WITHIN AND BETWEEN OBSERVER VARIABILITY OF CONTOUR LINE DRAWING FOR SCANNING LASER TOMOGRAPHY

Title WITHIN AND BETWEEN OBSERVER VARIABILITY OF CONTOUR LINE DRAWING FOR SCANNING LASER TOMOGRAPHY
Author, Co-Author Jr. Dul, Richard Madonna, Tonya Babjak, William Swanson
Topic
Year
2004
Day
Program Number
Poster 106
Room
Affiliation
State University of New York, College of Optometry
Abstract PURPOSE: To assess the variability in scanning laser tomography (SLT) indices caused by between-clinician and within-clinician variability during the interpretation and drawing of the contour line. Prior work has focused on the variability of the SLT global indices and reported variability across subjects in mm2. We analyzed data for individual sectors and expressed variability as percent of mean normal.

METHODS: SLT images were obtained using an HRT II from 24 eyes of 24 subjects free of eye disease, who ranged in age from 24 to 67 years (mean ± SD = 47 ± 14). Three clinicians, with varying degrees of experience, drew disc outlines for each image, and one repeated the assessment 2 weeks later. For each disc outline, indices were stored electronically then the outline was removed, so subsequent analyses were drawn independently of the others. Analysis of agreement within and across clinicians was performed for global and sector values for disk area, rim area, and rim volume, and for reference height, giving a total of 22 indices. For each index, both accuracy and precision were expressed as percent of mean normal value for the index. In addition, we computed the ratio of the standard deviations of the temporal versus nasal indices (rim disc area, rim area and rim volume).

RESULTS: Results for individual sectors were comparable to those for global indices. Accuracy was considerably better for within-clinician than between-clinician variability (3% vs 7-17%). Despite the common differences in the appearance of the neural retinal rim between nasal and temporal sectors, variability was similar: the median ratio of variabilities for the temporal versus nasal indices was 1.1 (range 0.7-1.5).

CONCLUSIONS: In clinical research settings, where more than one clinician may be involved in the care of individual patients, variability of SLT data can be reduced by having a single clinician produce disc contour lines.
Affiliation of Co-Authors State University of New York, College of Optometry, State University of New York, College of Optometry, State University of New York, College of Optometry
Outline