advertisement

---

Editors Selection IGR 18-3

Examination methods: Event- versus trend-based criterion for progression

Comment by Anders Heijl on:

20967 Detection of visual field progression in glaucoma, Kovalska M; Grieshaber MC; Schötzau A et al., Klinische Monatsblätter für Augenheilkunde, 2008; 225: 342-345

Find related abstracts

---

In a large retrospective study of 460 eyes of 251 glaucoma patients Kovalska et al. (689) compared detection of visual field progression with an event-based criterion and trend-based criterion. For inclusion patients had to established glaucoma with a manifest SAP visual field defect (defined by a conventional cluster criterion) plus a characteristic glaucomatous optic disc appearance. Studied eyes had to have at least nine visual field tests. Since the definition of criteria is so very important for the results, here is an exact quote of the authors' definition of progression: 'Progression was determined based on an event analysis. The threshold value of each test point location in every follow-up field was compared to an average of the threshold value¹ of the first field. The definition of visual field progression consisted of deepening of an existing scotoma, expansion of an existing scotoma, or a fresh scotoma in a previously normal part of the visual field in three consecutive fields. A deepening of expansion of an existing scotoma was diagnosed if two contiguous test point locations had declined 10 dB from their original values, and a new scotoma was diagnosed if an alteration meeting the criteria for a visual field defect occurred in a previously normal part of the visual field. In addition, progression was assessed by means of trend analysis, based on a point-wise linear regression analysis. The definition of visual field progression with this criterion consisted of a negative slope of local threshold values reaching -1dB per year, significant at an alpha level of 1%, and confirmed in two out of three consecutive fields for at least three test locations.' The authors found progression in 11% of studied eyes, in 8% of the eyes progression was shown only by the event analysis, in 1% only by the trend analysis and in 2% by both methods. In seven of those nine eyes that showed progression with both methods, progression was demonstrated earlier with the event criterion, and in only one eye the trend criterion identified progression first. How does this compare with our present knowledge? Certainly we have had clear indications that event analyses generally identify progression sooner than trend analyses for quite some time,² and the present study provides valuable confirmation of this. So event tends to be more sensitive than trend, but what about specificity? Generally, sensitivity can be increased at the cost of decreased specificity, and, therefore, it is certainly less than optimal to select diagnostic criteria without any knowledge about specificity. It is there-fore a shortcoming of the present study, and of most similar studies, that we do not get any idea of the specificity of the two different approaches. Estimating specificity is not easy, but it can be done, by selecting certain field series and randomize time sequence,³ or through computer simulations.²

Event analyses generally identify progression sooner than trend analyses

It is interesting that progression was only identified in 11% of the eyes. The length of follow-up is not reported, and certainly 11% would be very reasonable if the period of observation was relatively short, maybe three years or so. But with a minimum of nine field tests as an entry criterion and a clinical material I suspect that follow-up was considerably longer than that. The average damage at baseline was also small ‐ 3.8 dB MD at baseline, which is considerably less than we see in our rather population-based glaucoma patient group or even among persons with undiagnosed glaucoma.⁴ So it seems likely that many of the patients in the current had manifest glaucoma in only one eye and that most eyes with manifest glaucoma were at an early stage of the disease. This could be at least part of the explanation of the low percentage of progressing patients. So this paper again strengthens the evidence that event-based analyses in general may demonstrate glaucomatous visual field progression earlier than point-wise regression techniques. We must realize, however, that variations of these types of techniques can produce very different results, so that both event-based and regression techniques can be more or less sensitive and specific.^2,3 We must also realize that only if fields are obtained with reasonable time intervals will progression be detectable within a reasonable time ‐ demonstration of progression will require a minimum of four or five tests. It is very positive that the interest in demonstrating and measuring perimetric glaucoma progression seems to be increasing. We have useful techniques for this, but at the same time definition of criteria are very critical, and there is still room for much more research in this area. Repeated perimetry produces massive amounts of data, and every improvement of techniques that were implemented in the software packages of the perimeters would be of value in today's individualized glaucoma management.

---

• ← Previous
• Next →

---