Comparison between two fast threshold strategies: SPARK and SITA in normal subjects

Numerous fast threshold strategies have been developed in perimetry which use maximum likelihood approaches to estimate the threshold. A recent approach to threshold estimation has been developed estimating the threshold from a limited number of test points which further reduces examination time. This strategy, SPARK, has not been compared to the SITA strategy. The aim of this study was to compare SPARK with SITA in a normal cohort to evaluate within and between strategy agreement in threshold estimates.

A total of 83 normal subjects each underwent two visual field examinations with SITA and SPARK on two separate occasions on a randomly selected eye. The eye examined and the order of strategy examined first was randomised but remained constant over the two perimetry visits.

Visual field examination with SPARK Precision was on average 33% faster than SITA Standard. A positive correlation between group mean sensitivities of SITA Standard and SPARK Precision (rho = 0.713, p < 0.001) was found. In total, 95% of stimulus locations were located within the 95% limits of agreement and linear regression on the differences in sensitivities showed no statistically significant proportional bias (t = 1.713, p = 0.09). Pointwise analysis showed SITA Standard had significantly larger variability for individual stimulus locations examined over two visits when compared to SPARK (t = 9.175, p < 0.001).

The clinical examination of SPARK yields a sensitivity profile similar to SITA but in a faster examination time. The lower threshold variability of SPARK may be as a result of data smoothing in the threshold estimation process.