PRM92 - A Comparison of Discrete Event Simulation (Des) Versus Markov Models With A Practical Application to Human-Immunodeficiency Virus (Hiv)

Abstract

When conducting an economic evaluation, it is important to select an appropriate model type and provide a justification. Many analyses utilise Markov models but these are associated with a number of limitations. Discrete Event Simulation (DES) models, in which events are estimated using discrete time intervals rather than regular cycles and patients are simulated individually rather than as a cohort, can overcome many of the Markovian limitations. The aim of this study was to assess the advantages and disadvantages of DES and Markov models; utilising an application to HIV. A systematic literature review was conducted to identify modelling approaches assessing the cost-effectiveness of HIV treatments. Additionally, the use of DES models within Health Technology Assessments (HTA) was evaluated. A de novo DES was developed in Microsoft Excel® with VBA, based on assumptions and data from an existing cost-effectiveness Markov model assessing HIV treatments. Of the HIV publications identified, 4% used a DES and 42% used a Markov model. Only 17% provided a discussion around their choice of model type. DES models have not yet been used in HTAs for HIV in the UK but nine were identified within other disease areas. The de novo DES and those in the published literature demonstrated a realistic modelling approach due to the discrete timing of events and accounting for patient heterogeneity. The DES is a flexible model which can accommodate future adaptations; however, it relies heavily on data requirements in order to maximise its potential benefit. Neither Markov models nor DES are superior; the key is to choose the most suitable model for the decision problem and provide a clear rationale. In the context of HIV, DES is likely to be a good choice of model providing sufficient data is available.