Agreement to detect glaucomatous visual field progression by using three different methods: a multicentre study.

AIM: To examine the level of agreement among nine clinicians in assessing progressive deterioration in visual field (VF) overview using three different methods of analysis. METHODS: Each visual field was assessed by Humphrey Field Analyzer (HFA), program 24-2 SITA Standard. Nine expert clinicians assessed the progression status of each series by using HFA 'overview printouts' (HFA OP), the Guided Progression Analysis (GPA) and the Guided Progression Analysis (GPA2). VF series were presented in random order, but each patient's VF remained in chronological order within a given field series. Each clinician adopted his personal methods based on his knowledge to evaluate VF progression. The level of agreement between the clinicians was evaluated by using weighted κ statistics. RESULTS: A total of 303 tests, comprising 38 visual field series of 7.9 ± 3.4 tests (mean ± SD), were assessed by the nine glaucoma specialists. When the intra-observer agreement was evaluated between HFA OP and GPA, the mean κ statistic was 0.58 ± 0.13, between HFA OP and GPA2, κ was 0.55 ± 0.06 and between GPA and GPA2 it was 0.56 ± 0.17. When the inter-observer agreement was analysed κ statistic was 0.65 for HFA OP, 0.54 for GPA and 0.70 for GPA2. CONCLUSIONS: Using any procedure for evaluating the progression of a series of VF, agreement between expert clinicians is moderate. Clinicians had higher agreement when GPA2 was used, followed by HFA OP and GPA printouts, but these differences were not significant.

Evaluation of threshold estimation and learning effect of two perimetric strategies, SITA Fast and CLIP, in damaged visual fields.

PURPOSE: The threshold estimation, learning effect, and between-algorithm differences of the Fast Swedish Interactive Thresholding Algorithm (SITA Fast), of the Humphrey Field Analyzer (HFA), and the Continuous Light Increment Perimetry (CLIP) strategy of the Oculus Twinfield perimeter were evaluated in damaged visual fields. METHODS: Twenty-one glaucomatous patients with damaged visual fields (MD worse than -8 dB) underwent Oculus Full Threshold (FT), Humphrey FT, SITA Fast, and CLIP 30-2 perimetric examinations. All the tests were repeated in a second session at least 3 days later. The point-wise differences in absolute sensitivity and of the total deviation plot values between FT and fast algorithms, between fast algorithms and the learning effect were evaluated (Wilcoxon test and Bland-Altman analysis). RESULTS: The average point-wise sensitivity difference between SITA Fast and HFA FT strategy (0.84 dB) was significantly lower than that found between CLIP and Oculus FT strategy (1.71 dB). Between-algorithm point-wise differences of the total deviation plot values of the fast strategies were not significantly different. Learning effect for SITA Fast (0.67 dB) was higher than that found for CLIP (0.39 dB). Test time for SITA (367+/-71 sec) and CLIP (453+/-98 sec) were about 55% and 35%, respectively, shorter (p<0.001) than those found with FT algorithms. The acceptance for fast algorithms and particularly for CLIP was significantly better. CONCLUSIONS: The two fast strategies, even though using very different algorithms, showed good threshold estimation compared to FT strategies with a consistent time saving in damaged visual fields.