How to output results from Harrell concordance statistic

Reply
Occasional Contributor
Posts: 7

How to output results from Harrell concordance statistic

Hello,

 

I have the following code:

 

proc phreg data = cstat concordance  = harrell (se); 

  model survival * censored (1) = age sex;

run;

 

I would like to output in a new dataset the results of this procedure as to have a column for: estimate, standard error, concordance, discordance, tied in predictor and tied in time.

 

Basically, I'd like to be able to create a new dataset with the outcome of the "concorance = harrell (se)" as we can get the maximum likelihood estimates of model parameters by running:

 

proc phreg data = cstat ; 

  class sex (ref = '1');

  model survival * censored (1) = age sex;

  ods output parameterestimates = parms_cstast;

run;

 

Thank you very much!!

 

run;

Super User
Posts: 5,987

Re: How to output results from Harrell concordance statistic

I can't give you the answer, but I can set you on the road toward finding it.  Add this statement before the PROC:

 

ods trace on;

 

When you run the PROC, it will give you the names of all the ODS data sets that are available.  You might have to inspect them to find which one contains the information you are looking for, and then the proper data set to the ODS OUTPUT statement.

Occasional Contributor
Posts: 7

Re: How to output results from Harrell concordance statistic

Posted in reply to Astounding
Thank you for the very prompt reply! Thanks to you, I found it. It's
"concordance" (I should have thought of it!!). So it is

proc phreg data = cstat ;
class sex (ref = '1');
model survival * censored (1) = age sex;
ods output *concordance* = parms_cstast;
run;

Thank you SO MUCH!!
Milli

Super User
Posts: 21,527

Re: How to output results from Harrell concordance statistic

For future reference you can also find the relevant table names in the documentation under ODS Table Names for each procedure.

Ask a Question
Discussion stats
  • 3 replies
  • 84 views
  • 0 likes
  • 3 in conversation