 |
| Professor Aki Tsuchiya |
The EQ-5D instrument consists of a classification system, which
classifies a person’s health based on five simple questions (or “items”), and a
population value set, which allocates a relative value to each of those health
states based on average population preferences.
Published this month in Medical Decision Making, this paper’s most striking finding concerns the “composite” items where
two of the EQ-5D instrument’s five dimensions are covered, for example pain and
discomfort. Would composite items mean the same thing across the
classification and the valuation contexts?
Says Aki, “Imagine somebody had some discomfort but not pain, and they were asked whether they had any pain or discomfort – they would probably answer yes. Now imagine somebody else was asked to imagine having some pain or discomfort – they may imagine having pain rather than discomfort. And this is what the study shows: the composite items mean different things depending on the context. In other words, the health state values of EQ-5D(-5L) in population value sets may not correspond to the states that patients self-report themselves in. Taken at face value, the results would suggest economic evaluations that use EQ-5D(-5L) are systematically biased.”
This has major implications for the EQ-5D-5L and beyond, and
further research on splitting and dropping items is called for by the team which comprised Professor Tsuchiyo who works jointly between ScHARR HEDS and the Dept. of Economics, Brendan Mulhern, honorary member of ScHARR HEDS, from the Centre for Health Economic Research and Evaluation, University of Technology, Sydney, Arne Risa Hole, Dept. of Economics, University of Sheffield, and Nick Bansback, School of Population and Public Health, University of British Columbia, Vancouver.
For the full paper please click here -