Reliability of corneal thickness and endothelial cell density measures

PURPOSE: We evaluated the reliability and agreement between the Orbscan, an ultrasonic pachymeter (Humphrey 855), and the Konan SP 9000-LC in terms of central corneal thickness. The Konan was also used to study the reliability and agreement between endothelial cell density measures. METHODS: Twenty-five normal subjects were examined on two occasions (mean separation = 9 +/- 5 days) by a single examiner using all three instruments for central corneal pachymetry. The Konan Center Method and a manual counting method were performed by two examiners to determine endothelial cell density. Reliability and agreement were assessed by calculating the 95% limits of agreement (LoA) and intraclass correlation coefficients (ICC). RESULTS: For corneal pachymetry test-retest reliability, the 95% limits of agreement were -20 to +17 mum for the ultrasound, -27 to +22 mum for the Konan, and -13 to +13 mum for the Orbscan. There was fair-to-good agreement between the pachymeters (intraclass correlation coefficients range = 0.85 to 0.92). For endothelial cell density test-retest reliability, the 95% limits of agreement for the Konan Center Method was -498 to +530, and -482 to +333 cells/mm(2) for examiners 1 and 2, respectively. The test-retest 95% limits of agreement for the manual overlaid grid method was -355 to +355, and -535 to +670 cells/mm(2) for examiners 1 and 2, respectively. CONCLUSIONS: The reliability and agreement of the Orbscan and Konan corneal pachymeters was good, although the reliability of the Konan for estimating endothelial cell density was fair, at best.