Purchase this article with an account.
Konrad Pesudovs, Jyoti Khadka, Eva Fenwick, Ecosse Lamoureux; Rasch analysis of the glaucoma-specific module of the Eye-tem Bank project. Invest. Ophthalmol. Vis. Sci. 2013;54(15):5313.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
To assess the psychometric properties of a glaucoma-specific quality of life (QOL) item bank module of the Eye-tem Bank project.
The glaucoma-specific QOL item bank has 342 items distributed across 10 QOL domains: Visual symptoms, SY; Ocular surface symptoms, OS ; General Symptoms, GS ; Activity limitation, AL ; Mobility, MB ; Emotional, EM ; Health Concerns, HC ; Social, SC ; Convenience, CV ; and Economic, EC. The item bank was interviewer-administered to patients with glaucoma. Each QOL domain of the item bank was individually evaluated using Rasch analysis to assess its psychometric properties, namely precision (adequate if the person separation index (PSI) >2.0); dimensionality (principal component analysis of residuals, PCA: unidimensionality is considered if the (1) raw variance explained by the measure >50%, (2) unexplained variance of the 1st contrast <3.0 eigenvalues or ≤5%); item fit to the model (infit mean square ≤ 1.5); and differential item functioning (DIF) by age and gender (no DIF if the DIF contrast is <1.0).
The glaucoma item bank was administered to 293 patients (male, 55%; median age, 70 yrs; range 20-91 yrs). The majority (between 80% and 93%) of the participants endorsed the two lower end categories of all questions, indicating that most participants had low levels of QOL impact across the domains. Among the 10 QOL domains, five domains (AL, MB, EM, HC and CV) demonstrated excellent measurement precision, whereas, five domains (SC, EC, VS, OS and GS) had borderline precision (PSI ranging from 1.50 to 2.50). The PCA of residuals confirmed that eight domains (MB, EM, HC, CV, SC, OS, VS and GS) were unidimensional. Two domains (AL and CV) demonstrated some degree of multidimensionality (1st contrast >3.0 eigenvalues), which could be repaired by rearranging the items into subdomains. Of the 342 items, only 27 (7.9%) displayed misfit, and 7 (2.0%) showed DIF for age or gender.
The 10 QOL domains of the glaucoma-specific item bank demonstrated good psychometric properties against most metrics. Despite using a large sample, the preponderance of participants with low QOL impact of glaucoma impairs the establishment of precise measurement. Ongoing data collection in patients across the spectrum of glaucoma will strengthen the assessment of the psychometric properties and validity of these glaucoma-specific QOL domains.
This PDF is available to Subscribers Only