5-Level Bolt-Ons Have an Improved Psychometric Performance Compared to 3-Level Bolt-Ons in a Large, Multinational, Longitudinal General Population Sample

OBJECTIVES: To compare the psychometric properties of the 5-Level versus 3-Level bolt-ons for vision, breathing, tiredness, sleep, social relationships and self-confidence when added to the EQ-5D in a general population sample.

METHODS: POPUP is an observational, longitudinal study collecting general population data across 8 countries. Respondents entered demographic and disease characteristics, and completed simultaneous measurement of the EQ-5D-5L and 5-Level bolt-ons, and the EQ-5D-3L and 3-Level bolt-ons. The following psychometric properties of the bolt-ons were assessed: redistribution properties, ceiling/floor, informativity, convergent and divergent validity, agreement, known groups validity and Pareto changes.

RESULTS: 9758 respondents completed the survey in 2020, and 4839 completed it in 2023, with 3571 respondents having data at both timepoints.

Agreement between 3-Level and 5-Level bolt-on responses was generally excellent, with a low percentage of inconsistent response pairs. This percentage increased with the severity of problems reported (<1% for level 1 versus >12% for levels 2-3). Agreement was further confirmed with high intra-class correlation coefficients between the rescaled Level Sum Scores of the EQ-5D-5L+bolt-ons and the EQ-5D-3L+bolt-ons, showing strong concordance (ICC range 0.92-0.97, all p<0.001).

All bolt-ons reduced the ceiling of their respective EQ-5D by 14% and 22% respectively, but 5-Level bolt-ons showed larger reductions than their 3-level pairs. All 5-Level bolt-ons also consistently showed higher absolute (+53%) and higher relative informativity (+5%) than their 3-level pairs.

Correlations of vision, tiredness and breathing with EQ-5D dimensions were weak-moderate, showing good divergent validity for these bolt-ons. Breathing also demonstrated good convergent validity. Furthermore, only breathing improved known groups validity in groups defined by chronic conditions (relative efficiency 1.10-1.12).

Longitudinally, more Pareto changes (improvements, worsening, mixed changes) were found over time with the 5-level bolt-ons implying higher sensitivity with the 5-level descriptive system.

CONCLUSIONS: 5-Level bolt-ons demonstrate superior psychometric performance in assessing the health status of the general population.