METHODS In the course of a feasibility study on using an internet panel to obtain preference data from members of the general public within limited timescales, 73 health states (in 13 conditions and the EQ5D) were valued. These were generally based on summary data from relevant patient samples using disease specific quality of life measures. In several cases, existing descriptions were re-valued by the panel and for one condition, individual patient data were used to develop descriptions using cluster analysis. A simple questionnaire was developed for use with clinicians and patients to test credibility and completeness of descriptions.

RESULTS Although using disease specific QoL outcome measures brings efficiency gains in developing health states, there are considerable challenges where only summary data are available (clinical heterogeneiry, item reduction, and response distribution). Individual patient data may overcome some of these but presents other challenges, notably in choice of wording. Whether content validation is a necessary step in health state description is open to question, and little regard is taken of this in current methods. Content validation with patients presents a range of challenges, and responses in this study were poor. Overall responses from clinicians and patients on credibility were similar (rated as 59 on 0-100 scale) but with marked inter-rater variation. Completeness ratings differed qualitatively between clinicians and patients, with clinicians identifying symptoms that were not covered in the descriptions and patients reporting functional and social impacts.

CONCLUSION The development of health state descriptions from disease specific quality of life measures has advantages but important challenges remain. Content validation may be seen as a formative or summative step in health state development and further research into methods for assessing this is needed.