one sample survival power calculation

[#1]

Hi All,

There is an open label single treatment registry study that will enroll 1000 subjects who have had an initial intervention and then will be followed for 18 months. The event is days to second intervention. I’d like to find out statistically what 1000 buys, i.e. what can be tested such as days to 2nd intervention GT 182, in a survival analysis.

Is anyone aware of a program for one sample survival power calculation? Jill Tao at SAS Technical Support said SAS does not have a procedure for this calculation. She suggested contacting this forum.

Best,

John

john.jones@bioventusglobal.com

[#2]

Don't know if this helps as you don't say much about what type of survival analysis you are doing.

Proc Power documentation:

• COXREG <options>;

The COXREG statement performs power and sample size analyses for the score test of a single scalar predictor in Cox proportional hazards regression for survival data, possibly in the presence of one or more covariates that might be correlated with the tested predictor.

[#3 · Accepted solution]

If Jill says SAS doesn't have something for single arm survival, it is time to ask Mr. Google.  I found this site, for an interactive on-line calculator:

https://stattools.crab.org/Calculators/oneNonParametricSurvival.htm 

SteveDenham

[#4]

Hi Steve,



Jill is awesome isn't she; like a partner. Tech Support at SAS is the biggest justification for the software's cost.



I too Googled, before contacting SAS tech support, but could not find anything. Thank you very much for the link to CRAB. It is a one sample survival power and sample size calculator from a reliable source too. I'll let Jill know and then try it out some more.



My gratitude,



John

[#5]

Thank you very much BallardW.

The clinical team has decided they want 1000 in the registry. The registry is entirely exploratory so I am planning for both Kaplan-Meier and Cox PH. The intent is to better understand what the power of 1000 subjects will be under different assumptions. One thousand kinda says it all.... Many many thanks again. John