Economic Modeling and Simulation

Economic modelling and simulation in health uses mathematical structures to synthesise evidence and project the long-term costs and health consequences of competing interventions. Because trials rarely capture every comparator, every outcome, or a lifetime horizon, decision-analytic models extrapolate and combine data from many sources to produce the cost-effectiveness estimates that inform resource decisions.

Definition

An economic model is a mathematical framework that synthesises evidence on costs and health outcomes from multiple sources and uses simulation to estimate and compare the expected costs and effects of alternative interventions over a defined time horizon.

Scope

The entry covers the role and main families of economic models—decision trees, state-transition (Markov) models, and individual-level microsimulation and discrete-event simulation—together with the principles of model conceptualisation, validation, and transparent reporting. It is methodological reference material describing how models are built and judged, not advice on any particular intervention.

Core questions

• When is a model needed rather than a single trial-based analysis?
• Which model structure best represents the disease and the decision problem?
• How are transition probabilities, costs, and utilities estimated and entered into the model?
• How is a model validated and reported so others can trust its conclusions?

Key concepts

• Decision tree
• State-transition (Markov) model
• Microsimulation
• Discrete-event simulation
• Transition probability
• Cohort versus individual-level simulation
• Model conceptualisation
• Internal and external validation

Mechanisms

A model first conceptualises the decision problem and chooses a structure: a decision tree for short-horizon problems, a state-transition model for conditions described by health states and recurring cycles, or individual-level simulation when patient history or interacting events matter. Evidence on transition probabilities, costs, and health-state utilities populates the structure, and the model is run—analytically for cohort models or by Monte Carlo simulation for individual-level models—to produce expected costs and outcomes for each option. The model is then verified, validated against external data, and reported transparently so that its assumptions and limitations are visible (Caro et al., 2012; Siebert et al., 2012; Eddy et al., 2012).

Clinical relevance

Model-based economic evaluations are central to health technology assessment and frequently determine which interventions a health system funds, so understanding how models work supports critical appraisal of such evidence. This topic explains modelling methodology and is not a source of individual clinical or treatment recommendations.

Evidence & guidelines

The ISPOR-SMDM Modeling Good Research Practices Task Force series provides the principal methodological guidance, with dedicated reports on model conceptualisation, state-transition modelling, individual-level simulation, and model transparency and validation; standard textbooks by Drummond et al. and by Briggs, Claxton, and Sculpher give the foundational treatment (Caro et al., 2012; Siebert et al., 2012; Eddy et al., 2012; Drummond et al., 2005; Briggs, Claxton, & Sculpher, 2006).

History

Decision-analytic modelling moved from clinical decision analysis into health economics during the 1980s and 1990s as evaluators sought to extrapolate trial results to lifetime horizons and to compare interventions not studied head-to-head. State-transition models became the workhorse structure, individual-level simulation expanded with computing power, and the 2012 ISPOR-SMDM good-practice reports consolidated shared standards for building and reporting models (Caro et al., 2012; Siebert et al., 2012).

Debates

Cohort state-transition models versus individual-level simulation

Simple cohort Markov models are transparent and fast but cannot easily represent patient history or interacting events; individual-level microsimulation and discrete-event simulation are more flexible but harder to validate and more demanding of data, and choosing between them is a recurring modelling judgement.

Key figures

• Andrew Briggs
• Karl Claxton
• Mark Sculpher
• Uwe Siebert
• David Eddy

Seminal works

• caro-2012-overview
• siebert-2012-statetransition
• briggs-claxton-sculpher-2006

Frequently asked questions

Why use a model instead of analysing a clinical trial directly?

Trials usually have a limited follow-up time, omit some relevant comparators, and may not measure final health outcomes; a model extrapolates beyond the trial, links intermediate to final outcomes, and combines evidence from several sources to address the full decision problem.

What is a state-transition (Markov) model?

It represents a disease as a set of mutually exclusive health states, with patients moving between states over fixed time cycles according to transition probabilities; accumulating costs and outcomes across cycles yields expected costs and effects for each strategy.