Fluctuations of visual field interpretation related to cluster and scotoma analysis by one interpreter.

A total of 189 numerical printouts of the visual fields of 189 eyes of individuals either suspected of or verified as having glaucoma and displaying varying degrees of glaucomatous visual field damage were analyzed by one expert interpreter. The amount of clusters (or scotomas) was counted and delineated in each visual field under both a more stringent and a more liberal criterion. The interpretations of the visual fields differed significantly in both groups, as did the subjective certainty the interpreter allotted to his assessment. In the group of certain decisions, only 144 defects were assessed as opposed to 364 when a more stringent rather than a more liberal criterion was adopted. In uncertain cases, a reverse trend was observed, since only 11 as opposed to 324 defects were assessed when a more liberal rather than a more stringent criterion was adopted. Both trends may be explained by an increase in sensitivity at the expense of specificity (i.e., a change in the receiver-operating characteristic of the expert when the latter is pressed to change the criterion in a subjective cluster or scotoma assessment task). The ambiguity of the task and the difference in the criterion adopted by the interpreter significantly influences the assessment of normality and pathology, in particular its extent and spatial distribution.