Risk and Medical Decision Making
本书回顾并扩展了医疗决策理论，探讨在诊断和治疗风险下如何做决策，适合对医疗决策、风险分析感兴趣的读者。
Risk and Medical Decision Making, by Louis Eeckhoudt, 2002, Norwell, MA: Kluwer Academic Publishers Risk and Medical Decision Making is a new addition to the Studies in Risk and Uncertainty series published by Kluwer Academic Publishing. Books in this series cover a variety of topics relevant to insurance research including risk, valuation, and decision making under uncertainty. This latest addition reviews and expands on the existing theories of decision making in the medical care context. This focus is motivated by the fact that, unlike many financial risks that can be either divided or transferred to others (e.g., through diversification, insurance or social security) the personal aspects of medical risks are by essence indivisible and nontransferable. While the models are developed in the context of diagnostic and therapeutic risks, they may be generalized for decision making in a variety of contexts. The book contains an Introduction, followed by eight chapters divided into two parts. In the first part, the author focuses on medical decisions within the expected utility (EU) model. Part 2 addresses new non-EU models of choice including the dual theory of choice under risk. The reader should be familiar with the basic concepts of decision sciences, the use and notation of decision trees and the notion of aversion in the EU model. At the end of each chapter the author provides a summary of the key findings and several review questions (some solutions are provided). In the Introduction (Chapter 1), Eeckhoudt provides a literature review to place his work in context. He notes that much of the book follows the seminal work in medical decision making by S. Pauker and J. Kassirer (PK) in the 1970s. These works included (1) optimal therapeutic decisions in the presence of diagnostic risks, and (2) the valuation of diagnostic tests, and relied on the EU model to evaluate risky situations. Eeckhoudt recognizes and classifies the developments that followed PK's work, and structures this book along two categories of research. First, he notes that models of medical decision making have been expanded to include background risks, or comorbidities. These models develop a concept of prudence to complement risk aversion. Eeckhoudt provides several examples in the following chapters to illustrate this new concept. The second category of research involves new models of choice under uncertainty. Eeckhoudt notes that these new models deserve more attention because they give new intuitions about optimal behaviors. The last two chapters of the book provide a basic introduction to these areas of research. Chapter 2 is devoted to a review of the 1975 PK paper, in which the decision is to treat or not to treat, when the patient may or may not have a disease. Thus, the decision tree involves four outcomes, which are stated in terms of quality adjusted life years or QALYs. The choice of whether or not to treat depends on the benefit of treatment (known), the loss of QALYs because a healthy patient is treated, and the utility function of the decision maker. Eeckhoudt evaluates the decision tree in the case of a neutral decision maker to obtain a threshold for treatment. He shows that this threshold falls if the benefit of treatment increases or if, for a given benefit, the loss of QALYs for treating a healthy patient falls. When the problem is evaluated for a averse individual, the threshold falls because the treatment option is reducing; averse patients demand treatment more often than neutral patients. Chapter 3 introduces the notion of prudence. The decision maker's treatment decision is confounded with background risk, which is described as a co-morbidity risk. In other words, it is assumed that the patient faces an uncertain of developing another health problem if the disease is present. The model may also be interpreted to allow for an unknown severity of the disease. …