Purchase this article with an account.
Maxwell Pistilli, Gui-shuang Ying, Maureen G. Maguire, Ellen P. Peskin, Peter G. Dentone, Penny A. Asbell, DREAM Research Group; Symptom Questionnaires For Dry Eye Disease: A Comparative Rasch Analysis Of The OSDI, BODI And IDEEL. Invest. Ophthalmol. Vis. Sci. 2011;52(14):3865.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
To evaluate the psychometric properties of two questionnaires on dry eye disease (DED) severity as alternatives to the Ocular Surface Disease Index (OSDI).
The OSDI, Brief Ocular Discomfort Inventory (BODI), and Impact of Dry Eye on Everyday Life (IDEEL) Symptom Bother Module were given to 241 patients with a history of DED.Rasch analysis (Andrich rating scale, WINSTEPS) was used to assess category function (threshold and observed category order), dimensionality (item fit and principal components of residuals), targeting (mean person ability [MPA] and non-uniform differential item functioning [NUDIF]), and reliability of the measures.
Only the BODI exhibits disordered thresholds and responses, indicating that people are unable to distinguish between adjacent categories. The OSDI has the most misfit items, but smallest principal component; more misfit items indicates a more clear distinction between the dimensions indicated by the principal component, while the IDEEL’s fewer misfit items and similar component indicate a broad or "fuzzy" dimension. The BODI has high misfit and component. The BODI has the lowest MPA making it appropriate for more severe DED, and the IDEEL has the highest MPA making it best for milder dry eye. However both the OSDI and BODI have no meaningful and significant NUDIF, and would therefore perform better than the IDEEL across a range of dry eye severity. The BODI and IDEEL have slightly higher reliability than the OSDI, but all three are acceptable.
Despite their slightly higher reliability, the BODI is more susceptible to random variation due to poor category function, and the IDEEL is more susceptible to confounding due to unclear multidimensionality. The OSDI is the most common of the three, the most unidimensional, and could be further broken down based on established subscales if desired. Although any decision must be partially based on the population being studied, the OSDI is the most psychometrically sound of the three measures.
This PDF is available to Subscribers Only