Purpose: To formulate and test an algorithm using minimal topographic criteria for accurately diagnosing clinical keratoconus.
Setting: Subspecialty cornea practice and Keratoconus Genetic Research Project.
Methods: Both eyes of 86 keratoconic patients who had never worn contact lenses and 195 normal participants were studied with the TMS-1 videokeratoscope to evaluate the KISA% index, an algorithm that topographically quantifies the phenotypic features of keratoconus. The diagnostic efficacy of the KISA% index was compared with that of the modified Rabinowitz/McDonnell (K- and I-S values) and the Maeda/Klyce (KCI% and KPI) indices. The same indices were calculated for an additional 8 eyes with keratoconus-suspect topography and 12 eyes with early keratoconus.
Results: The mean KISA% was significantly greater in the keratoconus group (10,382%) than in the normal control group (20.44%) with minimal overlap. At a cutoff point for KISA% of 100, 280 of 281 participants (99.6%) were correctly classified. In contrast, the correct classification rate for the other indices were KCI%, 274 of 281 (97.5%); KPI, 249 of 281 (88.6%); K, 272 of 281 (96.8%); I-S, 269 of 281 (95.7%). Six of the 8 eyes with keratoconus-suspect topography had a KISA% between 60% and 100%, and 11 of the 12 eyes with early keratoconus had a KISA% greater than 100%.
Conclusions: The KISA% index set at 100 was highly sensitive and specific for diagnosing keratoconus; a range of 60% to 100% may be useful for designating suspects. This index is more useful than any of the other currently available tools for classifying patients with keratoconus for computerized segregation analysis and for distinguishing eyes with keratoconus from normal eyes in topographic screening of refractive surgical candidates.