Programming the statistical procedures from SAS

Estimate statements in PROC MIXED with categorical and continuous variables

Reply
New Contributor
Posts: 3

Estimate statements in PROC MIXED with categorical and continuous variables

Greetings! I am trying to estimate the social support (SS) score of patients on a particular day in a study, and contrast these estimates in two treatment (trt) groups.

This is a longitudinal study with 1 to 4 measurement occasions for each patient. For various reasons, I am modeling time (in days) as continuous in PROC MIXED. I also am also controlling for categorial and continuous covariates. It's not clear to me how to include continuous variables in my estimate and contrast statements. The below model produces non-estimable (non-est) results. Does anyone have suggestions or examples for inclusion of continuous variables in estimate or contrast statements? Thanks!

proc mixed data=MyProject method=ml order=internal;

    class EMRID trt occ H L;

    model SS=trt days trt*days H L M/ s chisq;

    repeated occ /type=un subject=EMRID r rcorr;

    estimate "trt=1 at baseline" intercept 1  trt 1 0  days 180  trt*days 180 0  H 1 0 0 0  L 1 0  M 37;

    estimate "trt=2 at baseline" intercept 1  trt 0 1  days 180  trt*days 0 180  H 1 0 0 0  L 1 0  M 37;

    run;

*where    trt          treatment                           categorical (1, 2)

               days     day of measurement          continuous (from 0-400)

               occ       measurement occasion      categorical (1, 2, 3, 10) 

               H          a covariate                         categorical (1, 2, 3, 10)

               L          a covariate                         categorical (1, 2)

               M          a covariate                         continuous (from 0-100, mean 37)    

;

Ask a Question
Discussion stats
  • 0 replies
  • 384 views
  • 1 like
  • 1 in conversation