Precision of autokeratometry expressed as confidence ellipses in Euclidian 2-space

A number of studies have examined the repeatability or precision of keratometers. All studies have found exceptionally high precision in keratometry of the spherically equivalent proportion, while the joint variation in astigmatic direction and magnitude barely has been explored. The purpose of this study was to examine the simultaneous variation in astigmatic direction and magnitude between two single autokeratometries, using a Nidek ARK 2000-S autokeratorefractometer. A single autokeratometry was performed on two occasions on 50 right eyes of patients with cataracts. The precision was defined as the paired difference between these two measurements. Each net astigmatism was transformed to polar values, and uni- and bivariate polar value analysis was performed. Precision of autokeratometry, expressed as a mean difference and a 95% confidence area, was calculated as a confidence ellipse in Euclidian 2-space. The accompanying net astigmatisms with confidence perimeters were derived by point-for-point transformation from polar values. Clinical variability may overwhelmingly be attributed to patient-related factors and only to a small degree to autokeratometry itself. Clinicians should know the precision of their keratometers, both for daily use and in the planning and dimensioning of scientific studies. Our results indicate that a minimum number of 25 eyes in each group is necessary in order to detect a difference of 0.25 diopters in surgically induced astigmatism between two surgical techniques. Precision of keratometry can only be determined from polar values or similar entities and never by the use of net astigmatisms.