Related Experiment Video
Updated: Jul 6, 2026
Three Different Protocols of Corneal Collagen Crosslinking in Keratoconus: Conventional, Accelerated and Iontophoresis
Published on: November 12, 2015
Agreement in the classification of keratoconus severity: a comparative study based on anterior corneal surface
Cesar Villa-Collar1, Jacinto Santodomingo-Rubido2, Julia Bodas-Romero3
1Faculty of Biomedical and Health Sciences, Universidad Europea de Madrid, Madrid, Spain.
Objective:
This study aimed to analyse the agreement in the clinical classification of keratoconus using four grading systems based exclusively on anterior corneal surface topographic parameters.
Materials And Methods:
A retrospective, descriptive, analytical, cross-sectional study was conducted to evaluate agreement and differences among the keratometric scale, Amsler-Krumeich (A-K), Alió-Shabayek (A-S), and Keratoconus Severity Score (KSS) classifications. Retrospective, anonymised data were obtained from anterior corneal surface tomography, including keratometry, pachymetry, corneal aberrations, and the location of the thinnest corneal point, measured with a corneal tomographer (Pentacam®, Oculus Optikgeräte GmbH, Wetzlar, Germany). Agreement between each pair of ordinal scales was assessed using weighted Cohen's kappa coefficient with quadratic weights.
Results:
A total of 455 eyes with keratoconus were analysed. The mean age was 39.3 ± 12.2 years, with no significant sex differences (p > 0.05). Mean keratometry was 46.5 ± 5.9 D and mean pachymetry was 474.8 ± 70.2 μm. Only 111 eyes could be simultaneously classified across all four systems. Pairwise agreement analysis using weighted kappa revealed generally low concordance between classification systems. The highest agreement was observed between the keratometric and Amsler-Krumeich classifications (moderate agreement), whereas comparisons involving aberration-based systems showed weak to slight agreement. When classifications were based solely on each system's differential parameter, a similar pattern was observed, with moderate agreement between keratometric and pachymetry-based classifications and fair agreement in the remaining comparisons.
Conclusion:
Keratoconus classification systems based on anterior corneal surface parameters show significant discrepancies and cannot be used interchangeably. Even when reduced to their primary defining parameters, agreement remains limited, indicating that these systems capture different dimensions of keratoconus severity.