Reproducibility of Visual Field End Point Criteria for Standard Automated Perimetry, Full-Threshold, and Swedish Interactive Thresholding Algorithm Strategies: Diagnostic Innovations in Glaucoma Study

Abstract

To compare the interthreshold and intrathreshold strategy agreement of visual field end point criteria for standard automated perimetry (SAP) with the full-threshold (FT) algorithm and the Swedish interactive threshold algorithm (SITA). Prospective, longitudinal cohort study. The interstrategy group included a randomly selected eye of 173 participants in the Diagnostic Innovations in Glaucoma Study who had undergone FT algorithm and SITA analysis within three months (sequence 1, FT + SITA). Intrastrategy agreement for the FT algorithm (sequence 2, FT + FT) was tested for 44 (25.4%) participants who had undergone FT analysis within one year of the FT used in the interstrategy pairing, and for 89 patients (51.4%) who had undergone SITA analysis within one year before (sequence 3, SITA + SITA). Four different end point criteria using Statpac II indices were tested. Interstrategy agreement was compared with intrastrategy agreement using kappa statistics. FT + SITA agreement (kappa) for pattern standard deviation (PSD) < 1% was 0.82; for PSD < 5%, the kappa value was 0.64; and for four or more pattern deviation plot points, the kappa value was 0.43. Agreement with glaucoma hemifield test (GHT) results was significantly higher (P < .01) for FT + FT (kappa = 0.94) than FT + SITA (kappa = 0.67), and approached significance (P = .07) when comparing FT + FT with SITA + SITA (kappa = 0.77). GHT results were more likely to be abnormal on the SITA analysis than on the FT analysis. No other significant differences were found. To minimize misinterpreting abnormal GHT results on SITA as evidence of change when switching strategies, both SITA and FT should be performed and compared within a short period. Other indices are comparable between the two strategies.