Re: st: simulation of survival datasets in Stata

["E. Paul Wileyto" <epw@mail.med.upenn.edu>]

You need to simulate time-to-event followed by censoring.  The easiest 
is exponential or Weibull. 
Weibull is :
S=exp(-lambda*t^p)
where lambda is the hazard ratio:
lambda=exp(sum beta x)

beta_0 is a constant reflecting the baseline hazard.  beta_1 ... reflect 
the effects of covariates.
p is the Weibull shape parameter (positive number), and if valued p=1 
reduces to exponential.

Generate lambda with hazard ratio 1.9 for drug treatment (0,1):

- lambda=-2.5+ln(1.9)*drug -

To simulate time-to-event, substitute the random unit variate R for S, 
and solve for t. 
t=(-ln(R)/lambda)^(1/p)

or in Stata:

- gen t=(-ln(uniform())/lambda)^(1/p) -

Some of the t values will be very large, and you will need a censoring 
variable to reflect the end of study (say 90 days). 
- gen fail=1 -
- replace fail=0 if t>90 -
- replace t=90 if t>90 -


Then, look at your handiwork-
- stset t, fail(fail) -
- sts graph, by(drug) -
- stcox drug -

Paul







Michael McCulloch wrote:

> Hello Statalist members,
> I'm seeking examples of simulation of survival datasets in Stata. Can 
> anyone point me in the right direction? My goal is to simulate a 
> dataset to validate a Stata survival program I've written.
>
> Thank you.
--
E. Paul Wileyto, Ph.D.
Assistant Professor of Biostatistics
Tobacco Use Research Center
School of Medicine, U. of Pennsylvania
3535 Market Street, Suite 4100
Philadelphia, PA  19104-3309

215-746-7147
Fax: 215-746-7140
epw@mail.med.upenn.edu
http://mail.med.upenn.edu/~epw/ 
*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/