R help - Feb 2009 - help: calculations for causespecific hazard ratios in a competing risks analysis with timedependent covariates

Dear R users:
    Analysis of the impact of a time-dependent covariate (GVHD or use of steroid
after bone marrow transplantation) on two competing endpoints (invasive fungal
infection and death) is frequently encountered in the setting of BMT data. Coxph
package can be used as the following:
    for the analysis of GVHD:
 > gvhd -> coxph(Surv(start,stop,status = =1) ~ GVHD, data=bmt.data)
    and the data should be edited as the following:
          patient start stop status(IFI) GVHD
             1     0      15     0         0
             1     15     35     1         1
             2     0      12     0         0
             2     12     20     2         1
             3     0      30     2         0
             4     0      100    0         0
             5     0      25     0         0
             5     25     100    0         1
   similarly, for steroid:
     > steroid -> coxph(Surv(start,stop,status = =1) ~ steroid,
data=bmt.data)
    and the data was edited as the following:
          patient start stop status(IFI) steroid
             1     0      16     0         0
             1     16     35     1         1
             2     0      13     0         0
             2     13     20     2         1
             3     0      14     0         0
             3     14     30     2         1
             4     0      100    0         0
             5     0      21     0         0
             5     21     100    0         1
 
I wonder how can I analyze the two risk factors (GVHD and steroid) concurrently?
And how to edit the data.
How can I manage it if a third baseline risk factor(age for example) was
included?
 
 Thanks 
 Yours sincerely Ping Zhang 
 Feb 24 2009



	[[alternative HTML version deleted]]