Interobserver and intraobserver reliability of a classification scheme for corneal topographic patterns.

AIMS To determine the interobserver and the intraobserver reliability of a published classification scheme for corneal topography in normal subjects using the absolute scale. METHOD A prospective observational study was done in which 195 TMS-1 corneal topography maps in the absolute scale were independently classified twice by three classifiers--a cornea fellow, an ophthalmic technician, and an optometrist. From these observations the interobserver reliability for each category and the intraobserver reliability for each observer were determined in terms of the median weighted kappa statistic for each category and for each observer. RESULTS For interobserver reliability, the median weighted kappa statistic for each category varied from 0.72 to 0.97 and for intraobserver reliability the range was 0.79 to 0.98. CONCLUSION This classification scheme is extremely robust and even in the hands of less experienced observers with minimal training it can be relied upon to provide consistent results.