Purchase this article with an account.
Rocco Palumbo, Lauren N Ayton, Amy Nau, Russell L Woods; Using Rasch analysis with measures of visual ability of people with ultra-low vision. Invest. Ophthalmol. Vis. Sci. 2016;57(12):1981. doi: https://doi.org/.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
There is a lack of consensus within the field on how to measure the visual ability of people with ultra low-vision (ULV) and changes in visual ability following interventions. We investigated a new approach: using Rasch analysis to combine the results of diverse measures of visual ability into a single outcome scale. The concept is that visual ability, especially with ULV, can be represented on a single dimension, and various measures of visual ability (items) can be associated with that scale. It is important that such a scale have a wide dynamic range.
Data from (1) 42 subjects (persons) with retinitis pigmentosa (RP) with 36 items; and (2) 52 subjects using the BrainPort device (sensory substitution) with 6 items, were analyzed with Winsteps software. The items (measures) included visual acuity, the basic assessment of light motion (BALM) test, visual fields, electroretinogram, orientation-and-mobility, activities-of-daily-living (e.g. object recognition, search) and a mobility questionnaire.
(1) Rasch analysis of the RP-subject data found person separation of 2.08 (reliability 0.81) and item separation of 3.29 (reliability 0.92), the standard error of the measure was 0.06 and the item correlation with the measure was >0.23 across the 31 included items. (2) Rasch analysis of the Brainport-user data showed person separation of 1.35 (reliability 0.65) and item separation of 3.24 (reliability 0.91), the standard error of the measure was 0.01 and the item correlation with the measure was >0.28 across the 6 items.
Our results indicate that the diverse items used in the RP-patient study provided a precise measure of visual ability. The low (<2) person separation in the Brainport-user study indicates insufficient diversity in the few (6) items to discriminate ideally between the visual abilities of the users. Items with less difficulty would improve discrimination of the Brainport users. Overall, these analyses suggest that Rasch analysis can be used in studies of the visual ability of people with ULV, while covering a wide range of visual abilities, which will be important for use in studies of future interventions.
This is an abstract that was submitted for the 2016 ARVO Annual Meeting, held in Seattle, Wash., May 1-5, 2016.
This PDF is available to Subscribers Only