Abstract
Purpose :
To report on progress of a major international project to develop and validate item banks for the comprehensive measurement of quality of life (QOL) all eye diseases (the Eye-tem Bank).
Methods :
14 disease-specific item banks are at various stages of development. We employ a systematic 4-phase method for the development of each disease-specific module: Phase I: Content identification (items from existing instruments, disease-specific patient focus groups and machine learning methods of patient data); Phase II: Pilot testing the initial item sets for item calibration using Rasch analysis; Phase III: Validation of the item banks as applied by computer adaptive testing (CAT); and Phase IV: Evaluating ophthalmic QOL.
Results :
Phase I (Australia, n=780, median age, 56.5 yrs; Nepal, n=101, median age, 29 yrs ; India, n=30, median age, 23.5 yrs) has identified 3226 unique items for 7 modules across 8 to 13 ophthalmic QoL domains (activity limitation, driving, mobility, lighting, visual symptoms, ocular comfort symptoms, general symptoms, convenience, concerns, social, emotional, economic and coping). Overall 30-50% of items are common between modules and 60-79% between countries. Phase II (total n=1910) has been completed for glaucoma (n=293), diabetic retinopathy (DR, n=514) and hereditary retinal disease (HRD, n=233) in Australia; and refractive error (n=305) in Nepal. Phase II is ongoing for age-related macular degeneration (n=100), acquired retinal diseases (n=20), refractive error (Australia, n=165) and amblyopia & strabismus (India, n=304, Australia, n=55). Items of 4 modules (glaucoma, HRD, DR and refractive error (Nepal)) have been calibrated using Rasch analysis. Phase III data collection for DR module has been completed in Singapore (n=183). Four new domains of QoL (driving, lighting, coping and treatment-related convenience) have emerged. Items are being extracted and refined using qualitative methods to develop cornea (n=41), retinal detachment (n=39), uveitis spectrum of diseases (n=41), and inflammation (n=39)-specific pilot item banks. A machine learning approach to patient data is being tested in uveitis.
Conclusions :
Comprehensive item banks, with up to 13 QoL domains for have been created for groups of diseases. Disease-specificity appears important with only one-third of items common across disease modules. Internationalisation appears viable with more than two-thirds of items common across countries.
This abstract was presented at the 2019 ARVO Annual Meeting, held in Vancouver, Canada, April 28 - May 2, 2019.