Skip to main content

Axiomatic Foundations for Person-Tradeoffs

Current thinking in medical decision making and health policy analysis is strongly influenced by the Quality Adjusted Life Years (QALY) utility model. This model can be used to assign subjective values to health outcomes for purposes of evaluating the overall worth (expected utility) of alternative risky decisions or policies. More recently, a fundamentally different approach has been proposed that attempts to quantify the relative worth of health outcomes in terms of hypothetical social impacts. This approach, called Person Tradeoff (PTO) measurement, is based on the following PTO judgment task. Suppose we wish to compare the worth of an intervention that potentially saves lives to one that potentially improves hearing. In particular, suppose that Intervention A could save the lives of 100 people chosen at random from our society; suppose that Intervention B could completely restore the hearing of X people also chosen at random who would otherwise be deaf. The PTO judgment task asks: How large would X have to be in order for Interventions A and B to be equally desirable? If the average response in a large sample of subjects is that restoring X = 6,000 deafnesses to full hearing is equally desirable as 100 deaths averted, one might infer that restoration of hearing is worth 100/6000 = .017 the worth of saving a life. I will refer to this number, .017, as the PTO ratio for restoration of hearing on a scale that spans "death" to "full health." (Of course, the questions have to be explained more carefully in order to be meaningful.) By pursuing similar questions for a variety of health state changes, one could determine PTO ratios for many different health state changes. The Norwegian health economist, Eric Nord, has argued that PTO ratios should be used as a measure of health state utility because PTO ratios take into account preferences for fairness and equitable distributions as well as personal preferences for health.

In this talk, I will report work in collaboration with Jason Doctor (UW Department of Medical Education) on an axiomatic foundations for PTO measurement, and work with Michael Perry (UW Psychology) on testing PTO axioms. Although the specific assumptions differ from those of the QALY model, the axiomatic analysis plays an analogous role in theory construction and testing. An axiomatic analysis of PTO measurement identifies behavioral assumptions that are jointly sufficient for the PTO representation. It is theoretically desirable that one should formulate assumptions that are necessary for the representation, to the extent that this is possible. Although this talk will have mathematical aspects to it, it will not focus on the mathematical problem of axiomatization and proof of a PTO representation theorem. Rather the talk will focus on (a) the role of axiomatic behavioral theories in social science theory construction; (b) the problem of testing critical assumptions of the PTO representation in behavioral experiments; (c) some preliminary data on tests of axioms will be discussed.