Abstract
Purpose: :
to compare the ability to detect progression of three different
Methods: :
expert visual field evaluation, Glaucoma Progression Analysis (GPA) by event (GPA I) and tendency (GPA II) algorithms of the Humphrey Visual Field Analyzer (HVFA)Methods: fifty four eyes with diagnosis of open angle glaucoma were recruited from 2000 until 2004 in a prospective study. All patients were examined by static perimetry (HVFA, 24-2 SITA Standard) six monthly being followed for no less than 36 months. The two tests chosen as GPA baseline required a maximum time difference of four months, less than 30% fixation losses, 30% false negative or 15% false positive errors.Main outcome measures; presence or absence of progression as evaluated subjectively by experts instructed to follow the European Glaucoma Society Guidelines, where progression required confirmation in at least two fields; presence, suspicion or absence of progression evaluated by the GPA I algorithm where two black triangles with in the same scotoma were considered suspicious and three black triangles were considered progression; presence or absence of a significant tendency of the Visual Field Index (VFI) analyzed by the GPA II algorithm, a significant tendency was considered when VFI was worse than -1,3% per year with a standard deviation inferior to 3,5%. Coincidence in detection of progression was studied
Results: :
thirty-eight series of visual fields were finally considered. Sixteen were disregarded due to low reliability or baseline unavailability. Twenty-eight patients (73,7%) showed no progression by any method. Ten patients (26,3%) showed progression by one method or more. Three patients (7,8%) were detected by all three methods. Six patients were detected by both GPA I and expert evaluation (k=0.29). Four patients were detected by VFI significant change and GPA I (k=0.04). Four patients were detected by VFI significant change and expert evaluation (k=0.04). Two further patients showed VFI significant change alone
Conclusions: :
event and tendency progression detection algorithms agreement was poor. Combination of methods may increase the clinical relevance of the findings in both progressing and non-progressing patients