Decision analysis for public health: principles and illustrations

Abstract

Over the past decade decision analysis has been applied increasingly to choices in the domain of public health. The technique allows policy makers to examine alternative health strategies in an explicit and logical manner. Each problem yields a model in standardized form so that disparate opinions and goals can be incorporated into the policy. The most common notation uses a decision tree to represent the model. Events are separated into those requiring a choice and those occurring by chance. The relative worth or utility of each potential outcome is specified separately from the probability of its occurrence. All possible outcomes are valued on a single coherent scale, although that scale may include many different attributes. The resource costs and the public health effectiveness (in terms of quality adjusted years of life saved) are two of the potential attributes commonly used in such analyses. The use of sensitivity analyses allows one to examine how changes in soft data can influence policy decisions. In this chapter, we present some of the basic principles that underlie this technology and illustrate its application to several recent controversies.