Abstract
Purpose: :
Expert interpretations of visual field tests are commonly used as a gold standard for the diagnosis of glaucoma and the determination of disease progression. The purposes of this study were to (1) generate custom, but familiar representations of visual field test data; (2) develop a web-based application that would permit remote experts to review and classify visual field test results; (3) evaluate the test-retest reliability of expert graders using this system.
Methods: :
Previous studies with normal populations have allowed us to faithfully approximate age-adjusted visual field reports of the Humphrey Visual Field analyzer (HFA). Custom Matlab routines were written to analyze raw numeric data exported from the HFA. A secure web interface provided remote experts access to the graphical visual field reports. Results from their reviews were posted to a SQL database and analyzed. Data from the worst eye (largest MD) of high-risk glaucoma suspects or early glaucomatous patients were selected for study (n=200). A longitudinal visual field sequence (up to 8 years) was graded for the presence of a visual field defects and for disease progression. A common set of reports were graded by two experts on two occasions separated in time by at least six months. Intra and inter-rater reliability was evaluated using weighted kappa statistics.
Results: :
Graphical reports were generated to display the calculated mean deviation (MD), total deviation (TD), pattern standard deviation (PSD), their associated probabilities as well as the glaucoma hemi-field test (GHT). Each expert successfully reviewed and evaluated more than 1800 individual visual field reports. Test-retest reliability was excellent for both experts (98% and 97% respectively, both p <.0001). Agreement between graders was excellent as well (98%, p < .0001).
Conclusions: :
We successfully implemented a web-based visual field evaluation system that transforms raw exported data into familiar summary reports. This system facilitates distance collaboration with remote experts and provides a reliable and efficient means of capturing expert assessments for further quantitative analysis.
Keywords: visual fields • clinical (human) or epidemiologic studies: biostatistics/epidemiology methodology