Standard or Fast? - Differences in precision between SITA Standard and SITA Fast testing algorithms and their utility for detecting visual field deterioration

Abstract

 

Purpose

 

In Humphrey visual field (VF) testing measurements from SITA Fast (Swedish Interactive Testing Algorithm) are thought to be more variable than those from SITA standard, yet some clinical centers routinely use SITA Fast on patients because the test is quicker to administer. This study sought to examine the relative precision of the two strategies across a range of VF sensitivities.

 

Methods

 

This study retrospectively analyzed 45,214 24-2 SITA Standard (7,335 eyes) and 13,808 SITA Fast VFs (2,134 eyes) from patients attending clinical centers in England (1997 to 2012). Pointwise linear regression (PLR) of measured sensitivity over time was conducted using VF series from a randomly selected eye of each patient. PLR residuals were pooled according to fitted sensitivities. The standard deviation of the residuals was used as an estimate of ‘measurement precision’ at different sensitivities and these were compared for the SITA Standard and Fast algorithms. Simulations of VF progression (Russell et al 2013 PLoS ONE) were used to evaluate the impact of this difference in precision on time to detect VF loss.

 

Results

 

Measurement precision worsened as sensitivity decreased for both test strategies (see Figure). The residuals for SITA Fast were significantly lower than for SITA Standard (p < 0.001) though this difference was only apparent at sensitivities below 20dB as variability in both methods peaked. Differences in measurement precision, once translated to time to detection, were modest resulting in minimal differences in diagnosis of progression.

 

Conclusions

 

SITA Standard is more precise than SITA Fast at lower VF sensitivities, but this difference is unlikely to make a sizeable difference in monitoring deterioration. SITA Fast may have lower accuracy than SITA Standard, but this was not tested here. This study highlights the usefulness of ‘big’ VF data in allowing investigations to be based on thousands of clinic patients rather than the tens of patients in test-retest studies.

 

 

Relationship between visual field variability and fitted pointwise sensitivity measured using the SITA Standard (black points) and SITA Fast (red points) testing algorithms. The dotted lines represent 95% confidence intervals of the standard deviation of residuals. The blue background is a smoothed colour density representation of the scatterplot for all residuals.

 

Keywords: 758 visual fields • 642 perimetry