Background: Typical health state valuation exercises use trade off methods, such as the Time Trade Off or the Standard Gamble, involving a series of iterated questions so that a value for each health state by each individual respondent is elicited. This iterative process is a source of potential biases, but this has not received much attention in the health state valuation literature. The issue has been researched widely in the contingent valuation (CV) literature which elicits the monetary value of hypothetical outcomes.
Methods: The lessons learnt in the CV literature are revisited in the context of the design and administration of health state valuations. The paper introduces the main known biases in the CV literature, and then examines how each might affect conventional iterative health state valuations.
Results: Of the eight main types of biases, starting point bias, range bias, and incentive incompatibility bias are found to be potentially relevant. Furthermore, the magnitude and direction of the bases are unlikely to be uniform, and depend on the range of the value (e.g. between 0 and 0.5). Limitation: this is an overview paper and the conclusions drawn need to be tested empirically.
Conclusions: health state valuation studies, like CV studies, are susceptible to a number of possible biases that affect the resulting values. Their magnitude and direction are unlikely to be uniform, and thus empirical studies are needed to diagnose the problem and if necessary to address it.
There are two versions of this paper. The longer version was published as a HEDS discussion paper in September 2012. The shorter version was published from Medical Decision Making (online before print) in March 2013.