Understanding the relationship between the EQ-5D, SF-6D, HAQ and disease activity in inflammatory arthritis

The growth of economic analyses and in particular cost-utility analyses (CUA), which use the QALY as a measure of outcome, has heightened the interest in the methodologies used to calculate the QALY. The EQ-5D has produced quite different utility values from that of the SF-6D. This article seeks to understand these differences using a cohort of patients with inflammatory arthritis.

To examine the relationship between the disease-specific measure, Health Assessment Questionnaire (HAQ) disability index (DI) and the preference-based measures, SF-6D, EQ-5D and European League Against Arthritis (EULAR) Disease Activity Score (DAS) in patients with rheumatoid arthritis (RA) and psoriatic arthritis (PsA).

Patients with RA and PsA (n = 504) attending a tertiary rheumatology referral centre completed the HAQ, SF-6D and the EQ-5D before starting biological therapy and again 12 months later. The SF-36 was converted into a utility using the preference-based SF-6D. Clinical outcomes such as the DAS, joint counts and laboratory measures were also recorded. We calculated single index utility scores from the preference-based instruments using UK population norms. We used regression analysis to derive a mapping function and calculated utility scores from the HAQDI and the DAS 28.

The mean utility observed at baseline for RA was 0.43 for the EQ-5D and 0.54 for the SF-6D and for PsA was 0.49 for the EQ-5D and 0.57 for the SF-6D. The utility gain demonstrated by the EQ-5D was over twice that of the SF-6D. The EQ-5D scored 17% of the RA group as less than 0 (state defined as worse than death); 7% of this group remained less than 0 at follow-up. The distribution of the utility estimates was similar for both RA and PsA.

Our findings draw attention to the impact of states worse than death on the overall distribution for the EQ-5D derived utilities and how these impact on its use in practice. EQ-5D-derived QALY changes are over twice that of the SF-6D. The implication of this for decision makers is that cost-effectiveness evaluations for treatments in this disease class are likely to be very sensitive to the choice of utility measure.