Abstract
Purpose :
There is currently no consensus on visual field (VF) criteria that define glaucoma. The purpose of this study was to compare the hit rates and consistency of 5 VF criteria for glaucoma: Glaucoma Hemifield Test (GHT), Hoddap-Anderson-Parrish 2 (HAP2), Foster, United Kingdom Glaucoma Treatment Study (UKGTS), and Low-pressure Glaucoma Treatment Study (LoGTS).
Methods :
We retrospectively paired VF (Humphrey 24-2 SITA) and OCT (Spectralis) exams that were performed within 4 months of each other from a 2-year period in our Glaucoma Clinic, with non-glaucoma pathologies excluded. One eye per patient was randomly selected. We measured global and sectoral averages of optic nerve head minimum rim width, circumpapillary retinal nerve fiber layer thickness, and macular ganglion cell layer thickness with OCT, and considered them abnormal when thinner than the 1% percentile of the age-adjusted normative value. We created an OCT abnormality score ranging from 0 (no average abnormality in any structure) to 6 (all three structures with 2 or more averages abnormal). Criteria specificities were inferred from hit rates in patients with score of 0, while sensitivities were inferred from hit rates in patients with higher scores. Criteria consistencies were inferred from follow-up mismatch rates, defined as the proportion of patients with an initial VF having positive results and the following VF having negative results.
Results :
We obtained 1230 eyes with a mean (standard deviation) age of 67.1 (12.3) years, and a mean VF mean deviation of -3.34 (4.78) dB. In eyes with OCT score of 0, HAP2 and UKGTS had higher hit rates indicating low specificities, while GHT, Foster, LoGTS had lower hit rates indicating high specificities (Fig. 1). In eyes with OCT score of 6, HAP2 and UKGTS had higher hit rates indicating high sensitivities while GHT, Foster, and LoGTS had lower hit rates, indicating low sensitivities. The hit rates at other OCT scores followed the same trend. There was a large overlap among criteria (Fig. 1), however, this decreased with lesser VF damage. Follow-up mismatch rates were higher in eyes with lower OCT scores (Fig.2).
Conclusions :
Our results highlight the lack of a criteria with a uniformly superior diagnostic performance compared to others. The selection of criteria for study design would therefore depend on the degree of damage anticipated, and need for either better sensitivity or specificity.
This is a 2020 ARVO Annual Meeting abstract.