Abstract
BACKGROUND: Discrete choice experiments (DCEs) are increasingly used in health preference elicitation studies. However, few studies have explored applying a DCE to value long health and wellbeing measures. This study evaluates feasibility, examines the impact of attribute ordering and explores if similar preference exists between countries. METHODS: A health and wellbeing classification system was derived from the EQ Health and Wellbeing (EQ-HWB) measure based on dimensionality, item performance, stakeholder preference and cultural feasibility. Representative samples of UK and Australian general population completed 13 DCE(TTO) tasks. Feasibility was assessed using data quality, time spent on the survey and each task, logical consistency and respondent understanding. Data were modelled using conditional logit model, to evaluate feasibility and impact of attribute ordering (health or other attributes ordered first). The UK and Australian value sets were compared on key characteristics, such as the relative importance of attributes, value set length and distribution. RESULTS: 2489 UK and Australian general public respondents completed the online DCE(TTO) survey. Participants reported good understanding of the DCE(TTO) questions and the attributes. Most of the more severe dimension levels had increasing disutility, with a higher proportion of insignificance observed with the wellbeing attributes. Physical health attributes had larger disutility than other attributes, with anchored utility values ranging from - 0.791 to - 0.588 to 1 for UK and Australian population. The preference between the two countries differed, with mixed evidence for ordering effects. CONCLUSIONS: DCE(TTO) is a viable method for health and wellbeing preference valuation. However, health and wellbeing preference can be influenced by attribute ordering and national setting. The results have implications for the development of future health and wellbeing valuation studies.