[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: Person-period dataset, stcox, subjects in analysis

From   Jeph Herrin <>
Subject   Re: st: Person-period dataset, stcox, subjects in analysis
Date   Sun, 17 May 2009 21:53:23 -0400

After the regression, you want to select

  keep if e(sample)


Dear Statalisters,

This may be a straightforward question with an obvious answer, but I need your help on it. I have a person-period dataset and I used stset command specifying origin and id as well as failure and exit. This gave me a specific number of failures and number of subjects at risk. I ran stcox using a variable that has missing values for a number of those subjects. I am trying obtain a selection variable so I can look at baseline characteristics of those subjects that were selected for this analysis. Which of the variables that were generated through the stset command should I be using for this purpose?

 stcox x

         failure _d:  failuredem == 2
   analysis time _t:  (age50dem-origin)
             origin:  time origin==1
  enter on or after:  time origin2==1
  exit on or before:  failuredem==1
                 id:  id

Iteration 0:   log likelihood = -1697.8111
Iteration 1:   log likelihood = -1696.3502
Iteration 2:   log likelihood = -1696.3477
Refining estimates:
Iteration 0:   log likelihood = -1696.3477

Cox regression -- Breslow method for ties

No. of subjects =      2321                     Number of obs   =     15721
No. of failures =      270
Time at risk    =  52641.86847
                                                   LR chi2(1)      =      2.93
Log likelihood  =   -1696.3477                     Prob > chi2     =    0.0871

          _t | Haz. Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
         x|   1.247601    .159471     1.73   0.084     .9711212   1.602796

So, basically, I want to be able to describe the 2,321 subjects in that analysis. I did that before, but I can't remember how anymore. Thanks for any help you can provide,



*   For searches and help try:

*   For searches and help try:

© Copyright 1996–2017 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index