How to specify event of outcome variable in Proc GLIMMIX

[UcheOkoro]

Please, I need help specifying the direction of the event in proc GLIMMIX for a longitudinal study. I used the (event='1') statement but when I flipped it the results were the same. It appears that the 'event' statement does not work. The outcome is binomial. 

The following is my SAS code

PROC GLIMMIX DATA=COVERED.Demographics method=laplace;

CLASS  record_id wkly_contact_simsx(ref='0');

MODEL hospital_adherence(event='1') =  wkly_contact_simsx
/SOLUTION DIST=bin LINK=Logit ;

random int / subject=record_id;

lsmeans  wkly_contact_simsx/  diff e cl oddsratio ;
  run;

Thank you so much for your assistance.

[Reeza]

Can you please show the log from each run?

Does your variable

hospital_adherence

have a format attached to it?

[UcheOkoro]

Thank you for your prompt response the  following are the log and results.

Thank you again

[Reeza]

Well, the warning does say it's ignoring the EVENT statement because the outcome isn't binary.

Can you show an proc freq of the hospital_adherence variable?

proc freq data=covered.demographics;
table hospital_adherence*wkly_gathering / missing;
run;

---

@UcheOkoro wrote:

Thank you for your prompt response the  following are the log and results.

Thank you again

---

[UcheOkoro]

Here it is . Thank you

[Reeza]

Ok, it looks you have some missing values across your outcome and fixed effect. Try removing those and re-running your proc to see what happens. 

I would expect the warning to disappear and if not that's part of the problem. 

[UcheOkoro]

Thank you for your response. I deleted the missing observations but the results still remained the same

[PaigeMiller]

GLIMMIX produces output of Class Level Information and Number of Observations like this:

Please show this to us from running PROC GLIMMIX using your data set.

[Reeza]

Is there a warning in the log still?

[UcheOkoro]

[Reeza]

I don't know enough about proc glimmix to be 100% certain but I'm guess it has to do with the subject being each record id. 

Is each record unique with a single value of adherence?

[Ksharp]

It should be "dist=binary " ,not "dist=bin " . "dist=bin" would make sas take it as "dist=binomial"