Building Better Models: A Simulation Case Study Demonstrating the Crucial Role of Ergodicity in Building Robust Markov Chains

Speaker(s)

Cornell M1, Baines D2, Howells R3
1Clarivate, Philadelphia, PA, USA, 2Clarivate, London, England, UK, 3Clarivate, Manchester, UK, UK

Presentation Documents

OBJECTIVES: We aim to illustrate the importance of ergodicity when building health economics models.

Our objective is to demonstrate when to build ergodic Markov models (MM), their benefits in terms of model stability and the reliability of results, and how ergodic MMs may be built.

METHODS: We build a series of MMs in R to demonstrate how ergodicity affects model structure, operation, and reliability. The research draws on foundational work by Briggs and Sculpher (1998) and insights from the ISPOR-SMDM Modelling Good Research Practices Task Force.

RESULTS: Where viable, ergodic models are more reliable as they can better reflect the treatment processes experienced by patients. Non-ergodic models may have too many health states or sub-models that affect model results negatively.

CONCLUSIONS: Health economists commonly employ MMs, especially in chronic disease modelling, due to their simplicity and ease of development. However, our investigation reveals the importance of ergodicity in model execution where viable. This simulation case study successfully illustrates the critical role of ergodicity in MMs for health economics studies. Even where it is not viable for the MM to be ergodic, lessons from ergodic MMs, such as the elimination of artificial linkages and sub-systems, can be implemented to improve non-ergodic MMs.

Our work underscores the need for enhanced awareness and education on the fundamental concept of ergodicity within the health economics community. Establishing industry-agreed standards on implementing the principles of ergodicity to elevate the quality of MM construction, particularly in the context of CEMs & BIMs.

Code

EE482

Topic

Economic Evaluation, Methodological & Statistical Research

Topic Subcategory

Budget Impact Analysis, Cost-comparison, Effectiveness, Utility, Benefit Analysis

Disease

No Additional Disease & Conditions/Specialized Treatment Areas