have the accuracy sometimes associated with formal mathematical calculations. The results are simply the consequence of combining both factual and uncertain quantities, both objective and subjective elements, that are an inescapable part of reaching conclusions about the preferred investments in new vaccines.

A quantitative structured model facilitates examination of the effect of uncertainty (in data and estimates) in a fashion that intuitive integration of such components does not. This is expressed in the sensitivity analysis reported in the study.

Value judgments on disease conditions are involved in all processes in which the importance of different diseases is ranked; incorporation of value judgments may be explicit, implicit, or unrecognized. These judgments are more subjective than those of a scientific nature. Providing a specific point at which the required value judgments are described and incorporated furnishes one means of isolating these differences of opinion (which are often incorporated into decision making in an ill-defined fashion) and determining if they affect the ultimate priorities.

The committee resolved many internal differences of opinion over these problems and others it as sought agreement on the approach it would follow in this complex area. Where information was incomplete or where quantitative prediction was complicated by many as yet unresolved issues, it chose to lay out what it felt was the most rational approach to selecting priorities, recognizing that exact data on all components required by the system would not be available before decisions had to be taken. Because of the many uncertainties involved in data and estimates used in the calculation of health benefits and costs, the usefulness of the numbers used in the final rankings lies in their relation to each other rather than their absolute precision, i.e., the system facilitates comparison of vaccine projects in a fashion that is open to revision if different estimates or assumptions seem appropriate and as new data becomes available.

The model is based on comparisons of expected health benefits and expected net costs (or savings) calculated for candidate vaccines. This approach combines elements of decision analysis and cost-effectiveness analysis. It was developed by the committee because it identifies each logical component contributing to vaccine benefits and costs but does not require placement of a monetary value on human life or suffering, in addition, it requires substantial amounts of information about diseases and vaccine characteristics. Committee members believe that the activity of gathering this information is beneficial in itself; it strengthens the decision-making process and highlights areas in which more research is needed. Chapter 2 describes four other approaches to establishing priorities that were considered but judged less satisfactory.

It should be emphasized that the proposed system is designed as an aid to decision making and not as a definitive answer to the selection process. Rather than provide a single list of priorities, the committee intends to demonstrate how different rankings could result from the adoption of different viewpoints on the undesirability of illness or death in specific age groups, or from assumptions about utilization and