A comparison of scoring weights for the EuroQol derived from patients and the general public

General health state classification systems, such as the EuroQol instrument, have been developed to improve the systematic measurement and comparability of health state preferences. In this paper we generate valuations for EuroQol health states using responses to this instrument's visual analogue scale made by patients enrolled in a randomized clinical trial evaluating tirilazad mesylate, a new drug used to treat subarachnoid haemorrhage. We then compare these valuations derived from patients with published valuations derived from responses made by a sample from the general public.

The data were derived from two sources: (1) responses to the EuroQol instrument from 649 patients 3 months after enrollment in the clinical trial, and (2) from a published study reporting a scoring rule for the EuroQol instrument that was based upon responses made by the general public. We used a linear regression model to develop an additive scoring rule. This rule enables direct valuation of all 243 EuroQol health states using patients' scores for their own health states elicited using a visual analogue scale. We then compared predicted scores generated using our scoring rule with predicted scores derived from a sample from the general public.

The predicted scores derived using the additive scoring rules met convergent validity criteria and explained a substantial amount of the variation in visual analogue scale scores (R(2)=0.57). In the pairwise comparison of the predicted scores derived from the study sample with those derived from the general public, we found that the former set of scores were higher for 223 of the 243 states. Despite the low level of correspondence in the pairwise comparison, the overall correlation between the two sets of scores was 87%.

The model presented in this paper demonstrated that scoring weights for the EuroQol instrument can be derived directly from patient responses from a clinical trial and that these weights can explain a substantial amount of variation in health valuations. Scoring weights based on patient responses are significantly higher than those derived from the general public. Further research is required to understand the source of these differences.