.856) in identifying the same 18 eyes with CCT < 500 μm. Multivariable regression confirmed that axial length (AL) and intraocular pressure (IOP), but not spherical equivalent (SE), were independent positive predictors of corneal thickness. While strong linear associations exist among the three devices, their wide LoA and proportional biases preclude direct clinical interchangeability for absolute pachymetric calculations. A measurement discordance exists where single-point pachymetry frequently yields values thinner than the Pentacam-derived TCT. Nevertheless, the devices demonstrate substantial classification agreement at the prespecified CCT cutoff.