Rasch analysis in the development of a simplified version of the National Eye Institute Visual-Function Questionnaire-25 for utility estimation

Abstract

Purpose: Preference-based health measures value how people feel about the desirability of a health state. Generic measures may not effectively capture the impact of vision loss from ocular diseases. Disease-targeted measures could address this limitation. This study developed a vision-targeted health state classification system based on the National Eye Institute Visual Function Questionnaire-25 (NEI VFQ-25).

Methods: Secondary analysis of NEI VFQ-25 data from studies of patients with central (n = 932)- and peripheral-vision loss (n = 2,451) were used to develop a health state classification system. Classical test theory and Rasch analyses were used to identify a smaller set of NEI VFQ-25 items suitable for the central- and peripheral-vision-loss groups.

Results: Rasch analysis of the NEI VFQ-25 items using the peripheral vision-loss data indicated that 11 items fit a unidimensional model, while 14 NEI VFQ-25 items fit using the central-vision-loss data. Combining peripheral-vision-loss data and central-vision-loss data resulted in 9 items fitting a unidimensional model. Six items covering near vision, distance vision, social vision, role difficulties, vision dependency, and vision-related mental health were selected for the health-state classification.

Conclusions: The derived health-state classification system covers relevant domains of vision-related functioning and well-being.