A comparison of utility measurement using EQ-5D and SF-6D preference-based generic instruments in patients with...

The purposes of this study were to analyse and compare aspects of validity (concurrent and discriminant) of the two widely used indirect utility instruments, the EuroQol-5D (EQ-5D) and the Short Form-6D (SF-6D) in a representative cohort of patients with rheumatoid arthritis (RA).

Five hundred and eighty-three consecutive adult patients (435 women, 148 men) with RA and referred to the outpatient Clinic were evaluated. Patients were asked to complete EQ-5D and SF-36. SF-6D utility scores were calculated using the eight mean SF-36 scores, according to published algorithms. Disease-related characteristics included disease duration, co-morbidities, a measure for disease activity [Disease Activity Score-28 joint (DAS28)] and for radiographical damage (Sharp van der Heijde scoring method, SHS). The agreement between the utility instruments was evaluated by Bland-Altman analysis. Construct validity was assessed using the Kruskal-Wallis test, Mann-Whitney U-test, Spearman's correlations, and receiver operating characteristic (ROC) curves. Multivariate analyses were used to assess the relationship among HRQoL and disease-related characteristics and socio-demographic data.

A comparison of means showed that SF-6D values exceeded EQ-5D values (p<0.0001). Agreement between both measures was only moderate. Utility scores and domains and summary scores of the SF-36 were highly correlated. The EQ-5D and SF-6D both detected change in different health status (<0.0001). The discriminatory power of both indexes was good, without significant difference, with an AUC of 0.869 and 0.820, respectively for EQ-5D and SF-6D. The EQ-5D and SF-6D both detected change over different health status among RA patients (both al level of p<0.0001) although EQ-5D was more efficient in detecting differences between groups in almost all cases. Comparison of EQ-5D and SF-6D scores within VAS groups showed that, for less healthy individuals (VAS scores 0-50), the median EQ-5D score was significantly lower than the median SF-6D score. The multivariate regression models for EQ-5D and SF-6D included both SHS and DAS28 (p=0.0001). The relative contribution of these domains differed substantially between patients with short and long standing disease duration. The presence of multiple chronic conditions also appeared to contribute to reduce the levels of utility of both instruments.

Although EQ-5D and SF-6D appeared to measure similar constructs, these instruments are quite different from each other in the assessment of RA. For worse health status the median EQ-5D scores were significantly lower than the median SF-6D scores. Moreover, EQ-5D and SF-6D appeared both significantly influenced by disease activity, radiological damage and co-morbidity. For that reason, we advise caution in the employment of these preference-based instruments, especially in RA patients with severe disease.