Purchase this article with an account.
Megan Tuohy, Brian C Samuels, Lyne Racette; An Investigation of the Spatial Weighting Procedure Used in the Visual Field Index. Invest. Ophthalmol. Vis. Sci. 2014;55(13):5630.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
The Visual Field Index (VFI) uses a weighting procedure in which more weight is given to central and paracentral locations compared to peripheral locations. This procedure was applied to reflect the higher importance of central vision. The VFI remains relatively stable during early disease before worsening in the later stages of disease. This ceiling effect may be due in part to the weighting procedure used in the VFI. The purpose of this study was to investigate the impact of the VFI weighting procedure on the correlation between the VFI and other measures of severity.
This cross-sectional study included 1,921 eyes from 1,137 participants enrolled in the Diagnostic Innovations in Glaucoma Study or the African Descent and Glaucoma Evaluation Study. Each participant had a reliable standard automated perimetry (SAP) test on the Humphrey Field Analyzer (HFA). We created a fully automated program to replicate the HFA VFI calculation and allow for the manipulation of its weighting procedure. Weights were applied according to the three patterns shown in Figure 1. The VFI derived with each of these weighting patterns was correlated with mean deviation (MD). These analyses were performed on the full sample and on a subset with early disease defined as an MD no worse than -4dB (n = 1624).
Our automated calculation of the VFI correlated well with the HFA VFI (R 2 = 0.99). As predicted, decreasing the central weight progressively increased the slope of the relationship of VFI as a function of MD (HFA weights: 2.65 vs no weights: 2.65 vs reverse weights: 3.00). In early glaucoma, a peripherally weighted VFI calculation system as a function of MD increased the slope of the linear model compared to the HFA weighting (1.657 vs 0.965) while maintaining the linear fit (R 2 0.456 vs 0.462, respectively).
Our study showed that increasing the weight given to peripheral locations in the VFI calculation of early disease increases the severity of VFI depression; however, it did not entirely eliminate the ceiling effect. This suggests the ceiling effect is due to more than the weighting procedure utilized by the HFA. The assignment of a value of 100% for test locations that do not show statistically significant depressed sensitivity may mask the impact of early defects in which only a few locations are affected and limit early disease detection.
This PDF is available to Subscribers Only