Dear all,

I am interested in evaluating if there is multiplicative interaction between one categorical and one ordinal variables. The first one InstrumentalSuportRESTRCITED (coded 1 and 0) where 1 means yes and 0 means no. The second variable is PatternsAbuse that has 4 options from 1-4 with 4 being the reference category.

I have ran the Poisson regression with the interaction term and adjusting for confounders (see below).


Array



My question is how do I calculate the IRR of the effect when support is 1 for each of the patterns of abuse. I understand that I have to multiply the effect of the exposure (1.InstrumentalSuportRESTRCITED ) by the interaction effect (InstrumentalSuportRESTRCITED#PatternsAbuse) for each of the exposures (1-3).

For example, 1.InstrumentalSuportRESTRCITED (0.2439615) X InstrumentalSuportRESTRCITED#PatternsAbuse 11 ( 3.5409) which equal to 0.863843275.

My questions is, do I also have to multiply the above for the model constant (0.5586895 ) ?

Thanks !