Abstract
Purpose :
SITA Fast (SF) is commonly used due to its shorter duration but SF measurements are thought to produce more variability compared to SITA Standard (SS). This study measures the visual field (VF) variability across a range of sensitivities of the 2 SITA strategies.
Methods :
A total of 1376 30-2 SS VFs and 1376 30-2 SF VFs from 344 glaucomatous eyes (344 subjects) previously recruited from 9 glaucoma clinics around the world were analyzed retrospectively. Each study eye had 4 SF repeated VFs and 4 SS repeated VFs, taken over a period of 4 weeks (both tests on each study eye per week). The sensitivity at each VF test location for each subject’s eye was fitted with an ordinary least squares regression to find the regression model and the least squares residuals. The 2 test locations near the blind spot were excluded, thus giving 74 threshold locations for each VF. Fitted sensitivity values were obtained from each regression model. The least squares residuals were then grouped according to the fitted sensitivity value (rounded to the nearest decibel). This method was used separately on SF VFs and SS VFs. To measure the precision of the 2 SITA strategies, the standard deviation (SD) of the residuals at each binned/rounded sensitivity was plotted. All statistical analyses were performed in MATLAB.
Results :
Average MD, PSD, and VFI values of the study population are reported in Table 1. Overall, there is little difference in precision between the 2 SITA strategies. The precision of SF was slightly lower in the mid-range sensitivities from approximately 7 dB to 20 dB compared to SS, according to Figure 1. The largest difference in precision was at 18 dB, where the magnitude of difference in SD was approximately 0.8 dB.
Conclusions :
The results obtained were comparable to the results Saunders et al reported in their analysis of precision measurements of the 2 SITA strategies in a longitudinal study [1]. Since our data came from a repeatability study, it is not as dependent on specific longitudinal modeling choices. However, we can draw a similar conclusion that SS is slightly more precise of a testing algorithm than SF at lower VF sensitivities, but the magnitude of difference is small. [1] Saunders et al., 2015. JAMA Ophthalmol., 133(1):74-80.
This is an abstract that was submitted for the 2016 ARVO Annual Meeting, held in Seattle, Wash., May 1-5, 2016.