Repeatability of the glaucoma hemifield test in automated perimetry

Purpose. To examine the concordance of the Glaucoma Hemifield Test and other global visual field indexes between two consecutive automated visual field tests. Methods. Normal subjects, subjects with ocular hypertension, and subjects with glaucoma had two automated visual field tests on the Humphrey Field Analyzer. The Glaucoma Hemifield Test results, mean deviation, and corrected pattern standard deviation of the two consecutive visual field tests were compared. Results. Forty-one normal subjects were tested within 1 and 2 years of each other. Four hundred seven subjects with ocular hypertension and 95 subjects with glaucoma were tested 1 year apart. The proportion of normal subjects who met a criterion for abnormality on two consecutive tests was 2.4%. The proportion of subjects with glaucoma with normal results of two tests was 10.5%. The specificity of automated visual field testing was improved from 80.8% to 89.9%, with a modest loss of sensitivity if two rather than one abnormal test result was required for entry into a clinical trial enrolling patients with glaucomatous field loss. Similarly, specificity increased from 84.2% to 89.5% if two normal tests were required for entry into an ocular hypertensive clinical trial. Among subjects with more closely spaced tests, the agreement between consecutive tests was similar for tests spaced 4 versus 12 months apart. Conclusions. Although there is concordance of Glaucoma Hemifield Test results on consecutive testing, there is enough disagreement to result in improved specificity from the use of a second test in a clinical trial setting.