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
- /
- PHREG output statement question

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
- Permalink
- Email to a Friend
- Report Inappropriate Content

11-27-2012 10:57 AM

In phreg, using the output statement with survival=

option, what is the formula to recreate the survival= calculation? I cannot recreate it.

Thank you very much!

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

Posted in reply to msecic

11-27-2012 03:19 PM

There are a number of survivor functions available, and the calculations are available for each type in the documentation under Details: PHREG Procedure>Survivor Function Estimators. There is a Breslow estimator, Flemin-Harrington estimator, and the product-limit estimator.

You should be able to check the calculation based on the formulas provided.

Steve Denham

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

Posted in reply to SteveDenham

11-27-2012 03:31 PM

I'm trying to use the betas from the model, plug in the individual patient data into a formula with the betas and try to recreate the exact values that I get in the survival=

I think i would use this, but what do I plug in for alpha?

hi(t) = exp(α + β1xi1 + β2xik + ・ ・ ・+ βkxik)

Thoughts?

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

Posted in reply to msecic

11-27-2012 05:44 PM

I'm not sure, but here's what I'd try

The formula you have is the hazard function formula, h(t), not the survivor function.

The survivor is S(t)=S(baseline)(t)^exp(B1x.....)

from SAS docs.

You have your beta's from the output and you can get the baseline syntax in proc phreg

Baseline (out=)

I don't know if you'd need to set values for the covariates or not, but that's where I'd start.

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

Posted in reply to Reeza

11-28-2012 05:37 AM

Expanding on Reeza's comment.

Some math thoughts: In exp(α + β1xi1 + β2xik + ・ ・ ・+ βkxik), the first term exp(α) is a constant that ends up as a multiplier of exp( β1xi1 + β2xik + ・ ・ ・+ βkxik), and is thus just part of S(baseline). You do not need alpha to calculate survivor functions.

Steve Denham