Salazar et al. retrospectively compared three pointwise methods of detecting visual field (VF) progression in 729 eyes of 567 POAG patients that had at least six reliable VFs and at least three years follow-up. Methods evaluated were the Early Manifest Glaucoma Trial progression event analysis (EMGT), Pointwise Linear Progression (PLR) and their recently-described Glaucoma Rate Index (GRI).1-5
The authors describe the GRI as being a trend-based method to measure VF progression, in which an exponential regression is performed at individual test point locations. Pointwise rate of change was expressed as the fraction of each test point's entire perimetric range that is lost or gained per year, corrected for age and location. An overall rate index is generated by summing all statistically significant pointwise rates of change, which were then normalized relative to the maximum possible rates of decay or improvement.
On the other hand, one might hypothesize that considering eyes progressing so markedly as to be detected by the AGIS method may not provide a realistic assessment of sensitivity to subtle change. Similarly, eyes having visual fields that are so unvarying that they are found to be stable in a PoPLR analysis may be presenting a test of specificity that is unrealistically undemanding.
The main outcome measures of this study were the proportion of VF series detected as progressing, estimates of relative specificities, time to detect progression, and agreement among measures. The authors found the GRI to be a sensitive and specific method that can detect long-term visual field progression events in glaucoma earlier than pointwise linear regression and the EMGT method.
Perhaps the most novel aspect of this study is the manner in which the authors compared the specificity and sensitivity of each of the studied methods. The authors compared findings from the three methods to findings using the highly specific AGIS scoring system.5,6 Eyes progressing according to the AGIS method were defined as a reference group with likely progression that could then be used to find a surrogate measurement for sensitivity. Similarly, the authors used a highly sensitive progression analysis, PoPLR7 to establish a reference group with few false-negative findings as a surrogate measurement for specificity. While this approach may not provide accurate estimates of true sensitivity or specificity, it does allow preliminary comparison of the three methods. On the other hand, one might hypothesize that considering eyes progressing so markedly as to be detected by the AGIS method may not provide a realistic assessment of sensitivity to subtle change. Similarly, eyes having visual fields that are so unvarying that they are found to be stable in a PoPLR analysis may be presenting a test of specificity that is unrealistically undemanding. Certainly, this idea deserves further assessment.
We look forward to learning how GRI's estimations of progression rates compare to current methods, and how GRI might be used when making Quality of Life-based therapeutic decisions.