Confidence Intervals for Assessing Ectatic Changes Post-Crosslinking Using the ABCD Grading System

Abstract

Purpose : The ABCD progression display (Pentacam, Oculus GmbH, Wetzlar, Germany) monitors progression in corneal ectasia and reports statistically significant changes by displaying one-sided 80% and 95% confidence intervals (CI) of each measured parameter. The parameters are anterior (A) and posterior (B) radii of curvature taken from a 3.0 mm optical zone centered on the thinnest point and thinnest pachymetry reading (C). The current system does not display CI after corneal cross-linking (CXL), as measurement noise of this population has not been established. Our prior study on noise post-CXL showed large CI in eyes with less than a year follow-up period. This study aimed to measure noise of post-CXL eyes after a year or longer and provide CI for this group.

Methods : Patients from ELZA Institute (Zurich, Switzerland) and Homburg Keratoconus Center (Homburg, Germany) with a minimum of 12 months post-CXL were enrolled. Three separate Pentacam measurements were taken for each eye, removing the patient from the Pentacam device between each measurement. A minimum 7.5 mm of coverage and an acceptable quality score were required. Pooled variance, standard deviations (SD), and one-sided CI were computed. Site specific and time specific comparisons were made using nonparametric statistics.

Results : 60 eyes (38 Zurich, 22 Homburg) of patients aged 29.5 ± 12.3 (range 11-62) were enrolled. Patients were 26.1 ± 19.3 months (range 12-28 Zurich, 16-115 Homburg) post-CXL. The values of A, B, and C parameters and their respective noise measured as SD were all comparable between sites. A subgroup analysis of eyes from Homburg comparing 16-35 months to 38-115 months post-CXL showed no difference in SD. Spearman’s Rho demonstrated no correlation between time post-CXL and SD in all parameters. The 80% CI of A, B, and C parameters were 0.0283, 0.0492, and 3.682, respectively. The 95% CI were 0.0553, 0.0962, and 7.196, respectively.

Conclusions : Measurement noise of each parameter on the ABCD progression display was analyzed in eyes post-CXL with at least 12-month follow-up period. Noise, after 12 months post-operatively, was not correlated with time post-CXL in this group. CI of each measured parameter were computed as above and will be incorporated into the next iteration of the Belin ABCD progression display.

This is a 2021 ARVO Annual Meeting abstract.