Andhra Pradesh Eye Disease Study - Visual Function Questionnaire: further improvements in psychometric properties using Rasch analysis.

PURPOSE

Previous Rasch analysis of the Andhra Pradesh Eye Disease Study-Visual Function Questionnaire (APEDS-VFQ) lacked comprehensiveness, specifically, dimensionality (whether it measures single/multiple constructs). Therefore, using the Rasch model this study provides a detailed assessment of psychometric properties of the APEDS-VFQ.

METHODS

A total of 351 visually impaired adults (mean age, 43.3 years) were verbally administered the APEDS-VFQ. Rasch analysis was used to assess the psychometric properties of the instrument.

RESULTS

Participants could distinguish only three categories of difficulty, so response categories were reduced from five to three. A single item ("reading small prints in newspaper/magazines", infit mean square 1.54) misfit the model. The overall pattern of fit statistics for item and person measures suggested that the underlying construct (visual ability) is not unidimensional. When the items were grouped into subsets based on functional requirements (resolution, contrast sensitivity, illumination and peripheral vision) and separate person measures were estimated for each of these domains, the first principal component contained the visual ability and accounted for 72% of the variance. Item measure distributions could be divided into 18 strata, and item-separation reliability was 0.99. Person measures could be divided into three statistically distinct strata and the person-separation reliability was 0.81.

CONCLUSIONS

The APEDS-VFQ is a precise measure of visual ability in visually impaired adults in India. Similar to other visual function questionnaires developed for the Western population, our results demonstrate that visual ability is a two-factor composite latent variable; one dimension heavily influences reading and the other most heavily influences peripheral vision (mobility).