Number of stimuli as a reliability parameter in perimetry

Catch trials test patient performance during automated, static perimetry, but their adequacy to estimate reliability is uncertain even though up to 10% of the test time is reserved for catch trials. The 308 visual fields (program G1, all 3 phases, Octopus 201) of 308 eyes of 308 glaucoma, suspected glaucoma, and normal subjects were studied. The 108 visual fields (mean sensitivity > 10 dB; corrected loss variance < 50 dB2) without false responses to catch trials were considered reliable. A multiple linear regression analysis of these 108 fields was performed and revealed the following result (r2 = 0.751): Number of stimuli = 480 + (40.short-term fluctuation) + (8.8.the square root of the index corrected loss variance) - (2.2.mean sensitivity). This equation was used to estimate the number of stimuli required of a reliable subject to complete an examination. Excess stimuli would thus be a sign of reduced reliability. The difference between the estimated and the actual number of stimuli was called the 'stimulus discrepancy'. In 169 fields with false-positive and 58 fields with false-negative responses, the false-positive and false-negative responses correlated with the 'stimulus discrepancy' (r = 0.19, P = 0.014; r = 0.29, P < 0.026, respectively). The number of stimuli depends not only on reliability but also on the software and hardware of the perimeter. 'Stimulus discrepancy' may be an additional useful perimetric reliability parameter which does not require extra testing time.