PURPOSE

Little was known about the head-to-head comparison of psychometric properties between SF-6D and EQ-5D-5L or the different value sets of EQ-5D-5L. Therefore, this study set out to compare the psychometric properties including agreement, convergent, and known-group validity between the SF-6D and the EQ-5D-5L using the real value sets from Thailand and the UK in patients with chronic diseases.

METHODS

356 adults taking a medication for at least 3 months were identified from a university hospital in Bangkok, Thailand, between July 2014 and March 2015. Agreement was assessed by intraclass correlation coefficients (ICCs) and Bland-Altman plots. Convergent validity was evaluated using Spearman's rank correlation coefficients between SF-6D and EQ-5D-5L and EQ-VAS and SF-12v2. For known-groups validity, the Mann-Whitney U test and Kruskal-Wallis test were used to examine the associations between SF-6D and EQ-5D-5L and patient characteristics.

RESULTS

Agreements between the SF-6D and the EQ-5D-5L using Thai and UK value sets were fair, with ICCs of 0.45 and 0.49, respectively. Bland-Altman plots showed that the majority of the SF-6D index scores were lower than the EQ-5D-5L index scores. Both the EQ-5D-5L value sets were more related to the EQ-VAS and physical health, while the SF-6D was more associated with mental health. Both EQ-5D-5L value sets were more sensitive than the SF-6D in discriminating patients with different levels of more known groups except for adverse drug reactions.

CONCLUSIONS

The SF-6D and both EQ-5D-5L value sets appeared to be valid but sensitive to different outcomes in Thai patients with chronic diseases.