Markov cohort analysis

Hi, Andrew Munzer

In such a model framework (as shown in the figure below), we want to know the percentage of the final state 1 and state 2 are in the therapy of drug A, and how can we know this value in the treeage? (Note: State1 and state2 are available in Both Success and Failure branches)
If the model only includes Markov node, we can use Markov Cohort Analysis for analysis, but if our model is complex and includes both the decision tree and the Markov node, how can we quickly get the ratio of each state in the treeage software? And can we get the confidence interval at the same time? 

Many thanks.




  • Zijie

    Your question has many parts and it may be appropriate for you to send a support ticket and ask us directly so we can assist you with your specific needs.

    Generally in answer to your questions, detail of the Markov Cohort details are only saved/reported when running Markov Cohort Analysis (or indeed when Patient Reporting is turned on for Simulation models). Otherwise, the outputs of the analysis become part of the Decision Tree Analysis (and so the health state details are not reported).

    It is simple to create additional payoffs to capture the proportion of the cohort in each health state and report this at the end of the model. This would mean you can just run your usual analysis and the additional payoffs would provide you the final proportion in each state.

    Please let us know if you would like us to assist you further with one-on-one advice via


    Comment actions Permalink
  • Hi, Belinda Orme

    I'm sorry, let's simplify the question. if a strategy has two markov nodes, can the ratio of each health state simply add up?

    Thanks for your patient reply.


    Comment actions Permalink
  • There is no automatic way to sum those percentages. Also, the percentages by state for each Markov model will always sum to 100%.

    As Belinda suggested, if you use an extra payoff, you can get that information fairly easily.

    1. Add another payoff via Tree Preferences > Payoffs.
    2. Check the box to calculate all enabled payoffs via Tree Preferences > Payoffs.
    3. Enter custom names for all your payoffs including the new one via Tree Preferences > Payoffs > Custom Names.
    4. Enter a 1 for the the Final Reward for states 1 and 2 in both Markov models.
    5. Run Rankings.

    You should see the final percentage of people in the first 2 states. Each Markov model itself will return the percentage of people in those states at the end of the analysis independently. But the decision tree portion of the model will take the weighted average of those two percentages for the overall value for the strategy.


    Comment actions Permalink
  • Hi, Andrew

    Thanks for your reply.

    Best regards


    Comment actions Permalink

Please sign in to leave a comment.

Didn't find what you were looking for?

New post