turn on suggestions

Auto-suggest helps you quickly narrow down your search results by suggesting possible matches as you type.

Showing results for

Find a Community

- Home
- /
- Analytics
- /
- Stat Procs
- /
- Sample Size Calculation for Log-Rank Test: Technic...

Topic Options

- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Printer Friendly Page

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Highlight
- Email to a Friend
- Report Inappropriate Content

05-06-2016 07:05 PM

Dear all,

I hope this could the right place to raise a question regarding the technical details for sample size determination using proc power for comparing two survival curves:

http://support.sas.com/documentation/cdl/en/statug/63033/HTML/default/viewer.htm#statug_power_a00000...

My question is to understand the calculation of N(i+1) specified as

N_{j}(i+1) = N_{j}(i) * [1-h_{j}(t_{i})*(1/b) - \phi_{j}(t_{i})*(1/b) - (1/b)*(1/(T-\tao-t_{i}))* I_{t_{i}>\tao})]

It looks as if the very last term 1/(T+\tao-t_{i}) is for administrative censoring (i.e., patients who enrolled late in the accrual phase and survival longer than \tao could be censored due to study closure). For example, suppose that subject live forever and will never be censored, (i.e., h(t)=0 and \phi(t)=0 for all t), suppose that N(0)=100, T=4, \tao=0 and b=1, then:

N(1) = 100*(1-1/3).

However, assuming subject entry is uniformed distributed between [0,4] then only subject enters from [3,4] could be censored and should the right answer for N(1) is

75 = 100*(1-1/4)?

I really appreciate the insights from you .

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Highlight
- Email to a Friend
- Report Inappropriate Content

05-09-2016 05:26 PM

Problem solved based on Cantor 1997 pp 83-92 and Joanna H Shih 1995 (Sample Size Calculation for Complex Clinical Trials with Survival Endpoints).