Editors Selection IGR 21-2

Clinical Examination Methods: An algorithm to detect visual field progression

Kouros Nouri-Mahdavi

Comment by Kouros Nouri-Mahdavi on:

86675 Detection of Progression With 10-2 Standard Automated Perimetry: Development and Validation of an Event-Based Algorithm, De Moraes CG; Paula JS; Blumberg DM et al., American Journal of Ophthalmology, 2020; 216: 37-43

Find related abstracts

De Moraes and colleagues introduce a long-awaited algorithm for carrying out pointwise event analyses (PEA) on longitudinal central 10-2 visual fields (VF) to detect change. They validated this algorithm using a cohort of patients from the ADAGES study. Only patients with established glaucoma were included and variability data were defined based on three rings of eccentricity rather than single locations.

The investigators provided evidence that progressing eyes based on the PEA showed faster rates of change compared to stable eyes, displayed low improvement rates and that the proportion of deteriorating eyes increased as a function of follow-up time. The specificity of the above criteria was estimated based on the number of tests locations displaying improvement relative to the limits of variability. While using this approach is not infrequent in the perimetry literature, it is based on the assumption that the amount of noise is symmetric for both deterioration and improvement, which may not necessarily be true. Similar to the Humphrey Field Analyzer's Guided Progression Analysis (GPA) software criteria, a diagnosis of progression was established when 3 or more test locations were found to be deteriorating.

Validation of this approach in a larger longitudinal cohort with longer follow-up and ideally in a longitudinal series of normal subjects to establish true specificity would be desirable and would pave the way for clinical implementation of this approach. My wish list, when that happens, includes as much versatility as possible for the clinician using this algorithm. I would encourage incorporating a total deviation based approach in addition to the current pattern deviation based approach and the ability to choose the number of locations required to meet worsening or improvement criteria. While a minimum of 3 worsening locations for the 24-2 strategy reflects a fairly large area of deterioration, given the much denser grid of the 10-2 VF, flagging progression with a larger number of deteriorating locations may be desirable with 10-2 VF. Highlighting the number of improving points would be an advantage; the current GPA for 24-2 does not provide this.

As the authors mentioned, the small number of eyes with more severe visual field loss at baseline could be a limitation for detection of progression in more advanced glaucoma and enlarging the database before finalizing the algorithm would be advantageous.

The authors should be commended for addressing one of the unmet needs in the realm of glaucoma diagnostics; I look forward to a full clinical implementation of this algorithm without the usual constraints imposed by perimetry manufacturers!

Issue 21-2

Change Issue