Hello everybody.

I'm trying to test if a variable can be considered an effect modification in a svy poisson model.
I planned to evaluate the p-value of the interaction terms in a model to decide if age is an effect modifier.
However, I'm dealing with an interaction between a three-level categoric variable (Age) and a binary exposure (Binge).
Here are the results of the interaction terms:

---------------------------------------------------------------------------------------
| Linearized
NUP | IRR Std. Err. t P>|t| [95% Conf. Interval]
----------------------+----------------------------------------------------------------

age#binge |
25 0 | 1.052783 .0754358 0.72 0.473 .9148233 1.211548
40 0 | 1.095433 .074272 1.34 0.179 .9590996 1.251146
60 0 | 1.158218 .0832244 2.04 0.041 1.006044 1.333411

Look that just one term is significative. Should I consider that age is an effect modifier?
Another option for testing interaction significance would be to use LR test or Deviance but as far as I know this is not avaiable in svy commands. Am I correct?