Update to the -multistate- package now available on SSC

[Michael Crowther]

Thanks to Kit Baum, the -multistate- package for multi-state survival analysis has just updated on SSC. This is version 2.0.0, so a major update with lots of new features:

• cyclic transition matrix - observations can now return to previous states
• Aalen-Johansen non-parametric estimator of transition probabilities
• standardised/population-averaged predictions
• transition-specific timescales
• predict the probability of ever visiting each state within a time window
• user-defined predictions - provide your own Mata function to calculate bespoke predictions
• speed improvements
• bug fixes

More details here:

https://www.mjcrowther.co.uk/2018/03...version-2.0.0/

Thanks,
Michael

[Chen Samulsion]

Dear Michael Crowther, Thank you and Paul Lambert for the interesting job. I heard Lambert when I study social stratification and his CAMSIS.
Let me keep on the target. I install the -multistate- package (ssc install mutistate), but I find there's no help file for multistate command (no file named multistate.sthlp or multistate.hlp). Does this suggest that "multistate" itself is not a command, while relevant tasks will be completed by using msset, msaj, msgraph etc. successively? By the way, I must confess that I've never used multistate 1.0.0 before. Thank you.

[Chen Samulsion]

The description put that the package provides a set of commands, including msset, msaj, msboxes, stms, predictms. Maybe that's the very answer.

[Steve Samuels]

This is wonderful update, Michael! Thank you!

[Michael Crowther]

Thank you both, hope you find it useful! Yes spot on Chen Samulsion there's no multistate command, but the list of commands you found in the description.

Thanks,
Michael

[Carole Rosenberg]

Thank you very much for this wonderful package.

I have a question about update about cyclic transition.
I have mutltistate and cyclic data but I don't manage to program them. The example I found is only about unidirectional models, is there an example somewhere with cyclic data?

Thank you.