[Sean Hardiman]

Thank you! I will give this a try!

[Sean Hardiman]

Paul, a follow-up question - is there any significance to using foo as the saved value (not sure what the technical term for that value is)? I have three comparisons amongst for groups (2-1, 3-1, 4-1) to wanted to be sure I hadn't missed something. Thanks!

[Sean Hardiman]

Hi Paul Dickman, everyone,

Bumping an old thread...as I have a new question!

I've used stpm2 to plot survival curves and predict point estimates and confidence intervals as was suggested earlier. This has worked really well.

I'm now finishing up a publication submission and a journal has asked for a table with the number of patients at risk at each time point (i.e., a risk table). Looking at the stpm2 and stpm2cr documentation, I'm not seeing a way to do this as described that is similar to what is possible for figures plotted using sts graph. Does anyone have any suggestions for the correct approach to establish a risk table from an stpm2 model?

Thanks!

[Bjarte Aagnes]

Technically if you want a risk table based on the data used for both KM and model based survival curve: either use the sts option addplot() to add your tw plots of the model based survival curves, or after sts graph use -graph describe- and copy the relevant parts of the command returned to your tw plot command.

One consideration might be to show the unweighted or the weighted data.

The best way to present a risk table for the survival curve from a specific model with interactions and/or continuous variables is easier to comment if the model, and data are described.