Dear all,

I am using Fine & Gray proportional subhazard models to calculate the effect of some health behaviours on my outcome variable. I use the following model (after following the steps of imputation):

mi estimate, shr: stcrreg i.smoking##i.edulevel, compete(event==2 3 4) (and some extra values but they are not relevant here).

I am interested in the interaction between smoking and educational level and I would like to get one overall p-value to see whether this is significant. I do however only get two values because both are factor variables and educational levels has three levels (smoking has only two). Does anyone know how to get one overall p-value for this interaction and then perform post-test to see where the differences are? I have tried to do post-test (contrasts) to get this value, but this doesn't work because of the imputed data. I also tried mi test / mi estimate, cmdok / mimrgns (here I only get predictive margins not contrasts), but so far I wasn't succesful.

Thank you in advance!