Comparative exploration of EQ-5D-5L bolt-on variants for fatigue, sleep, and vision in a general population sample

Fatigue, sleep, and vision are among the most commonly used EQ-5D-5 L bolt-ons, with multiple alternative wordings. This study aimed to evaluate and compare the psychometric performance of the following bolt-on variants: (1) fatigue (FA) vs. tiredness (TI); (2) vision (VI) vs. VI with reference to vision aids (VI-A); and (3) sleep (SL) vs. sleep with examples in the dimension title (SL-E).

A representative sample of 1,262 adults in Hong Kong completed an online survey including the EQ-5D-5 L and bolt-ons. The following psychometric properties were assessed: ceiling/floor, informativity (Shannon's indices), convergent validity (correlations with conceptually related items of PROMIS-29 + 2), agreement (Kendall's tau), explanatory power (regression on EQ VAS and PROMIS-29 + 2 Level Sum Scores), and known-group validity (effect sizes).

Bolt-ons reduced EQ-5D-5 L's ceiling from 35% to 20-24% and improved relative informativity from 0.33 to 0.35-0.36. FA, TI, SL and SL-E demonstrated moderate-to-strong convergent validity (r = 0.55-0.67) with PROMIS-29 + 2 items. All bolt-ons with the exception of VI-A enhanced explanatory power for both EQ VAS and PROMIS-29 + 2. Agreements between TI and FA (0.71), SL and SL-E (0.74), and VI and VI-A (0.78) were strong. For each of the three bolt-on pairs, differences in reducing the EQ-5D-5 L's ceiling, relative informativity and convergent validity were generally small. In known-group comparisons, FA and SL-E outperformed their counterparts, whereas the difference between VI and VI-A was minimal.

All six bolt-on items showed good validity in a general population sample, with minimal differences between items belonging to the same bolt-on pair. These findings inform item development for the experimental EQ-5D Bolt-on Toolbox.