Comparing Three Software Tools for Implementing Markov Models for Health Economic Evaluations

ISSN： 1170-7690

Source： PharmacoEconomics, Vol.27, Iss.9, 2009-09, pp. : 745-753

Disclaimer: Any content in publications that violate the sovereignty, the constitution or regulations of the PRC is not accepted or approved by CNPIEC.

Previous Menu Next

Abstract

Background Various software packages are commonly used for the implementation and calculation of decision-analytic models for health economic evaluations. However, comparison of these programs with regard to ease of implementing a model is lacking.Objectives (i) to compare the assets and drawbacks of three commonly used software packages for Markov models with regard to ease of implementation; and (ii) to investigate how a technical model validation can be conducted by comparing the results of the three implementations.Methods A Markov model on chronic obstructive pulmonary disease was implemented in TreeAge, Microsoft® Excel and Arena® with the same assumptions on model structure, transition probabilities and costs. A hypothetical smoking cessation programme for patients in stage 1 was evaluated against usual care. The packages were compared with respect to time and effort for implementation, run-time, features for the presentation of results, and flexibility. Agreement between the packages on average costs and life-years gained and on the incremental cost-effectiveness ratio was considered for technical validation in the form of expected values (between TreeAge and Excel only) and Monte Carlo simulations.Results Ease of implementation was best in TreeAge, whereas Arena® offered the highest flexibility. Deterministic results were in agreement between TreeAge and Excel, as were simulated values between all three packages.Conclusions Excel offers an intuitive spreadsheet interface, but the acquisition of and the training in TreeAge or Arena® is worthwhile for more complex models. Double implementation is a practicable validation technique that should be conducted to ensure correct model implementation.