Purpose: To compare the scaling properties of the English, Spanish, French, and Chinese versions of the EQ-5D-3L and EQ-5D-5L descriptive systems.
Methods: Members of the general populations in the UK, Spain, France, and China were interviewed to measure the severity of health problems represented by the response labels used in the EQ-5D descriptive systems using a visual analog scale. Multiple linear regression models were used to compare the perceived label severity across the four language groups. Severity scores for labels from each EQ-5D-5L dimension scale were compared with each other to assess ordinality.
Results: EQ-5D-5L and EQ-5D-3L labels used for describing different levels of health problems were rated differently, while those describing the same level of health problems were rated similarly. For example, the deviation of any group mean from the grand severity mean score for the label 'slight(ly)' was no larger than 5 points on a 0-100 scale for all five EQ-5D dimensions (p > 0.05 for all, t tests). Label ratings violating hypothesized ordinality of the EQ-5D-5L scales were observed in only a small proportion of respondents.
Conclusions: Our study provided some preliminary evidence supporting the ordinality and equivalence of the EQ-5D-5L and EQ-5D-3L descriptive systems across four major languages.