Building a Decision Tree Cost Effectiveness Model

Abstract

Decision analytical modelling as a vehicle for cost effectiveness analyses may use various modelling approaches including decision trees and Markov models. Determining when to use a particular modelling approach and choice of model will depend on a number of different factors. For example, decision trees are most useful when health events happen close together and don't repeat; when health events happen quickly or not at all; and when uncertainty over the effects of treatment is resolved quickly. This chapter guides you through choice of model with focus lying on how to develop a decision tree to assess cost effectiveness. 3.1 Introduction Decision analytical modelling is often described as a vehicle for cost effectiveness analysis (CEA) (Saramago et al. 2012). As highlighted in the previous chapter, whilst CEA may be undertaken within the context of a single randomised controlled trial (RCT), use of single trial data alone limits analyses to the population within that trial, the interventions assessed and the time over which the trial participants are followed up (NICE 2013). Decision analytical modelling allows us to tailor our analyses to the question we are trying to answer, facilitated by use of multiple sources of data to answer questions relating to resource allocation that are typically beyond the scope of a single RCT. This chapter introduces decision trees in cost effectiveness models and is struc-tured as follows: In Sect. 3.2 we define decision modelling and discuss when decision modelling in CEA is appropriate and what factors influence choice of model (decision tree or Markov model). Section 3.3 then guides you through how to develop a decision tree to assess cost effectiveness. Using an example, we show you how to construct and populate a model and how to interpret the results. Section 3.4 discusses potential costs and outcomes that you might want to think