topic Re: Model selection and saving probabilities using fine and gray model in Statistical Procedures

Model selection and saving probabilities using fine and gray model

DavidsM — Sat, 06 Apr 2019 13:06:43 GMT

Hello our esteemed advisers,

I am trying to select a parsimonious model from my predictors using Fine and Gray model and saving the output data. The problem is that SAS is telling me that the selection procedure is ignored and the time variable is excluded from the output data and I want to use the time variable in that output data set.

Below is the code am trying

proc phreg data=Disease plots(overlay)=cif;
class Group BMI (ref='1') Smoking (ref='1') Alcohol (ref='1') Exercise (ref='1') DB (ref='1')
model Time*Status(0)=BMI Smoking Alcohol Exercise HTN DB / eventcode=1;

Selection=Backward Slstay=0.05 details;
baseline covariates=Predictors out=Out1 cif=_all_ timelist=6 9;
run;

I will be glad to be advised.

Re: Model selection and saving probabilities using fine and gray model

Reeza — Sat, 06 Apr 2019 22:46:46 GMT

Can you post the log as well, including the portions where it says time is excluded.

Re: Model selection and saving probabilities using fine and gray model

DavidsM — Sun, 07 Apr 2019 11:28:54 GMT

This is the log file from the code I tried. The variable Time is the follow up time but its excluded from the output data created and selection procedure is ignored.

Below is the log file.

NOTE: The SELECTION= option (MODEL statement) is ignored for the Fine and Gray competing-risks analysis.
NOTE: SLSTAY= is ignored since SELECTION=NONE.
NOTE: DETAILS is ignored since SELECTION=NONE.
NOTE: Data for variable Time in the COVARIATES= data set are not copied to the OUT= data set in the BASELINE
statement.

Re: Model selection and saving probabilities using fine and gray model

Reeza — Mon, 08 Apr 2019 04:08:04 GMT

I suspect it's not implemented then, or at least not in the version you're using. I can't find any exclusions noted on the STAT 15.1 docs. I'm assuming your code as posted isn't correct, because you have the selection after the semicolon, not before. I've moved your question to the Stats forum so someone with more helpful answers can hopefully chime in.