Item banks for measurement of refractive error-specific quality of life.

PURPOSE: To evaluate the psychometric properties of refractive error-specific quality of life (QoL) item banks and assess their performance using computerised adaptive testing (CAT) simulations.

METHODS

In this cross-sectional study a 392-item questionnaire, grouped under 11 QoL domains, was interviewer-administered to 305 people with refractive error [mean age ± S.D., 30.5 ± 14.1 (range (18 to 83) years; male, 50.5%; mean ± S.D. spherical equivalent refractive error -2.4 ± 2.9 (range: -15.0 to +11.0) Dioptres; spectacles (n = 257), contact lens (n = 37), refractive surgery (n = 25), uncorrected refractive error (n = 57)]. Rasch analysis was conducted on each QoL domain using the Andrich rating scale model to investigate parameters including response category functioning, person- and item-reliability, infit and outfit statistics, unidimensionality, targeting, differential item functioning and local item dependency. The resulting item banks underwent CAT simulations in 1,000 cases with 'high' and 'moderate' precision stopping rules.

RESULTS

Rasch analysis iterations resulted in 13 refractive error-specific item banks (Convenience, Health concerns, Economic, Activity limitation, Mobility, Emotional, Social, Visual symptoms frequency, Visual symptoms severity, Visual symptoms bothersome, Comfort symptoms frequency, Comfort symptoms severity and Comfort symptoms bothersome), containing a total of 366 items. The item banks had good psychometric properties including satisfactory measurement precision, infit and outfit statistics and unidimensionality. In CAT simulations, the mean number of items required to achieve high and moderate measurement precision was 9.4 and 4.7, respectively.

CONCLUSION

Overall, refractive error-specific QoL item banks show promise in their ability to comprehensively and precisely evaluate a range of quality of life parameters. These items banks when administered using a CAT system offer unique outcome tools for implementation in clinical trials, healthcare and research.