Sampling rates

Please explain the difference between the sampling rates for distributions - particularly stage vs. trial.

If I choose to resample per Markov stage does it mean that all trials will have the same value per _stage?

If I choose to resample per individual trial will each trial have the same value across _stages?

Is there a way to resample per trial per stage such that, for example, the same drug has a different cost for each individual which varies every cycle?



  • Aliza

    Distributions can be created for Probabilistic Sensitivity Analysis and Individual Microsimulation. Depending on what you are using the distribution for will determine which of the re-sampling rates you choose.

    For distributions which are used for uncertainty (PSA) you would re-sample per EV.

    For distributions which need to change for every individual - such as patient characteristics - you would re-sample per Trial. For example, a distribution for Age where each trial can have a different age would be sampled per Trial.

    Lastly, distributions which you want to give a different value each Markov cycle select the re-sample per Markov Cycle. This is less common.

    Can I encourage you to read the Help files, Chapter 20 specifically talks about many options for creating distributions.

    If you have a very specific question, feel free to send it to support.

  • Thanks Belinda. I don't have a specific question in this case. I have read Chapters 20 and 21 but either missed the explanations or didn't understand them. For example, what does EV even stand for in this case? Expected Value?

    I guess I am just asking for a recapitulation of the manual's explanation in simpler, more nuts-and-bolts terms.

  • EV refers to Expected Value. So when you sample a distribution "per EV", you are really sampling it once for each calculation of the model. This is appropriate for PSA.

Please sign in to leave a comment.

Didn't find what you were looking for?

New post