ADVANCED GLAUCOMA INTERVENTION STUDY .2. VISUAL-FIELD TEST SCORING AND RELIABILITY

Background: The Advanced Glaucoma Intervention Study (AGIS) is a multi center, randomized clinical trial designed (1) to determine the better of two surgical management strategies for glaucoma when medical treat ment alone no longer adequately controls the disease and (2) to clarif y the clinical course and prognosis of open-angle glaucoma after surgi cal interventions have begun. The investigators of this study have dev eloped quantitative methods to assess the test reliability and measure the severity of glaucomatous visual field defects with the 24-2 thres hold program of the Humphrey Visual Field Analyzer. This report detail s these methods and presents information about long-term fluctuation d uring a time interval short enough to render change due to disease unl ikely. Methods: In AGIS, reliability assessment of automated visual fi eld tests is based on the number of questions asked; percent of fixati on losses, false-positive responses, and false-negative responses; and amount of short-term fluctuation. The AGIS visual field defect score is based on the number and depth of clusters of adjacent depressed tes t sites in the upper and lower hemifields and in the nasal area of the total deviation printout of the threshold program single-field test S TATPAC-2 analysis. The score ranges from 0 (no defect) to 20 (all test sites deeply depressed). The fluctuation in score between two tests o btained during an interval preceding AGIS surgical interventions, rang ing from 1 day to 6 weeks on 756 eyes of 562 patients, is analyzed. Re sults: The scores of the first visual field tests, which determined el igibility for the study, range from 1 to 17. Large positive or negativ e intertest fluctuations (absolute changes of 4 or more) occurred in t he second tests for 16% of eyes; more scores improved (11%) than worse ned (5%). The frequency of large fluctuations tended to increase as th e time interval between tests increased. Absolute intertest fluctuatio ns were independent of age. Conclusion: For visual field tests obtaine d with the automated perimeter, AGIS investigators have developed obje ctive, quantitative methods of scoring test reliability and severity o f glaucomatous field toss. For 16% of eyes, long-term fluctuations wer e large enough (greater than or equal to 4 points) to suggest improvem ent or worsening of the glaucomatous defect even though the interval i n which the fluctuations occurred was short enough (median time, appro ximately 1 week) to render changes in disease status unlikely.