Abstract
Purpose: :
To compare the statistical properties of the visual field index (VFI) to those of Mean Deviation (MD), in patients with glaucoma followed over time.
Methods: :
Mean Deviation and VFI were calculated for visual fields from 109 patients followed for 10 years (median, 22 tests). Variability, rates of change, their statistical significance, and evidence for non-linear linearity in progression were compared between the two indices.
Results: :
The relationship between the MD and VFI (r=0.83, p<0.001) appeared linear, except with MDs better than -5.0 dB where 7/132 eyes (5.3%) had a VFI of 100% (ceiling effect). The predicted VFIs for fields with MD values of 0 dB, -10 dB, and -15 dB were 100%, 77%, and 60%, respectively. Rates of change of the two indices were moderately closely related (r=0.75, p<0.001), and statistically significant trends over time (p<0.05) occurred in the same number of eyes (87/204, 43%). Of the 104 eyes with significant trend on either VFI or MD, 70 eyes showed trends in both indices (kappa, 0.66). The variability over time of the VFI increased more steeply with visual field damage than that of the MD (Spearman r, -0.67 and -0.56), and there was no evidence that change over time was more linear with the VFI than the MD (pairwise comparison of Durbin-Watson statistics, Wilcoxon test, p=0.67).
Conclusions: :
The VFI provides a simple and understandable metric of visual field damage. However, it does not appear to offer substantial advantages over the MD for estimating a patient’s rate of change over time. The reliance of the VFI on pattern deviation probability values causes a ceiling effect which may reduce its sensitivity to change in eyes with early damage.
Keywords: visual fields • perimetry