Interrupted time series

[Bill Smith]

Using Stata SE 13.1 under Windows 7E. I've used the user written module itsa (http://ideas.repec.org/c/boc/bocode/s457793.html) to run a series of interrupted time series analyses on frequencies of several variables rolled up into quarters as outcomes. Now I want to add covariates, e.g., sex, gestational age. I think I can use the mean during each quarter for gestational age, but I'm not sure what to do with the sex variable. Should I just use the count of say, all females during each quarter as the covariate? Thoughts?

[Bill Smith]

Gave this some more thought. Not sure itsa is appropriate for poisson models. Perhaps the arpois module (http://www.stata.com/stb/stb46/sts13/arpois.hlp) might be an option. Does anyone know how to incorporate the intervention into this analysis? Another option is a mixed effects poisson model using xtmepoisson described here; http://www.bmj.com/content/bmj/344/bmj.e3005.full.pdf. It appears they used hand rub procurement as the marker for intervention, but it's not exactly clear how the segmented regression was implemented. I think I could live with arpois if I could figure out how to incorporate the change point. Anny suggestions?