Combining EQ-5D-5L items into a level summary score: demonstrating feasibility using non-parametric item response...

Area of Science:

• Health Outcomes Research
• Psychometrics
• Health Economics

Background:

• The EQ-5D-5L is a standard health questionnaire for estimating health utilities.
• Current scoring relies on preference-based weights, with limited exploration of alternative methods.
• Investigating alternative scoring is crucial when utility weights are unavailable or inapplicable.

Purpose of the Study:

• To assess the appropriateness of the EQ-5D-5L level summary score using Mokken scaling.
• To explore the interpretability of the EQ-5D-5L score through item ordering.
• To evaluate the psychometric properties of the EQ-5D-5L under alternative scoring.

Main Methods:

• Utilized the R package "mokken" for psychometric analysis.
• Assessed monotonicity and manifest invariant item ordering (MIIO) on EQ-5D-5L data.
• Employed data from the Multiple Instrument Comparison (MIC) study across six Western countries.

Main Results:

• All EQ-5D-5L items demonstrated monotonicity; however, the anxiety/depression item showed weak scalability.
• Excluding the anxiety/depression item significantly improved overall scale scalability (Hs from 0.559 to 0.714).
• Mokken scaling confirmed a moderate to strong scale, with improved accuracy (HT from 0.463 to 0.743) when the anxiety/depression item was excluded.

Conclusions:

• The EQ-5D-5L items form a moderate to strong Mokken scale, supporting the use of a level summary score for ordering individuals.
• Item ordering provides a basis for interpreting score ranges, linking lower scores to pain/anxiety/depression and higher scores to self-care limitations.
• Excluding the anxiety/depression item enhances the psychometric properties and interpretability of the EQ-5D-5L scale.