Conclusions
EQ5D and SF6D measure different aspects of HRQoL. The difference in psychometric properties between them and the lack of agreement is probably clinically significant. Because the ability to detect a change in the underlying construct and similarity to a disease-specific instrument is quite different, the choice of instrument should probably be guided by diagnosis and/or treatment choice. In our study of patients with chronic low back pain, the SF6D had the best ability to detect change and correctly identify patients as improved or non-improved beyond a 95% confidence level of measurement error.
Finally, our study supports the findings of Soegaard et al.. They concluded that the SF6D and EQ5D cannot be used interchangeably for measurement of preference value and that sensitivity analysis examining the impact of between-measure discrepancy remains a necessary condition for cost-utility evaluation results.