Deterministic Markov Model — Fixed-parameter cohort state-transition analysis
Deterministic Markov Model — Fixed-parameter Markov chain for cohort-level state transitions · Also known as: DMM, Deterministic Markov Chain, Cohort Markov Model, Fixed-Parameter Markov Model
A Deterministic Markov Model is a cohort-level state-transition model in which all transition probabilities, state utilities, and costs are assigned single fixed values and the model is solved analytically in a single pass. Widely used in health technology assessment, policy analysis, and operations research, it traces a hypothetical cohort through mutually exclusive health or system states over discrete time cycles, accumulating expected outcomes such as quality-adjusted life years (QALYs) or costs.
Read the full method
Sign in with a free account to read this section.
Method map
The neighbourhood of related methods — select a node to explore.
When to use it
Use a Deterministic Markov Model when the decision problem involves transitions among a manageable number of discrete states over time, parameter estimates are available as point values from the literature, and a fast, transparent base-case result is needed before investing in probabilistic analysis. It is the standard starting point for health economic evaluations submitted to HTA bodies such as NICE. Avoid it when individual-level heterogeneity is essential to the research question (use microsimulation instead), when memory of prior states matters (tunnelling states or discrete-event simulation may be needed), when the number of states grows combinatorially, or when the analysis requires a full probabilistic uncertainty characterisation as the primary output (use probabilistic sensitivity analysis layered on top, or switch to a stochastic Markov model).
Strengths & limitations
- Transparent and fully reproducible: a single deterministic run produces results that can be traced step by step and audited.
- Computationally fast: no sampling required, so even complex multi-state models run almost instantaneously.
- Well-accepted by regulators and HTA bodies as a standard health economic modelling framework.
- Easy to validate by hand calculation for simple models, improving confidence in implementation.
- Naturally structures the problem into states and transitions, forcing explicit elicitation of all key parameters.
- All parameters are entered as fixed point estimates, so the base-case result ignores parameter uncertainty by design — probabilistic sensitivity analysis must be added separately.
- Assumes the Markov property (memorylessness): future transitions depend only on the current state, not on how long a patient has been there or prior history.
- Models population averages; individual heterogeneity is not captured unless the cohort is stratified into sub-groups.
- State-space explosion: capturing complex clinical pathways may require a large number of states, making the model unwieldy.
- Tunnel states are needed to model time-in-state dependency, which complicates model structure.
Frequently asked
What distinguishes a deterministic Markov model from a probabilistic (stochastic) one?
In a deterministic model every parameter is a fixed point estimate and the model is run once, producing a single expected outcome. In a probabilistic sensitivity analysis (PSA) extension, parameters are drawn from distributions across thousands of simulations to characterise second-order uncertainty. The underlying cohort transition mechanism is the same; 'deterministic' refers to the model run, not the clinical process.
How many health states are typically needed?
There is no universal rule, but parsimony is valued. Models commonly use 3–10 states. Adding states to capture tunnel effects or sub-stages increases precision but also increases data requirements and validation burden. Each additional state requires reliable transition probability estimates.
What is the half-cycle correction and when should it be applied?
The half-cycle correction assumes that, on average, transitions occur at the midpoint of a cycle rather than at the start or end. It is recommended when cycle lengths are long (e.g., annual cycles) and event rates are moderate to high, as it reduces systematic bias in accumulated rewards. It is less important with short cycle lengths.
Can a deterministic Markov model handle age-dependent transition probabilities?
Yes. Time-varying transition probabilities can be incorporated by using age- or cycle-dependent probability matrices that are updated at each cycle. This makes the model more realistic for age-related conditions at the cost of additional parameterisation.
When should I use discrete-event simulation instead?
Discrete-event simulation (DES) is preferable when individual-level resource use, patient heterogeneity, queuing, or event histories matter. If the Markov property (memorylessness) is violated and tunnel states become unwieldy, DES offers a more natural modelling framework despite its higher computational and complexity cost.
Sources
- Sonnenberg, F. A., & Beck, J. R. (1993). Markov models in medical decision making: a practical guide. Medical Decision Making, 13(4), 322–338. DOI: 10.1177/0272989X9301300409 ↗
- Briggs, A., Sculpher, M., & Claxton, K. (2006). Decision Modelling for Health Economic Evaluation. Oxford University Press. ISBN: 9780198526629
How to cite this page
ScholarGate. (2026, June 3). Deterministic Markov Model — Fixed-parameter Markov chain for cohort-level state transitions. ScholarGate. https://scholargate.app/en/simulation/deterministic-markov-model
Which method?
Set this method beside its closest kin and read them side by side — the library lays the books on the table; the choice is yours.
- Discrete-Event SimulationSimulation↔ compare
- Markov ModelSimulation↔ compare
- MONTE-CARLO-SIMULATIONDecision-making↔ compare
- SENSITIVITY-ANALYSISDecision-making↔ compare
- Stochastic Markov ModelSimulation↔ compare