Peter A Ubel1, Jeff Richardson, Jonathan Baron. 1. Health Services Research & Development, Veterans Affairs Medical Center, University of Michigan Health Care System, 300 North Ingalls, Room 7C27, Ann Arbor, MI 48109-0429, USA. paubel@umich.edu
Abstract
BACKGROUND: The person trade-off (PTO) has been advocated by some as an alternative measure for the purposes of cost-effectiveness analyses. However, the measurement properties of PTO elicitations are still being defined. METHODS: We presented subjects with two PTO scenarios. In the pre-existing paraplegia scenario, they were asked how many paraplegics' lives would have to be saved to be just as important as saving 100 'normal' people's lives. In the paraplegia onset scenario, they were asked how many patients who would experience the onset of paraplegia need to be saved to equal the benefit of saving 100 'normal' lives. We varied the order of the two scenarios across subjects to test whether PTO elicitations are susceptible to order effects. In addition, we varied whether subjects were required to provide a numerical response to the first elicitation. RESULTS: Subjects' PTO indifference points for the two scenarios varied dramatically depending on the order with which they received the scenarios, and according to whether the first elicitation required a numerical response. For those subjects providing numerical responses to both elicitations, median PTO responses varied by a factor of close to two in the pre-existing paraplegia scenario and by a factor of eight in the paraplegia onset scenario. However, the magnitude of the order effect was significantly reduced when subjects were not asked to provide a numerical response to the first PTO elicitation. CONCLUSION: PTO elicitations are susceptible to order effects. These order effects are partly due to numerical anchoring. However, other cognitive factors contribute to the order effect. Further research should clarify whether these order effects can be reduced.
BACKGROUND: The person trade-off (PTO) has been advocated by some as an alternative measure for the purposes of cost-effectiveness analyses. However, the measurement properties of PTO elicitations are still being defined. METHODS: We presented subjects with two PTO scenarios. In the pre-existing paraplegia scenario, they were asked how many paraplegics' lives would have to be saved to be just as important as saving 100 'normal' people's lives. In the paraplegia onset scenario, they were asked how many patients who would experience the onset of paraplegia need to be saved to equal the benefit of saving 100 'normal' lives. We varied the order of the two scenarios across subjects to test whether PTO elicitations are susceptible to order effects. In addition, we varied whether subjects were required to provide a numerical response to the first elicitation. RESULTS: Subjects' PTO indifference points for the two scenarios varied dramatically depending on the order with which they received the scenarios, and according to whether the first elicitation required a numerical response. For those subjects providing numerical responses to both elicitations, median PTO responses varied by a factor of close to two in the pre-existing paraplegia scenario and by a factor of eight in the paraplegia onset scenario. However, the magnitude of the order effect was significantly reduced when subjects were not asked to provide a numerical response to the first PTO elicitation. CONCLUSION: PTO elicitations are susceptible to order effects. These order effects are partly due to numerical anchoring. However, other cognitive factors contribute to the order effect. Further research should clarify whether these order effects can be reduced.