Interaction means that effect is different per category, e.g. for gender - different for males and for females. But interaction has to be with another explanatory variable, e.g, treatment, age, etc. Typical examples of feasible interactions are

1. Treatment and Severity of Disease

2. Age and Risk Factors

3. Age and Type of Disease

4. Measurement and State of Subject during Measurement

5. Ethnicity and Disease

6. Study Center and Treatment

etc., etc.

Realisation in SAS is very simple, e.g. as an example below where interaction between sex and treatment effect as included

proc logistic data=xx;

model response=sex trt sex*trt;

run;

Dear Ladislav Pecen thanks for your kind answer.

Indeed, I made a table with the variables of interest and I show it as univariate associations plus the interaction p value. However, I was recommended to adjust them by gender, so I got confused regarding whether i should add the variable 'gender' in both univariate and interaction term analysis. Perhaps should be this the way how it has to be represented?

Proc logistic data=temp;

model dep=A B A*B gender ;

run ;

I would appreciate your insights.

Diana, it is enough to include gender into your model as you did.

Thanks Dear Mehmet.

Your model is an adjustment for gender effect:

Proc logistic data=temp;

model dep=A B A*B gender ;

run ;

But if also an interaction of A and B effects ith gender could be realistic scenario, than the model is

Proc logistic data=temp;

model dep=A B A*B gender gender * A gender * B ;

run ;