Plot of hazard ratios by subgroup

[Lili Bulfone]

Hi
I have survival data for two interventions. There are several covariates of interest. I would like to produce a plot like the one shown (as would be commonly presented in publications). Is there a straightforward way of doing this in Stata?
Thank you.
Lili

[David Fisher]

Hi Lili Bulfone ,

There are probably various ways of doing this. However, I have written a package called ipdover which automates this. To find and install it, type into Stata: ssc describe ipdover. With a bit of additional work, you can also incorporate extra details like the p-values and median survival times shown in your attached figure.

BW,
David.

[Lili Bulfone]

Thanks very much David. I'm off to read and try out the package!
In case this is useful to anyone else with a similar question, giving Stata the command 'ssc describe ipdover' gives no joy but if you give Stata the command 'search ipdover', it will direct you to David's article about "Two-stage individual participant data meta-analysis and generalized forest plots" in the Stata Journal (Vol 15 #2 ) and to where you can install all of the files (including the ipdover.ado ) in the st0384 package.
Thanks again, David. I really appreciate your help!

[David Fisher]

You're quite right, sorry about that! The package name is ipdmetan, which contains the programs "ipdmetan" (which performs two-stage IPD meta-analysis) and "ipdover" (which performs subgroup analysis, as above) -- the two commands are closely related. So I should have said to type "ssc describe ipdmetan". Please use the SSC installation, as it will be more recent than the Stata Journal download.

Also, please note that the metan package (also from SSC) also needs to be installed in order to produce forest plots.