04-11-2013 11:39 PM
I have some questions. We were using a published case-controlled study data to do some analyses. We need to do conditional logistic regression, using the matching id as 'strata' variable. My instructor told me to use proc phreg to do the conditional logistic regression, following the way of the example found in SAS documentation. (You can easily find it.) In that example, the output gives estimated HR of the explorotary variables.
My questions are:
1. Since we are doing the logistic regression, though we are using proc phreg, should we explain the 'HR' as Odd Ratio? Or it is just HR?
2. I was told, the HR and OR are the same. Why? under what condition? I know in low probability, these two are approximately equal. But I feel their meaning are not the same.
3. Why can we use the proc phreg to do the logistic regression? Or for what reason, we want to use proc phreg to do the conditional logistic regression?
4. If I use the proc logistic to do the conditional logistic regression, will I get the same result? (Sorry, I did not try it before.)
04-13-2013 03:41 PM
Read the following article:
Prentice RL, Breslow NE. Retrospective studies and failure time models. Biometrika 1978 Apr;65(1):153-158.
Matched case-control samples correspond to risk sets in proportional hazards models.