A model is only as good as its assumptions: a reply to Peele

The physician-patient encounter: the physician as a perfect agent for the patient versus the informed treatment decision-making model

Assuming a goal of arriving at a treatment decision which is based on the physician's knowledge and the patient's preferences, we discuss the feasibility of implementing two treatment decision-making models: (1) the physician as a perfect agent for the patient, and (2) the informed treatment decision-making models. Both models fall under the rubric of agency models, however, the requirements from the physician and the patient are different. An important distinction between the two models is that in the former the patient delegates authority to her doctor to make medical decisions and thus the challenge is to encourage the physician to find out the patient's preferences. In the latter, the patient retains the authority to make medical decisions and the physician role is that of information transfer. The challenge here is to encourage the physician to transfer the knowledge in a clear and nonbiased way. We argue that the choice of model depends among other things on the ease of implementation (e.g., is it simpler to transfer patient's preferences to doctors or to transfer technical knowledge to patients?). Also the choice of treatment decision-making model is likely to have an impact on the type of incentives or regulations (i.e., contracts) needed to promote the chosen model. We show that in theory both models result in the same outcome. We argue that the approach of transferring information to the patient is easier (but not easy) and, hence, more feasible than transferring each patient's preferences to the physician in each medical encounter. We also argue that because better "technology" exists to transfer medical information to patients and time costs are involved in both tasks (i.e. transferring preferences or information), it is more feasible to design contracts to motivate physicians to transfer information to patients than to design contracts to motivate physicians to find out their patients' utility functions. We illustrate our arguments using a clinical example of the choice of adjuvant chemotherapy versus no adjuvant chemotherapy for women with early stage breast cancer. We also discuss issues relating to the current realities of clinical practice and their potential implications for the way that economists model physician-patient clinical encounters.

Time in health: can we measure individuals' "pure time preferences"?

The time dimension in health is assessed and/or explained using the concept of time preference, i.e., preferences about when things occur or the timing of an outcome. It is believed that an individual's time preference can be relatively easily measured. This paper argues that individuals' responses to time-preference-type questions may represent not only their attitudes toward the timing of events (i.e., the points in time at which they are going to occur) but also their attitudes toward other things, such as the sequence of events (i.e., the order of good and bad events over an individual's lifetime health profile). This paper explains the distinction between the time-preference and sequence-preference concepts. Using two recent empirical studies, it demonstrates the inability to measure individuals' pure time preferences. The implications of this empirical obstacle in the context of medical decision making and health-care program evaluation are discussed.