Non-inferiority Proc Mixed Repeated Measures ANOVA

[PeterBr]

Hi All,

I have a repeated measures design dataset where I would like to test for non-inferiority across two groups over 4 time periods, specifically I am interested in the "group" and "group*time" interaction. I have written the following code but am wondering if there is a way to change it to test for non-inferiority?

proc mixed data=mydata;
  class id group;
  model outcome = time group time*group time*time / solution cl;
  random intercept time / subject = id;
  estimate '1-2' group 1 -1 / cl;
run;

[SteveDenham · Accepted solution]

You can use the options LOWER or UPPER to get one tailed tests of the differences in any of an LSMEAN, ESTIMATE or best of all, an LSMESTIMATE statement.

If you know the margin of inferiority,, that can be entered with the TESTVALUE option in the LSMESTIMATE statement.

SteveDenham