Assessing the utility of reliability indices for automated visual fields. Testing ocular hypertensives.

Monocular (right eye) visual fields were recorded with the Humphrey Visual Field Analyzer (30-2 Program) at baseline as well as 6 and 12 months later in 120 patients with established ocular hypertension. Indices of field reliability (fixation loss, less than 20%; false-positives and false-negatives, less than 33%) and field sensitivity (mean deviation [MD] and pattern standard deviation [PSD]) were examined. At baseline, 35% of patients exhibited low reliability (LR) fields, a figure which decreased to approximately 25% at 6 and 12 months, respectively. During this period, over 50% of patients produced at least one LR field, whereas 8.3% were unable to produce even one reliable field. Exhibition of a LR field appeared to be independent of patient age. Fixation errors, the major cause of LR fields, decreased by approximately 10% over the 12-month period; most patients had between 20 and 32% fixation errors. The incidence of significant defects identified by PSD was greater than that for MD; this was true for both reliable and LR fields. It is suggested that increasing the fixation loss criteria for assessing patient reliability to a 33% cutoff might substantially increase the percentage of fields graded reliable with minimal effect on the sensitivity or specificity of the test.