14.9. Summary

Decision analysis is a general approach to decisions, based on problem structuring, quantifying and optimising. In principle, it can be applied to any decision. We have seen a number of examples from drug development, with emphasis on clinical programs and trials. The examples range from the design of a single trial (sample size, sequential design), over the question about optimal dose, to the design of a program and project prioritization. The list could of course be expanded. In particular, more examples could be taken from discovery and commercial perspectives. An attractive feature of decision analysis is that it often promotes cross-skill cooperation and is a useful tool to facilitate mutual understanding of each others' ideas.

It must be stressed, however, that decision analysis must be handled with care. Within a decision model, it is straightforward to optimise the decision. However, this decision may be quite bad if the decision model is not a good enough model for the real problem. There is a clear risk of getting a suboptimal decision as a result of a model that is too simplified. On the other hand, making models too complicated give a similar risk. It is then hard to see through the assumptions and to assess their validity. The inherently quantitative nature of a decision analysis may constitute a temptation to overvalue aspects that are easily measured and ignore qualitative aspects. It is important to realize that ethics must get priority over profit, that a project cannot always be analysed separetely from long-term company strategy, that psychology sometimes is more important than cold rationalism. The conclusion is that decision analysis by all means should be applied in pharmaceutical development but that the applications should be made with care.