Methods

1. **Markov model of disease**: 1) define the disease in terms of states (clinically and economically). The states should be mutually exclusive.

2) If a model comprises k states, there are k * k transition matrix. In practice, many of these may be set to 0.

3) An important assumptio of Markov model: the probability of moving out of a state is not dependent on the states a patient may have experienced before entering that state. (Markovian assumption).

4) Two different types of Markov model can be characterised by the form of transition probabilities. Markov chain: transition probabilities are constant  Time-dependent Markov process: flexible in modelling of chronic disease

5) The problem is probabilities available in the literature may not refer to the same period of time as the chosen Markov cycle.

6) Attach weights to the model for the cost and health outcome quantities to be estimated. A weight of 1 is attached to each state of the model in which a patient is alive and a weight of 0 is attached to the dead state.

7) Markov models are particularly suited to the calculation of quality-adjusted life-years (QALYs), since the construction of QALY involves weighting the length of time spent in a particular state of health by a value representing quality of life experienced in that state of health.

8) When calculating cost, it is often useful to attach costs not only to the states of the model themselves, but also to transitions between states.