Jt. Barr et al., Corneal scarring in the Collaborative Longitudinal Evaluation of Keratoconus (CLEK) Study: Baseline prevalence and repeatability of detection, CORNEA, 18(1), 1999, pp. 34-46
Purpose, The multicenter Collaborative Longitudinal Evaluation of Keratocon
us (CLEK) Study is a prospective, observational study of 1,209 keratoconus
patients. We report on the prevalence of corneal scarring in these patients
. We also report on the test-retest repeatability of corneal scar documenta
tion at the slit-lamp biomicroscope by trained clinicians and by masked pho
tograph readers and on the scarring-status agreement at baseline between cl
inicians and photograph readers. Methods. Clinicians and masked photograph
readers graded each cornea as to scarring status. Patients were examined by
clinicians, and their corneas were photographed at baseline (2,297 nongraf
ted eyes of 1,209 patients) and at a repeated visit (258 nongrafted eyes of
138 patients). These photographs were evaluated by the masked readers at t
he CLEK Photography Reading Center (CPRC). Clinicians reported corneal scar
s in right eyes at baseline as "definitely not present" in 53.9%, "probably
not present" in 8.4%, "probably present" in 8.2%, and "definitely present"
in 29.4% of patients. A weighted kappa statistic of 0.83 (95% confidence i
nterval from 0.78 to 0.88) indicates that agreement is excellent between ba
seline and repeated assessments for the presence of a corneal scar by clini
cians. Results, Agreement is very good between baseline and repeated photog
raph-reader assessments for the presence of a scar, with a weighted kappa o
f 0.77 (95% confidence interval, 0.72-0.82). The kappa statistic comparing
photograph-reader scarring assessments with clinician results was 0.69 (95%
confidence interval, 0.66-0.71). Conclusion, The data also suggest better
agreement between clinicians and readers when Vogt's striae and corneal ner
ves were observed. The data also suggest better agreement when corneal stai
ning was not observed by the photograph readers. The CLEK Study protocol fo
r determining the presence of scars is highly repeatable.