Value measurement in cost-utility analysis: explaining the discrepancy between rating scale and person trade-off elicitations

Previous studies have shown a discrepancy between common utility elicitation methods, such as rating scale (RS) elicitations, and person trade-off (PTO) elicitations. This discrepancy has generally been felt to be due to the fact that RS elicitations ask people to compare conditions to each other in terms of numbers on a visual rating scale, while PTO elicitations ask people to think in terms of numbers of people needing to be treated. However, there are three other cognitive differences between PTO and RS elicitations that might contribute to the discrepancy: In PTO elicitations, as opposed to RS elicitations: (1) people are asked to think about how conditions affect people other than themselves; (2) people are asked to think about the benefit of treating conditions; and (3) people are asked to chose who to treat among patients with different conditions. In this study, we attempt to determine whether any of three cognitive differences contribute to the discrepancy between RS and PTO utility elicitations. We randomized prospective jurors to receive one of several survey versions in which we manipulated the RS and PTO elicitations to reduce the differences between them, while preserving their underlying structure. In the RS-self survey, we asked subjects to rate a health condition on a scale from 0 to 100 as if they had the condition. In the RS-other survey, we asked people to rate the health condition as if someone else had it. In the RS-cure survey, we asked subjects to rate the benefits of curing someone else of the health condition. In the PTO-comparison survey, we asked people to determine equivalent numbers of people needing to be cured of two conditions in order to produce equal benefits. In the PTO-choice survey, we presented people with a budget constraint and asked them to determine equivalent numbers of people needing to be cured of two conditions to help decide which group should be treated within the budget constraint. Two hundred and twenty two subjects completed surveys. Median utilities for health conditions were higher in the two PTO elicitations than in the three RS elicitations. There were no differences in utilities across the two PTO surveys, nor across the three RS surveys. In addition, the PTO elicitations allowed people to make finer distinctions among non-life threatening conditions than did the RS elicitations. The discrepancy between RS and PTO elicitations cannot be explained by shifting subjects attentions from themselves to others, nor from conditions to the benefit of treating conditions. Instead, the discrepancy occurs because of the general format of RS and PTO elicitations.