OBJECTIVES: To construct a health economic model for a chronic condition with slow irreversible disease progression (slow process) and recurrent periods of severe symptoms (rapid process), processes which are dependent. We wish to study changes to either process or both simultaneously. METHODS: The processes are dependent; more progressed patients have more frequent and worse attacks, more frequent attacks lead to accelerated progress. Estimating short-term transition probabilities for the slow process is difficult, suggesting a time-scale with long cycles. However, long cycles are irrelevant for the rapid process, thus suggesting short cycles. Denote the progress with a discrete scale of increasing progress, 1, 2, ..., and attacks 0=no attack, and 1=ongoing attack. RESULTS: We use a two-dimensional discrete time Markov model, disease progression in dimension Y and attacks in dimension X. The probability of X depends on the previous X and Y, and vice versa for Y. Hence the dependence between X and Y can be modelled. Costs and outcomes are estimated using state and transition rewards, enabling health economic analysis. Updating X more frequently (e.g. weekly) than Y (e.g. yearly) gives relevant cycle lengths. Y updates in outer cycles, and X updates in inner cycles. X's dependence on Y is straight-forward, but Y's dependence of X makes it necessary to summarise outcomes in the inner cycle, to inform update in the outer cycles. The type of summary depends on the evaluation method, e.g. Monte Carlo simulation requires total time in X=0 and X=1 to estimate the progression risk since it depends on X. CONCLUSIONS: Using cycle lengths relevant for each of the processes improves transition probability estimates in terms of predicting observed outcomes. This compensates for increased error due to different update rates. The face validity of the model may be improved by using relevant cycle lengths for the processes.