How to simulate survival data using given real dataset

[HappySASUE]

Hi 

I need to do simulation analysis for survival study. I have one real clinical dataset with only 40 patients. need to create 1000 simulation samples.   

Thanks

[mkeintz]

---

@HappySASUE wrote:

Hi 

I need to do simulation analysis for survival study. I have one real clinical dataset with only 40 patients. need to create 1000 simulation samples.   

Thanks

---

How would you create one simulation sample?   And BTW, what sample size do you intend to generate?

I presume by "simulation" that you intend to start out with a sequence of hazard rates, then randomly apply those hazard rates to get period-by-period survivals, yes?  What is the relevance of your clinical dataset?  Is it the source of your hazard rates?

[HappySASUE]

Thank you for your reply.

Yes the clinical dataset is the source of my hazard rates, also with independent variables, such as age, gender.... "Simulation" here I mean is to create samples which "mimic" the real clinical dataset.   The sample size could vary, 30, 40, 400...etc. 

[Reeza · Accepted solution]

I'm not aware of any quick user friendly references. I believe there's a few answered question on here about this and the topic is covered in Rick Wicklin's Simulating data book. 

It also depends on if you're simulating via or just randomly repeating data. You could expand your 40 to 1000 using PROC SURVEYSELECT for example, obviously sample with repetition. 

---

@HappySASUE wrote:

Hi 

I need to do simulation analysis for survival study. I have one real clinical dataset with only 40 patients. need to create 1000 simulation samples.   

Thanks

---

[HappySASUE]

Thank you. I will take a look of this book.  This is a very useful book!