Anchoring in the Lead-Time Time Trade-Off : does the “Starting-point” influence preference elicitation?

Background: The standard TTO’s different treatment of states worse than death (SWD) and states better than death (SBD) is by many viewed as a prima-facie cause for concern related to potential bias caused by using two different elicitation regimens. In response to this concern, Robinson and Spencer (2006) described an approach that made it possible to value SWD and SBD in the same exercise by adding a “lead-time” to the valuation tasks. There is a considerable literature suggesting that theoretically irrelevant fac- tors influence preference elicitation trough heuristic processes, and that these processes may bias the resulting TTO values. If the lead-time extension serve as a heuristic anchor, the a priori choice of the lead-times initial position might have consequences on the respondents elicitation process, and if pro- nounced, cause of bias and error. Assessing the existence, the magnitude or the direction of the potential influence might provide researchers with valu- able information for future LT-TTO studies. Purpose: To explore potential influence caused by the added “lead-time” component in the elicited LT-TTO values. Methods: A LT-TTO survey with manipulated “lead-time” starting points were run on a sample of the Norwegian population. The manipula- tions added starting points ranging from 0 to 10 years above the starting point of the comparator health state. The respondents received eight EQ-5D descriptions of hypothetical health states and valued them using the manip- ulated LT-TTO survey. The Kruskal-Wallis one-way analysis of variance by ranks was applied, and a robust regression with Huber and Tukey bi weights was fitted to detect potential influence attributable to the starting points. Results: A significant between group variation in the elicited results were detected by the statistical testing, and the robust regression revealed a pos- itive linear relationship between the TTO-values and the lead-time starting points with a coefficient of 0.039. Conclusion: The findings suggest that the LT-TTO values are influenced by the starting point of the “lead-time” and that the method is sensitive to heuristic processes mediated by the “lead- times” initial positioning. The observed variation raises doubts w ether the LT-TTO leads us closer to an unbiased preference elicitation tool, or if it only trades one type of heuristic effect off for another. Since it is difficult to perceive that the dependency is a result of an a priori property of peo- ples’ perceptions of health and time, anchoring should be considered when designing and interpreting LT-TTO studies