Effect of the number of threshold determinations on short-term fluctuation in automated perimetry

The authors studied the effect of multiple threshold determinations on local and global (root-mean-square) short-term fluctuation in 30 subjects (10 controls, 10 glaucoma suspects, and 10 glaucoma patients) with a mean age of 56.70 years. Five custom programs were designed on the Octopus perimeter in which the total number of threshold determinations was equal but in which thresholds at 2 principal locations were determined between 2 and 15 times. One randomly selected eye of each subject was first tested with the standard Octopus programs G1 and 31 followed by the custom programs. Local fluctuation initially increased with the number of threshold determinations and then stabilized after five determinations. Global fluctuation was not influenced by the number of determinations and the values obtained by the shorter custom programs were not significantly different from those obtained by the standard programs. Programs that determine thresholds only twice at a given location may underestimate local fluctuation. Furthermore, local fluctuation is de-emphasized by the "watering-down" effect of estimating global fluctuation.