Abstract
Purpose :
Visual acuity (VA) and contrast sensitivity (CS) are among the most used measures of visual abilities. Given their widespread presence in both clinical and experimental settings, a number of tests have been developed to measure these visual functions. However, different tests have different reliability, and are typically used once to evaluate a visual deficit and are often not validated for use in intervention studies that involve repeated testing. Here we evaluate the reliability of a sample of tests to determine their relative suitability for assessments of visual interventions.
Methods :
Thirty-four participants with normal or corrected-to-normal vision participated in a 3-session study. Sessions 1 and 2 were conducted on sequential days, followed session 3, 6 days later. Tests were the Freiburg Visual Acuity & Contrast Test (FrACT) using Landolt C’s, M&S Linear Contrast, EDTRS, and an iPad based adaptive contrast sensitivity test. Test order was counterbalanced across participants.
Results :
VA: ETDRS showed significant correlation between sessions 1 vs 2 (r(32)=0.60, p=0.0002), 2 vs 3 (r(32)=0.65, p<.0001); FrACT revealed greater correlations than ETDRS between sessions 1 vs 2 (r(32)=0.66, p<.0001), 2 vs 3 (r(32) = 0.78, p<.0001). Inter-test reliability was high between ETDRS and FrACT for Session 1 r(32)=0.55, p=0.0008, 2 r(32)=0.76, p<.0001, and 3 r(32)=0.700, p<.0001.
CS: FrACT showed significant correlation between sessions 1 vs 2 (r(32)=0.64, p<.0001), 2 vs 3 (r(32)=0.77, p<.0001). M&S Linear Sine Contrast showed a significant correlation between sessions 1 vs 2 (r(28) = 0.79, p<.0001), 2 vs 3 (r(28) = 0.90, p<.0001) for 6 cycles per degree (CPD), whereas for 18CPD, there is significant correlation for 1 vs 2 (r(28)=0.91, p<.0001) and 2 vs 3 (r(28)=0.56, p=0.0007). The iPad test showed significant correlation for session 1 vs 2 (r(23)=0.86, p<.0001) for 16CPD and (r(23)=0.62, p=0.019) for 32CPD.
Conclusions :
Results suggest FrACT may be moderately more reliable than ETDRS in measuring VA, however, EDTRS showed moderate learning and FrACT showed a decrement in performance. For CS, M&S and the iPad test showed somewhat greater reliability. Overall, test-retest reliability in CS measurement is affected by spatial frequency, with lower spatial frequencies showing higher reliability.
This abstract was presented at the 2019 ARVO Annual Meeting, held in Vancouver, Canada, April 28 - May 2, 2019.