Related Experiment Video
Updated: Jul 26, 2026
Development of a Virtual Reality Assessment of Everyday Living Skills
Published on: April 23, 2014
Psychometric Performance of a New Condition-Specific Preference-Weighted Measure, Vision Impairment in Low
Donna Rowen1, Jill Carlton1, Emily McDool1
1Sheffield Centre for Health and Related Research, University of Sheffield, England, UK.
Objectives:
The Vision Impairment in Low Luminance-Utility Index (VILL-UI) is a novel preference-weighted measure for use in patients with age-related macular degeneration (AMD). No evidence exists on its psychometric performance nor its performance in comparison with the generic preference-weighted measure, EQ-5D-5L, commonly used in economic evaluation. This study compares the psychometric performance of VILL-UI with EQ-5D-5L in patients with AMD.
Methods:
Assessments of feasibility, convergent/divergent validity, and known-group validity of VILL-UI and EQ-5D-5L are undertaken using MACUSTAR data at baseline, 12, 24, and 36 months. Analyses are undertaken separately using UK and German preference weights for both measures.
Results:
The sample with complete responses (n = 586) had mean age 71.9 years (standard deviation 6.9), 65.2% women, with predominantly intermediate AMD (87.2%). VILL-UI and EQ-5D-5L are feasible for completion, although VILL-UI has fewer usable responses due to its response options (baseline 89% vs 100%). EQ-5D-5L has high ceiling effects, with around one-third of participants reporting the best health state compared with under 8% for VILL-UI. Convergent validity between EQ-5D-5L and VILL-UI utilities and dimensions in which a relationship is expected is low, with divergent validity demonstrated where expected. VILL-UI detected statistically significant differences in known groups for visual acuity, visual function, and AMD stage across most time points, with little evidence of known-group validity for EQ-5D-5L.
Conclusions:
VILL-UI is appropriate for use in future AMD studies to inform economic evaluation. VILL-UI has superior performance to EQ-5D-5L for known-group validity and has fewer ceiling effects but has fewer usable responses.
