American Medical Association 2-4 by the same authors (members of the Panel on Cost-Effectiveness in Health and Medicine) have become a standard of cost-effectiveness analysis.
The most significant feature of this volume is setting forth a clear and consistent set of rules for performing cost-effectiveness analysis. The usefulness of cost-effectiveness analysis lies in its ability to compare one program with another. The value of each cost-effectiveness analysis is analogous to the value of an intercom for communicating around a large building. A single intercom would be useless, as no one could hear the message. With two compatible intercoms, one conversation can at least begin. The number of two-way conversations then grows approximately with the square of the number of intercoms. With just ten compatible intercoms, 45 different two-way conversations would be possible. If the intercoms were not compatible, no conversations could occur. The field of cost-effectiveness analysis has been one with few 'conversations' because of incompatible approaches.
Gold et al. recommend that all cost-effectiveness analyses include the 'reference case' to help ensure consistency and relevance in future cost-effectiveness studies. The reference case specifies a standard comparator (existing usual care) against which new treatments are compared. The reference case uses a standard real discount rate of 3% per year-a rate that approximates the real long-term return in many public projects. Finally, the reference case adopts a societal perspective that includes not only the costs of health institutions, but also values the time of the patient, and unpaid time contributed by his friends and family.
One of the strengths of the book is the explicit discussion of modeling in linking the necessary data for cost-effectiveness analysis. Chapter 5 (Mandelblatt et al. 'Assessing interventions . . .') nicely catalogs the various types of model often used in costeffectiveness analysis. As most realistic cost-effectiveness studies require combining information from economics, epidemiology, biostatistics and policy, all the necessary information is rarely furnished from a single data set or clinical trial. Commonly, the need arises because most cost-effectiveness studies have a time horizon extending across several years, and sometimes across decades or an entire lifetime. Few real intervention studies have this length of follow-up. Thus, this chapter notes, cost-effectiveness analyses of management strategies for chronic diseases often involve state-transition models, which are well suited to analyzing situations in which the states recur. Analyses of the diagnosis and management of acute conditions are well represented by decision trees, for which software such as SOFTREE is helpful.
One of the important by products of the use of models is the ability to refine the plans for possible health programs to make them more cost-effective. In chapter 3 ('Framing and designing . . .'), Torrance et al. note that a program of mammography screening for breast cancer addresses not only whether to screen, but also which women to screen and how frequently to repeat the screening.
The book's closing chapter (chapter 9, 'Reporting') by Siegel et al. is an explicit recognition that cost-effectiveness analysis is an applied field intended to influence policy. To enable decision makers to judge the quality of a cost-effectiveness analysis, this