Due to copyright issues, an electronic copy of the thesis must be ordered from the faculty. For the faculty to have time to process the order, the order must be received by the faculty at the latest 2 days before the public defence. Orders received later than 2 days before the defence will not be processed. After the public defence, please address any inquiries regarding the thesis to the candidate.
Trial Lecture – time and place
See Trial Lecture.
Adjudication committee
- First opponent: Professor Jes Søgaard, University of Southern Denmark, Denmark
- Second opponent: Professor Bjarne Robberstad, University of Bergen, Norway
- Third member and chair of the evaluation committee: Associate Professor Emily Burger, Faculty of Medicine, University of Oslo
Chair of the Defence
Professor Arnstein Finset, University of Oslo
Principal Supervisor
Senior Researcher Kim Rand, Health Services Research Centre, Akershus University Hospital
Summary
Many countries estimate population-specific value sets for the most used instruments measuring health related quality of life, where the EQ-5D is the most used by far. With increasing numbers of values sets, more attention is put on representativeness and legitimacy of estimated values. Protocols for EQ-5D valuation provide guidelines for data collection, recommending administration of time trade-off (TTO) and discrete choice experiment (DCE) tasks by personal face-to-face interviews with minimum 1000 respondents representative of the general population.
This thesis explores themes related to the samples from which values are estimated and sources of possible error and bias in value sets. The included papers assessed whether respondent characteristics are related to TTO non-completion and time use, if having significant others is associated with the values elicited using TTO, and the impact of sample size on prediction accuracy of estimated values. Analyses were based on data collected for a Norwegian valuation study and supplemented with data from previously published valuation studies.
The findings suggest that:
• Respondents who are older, in poorer health or with an immigrant background had increased likelihood of drop-out and of needing more time to complete
• Respondents with significant others assigned on average higher utilities and were less likely to value health states as worse than being dead than those without significant others
• There is little to gain in terms of prediction accuracy beyond approximately 500 respondents using commonly applied regression models estimating health state values
If current methods and models are considered sufficient, valuation studies could be completed at lower costs with fewer respondents, without significant loss in prediction accuracy. With smaller samples, however, representation of important population groups in the final sample becomes even more important.
Additional information
Contact the research support staff.