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
- /
- SAS Programming
- /
- SAS Procedures
- /
- Median time and/or survival rate comparision in su...

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

11-09-2010 11:37 AM

Hello all,

In survival analysis (using, say, Kaplan-Meier curves) how can I do such comparisons between two arms? I know log-rank test, but that test is for comparison of two entire survival curves, not just medial survival time or survival rate at certain time point (say, survival rate at 12 months). Personally I don't think this type of comparison is reasonable, and actually I cannot find such single-point comparison in any survival books or statsitical journals, -- I only see this occasionally in certain poster presentations. However, we are sometimes requested to do this by some physicians (like difference or even 95% C.I. of difference of two 12-month survival rates). Any option(s) in, say, "lifetest"? Any suggestions? Thanks!

In survival analysis (using, say, Kaplan-Meier curves) how can I do such comparisons between two arms? I know log-rank test, but that test is for comparison of two entire survival curves, not just medial survival time or survival rate at certain time point (say, survival rate at 12 months). Personally I don't think this type of comparison is reasonable, and actually I cannot find such single-point comparison in any survival books or statsitical journals, -- I only see this occasionally in certain poster presentations. However, we are sometimes requested to do this by some physicians (like difference or even 95% C.I. of difference of two 12-month survival rates). Any option(s) in, say, "lifetest"? Any suggestions? Thanks!

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

11-09-2010 11:21 PM

Hi.

Maybe you should use Wilcoxon-rank-sum-test ,because it is nonparameter statistical method which is robust for any distribution. And log-rank is most powerful only for the survival curves both paralleled.

Ksharp

Maybe you should use Wilcoxon-rank-sum-test ,because it is nonparameter statistical method which is robust for any distribution. And log-rank is most powerful only for the survival curves both paralleled.

Ksharp

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

11-10-2010 08:39 AM

Logistic regression may do what you want. It will handle the failure at a single point in time. It won't handle censoring.

I agree that a complete survival analysis is more powerful, but there are times when it makes clinical sense to address the outcome at a specific point in time. Message was edited by: Doc@Duke

I agree that a complete survival analysis is more powerful, but there are times when it makes clinical sense to address the outcome at a specific point in time. Message was edited by: Doc@Duke

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

01-21-2011 05:54 PM

What about truncating the data to the time point of interest and comparing the differences of the curves up to that point?

I think that's really what's of interest, ie is the surivival different for the first year, second year, etc...

If you truncate at 12 months your estimate is the same as if you hadn't truncated but information afterwards is censoring/truncation in effect.

Does that make any sense?

I think that's really what's of interest, ie is the surivival different for the first year, second year, etc...

If you truncate at 12 months your estimate is the same as if you hadn't truncated but information afterwards is censoring/truncation in effect.

Does that make any sense?