Aggregation of scatter plot in PSA of microsimulation Markov model
lz
Dear community,
I am now doing a CEA using the microsimulation Markov model. I used analysis->Monto Carlo Simulation->sampling (10,000) and trials (10,000). After having the results summary, I found that the ICE Scatterplot (AvsB) shows obvious aggregation. The variables, tables and their distributions have been checked without error. Besides, When doing the subgroup analysis, I found that all the incremental QALYs of the subgroup analysis were less than that for total sample analysis.
What is the possible problem with the model (or analysis)?
Thanks for any help!
Lz
Comments
When you run Sampling/PSA and Trials/Microsimulation, the data will always be aggregated up to a single set of results for each model calculation, despite the fact that the results were generated from individual patients.
If you want to see variability among patients and/or subgroups, you need to run Trials/Microsimulation without PSA.
If you really need PSA for a subgroup, you will need to temporarily change the assignment of patient characteristics to generate only patients that fit withing the desired subgroup. Then the full PSA results would be for that subgroup.
Andrew
Please sign in to leave a comment.