ERROR: Variable in the RANDOM statement has only one level.

[fouke_o]

Hi all, 

I am trying to fit a frailty model in my survival analysis using this code:

PROC PHREG data=DataSurvival ;
CLASS Riluzol Center;
MODEL timeM*Deceased(0)=Riluzol ALSFRSRdecline Diagnosticdelay ;
RANDOM Center / noclprint;
HAZARDRATIO 'Frailty Model Analysis' Riluzol;
Format Center Sites.;
RUN;

And keep getting the error: 
ERROR: Variable Center in the RANDOM statement has only one level.

My center variable for riluzol looks like this: 

Table of Riluzol by CenterRiluzol Center1 2 3 4 Total01TotalFrequency Missing = 639

195 13.72	3 0.32	1030 35.99	181 12.46	1409
1226 86.28	926 99.68	1832 64.01	1272 87.54	5256
1421	929	2862	1453	6665

Anyone any idea? 

Thanks!

[PaigeMiller]

Could you please re-format the table so I can read it and understand it? And also please explain what this table is and how it was created.

[fouke_o]

NIV users

The FREQ Procedure

Sorry, it did something with the table, here you go: 

[ballardw]

I think that you may want to examine a few more variables in combination. I think that you can get a message like that when considering the DEPENDENT variable(s) if there is only one value of the random variable associated with a given outcome.

You might try:

proc freq data=datasurvival;
  table deceased*Riluzol*center / list ;
run;

And see if a combination of deceased and riluzol only have one value of Center .

From your table I suspect Center=Tours and Riluzol=0 all have the same value for Deceased.

[PaigeMiller]

ERROR: Variable Center in the RANDOM statement has only one level.

When SAS says there is only one level, I believe SAS. It may be that because of missing values somewhere in your data set, the model winds up with only a single level of CENTER. Anyway, given the model you are fitting, we would need to see frequency tables on all variables in the model, not just the one you show.