Re: st: competing risks with repeated ordered events

[Marina Daskalopoulou <med2140@gmail.com>]

Dear Christophe

If you want to fit the model for all your events simultaneously you need to
augment the data set by n number of events and then generate an identifier
for each failure type.

expand 3 /* if you have 3 events for example */

bys id: gen k=_n /* sort by your identifier and then the first record of
each subject will be their failure type (1,2,3...) */

Then generate a binary failure indicator:

gen failed=(event==k) /* make sure there are no missing values in "failed"
*/

Declare the data as survival data:

stset time, f(failed) /* include censoring/origin date etc */

And run your proportional hazards model

xi: stcox i.varname i.varname2 i.varname3 i.k

You could also run it in a loop:
forvalues i=1/3 {
stcox i.varname i.varname2 i.varname3 if k==`i'
}


The CASCADE collaboration has an excellent manual on competing risks on the
MRC CTU website.

Hope this helps,
Marina
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/