Chapter 14. Decision Analysis in Drug Development

Carl-Fredrik Burman^[]

^[] Carl-Fredrik Burman is Statistical Science Director, Technical and Scientific Development, AstraZeneca, Sweden.

Andy Grieve^[]

^[] Andy Grieve is Head, Statistical Research and Consulting Center, Pfizer, United Kingdom.

Stephen Senn^[]

^[] Stephen Senn is Professor, Department of Statistics, University of Glasgow, United Kingdom.

Decision analysis is a quantitative methodology for analysing and optimising decisions. It has its roots in Statistical Decision Theory as outlined by Raiffa and Schlaifer (1961). The basic idea is to structure the decision problem, quantify uncertainties and preferences and then solve the resulting optimality problem. In this chapter, we will present some typical decision problems in drug development: go/no go decisions, sample size calculations, dose-finding studies, sequential designs and project prioritisation. The examples will be displayed and analysed using the dedicated Operations Research module SAS/OR, together with optimisation routines in SAS/IML as well as the statistician's standard SAS arsenal.