How to change relative-risk denominator from default (e.g., with Poisson regression)?

[DaisyV]

When running my Poisson regression model (dep var [dth]), STATA defaults the denominator of the relative risk calculation to the 'lowest'/'first' group of the independent categorical variable of interest (i.e., [age] --> 4 categories).

I want to generate relative risks using the 'highest'/'oldest' group of my categorical variable (or others, potentially) in the denominator. How do I do this? I have gone through STATA help documentation, STATA YouTube channel etc., etc. ad nauseum -- no luck...

I am using STATA v 13.1.

Examples of my attempts thus far:

---------------------------------------------------------------------

poisson dth age, exposure(personyrs) irr
*this just provides the IRRs with the 'first'/'lowest' group as the denominator...
*how to make 'highest'/'last' group the denominator in IRR??

xi: poisson dth i.age, exposure(personyrs) irr
*same results as above

xi: poisson dth i.age_75, exposure(personyrs) irr
*produces an error

*trying to specify which group I want in the denominator vs. default using poisson model... have generated reverse categories for age (i.e., agegrp) using oldest age as 'first' group...
xi: poisson dth i.agegrp_75, exposure(personyrs) irr
*error
*don't know what else to do...

xi: poisson dth i.age==75, exposure(personyrs) irr
*error
xi: poisson dth _Iage_75 , exposure(personyrs) irr
*error

*going to try lincom command as per an example that looks similar
xi: poisson dth i.age, exposure(personyrs) irr
lincom _Iage_75, irr
lincom _Iage_50, irr
*produces exact same results as regression model i.e., 'first'/'lowest' category is denominator... (??!??)

---------------------------------------------------------------------

Thanks in advance for any insight anyone might be able/willing to offer!!
D.

[Rich Goldstein]

there is no reason to use -xi- and several reasons no to use it; see -h fvvarlist- including including the "base levels" section of the help file

[DaisyV]

Many thanks for your response! And for pointing me to the fvvarlist documentation! Which also led me to a YouTube video that shows exactly how to specify the denominator in the IRR.

**For anyone else interested in this, you can see a demo showing how to specify the denominator from watching this video: http://www.youtube.com/watch?v=Wa1Nd9epHmY**

Cheers and thanks again!
D.