Abstract
Purpose :
Functional vision assessment (FVA) in ultra-low vision (ULV) (VA≤20/1600) is important in developing outcome measures for evaluating efficacy of treatment and rehabilitation in people undergoing vision restoration therapies. The purpose of this study was to develop and validate a FVA that could be useful in the ULV population.
Methods :
47 individuals with ULV participated in the study. The test comprised 19 scenes presented in a virtual reality (VR) headset, each with three levels of difficulty created by varying contrast or speed of motion). Participants performed 3 trials/item. Responses were scored (1 for correct and 0 for incorrect) in a multiple alternate forced choice (m-AFC) design with m = 2 to 4 choices, depending on the scene. Data were analyzed as measures of a latent visual ability variable using a signal detection theory-based method that explicitly models the m-AFC task (Bradley & Massof 2019). Visual acuity was estimated using the Berkeley Rudimentary Vision Test.
Results :
Item measures in d’ units ranged from -1.6 to -0.05 (more negative = easier), with zero indicating chance level performance. The average 95% confidence interval for item measures spanned 0.55 d' units. There was a good spread in item scores validating a scale across different difficulty levels e.g. high contrast items settled near the bottom and low contrast towards the top. Person measures ranged from -1.1 to 2.5 (mean SE = 0.05) with positive scores indicating more ability and negative scores indicating less ability. There was significant positive association between visual acuity and person measures (R2 =0.8, P<0.05).
Conclusions :
There are currently limited options for assessing functional vision in people with ULV. Our results demonstrate the validity and flexibility of using VR to obtain reliable vision measures in ULV and is part of a larger development and validation project for VR-based tests for assessing functional vision in ULV. Such tests will be useful for baseline and follow up evaluations in vision restoration therapies or as an outcome measure of vision rehabilitation. The wide confidence intervals on the item measures indicate a need to gather data in a larger ULV sample. The low standard errors on the person measures show that this is a reliable approach to FVA in a wider ULV population. Using VR has multiple advantages, including portability for use in patients' homes.
This is a 2020 ARVO Annual Meeting abstract.