Abstract
Purpose :
The contrast sensitivity function (CSF) is compromised in many visual neuropathologies. An expert panel proposed acuity and contrast sensitivity as endpoints for public health surveillance of visual impairment[1]. Such endpoints should exhibit the range to measure the population and the precision to track individuals.The purpose of this study was to evaluate contrast sensitivity in normal or impaired vision, including patients with glaucoma, multiple sclerosis (MS), and retinitis pigmentosa (RP). Comparing the population variability of an endpoint with its test-retest variability will help determine its potential for health surveillance.
Methods :
We collected 1510 CSFs (954 monoc, 556 binoc; 469 MS, 446 glaucoma, 327 RP, and 268 controls) using the qCSF. We computed the Area under the Log CSF (AULCSF) between 1.5 and 18 cpd and CSF acuity, the high-frequency cutoff where sensitivity=0.0. The qCSF can generate credible interval (CI) estimates within single tests, which are closely related to variability across multiple tests [2].
Results :
The empirical cumulative distributions for AULCSF are presented in Fig. 1. Both control and patient cohorts exhibit a wide range of vision. Monocular AULCSF values range from .5 to 2.0 for control, 1.25 to 1.75 for MS, and 0.0 to 2.0 decimal log units for RP and glaucoma . There is overlap between the upper limits of the impaired samples and the normal sample. There is little overlap at the lower limits of the controls: no control subjects exhibit AULCSFs below .5. The reliability estimates provided by the credible intervals were only slightly smaller (i.e. tests more precise) for controls (median CI width mono/binocular .10/.11) than patients (glaucoma, .13/.1; RP, .13/.14; MS, .14). This suggests that the underlying assumptions of the qCSF are valid for normal and impaired vision [2].
Conclusions :
Our data show that qCSF can assess a broad range of vision, without the floor/ceiling effects of other tests[3]. To distinguish variability in the population and endpoint variability, we demonstrate that credible intervals are small compared to population variability, and comparable to test-retest variabilities in other studies. Further studies will examine the potential for tracking other vision loss, and likewise examine progression and remediation of vision loss over time.
[1] Lee et al (2012) Am J of Ophthal 154,6: S3-S7.
[2] Hou et al (2015) JOV, 15:2.
[3] Pesudovs et al (2004) BJO, 88:11-16.
This is an abstract that was submitted for the 2016 ARVO Annual Meeting, held in Seattle, Wash., May 1-5, 2016.