st: baseline adjustment in linear mixed models

[Formoso Giulio <GFormoso@regione.emilia-romagna.it>]

Hi. I am going to evaluate the effects of an information campaign on antibiotics, comparing the average antibiotic prescription in a 4 months follow-up in an intervention vs a control area, using a linear mixed model. My question is: would it be possible to introduce as covariate a "baseline prescription" term (e.g. the average prescription in 2 months before the campaign)? 

More specifically, I reshaped my database in the long format, with 30 health districts as unit of observation, each with 4 prescription data (one for each of the 4 months of follow-up). Therefore, I have 4 lines for each health district. In the baseline prescription column, baseline data would be fixed for each health discrict, as I try to show below with an example (hope it will be readable in the e-mail):

health_district 		month		 prescription		intervention 		baseline
	1		      1		          450			0		     326
	1		      2		          480			0		     326
	1		      3		          465			0		     326
	1		      4		          460			0		     326
	2		      1		          420			1		     300
	2		      2		          435			1		     300
	2		      3		          440			1		     300
	2		      4		          412			1		     300
ecc

Following is the syntax I used:

xtmixed prescription intervention baseline || health_district: || cf:, reml 

By the way, the baseline term was highly significant

Thank you very much for your help! (and excuse me in case this is too simple for this forum of if my approach is too naive)


Giulio Formoso
Regional Healthcare Agency, Emilia-Romagna, Italy


*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/