Tunnel state configuration for Markov cohort

I have a Markov model which requires tunnel states to account for different probabilities based on time spent in a health state. I was trying to understand how tunnel state works, so I created a small Markov model from the TreeAge manual, as shown in the image below:

I was under the impression that the _tunnel=3 under "Recover" label indicates a probability of 1, which means who did not die entered pre-cancerous state. So, I altered the model to look like the image below:

But the results are very different for these two models. I cannot figure out the difference in results between these two models. In the first image, what is the probability under "Recover" label (_tunnel=3), if it is not 1? Thank you.



1 comment
  • Official comment

    The example model uses tunnels in a typical way, but the probability expression may be confusing.

    _tunnel = 3

    The above is really a logical expression. If the logical expression is true, then it returns 1. If it is not true, then it returns 0.

    Therefore, from the Live event, the cohort will stay in the Cancer state for the first two cycles (_tunnel values 1 and 2), but will return to the Pre-Cancerous state in the third cycle (_tunnel value 3).

    Your version eliminates the opportunity to return to the Pre-Cancerous state, so you should get different results.

    To understand this further, I recommend that you run the Markov Extended Report on the original model and uncheck the option that consolidates tunnel states. You should then see the patient flow I highlighted above.


Please sign in to leave a comment.

Didn't find what you were looking for?

New post