Poisson regression with time varying covariates and counting process data structure

[shairstin]

Hello all

I am trying to model a count outcome (emergency room visits) with a primary predictor covariate of interest that is time varying (on or off a drug). Once the drug is started, they patient is considered "on" the drug going forward (ie they do not revert to being off the drug).

My data is structured in a counting process (multiple lines per individual) for each risk interval. 

Does anyone have any resources as to where I would find code to do this? Can proc genmod handle a counting process data structure?

I am not really sure where to start...

[Ksharp]

It seems you have longtitude data.

Try PROC GLIMMIX or PROC GEE

[shairstin]

Thanks

It seems those procedures are more suited to my data.