State Transition Model

A general term for a decision-analytic model, such as a Markov model, representing disease progression as movement between defined health states.

Source: Sonnenberg FA, Beck JR. Markov models in medical decision making: a practical guide. Medical Decision Making. 1993;13(4):322-338. doi:10.1177/0272989X9301300409.

A state-transition model is a general term for a decision-analytic model, such as a Markov model, that represents disease progression as the movement of patients between defined health states over time. Patients occupy mutually exclusive states and transition between them according to specified probabilities, with costs and effects accruing as they do. The term covers both cohort and individual-level forms and is used broadly for models built on health states and transitions, of which the Markov model is the most common example.

Source: Sonnenberg & Beck 1993

A state-transition model is a framework that represents disease progression as movement between a set of defined health states over time. Patients occupy one state at a time and may move to others at the end of each cycle according to transition probabilities. The approach covers both cohort Markov models, which track proportions of a group, and individual simulations, which follow one patient at a time. It is widely used in economic evaluation because many diseases can be described naturally in terms of distinct clinical states. Careful choice of states and cycle length shapes its accuracy.

Source: Siebert et al. 2012

A state-transition model is characterised by representing the disease as a set of defined health states and modelling patients' movement between them over time through transition probabilities, with costs and health effects attached to states or transitions. It captures disease progression as a sequence of state changes rather than as continuous processes or fixed pathways. This state-and-transition structure, whether followed by a cohort or by individuals, is the defining feature that distinguishes state-transition models from other modelling approaches such as decision trees.

Source: Sonnenberg & Beck 1993

A state-transition model can take a cohort form, in which a hypothetical cohort is tracked as proportions across states using average transition probabilities, or an individual, or microsimulation, form, in which patients are simulated one at a time through their own state histories. The cohort form is simpler and deterministic; the individual form captures heterogeneity and history at greater computational cost. Both share the state-and-transition structure, so the term state-transition model encompasses these related approaches to representing disease progression.

Source: Sonnenberg & Beck 1993

A Markov model is the most common type of state-transition model, one in which transitions depend only on the current state, the Markov assumption. The broader term state-transition model covers models built on states and transitions generally, including those that relax the Markov assumption, for instance by using individual simulation with memory. So all Markov models are state-transition models, but the state-transition category is more general, encompassing related models that represent disease as movement between states.

Source: Sonnenberg & Beck 1993

A state-transition model is used to represent conditions where disease progression can be described as movement between distinct health states over time, particularly chronic and recurring conditions where events repeat and outcomes accrue over years. It suits problems that a decision tree would represent awkwardly because of recurring events or long horizons. By capturing the states patients pass through and their transitions, a state-transition model efficiently represents ongoing disease processes, making it a standard approach in health economic evaluation of long-term conditions.

Source: Sonnenberg & Beck 1993