How do the EQ-5D, SF-6D and the well-being rating scale compare in patients with ankylosing spondylitis?

Purpose: To compare aspects of validity of EuroQol-5 Dimensions (EQ-5D) and Short-Form-6 Dimensions (SF-6D), two indirect utility instruments, and the well-being rating scale (RS) in ankylosing spondylitis (AS). EQ-5D, SF-6D and RS were available for 254 patients fulfilling modified New York criteria. 134 patients were part of an observational cohort and 120 were part of a randomised controlled trial (RCT). Aspects of validity assessed were truth (agreement and correlation with external health measures) and discrimination (differentiation between health states, repeatability and detection of treatment effect). Median (range) values were 0.69 (-0.08-1.00) for the EQ-5D, 0.65 (0.35-0.95) for the SF-6D and 0.65 (0.14-1.00) for the RS. Agreement (intraclass correlation coefficient) was moderate (0.46-0.55). Instruments correlated equally with disease activity, functioning and quality of life. The SF-6D showed smaller average differences in utility between patients with better and worse disease compared with the EQ-5D and the RS. The smallest detectable difference (SDD) (in the control group of RCT) was 0.36, 0.17 and 0.33 for EQ-5D, SF-6D and RS, respectively. The ability to detect treatment effect (in the intervention trial) showed standardised effect sizes that were moderate for EQ-5D and SF-6D (0.63 and 0.64) and low for the RS (0.23). In patients with AS, EQ-5D, SF-6D and the RS correlate equally well with external measures of health, but have different psychometric properties. The SDD is most favourable for the SF-6D, but it discriminates less well between patients with different disease severities. The RS has a poorer ability to detect treatment effects. It is difficult to recommend one of the instruments