The multistate package implements the methodology described in the project Multi-state survival analysis, and published in the Stats in Med article found here. This page contains a collection of tutorials and examples to fit a wide variety of models and predictions.

The current stable version of multistate can be installed from the SSC archive by typing:

ssc install multistate

Any updates can be picked up by typing adoupdate, update. Version history is available here. The development version of multistate in Stata, can be installed by typing:

net install multistate, from("")

and then run adoupdate as normal to pick up any updates, with version history available here.

I recommend you read the accompanying methods paper, and have a thorough look at the help files for the commands, in particular predictms, to get familiar with the different options and capabilities. Below are some examples of what we can do with multistate.

Each tutorial below is tagged to illustrate the level of completeness, as follows:

  • [Tutorial] - a completed page, with code and full worked example
  • [Draft] - still in draft, contains code but not fully documented
  • [Code only] - example code only, minimal description
  • [Sim] - example shows how to simulate data from the illustrated model
  • [TBA] - to be added
  • [EG] - indicates that the tutorial shows how to fit an example from the literature within the multistate framework

Things to read first:

  • Most importantly, have a look at the methods paper that introduces the modelling framework here
  • The difference between Markov and Semi-Markov models [TBA]
  • Methods for calculating predictions from a multistate model [TBA]


Markov models

Semi-Markov models

  • A semi-Markov illness-death model [TBA]


comments powered by Disqus