Is visual field evaluation using multiple correlations and linear regressions useful? An evaluation of Delphi perimetry

· Background: Delphi perimetry is a method of visual field examination which produces a statistical estimation of the visual field by testing only four critical points of the central visual field. This study was performed to evaluate this technique for the detection of glaucomatous field loss. · Method: Patients with glaucoma and ocular hypertension underwent Delphi perimetry and Humphrey visual field analysis (HVFA) program 24-2. The visual field results of both examination were compared. · Results: Of 262 eyes from 199 patients, 120 eyes showed glaucomatous defects by HVFA and 142 were normal. Delphi perimetry showed abnormal visual fields in 107 eyes, 13 of which were false-positive results as Humphrey visual fields were normal. Delphi classified 155 fields as normal, of which 26 were false negatives as Humphrey visual fields showed glaucomatous defects. Therefore, the sensitivity of Delphi perimetry for the detection of glaucomatous visual field defect was 78% and the specificity was 91%. In the 26 false-negative eyes, the most common defect missed was an isolated paracentral scotoma or an early nasal step. Furthermore, 27 of the 94 glaucomatous eyes classified as abnormal by Delphi had defects estimated by Delphi perimetry that corresponded poorly to the field loss demonstrated by Humphrey visual field analysis. Therefore, qualitative sensitivity and specificity of Delphi perimetry for producing an accurate representation of the location, extent and defect depth of glaucomatous visual field loss would be 48.8% and 72% respectively. · Conclusion: In this study Delphi perimetry failed to give an accurate statistical estimation of the visual field in an unacceptably high number of cases; therefore, it cannot be recommended for clinical use. Received: 12 June 1997 Revised version received: 11 November 1997 Accepted: 13 November 1997