Abstract
Purpose: :
The Impact of Vision Impairment questionnaire (IVI) had shown good reliability and validity (Weih et al. 2002). Whether its overall score, based on a summation of Likert values, is a true linear measure of difficulty performing daily activities has not been shown. Substantiating that the IVI has a valid measuring scale is fundamental to objectively assess rehabilitation outcomes. The purpose was therefore to rescale the IVI using Rasch analysis to generate a linear measure.
Methods: :
314 first–time referrals to low vision clinics completed the 32–item IVI. The six response categories ranged from "no difficulty" to "can’t do because of eyesight". Sociodemographic and vision data were also obtained. The data were Rasch analyzed with a partial credit model using RUMM 2020 software. The overall fit of the model, the response scale, individual item fit, differential item functioning and person separation index (PSI) were assessed. The original and Rasch–analyzed IVI were correlated with distance visual acuity to assess criterion validity.
Results: :
26 items displayed ‘disordered thresholds’ which indicated the presence of substantial overlapping between responses contributing to misfit. Collapsing the response scale to three categories (4 items) and four categories (28 items) produced ordered response thresholds for all items. Item reduction was then undertaken to improve overall model fit. Four items with high levels of missing responses, poor spread, high skewness and substantial deviation between observed and expected model curves were removed. This produced overall fit to the Rasch model (Item–Trait Interaction Χ2=118.3; p=0.32). The final mean person and item Fit Residual values were 0.06 (SD=0.85) and –0.20 (SD=1.45), respectively. The PSI score was 0.95, indicating high person separation reliability. The revised scale was well targeted to the participants with similar mean locations for items (0.00) and persons (0.16). Criterion validity was similar between the original (r= 0.34) and revised IVI (r=0.37).
Conclusions: :
Through response scale and item reduction strategies, an improved IVI has been created to assess the difficulty with daily activities for people with low vision. Importantly, with a confirmed interval measure, the revised IVI has the potential to assess low vision rehabilitation outcomes.
Keywords: clinical (human) or epidemiologic studies: biostatistics/epidemiology methodology • quality of life • low vision