Quantitative classification of visual field defects in early glaucoma.

Objective: While qualitative visual field (VF) grading systems enable the classification and description of VF defects in glaucoma, they may be prone to subjective interpretation and interobserver disagreement. In this study we characterise the types and frequencies of VF defects in a large cohort of subjects with early glaucomatous loss, using a quantitative method based on the Ocular Hypertension Treatment Study (OHTS) VF classification system.

Design: Cross-sectional study SUBJECTS: 1330 eyes of 733 subjects with healthy, suspect or early glaucoma METHODS: We translated the OHTS system into an objective, quantitative method to classify the VF of each eye into nerve fibre bundle (bundle) and non-bundle defects comprising 14 patterns. This was applied to VF tests from subjects who received two 24-2 SITA-Faster VF tests per eye on the same visit.

Main outcome measures: Distribution of initial and repeatable VF patterns across cohort, and the relationship with mean deviation and test reliability.

Results: The mean baseline MD and PSD was -1.08 (SD 1.83) and 2.26 (SD 1.26) respectively. The most common repeatable pattern was the inferior nasal step (22.6%), followed by the inferior enlarged blind spot (19.1%), superior nasal step (12.6%) and superior enlarged blind spot (11.6%). The frequency of occurrence of initial defects was significantly lower than repeatable defects for the majority of patterns. The frequency of nasal step defects decreased as MD worsened, with a corresponding increase in frequency of arcuate defects. There was a consistently higher frequency of bundle and non-bundle defects in unreliable versus reliable tests based on a false positive rate above 15%, including an inferior enlarged blind spot (30.0% vs 20.8%, P < 0.001), superior central defect (13.7% vs 5.5%, P < 0.001) and partial superonasal quadrant (14.4% vs 7.2%, P < 0.001) defects.

Conclusions: In a large cohort of suspect and early glaucoma, the most common VF defects were nasal step, enlarged blind spot and arcuate defects, with a significant difference in distribution between initial versus repeatable defects and reliable versus unreliable tests. Application of objective criteria may improve the accuracy and consistency of classifying VF defects.