Purchase this article with an account.
S. Okuyama, S. Takada, F. Tanabe, H. Nomoto, S. Hashimoto, E. Arimura, C. Matsumoto, Y. Shimomura; Estimation of Visual Field Loss Progression Using Linear Regression With Mean Defect, Corrected Mean Defect in Cluster, and Pointwise Changes in Polar Diagram. Invest. Ophthalmol. Vis. Sci. 2007;48(13):1638.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
To evaluate the performance of linear regression analysis with mean defect (MD), corrected mean defect in cluster (cluster CMD), and pointwise changes in polar diagram in EyeSuite perimetry (HAAG-STREIT) for detection of visual field (VF) loss progression.
Two hundred and fifty nine eyes with glaucoma or glaucomatous optic neuropathy were included in this study. We required all subjects to have six or more reliable VFs confirmed by Octopus program G2 with normal strategy. MD in the initial VF test of each eye was 5.0±5.1dB (-2.9~23.1dB). Linear regression analysis was performed in EyeSuite perimetry to estimate rates of progression of MD and cluster CMD based on nerve fiber bundles and pointwise changes in polar diagram. In the polar diagram, each test location corresponded to the optic disc angle of the respective nerve fiber.
Of the total 259 eyes, 101 eyes showed a significant change by the amount between -1.1 and 1.7dB/year in MD (p≤0.005). The remaining eyes had the MD slopes ranged from -0.8 to 0.5dB/year. MD increased by more than 0.5dB/year in 20 eyes and by 1.0dB/year or more in 9 eyes. In 168 eyes with a MD slope ranged from -1.0 to 1.7dB/year, at least one cluster CMD increased (p≤0.005; range, 0.1~5.7dB/year). The CMD slopes of the other clusters ranged from -5.0 to 1.0dB/year. In 61 eyes with a MD slope ranged from -0.9 to 1.7dB/year, at least one cluster CMD increased by more than 1.0dB/year. By polar trend diagram, 152 eyes (range of MD slope: -0.9~1.7dB/year) were judged with local or diffuse loss progression. Among them, 15 eyes did not show any cluster CMD increasing significantly (p≤0.005). Some of them appeared only diffuse loss, and the others appeared progression in a few test locations with or without long term learning effect. Thirty one eyes were seen with at least one increasing cluster CMD (p≤0.005) and yet, they were not determined as progressing by polar trend diagram. In most of them, long term learning effect was observed.
The cluster trend table based on the corrected defects and the polar trend diagram in EyeSuite perimetry appears more sensitive in detecting local progression than the MD slope. Comparison of these trend analyses is maybe useful to detect suspected progression and to estimate the VF changes more correctly.
This PDF is available to Subscribers Only