Gonioscopy in the dog: inter-examiner variability and the search for a grading scheme.

Abstract

To investigate inter-examiner variability in gonioscopic evaluation of pectinate ligament abnormality in dogs and to assess level of inter-examiner agreement for four different gonioscopy grading schemes. Two examiners performed gonioscopy in 98 eyes of 49 Welsh springer spaniel dogs and estimated the percentage circumference of iridocorneal angle affected by pectinate ligament abnormality to the nearest 5%. Percentage scores assigned to each eye by the two examiners were compared. Inter-examiner agreement was assessed following assignment of the percentage scores to each of four grading schemes by Cohen's kappa statistic. There was a strong positive correlation between the results of the two examiners (R=0·91). In general, Examiner 1 scored individual eyes higher than Examiner 2, especially for eyes in which both examiners diagnosed pectinate ligament abnormality. A "good" level of agreement could only be achieved with a gonioscopy grading scheme of no more than three categories and with a relatively large intermediate bandwidth (κ=0·68). A three-tiered grading scheme might represent an improvement on hereditary eye disease schemes which simply classify dogs to be either "affected" or "unaffected" for pectinate ligament abnormality. However, the large intermediate bandwidth of this scheme would only allow for the additional detection of those dogs with marked progression of pectinate ligament abnormality which would be considered most at risk of primary closed-angle glaucoma.